A Study of Electron Pair Production in 16 GeV/C Pi-Minus Proton Collisions Citation Marshall, Max (1980) A Study of Electron Pair Production in 16 GeV/C Pi-Minus Proton Collisions. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/5ayt-ay89. https://resolver.caltech.edu/CaltechTHESIS:10202025-214210581 Abstract Inclusive electron-pair production in pi-minus proton collisions at 16 GeV/c was studied using a large-acceptance spectrometer equipped with Cerenkov and shower counters. A two standard-deviation signal vas observed in a region of invariant pair mass greater than 140 MeV/c 2 and Feynman X between 0.15 and 0.4. The cross section, extrapolated to the full range of X, for masses greater than 140 MeV/c 2 , is estimated to be less than 2.5 microbarns, and the corresponding single electron to pion ratio to be less than 2.6 · 10 -5 . The observed electron pairs occurred mainly in the mass region 200 to 600 MeV/c 2 and are believed to have originated in the primary hadronic interaction rather than the decay of secondary particles. Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: (Physics) Degree Grantor: California Institute of Technology Division: Physics, Mathematics and Astronomy Major Option: Physics Thesis Availability: Public (worldwide access) Research Advisor(s): Pine, Jerome Thesis Committee: Unknown, Unknown Defense Date: 20 December 1979 Funders: Funding Agency Grant Number U.S. Department of Energy UNSPECIFIED Caltech UNSPECIFIED Record Number: CaltechTHESIS:10202025-214210581 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:10202025-214210581 DOI: 10.7907/5ayt-ay89 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 17726 Collection: CaltechTHESIS Deposited By: Benjamin Perez Deposited On: 27 Oct 2025 18:38 Last Modified: 27 Oct 2025 18:51 Thesis Files PDF - Final Version See Usage Policy. 60MB Repository Staff Only: item control page

A STUDY OF ELECTRON PAIR PRODUCTION IN 16 GeV/C PI-MINUS PROTON COLLISIONS Thesis by Max Marshall In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy California Institute of Technology Pasadena, California 1980 (Submitted December 20, 1979) -iiAcknowledgements I !cknovledge the u.s. financial support, Dept. of Energy and for both persoMl and for this research. !=knowledge and thank the following people: thesis Caltech Jerry Pine, I my adviser, for patient guidance and instruction: David • Bl,ckus, Bill Woody, who Dunwoodie, worked Ryszard Stroynovski, and Bill :raig on the data analysis, for contributions and companionship during difficult work; Bierce, and Johnson, for Bill assistance Walsh, Dick at SLAC; Hans Jansen, Jerry Pine, John Scheid, Ricardo Gomez, Tom Himel, ani Bill Friedler, for their contributions and companionship over various time periods while building the cylindrical chambers: David Leith and the others of SLAC Group B, past and present, for their contributions to planning and running tne experiment; Laura Peirce for personal support and help with the text; and the staff of the accelerator, !nd SLAC technical drawing groups at pleasant and product~ve environment. for computer, aaking it a -iiiAbstract Inclusive electron-pair production in collisions at 16 GeV/c was studied using a large-acceptance proton pi-minus spectrometer equipped with Cerenkov and shover counters. t If 0 standard-deviation signal vas observed in a region of invariant pair mass greater than 140 l!eV/c2 between 0. 15 and o. 4. and Feynman I The cross section, extrapolated to the full range of x, for masses greater than 140 l!eV/c 2 , the corresponding single electron to pion ratio to be less than to The be observed than 2.5 microbarns, is and estimated less A electron pairs occurred mainly in the mass region 200 to 600 MeV/cZ and are believed originated in to have the primary hadronic interaction rather tnan the decay of secondary particles. -ivT~ble of : ont e nts :BAPTER 1 -- Introd u c t ion Motivati~n ..•.•..•. ••••• • •••• ••• ••••• •••••• 1 1.B -- This Experiment•••••••o•••••••••••••••••••• 8 1. A :aAPTER 2 -- The Experimental Apparatus 2.A I ntroduc ti~n ..•••• •.• •••••••••••••••••••••• 12 2.B Overview of LASS ••••• 12 2.c Bea m Sys tem••••••••••••••••••••• •~••• •••••• 18 2.D Liquid Hydr~gen r~rget•••·••••·•••••••••••v 21 2.E Proportio na l Wire Ch ambers •• ~•••••••••••••• 23 2.F 25 2.G Ce renkov Counte~s.Ge ••••••••••• 32 2.H S hoYer Cou nters ..••. • • • • 9 Q e f b i a • i e • • s • • • o • e e o • 35 . . . .. . . . .. . . . 38 0 •••••• 2.I -- Phj tomu lt ipliers and $ • ., •• Do wnstream Scintillators. . . . :aAPT ER 3 -- Electroni cs: 3.A ~ ~ Trigger and Data Acquisition I ntroduction .. • • • e O O •• e • • • e 9 4 I e • • e II e • e •• ♦ e $ 41 42 3.B Triggers, Gating and Scalers .. 3.C Fast Logic: 3.D CA~AC and Cha mber Read~ut•••••••••••••••••• 49 :a&PTER 4 -- Trackfinding Program 4.A -- Introduction •.•.•••••••• e • e • 0 • • 0 a • • t • e • • • a • 53 4.B aathe ■ atical 4.C !atch Point ~aking•·••••••••••••••••••••••• 56 4.D Trackfinding Algorithm••••••••••••••••••••• 59 Fitting Procedure: Tracks and Vertices •••• 61 Chamber Alignment ••••••••••••••.••.•••••••• 62 Bubble Chamber Coaparison •••••••••••••••••• 64 Early Event Rejecti~n•••••••••••••••••••••• 72 Conventions ••••••.•••••••••••• :HAPTER 5 -- Data Processing: Normalization cuts, Background, and 5.A Introduction and summary••••••••••••••••••• 15 5.B Electron Identification •••••••••••••••••••• 78 B. 1 Shower Counters•••••••••••••••••••••••••• 79 B.2 C1 Cerenkov Counter•••••••••••••••••••••• 90 B.3 Electron Detection Efficiency •••••••••••• 99 B.4 Hadron Punch-Through .••.••.••.••••••••••• 102 B.5 Tvo Electron :at••••••••••••••••••••••••• 106 5.C -- Track Quality Cuts••••••••••••••••••••••••• 107 c.1 confidence Level cut••••••••••••••••••••• 107 c.2 "~atrix Flag"••••••·•••••••·••••••·•••••• 109 5.D -- Pair Vertex Position and the Beam Association Cut•••••••••••••••••••••••••••• 110 D.1 Mass, Geometry, and Conversion Pairs ••••• 110 o. 2 Photon Conversion Pairs at High Mass ••••• 111 D.3 Beam Association Distance Cut•••••••••••• 115 5.E -- Low Mass Pairs••••••••••••••••••••••••••••• 122 B.1 Geometry and ~ass Resolution .•••.•••••.•• 122 E.2 Low Mass Mate Cut•••••••••••••••••••••••• 129 E.3 Pair Vertex Related Cuts••••••••••••••••• 137 E.4 Low ~ass Pairs Appearing at High ~ass •••• 143 5.F -- Background Processes••••••••••••••••••••••• 152 P.1 Sources of Background•••••••••••••••••••• 157 P.2 sources of Uncorrelated Electron Tracks •• 158 F.3 sources of Negative Electrons •••••••••••• 159 F.4 Tvo Hadron Punch-Through ••••••••••••••••• 161 P.5 Empty Target Background•••••••••••••••••• 164 F.6 Background Subtracted Result ••••••••••••• 164 5.G -- Monte Carlo and Acceptance Correction •••••• 166 G.1 Monte Carlo Track Generation ••••••••••••• 166 G.2 Electron Identification and Beam Associatiion •••••••••••••••••••••••• 171 G.3 -- Final Acceptance Correction and Normalization .••••••••••••.••••••••••••••• 174 :BAPTER 6 -- Results 6.A Introducti~n •••••••••••••••••••••••••• ~•••• 180 6.B Geometrical Acceptance and Rho-o ~~ga Signal 181 6.c Comparison with Other Experi ments •••••••••• 191 6.D A Model for Lepton Pair Production ••••••••• 196 6.E Summarf•••••••••••••••••••••••••••••••••••• 201 A?~ENDIX 1 -- Cylindrical Spark Chamber System Al.A Introduction •••••••••••••••.•••••••••••••• 206 11.B Physical Description•••••••••••••••••••••• 206 11.c Electrical Description ..•••••.•••.•••••••• 212 11.0 Electronics Pabricati~n and Testing ••••••• 221 11.E Digital Readout••••••••••••••••••••••••••• 223 Al.E Autoaatic rester•••••••••••••••••••••••••• 225 11.F Gas Syste••••••••••••••••••••••••••••••••• 226 11.G Cosmic Bay Tests•••••••••••••••••••••••••• 228 -1- :aAPTER 1 -- Introduction 1.A -- Motivation s=ience is a process Jf observation and systematization. In physics, systematization often takes the form of fundamental laws of nature that phenomena. Once unierstanding are the auch simpler fundamental complex phenomena than the principles is obserTed are known, straightforward, but re=ognizing a nev principle is difficult unless the observed phenomenon depends on it in a particularly simple vay. In the area of so-called elementary interacti~ns behavior. nature of hadrons constituents physics, the display complex and multi-faceted This behavior seems where particle different to arise combinations from compound sore elementary are bound in a dynamic structure. It has been difficult to study these constituents of a because the binding f~rce seems not to allow them to be isolated. Blectron scattering has studying the structure of hadrons because electrons inner ace not affected by the ~ad because they interaction been particularly strong-interaction useful binding for f~rces exhibit no inner structure of their own. occurs by the well understood electromagnetic process, and when the energy is high enough, l~cge angle scattering inYolves the constituents -2- individually. Electron-positron related annihilation into hadrons to electron-hadron scattering. is closely The electromagnetic =~apling between a charged particle and a virtual photon the same whether the is charged particle is scattered or is annihilated (or created) with its antiparticle. A collision b~tveen an electron and positron has a probability of certain producing a virtual photon which may return to a pair of elactrons, create a pair of muons, or create a pair of the constituents of hadrons. has been inferred anon pair production in pair exist c~nstitu ent from can of any (The number of constituents the ratio of hadron production to mason}, or produce several electron as annihilations.) rhe a single hadron (a vector hadrons, vhere the additional c~nstituents are created in pairs fros kinetic energy. The third related process. the annihilation of hadron =~nstituents into a p~ir of leptons, occurs occasionally the collision of two hadrons. The aim of this experiment vis to detect electron pairs produced in ~nd to determine if they in occurred as hadron collisions the result of a c~nstituent annihilation process. In this experi ment, electrons were identified with c~unters and p1rticles having shower a counters. velocity Cerenkov greater than Cerenkov counters a tag const~nt -3daterained by the material in the counter. Electrons, having a •ery small aass, needed little momentum tha critical velocity, Cerenkov media ■ exceed while hadrons needed more aomentu ■ than was generally imparted to the to outgoing particles. Since was a gas, the particle trajectories usaally were not disturbed. Shover counters contain a large a ■ ount of dense material interact, their energy with being which incoming electrons converted into a detectible sh~wer of electron-positron pairs. Again it is the low that from distinguishes particles. since the electrons the shover other counters ■ ass types of absorbed the particles, they were placed last in the line of detectors. The major challenge in this experiment was to distinguish "direct" electrons produced in the hadronic interaction from hadrons and from secondary electrons created by hadrons made in the interaction. Electron pairs are produced in hadron c~llisions with a probability approximately ratio the electromagnetic to strong of the strengths of interactions squared, about 1/20,000. aisidentifying two hadrons needed 20,000 in order that the real signal The equal to probability the of to be less than one in was larger than the aisidentified hadron background. Secondary electrons can be produced when a hadron decays and produces either a lept~n (electron or electromagnetic interaction muon) pair via the or a single charged lepton via the veak interaction. the intermediate interaction. increased Such decays are greatly enhanced vhen particle cannot by decay the strong (Lack of a strong decay mode results in a lifetime ■ uch and thus a greater opportunity to decay into leptons.) Weak decays are often so slov that there ~ path discernible length before the decay, in which case tney vere ignored. This was the strange ka~ns, and charged pions. baryons, is case for the long-lived Weak decays of tne D mesons, heavy leptons, and the i-boson, however, would produce leptons interaction. coming The apparently and pi-zero fro• primary the eta mesons are stable with respect to strong interactions and decay into electron pairs on the order of o ne percent of the time by the Dalitz aode (electron pair plus a photon}. decay Other mesons may also decay this way but with less probability because they are sh:>rt-lived. rne most serious source of from the Bethe-Heitler background process electron pairs (photon conversion) where pbotons from the decay of pi-zeros t3o vas with matter close to the primary interaction to be resolved. These c:>uld usually be electron invariant pair recognized interacted and mass eli mina te d vas less H:>vever, the detector failed to find one of because the than 150 ~eV/c 2 • the tracks in 10-20 % of the pairs, in which case the other track could not ba eliminated on the basis of pair mass. p~irs were produced at an average rate Photon conversion of about 0.1 per -sh~dronic interaction, so that a siaultaneous occurrence of this phenomenon for tvo pi-zeros, which mimicked direct high •~ss electron pairs, occurred at a rate several times higher than the direct process. A source of direct lepton pairs mesons. vas the decay of vector These give rise to pairs at particular masses, and c~n be recognized c~nsidered in way. that resonance (Vector mesons may be effects that modify the amplitude for the constituent annihilation at certain masses, or they may be The considered particles that decay into lepton pairs). lepton production c~lculable sections from vector rate due to fairly well measured and branching ratios. meson decays inclusive is cross Thus, they provide a useful c~libration signal in the lepton pair mas s spectrum. The original motivation for observing direct leptons was the search for the W-boson through its weak decay, 1965. once aroused about determined. direct lepto ns were their source, which By 1974, more most observed, curiosity was is of only singly nov being and in pairs. interesting feature of these experiments was that tne ratio of leptons to pions was about r3nge in than ten experiments reported direct muons or electrons, detected The starting center 10-• over a vide of mass energy and trans ve rse momentum of t~e outgoing lepton/pion. ~ore recent measurements of ritio have all been 2 to 3 •10-s. this -6- The Drell-Ian model f~r into lepton pairs the was annihilation proposed in of 1970 . predicted a scaling lav for a fixed ratio of center of ■ odel This pair ■ ass hadrons momen t.u m distributions (which were measured in e1ectron-hadron and neutrino-hadron scattering). The assumptions are valid only in the limit of large center of mass energy and lepton •~ss (relative to the masses of hadrons). annihilate rather than quarks ~ which meson The unmodified model agrees poorly vith measurements at th e energy of experiment. pair This experiment is compared with a modified Drell-Yan mode1 in constituen ts to mass energy and a relationship between the cross section and the quark (constituent) inside constituents this Feynman diagrams are shown for th e original and modified models in fig. 1.A. The lepton to pion ratio was considered ano ~alously large in 1976 when this experiment was proposed. extra leptons vas originating as pa i rs. not known, nor The source of was fraction the The recent experi ments indicate the that most of the leptons occur as pairs, and that the extra pairs !re created by co nst ituent an nihilations at a rate more than !n order of ma gnitude larger t han predicted ~y the Drell -Yan model. A thorough history of dir e ct lep ton p hysics is given in the thesis of Woody(1), and the models and experiments are described in several review articles(2). -7- a) Drell-Yan model b) Pion annihilation mo del Figure 1.A: Feynman diagrams for lepton pair production -arllis experiment complements a recent p~ir experiment at high anon statistics the same energy, by observing electron p~irs at a lover value of I and at masses below the two ■ uon threshold. 1.B -- This Experiment Tbis experiment (SLAC E-127) triggered on produced in the interactions single electrons of 16 GeV/c pi-minuses with protons in a one meter liquid hydrogen target. Only events with electron-positron pairs were studied. The experiment vas the first use of most of of the LASS the components facility and vas chosen because of its ■ odest tr~ckfinding and resolution requirements in addition to intrinsic interest. its The electron pair requirement was well suited to LASS because of its ability to handle the large d1ta rates due to the relatively loose trigger, and because air had the right identification. not yet Cerenkov (The gas-tight.) ~=ceptance, LASS had threshold for large segmented Cerenkov counter vas Because the of the capacity full to solid o bserve fraction of the kinematic region of pair production study the electron accompanying hadrons. angle a large and to A small, but potentially interesting, acceptance in the region of X below 0.15 and above 0.4 was lost because the downstream detectors were not used in the data analysis. -9- was identification Electron accomplished using 38-segment Cerenkov counter and 13 off-axis shover which identified only the counters high transverse momentum tracks. count in at least one of the Cerenkov cells (pulse l height corresponding to a single photo-electron) and a pulse height in at least one shower counter corresponding to an energy deposition of 750 MeV was the trigger vith requi rement. single electrons were recorded in the h ope that hadron misidentification would be low enough to al l ow a of Events single electrons and pairs. It was later clear that aiequate hadron rejection was possible which both tracks comparison onl y for pairs in were identified in the Cerenkov counter and at least one identified also in a shower counter. The experiment was run in Nov.-Dec. 1976 an d Feb.-Apr. 1977. The first computer track reconstruction prog ram was Dec.-Jan. 1977-1978. A further in new, improved version was written and run in Jul.-Aug. 1978 on half the with run improvements data and was rerun on the full sample in Apr.-Jul. 1979. Approximately 10• pions before acceptance million trigg e rs ■ illion program. were were losses were received of recorded gi vin g sensitivity 4 events pe r nanobarn. on tape from which Five one preselected and processed by the trackfinding One hundred thousand were selected as possibly -10containing an electron pair. accepted, of which About tvo hundred vere finally one-third vas estimated to be genuine. ?he OYerall acceptance was 1.21, and the corrected inclusive electron pair cross section for pairs with ~eV/c2 was 1.1±0.7 microbarns. This mass above corresponds 140 to an electron to pion ratio of 2.6t1.6 •t0- 5 • rhe observed number of events whose mass was consistent vith the rho and omega vect~r mesons was smaller than the predicted from r~tios. Due relatively the to measured the large cross sections and branching small number background of events and the estimate, the measurement was consistent both with zero and the between number prediction. The signal the pi-zero and rho in mass, vhich is attributed to c~nstituent annihilation, significance of vith experiments other two The greater than zero with a standard deviations, and in agreement clnstituent-annihilation section. was lowest and the predictions model discussed in the of the results mass data point was not in agreement with the steep rise in cross section at low masses predicted by the model. :hapters 2 and 3 discuss the physical layout of the experiment and the data-acquisition system respectively, and f~r the most following part more need not be read interesting chapters. to understand the Sections 2.1 and 2.B give an overview of the layout and should be read. Chapter -11- 4 describes the track reconstruction quite interesting, but need not be read. real electron procedure which is The selection of pair events and estimation of the losses and backgrounds are described in chapter 5, and the results discussed in chapter 6. are The appendix describes the building and testing of the cylindrical spark chamber package and the associated electronics. The event selection procedure described in chapter developed independently of the main group effort. results of have different a E-127 5 vas The final may use different cuts and therefore may statistical significance than those reported here. References for chapter 1 1. c. L. Woody, A study of Electron Pair Production in 16. 1 GeV/c pi-minus proton Collisions Using a Large Aperture Solenoid Spectrometer, Ph.D. Thesis, Johns Hopkins Oniversity (1978}. 2 • . N. s. Craigie, Lepton and Photon Production in Hadron collisions., Phys. Rep. 47, 1 (1978 ) ; L. M. Lederman, Lepton Production in Hadronic Collisions, Phys. Rep. 26, 149 (1976) -12- :BAPTEB 2 -- The Experimental Apparatus 2.A -- Introduction This ch1pter experiment. describes the apparatus used in this LASS, the Large Aperture Solenoid Spectrometer f1cility at SLAC, was the framework, consisting of superconducting solenoid magnet surrounding a large the hydrogen t1rget, several sets of spark and proportional chambers, a multi-cell Cerenkov counter, and a large dipole spectrometer m1gnet preceded lengthening the sh~ver counters. line, the followed spark by LASS was modified for ch1mbers(1). by and Cerenkov and proportional electron identification counter and adding a wall of The following sections describe the beam numerous and diverse detectors, and their layout in LASS. 2.B -- Overview of LASS Fig. 2.B.1 shows a cross viewed from above. section of the The upstream end was disk with a hole for the beam to enter. vere saparate cryostat n1rrov s~lenoid apparatus The solenoid magnet was 4 meters long and 1 meter in radius. windings whole divided ring. into four solid a iron The superconducting coils, each inside a Plane chamber pac kages fit in the spaces between the rings, coveri n g the entire bore. A solid iron ring after the last coil acted , ,: ~ = C ~ Cl ..,,. Figure 2.B.1: Cerenkov Counter Proportional Chamber Capacitive -Diode Chamber Magnetostrictive Chamber Sci nt i !lotion Counter Hodoscope Shower Counter LH2 Target Ring Counter XY Countet - --.... ►1 BEAM~ Beam PWC s 1 Superconducting Cai Is and Vacuum Vessel ~ ~ LASS Spectrometer • • • I • I • !: 11 I I : : ::· . . . :• I I I • ! I : • ! I • I •I• C2 Counter Hodoscopes HA,HB I ..,,. Downstream MS ~Chambers Loi 1pop Counters 1m lmL J•26AI •-n \[ !I [ [\_ Sc inti I lotion ': , ,: CD4/Plug 4 JHU Proportional Hodoscope ~~l W+E s Downstream Twixt Shower Shower Counters Counter Dipole Magnet MS Chambers Twixt N I I IJ,j ~ -14~3 magnetic flux return path. Beam entered the solenoid magnet parallel to along the 22.4 kilogauss field. the proportional particle. magnet. The the transverse to radius the helix was momentum (Pt) of the 350 ~eV/c PL particle just skimmed the inside A of backward) (or of of the magnet (a helix with a radius pitch and Charged particles produced in the target followed a helical path forward through axis its the helix l ~ngi tudinal mo men tum 0.5 of meter). The (meters/turn) was proportional to the (P1. ) • A 500 through one radian in 70 cm., aev /c P.t particle rotated about the distance between plane chambers. Particles with large transverse momentum hit the inside wall of the solenoid unless a correspondingly large •~mentum GeV/c (1 P1 longitudinal was needed for a P~ of 350 aeV/c, 4 c;ev /c P,t. for 1 GeV/c pt. carried ) them out the end hit the wall enough. Those with low forward momentum without intersecting any plane chambers .. chambers surrounding the target measured these. forward particles, the solenoid package did determine the momentum. The dipole m3.gnet. intended to hori zont3.l plane. kG•m, and an the before momentum and n ot For very accurately after were it from the bend in the (The dipole had a field aperture Cylindrical These particles passed through the chambers determine soon i:nt eg ral of 20 of about 2 m. hori zonta lly by 1 m. -15- vertically.) Fig. 2.B.2 shows the magnetic field due to the s~lenoid and dipole. The field near the made solenoid conventional magnetostrictive read~ut of spark chambers impractical. The spark chambers (called the CD chambers) in the solenoid ased 1 capacitor-diode readout method. The center of these chambers was deadened because of their long memory time the high particle multiplicity near the bea m path. spot The dead each was covered by a small area proportional wire in chamber ("plug" chambers). (plug1/CD1, plug2/CD2, Three of these chamber plug3/CD3) packages slid between the coils. rwo large active area proportional chambers (called 1.5 2.5) were installed inside coils 2 and 3. hive since been added.) iatersected five The measuring readout proportional chambers with rings path of planes. fast a and (More chambers forward particle Three induced charge (trigger c hamb ers TA,TB,TC) of wedge shaped pads (polar coordina te readout) were also mounted inside the coils. that and (It vas once intended logic make these signals into a selectable Pi-P~ trigger.) The cylindrical chamber package surroun ded the target inside coil 1. It consisted of a cylindrical proportional ~nd concentric ten readout. spark chamb ers with chamber capacitor-diode -16- 30 (a) 20 10 0 -1 0 (f) (f) ::) <! (_') 0 -20 _J -~ ( b) 20 10 0 -10 -20 0 8 - 78 200 400 600 800 1000 CE NTI ME TE R S Figure 2.B . 2: Magnetic field components as a £unction of Z: a) al ong X=O , Y=O; b) along X=75 cm. , Y=35 cm. 3~26B37 -17- ?he region between the end of the solenoid and the dipole is called the twixt-region. radial The comp~nent reconstruction. solenoid field which here, develops Particles entering this region pass through spark chambers. counter wall (C1) and Some pass through the aperture of the dipole magnet and some hit its sho wer track complicates the 38 cell atmospheric pressure Cerenkov counter several a was front face. aain The mounted on the fr .o nt face of the dipole. A relatively low refractive index gas was needed to identify electrons in C1. Air was used, giving a pion 5.8 light output was small, necessitating the GeV/c. The threshold extension of the beginning of C1 from the last iron ring the middle of coil 4. chambers to CD4-plug4 was displaced and used in front of the shower c~unter twixt-region of wall were and two dipole. large other ma gnetostrictive readout spark chambers (MS1T and MS2T) and a proportional hodoscope (JH-OP) The vire vertical between the KS chambers. Four MS chambers (MS1D through MS4D) measured tracks behind tbe just in dip~le. Another proportional hodoscope (JH-DN} front of MS1D provided ti ming information. ch~mber were HB), a large c~unter two last scintillation counter hodoscopes (HA and Cerenkov (Group Behind the counter large shover B SC), not used in this ex per iment. The 40 inch bubble chamber stored behind (C2) the and a sho we r counter was -18- also not used in this experiment. aentioned here are discussed in the Other detectors not next secticns (aostly scintillators associated vith the beam system). rhe gas systems, power supplies, systems, and readout magnetic electronics field with the exception is associated spectro meter detectors are not described. An shielding the cylindrical chamber system which is de~cribed in greater detail in Appendix 1. The data acquisit ion system is described in chapter 3. 2.C -- Beam System SLAC beamline 20-21 (shown in fig. p!cticles produced spill lasted LASS by the interaction of the 20 GeV/c, 1.5 mA., primary electron beam with ba!m 2.C) tra nsported to a berylliu ~ target. The 1.6 microsec., repeate d up tv 180 times per second. :~llimators in the beam line were adjusted to allow about pions to reach the target per pulse. 4 (The u sable rate was limited to 10 pa rticles pee spill in order to aviod multiple interactions. i~ns from In practice, the long old lifeti me of pions residual tracks in the spark cha mbers necessitated a lJwer trigger rate, which was acco mp lished by number of per spill.) The beaa reducing the vas focused by ~lternating horizontally and vertically focusing quadrapole I ELECTRON BE A M TR AN SPORT 6- 78 20010 Figure 2 . C: \ 2 I OI VERT ICAL 2102 TR W ERS ING CCU NTER I ~Ylj~ f-j VERTICAL PLA NE i::J BL2O ~ _s~1-, -~__ _ 030 T C 3 1 e, STEERING SEMS F2 COLLI M A TOR ~~ ESB WALL ~ BEAM PWC's / LHz20F 4 I \ !'\.'-"-· XY-C0UNTER 21 F4 COLLIMATOR 8cp-H0D0SC0PE SLAC Beamline 21 ·~----v- ~ , LllJ/ ./l_c:j,/ ~--6 2007 \ 20 QS 2008 L'7] Fl FlX ED COLLI MATOR PC42 840 f,l 8~1(:l r-p-;35 PC43 BEAM POR T 3n6A I ~ ~ WE DGE 52 COUNTER Cl . BEAM SHUTOFF 51 2 \ ~ ,..,...-----.,.,.....RFS / 1 . 2o o ~ R F s \ / 10N CHAMBER ~ ! ~ rLL'.J i _ __J\ 14 i:LJ - - + 0 - 1/J f77! Dl 2004 2006 2002 ,l( V) F2 2001.1 2002 (Hl I sn ·soLs ( V) 3.1 3°N •1 PM,AP PULS ED MAGNET DC DIPOLE (er DUMP ) B,D QUADRUP OLE Q CURR ENT- ME A SU RING T0 R0ID j r p POS ITION MONITOR AL UMIN UM OX IDE SCREEN PR SL! T SL F2 TO F4 TRANSPORT TC TARG E T RF- SEPARATOR RFS ~~ REMOTELY VARIABLE Ft C0LLIMATORS P-H0D0 SC 0PE 2 0::·3 20D 3 ( H) 4 . ~. oN i :!: 3.65 mr COLLI MA TOR .,.,c- BE AM 38 5 ~ 31 PM1-5 /,',030 P30 832 133 P32 H V PM3g - rTT"j' ·,:_ __/',_ TT 1. __n___n ___ /\__ __ f : : l ~ n· --- - ---------ull.l.!h:----V ~~ • T MA~~ eI30 Q3l , 830 Q32 ~ 833 033 AP3 0 BE. 835 034 AP31 BEAM SWI TC:iYl, RD (BSY) TARGET TO F2 TRA NSP ORT A - BEAM I tO I-' -20■ agnets, the position of a particle at the first focus and depended variable slit lead collimator passed only particles with momentum of 16 GeV/c (Fl) on thin its momentum. A lead filter at F1 removed electrons from the t21. A beam. The magnets after F1 brought the ac~romatic focus (F2}. beam to a second The beamline has been described in detail elsewhere(2). Particle-type separators the separation was between F1 and F2. dispersion accelerator possible using the RF (The separation vas based on in space of particles bunch due to velocity from the same differences., The separators were not used since 941 of the particles produced were pions. Like Fl and F2, F3 was a momentum dispersed f ocus and F4 was a=hromatic. The P-hodoscope scintillation counter tagged the momentum of located at ?3, with six fingers overlappin g to form 11 bins, each particle to ±~251. F4 was l~cated at the far end of the hydrogen target. Tvo threshold Cerenkov counters chec ked the mass of the beam particles. k~ons. c-pi tagged pions, They were located C-k tagged pions and near the end of the beam line, just outside and inside rhese did not differentiate pions from muons, and counters the and LASS buildi n g 2.31 of the accepted beam particles were muons. respectively. -21- ?he theta-phi (named hodoscope located for in pinning the down x and y. counter fingers Downstream, near the target, vere tvo sets of proportional vire chambers one meter apart. and angle) front of c-pi tagged the position of the particles vith 0.127 cm. scintillation in track direction Beam position were calculated from the theta-phi hodoscope and these chambers. The main computer provided online histograms of the beam position which was adjusted manually vhen necessary. Scintillation counter SE, just inside :-k, generated a narrow (2 nsec.) pulse in the trigger logic circuit. beam position was 11ade The XY was a q~adrants 1.91 cm. building before pulse, the master timing A final check of the the XY and RI N·G scintillation by counter assembly. (each the square square, divided 0.64 counted equally if the beam was centered. into four cm. taick) which The RING counter, also with four sections, surrounded the XY and was used to veto particles outside the main beam spot. 2.D -- Liquid Hydrogen Target The target assembly was separate from the It was mounted on solenoid rails so that the hydrog en vessel slid into position inside the solenoid through t he iron end plate. magnet. Fig. hole in the 2.0 shows the target construction. 7-78 2 1(16 Dia. J__ 11 11 Figure 2.D: ~ LH 2 Return Flow Mylar Cylinders 36 Target Cell Platinum Resistor \ Mylar Window Vapor Pressure Bulb. Resistor Wire Leads Vapor Pressure Bulb Liquid hydrogen target Mylar Window LIQUID HYDROGEN TARGET ).,t20(1,t I I N N -23- rhe hydrogen vessel vas a cylinder 91.4 cm . ••• in diameter, with end windows long and 52.4 0. 13 ••• thick. The hydrogen vessel was enclosed in an evacuated .71 am. thick with aluainum tabe a 0.13 ••• mylar win do w at downstream end. 2.e -- Proportional Wire Chambers lll the proportional electronics. The chambers in LASS used electronics were unus ual recorded in a memory the time when a signal vice in nsec. 25 The bins(3). low in that they o ccurred d uty same the on a cycle of SLAC necessitated a high instantaneous particle f l nx which led to the need for feature the precise ti me resolution . This memory also avoided the need for delaying tbe signals with a cable to await a final trigger. I n this experiment, wrong ti me tracks and tracks without proportio n al cham ber points ce ntra l wire plane for eac h coordinate and Y coordi na tes. were rejected. Tbe planar proportional chambers had a surrounded measured. by two cathode planes The cha mbers meas ured the X and another coordinate (called E) at an a ngle bet ween X and Y. This measure ment was necessary to reso lve amb iguities in matching up the X and Y coordinates. v~s used In tract with the same weight as X an d Y. consisted of gold plated tungsten wires f i tting it The wire planes s t re t c hed on an -24sluminu• The cathodes were alu ■ inized aylar sheets frame. etched to prevent operation outside the cathodes were connected sense wires at ground. active area. The to negative high voltage with the The sandwich of three planes was enclosed between two aluminized mylar sheets vhich served as a gas bag and blocked light. In the plug chambers, each plane has 256 wires 1.016 at 35 degrees to the mm. The horizontal. solenoid E plane wires are spaced at The frames are large so as not to obstruct the the wires are 2 met ers long. The bore, so that cathodes are etched avay outside the 33 cm~ where X square region Y wires are supported with mylar strips perpendicular t o them, limiting the sensitivity active length is to desired. The ±14 cm. and The mylar strips for the E vices are placed the same as for the Y wires. The 1.5 and 2.5 chambers have planes with 64 0 wires each, apart. The wires are supported at two places inside the active area. The E 2.032 mm. plane wires run at 45 degrees. The beam chambers have four planes each. . his an degrees. Y' X plane, The first (BM-UP) a Y plane, and an E and a P plane at t45 The second (B M-DN) has an x, a Y, and an X' and a displaced half a vire spacing from X an d Y to double the resolution. There are 64 vires per plane spaced at 1.016 -25- ••• JH-OP and JH-DN have only one The active area with plane vertical vires. coYers the aperture of the dipole. There ~re 512 wires in each, spaced at 4.064 mm. The trigger chambers concentric rings, segments. The rings (in cm.) each cathodes were divided divided into 128 approximate were: track 3, 7, 23, 59 for TA; verification three wedge shaped radii of the boundaries of the 32, 48, 71 for TB; 17, 25, 38, 55 for TC. for into (no inner ring), They were used only because the resolution v~s so poor. The cylindrical proportional chamber co ns truction vas different due to the requirement of low density of matter in the path of scattered particles. It is shown in fig. 2.E. The basic structural support vas the 3.5 mm. thick between two The 160 anode mylar-styrofoam g~s seal. ~ctive dead area. of wires vere 5.188 cm. An tube served as the o th er cathode and There is a plexiglas ring in the length tube plexiglas rings glued to the mylar. Wire spacing vas 2.037 mm., at a radius o~ter styrofoam bonded to a layers of 0.005 inch aluminized mylar on the inside and outside. stretched inner center of the to stabilize the wires which caused a 15 mm. STYROFOAM 20µ. WIRES ?i'i~ = = '\.~ 2=.~-., . . J" ,~ 5 POSITIONING PIN · -.- ~ \ \ < / Figure 2.E: LUCITE SUPPORT RING 'd " · - \• >, > _ , , \ LUCl~E~PC BOARD POSITIONII\JG/ PIN CON N. WIRE --~ Af FEMALE PIN t:2::cc:cc:22~ '· • '2<2 , , 7 Cylindrical proportional chamber INNER CATHODE ", \\~ •c:== • 2 > '\ \ 3 ==z:z,,2z:::z:::/::::z:::/z:z,2,2/:::z:::z::::,='zz:zz2,2z:::z:::z::::z=Aa, OUTER CATHOD~ \ \ _c.:11;--· . . -··~ - =~~~~~~~~-- - - -·~ - ~ - -_,_L-..___ -_, I \ J:,, T L'-,c~ _,·- - ~ c:c\--=:c.:.c::c:c 1 -~ - - ::c::: - ~ . --:--,~- ~ PC BOARD. 5 mil. aluminized mylar 4 ;;,;,,'J BEAM I I N O'l -21- 2.F -- Spark Chaabers Four different cylindrical kinds of spark vere chambers used: capacitor-diode, plane capacitor-diode, and tvo sizes of magnetostrictive chambers. I, All the plane chambers had four readout planes: Y, E. ~nd P. The X and Y were one spark gap, the E and P another. The and E wires P vere ±30 degrees from vertical in the MS-downstream and CD chambers, 25 degrees in the ~S-twixt. The planes were a woven mesh of nylon filame nt s and aluminum wires. wire spacing was about 1 mm, which varied over The the vire cloth. It vas mapped and All chambers were deadened in the region of the the plane corrected in beam by the insertion of a polyurethane disk software. into the gap rhe twixt MS chambers were 4 m. wide by 2 m. high, and the (radius 10.8 cm. for CD's, 3.68 cm. for MS's). downstream chambers vere 3 by 1.5 m. The following scheme W3S devised to cancel the relatively large twixt were chamber wands: The wands field for wrapped with fine enameled wire which carried a large current { 10 amps, at v~ltage field drop of several hundred volts). longitudinal the a This created a that opposed the spectrometer field. This field, arising frora the solenoid, was radially symmetric; the component zero in co mponent along the of wand the was the center and in opposite directions at the ends. -28rhe canceling coil winding direction r eversed in the center. The spectroaeter field was somewhat o vercollipensa ted near the center and very undercompensated efficiency near the ends ne ar the ends. was therefore poor8 The (A varying pitch winding would have properly solved th9 problem.) I~ the capacitor diode chambers, each wire was connec ted its own small readout circuit. charged a capacitor which electronically to Spa r k c t rrent in the wire remained c harged and long after the spark was over. ~as read In the plane CD chambers, a network of diodes and da ta busses allowed the c~pacitors to be interrogated serially by a small number discriminators. This had of t h e advantage of a low c ost per wire and s mall size. The disadvantag e was the rectificat ion of inductive drops enor mous voltage rates of along t he busses caused by change of curren t in the sp21rk circuit'" This proble m has been solv ed ov er a period of several years9 When this experiment ran, the efficiency was still in the 60-80 % range, and crosstalk bet ween v i res vhere the cu rrents were large led to spar ks apparentl y s pan ning many wires and a distortion of the apparent positio n o f the spa rk. The cylindrical CD cha mber readout wa s more ~od also more bulky and discri minator for eac h wire and straightfor~ard expens ive. There ~as no was s ha re d. c ircuitry a The worst feature of this syste m was that each capacitor and its discriminator were separated by 25 feet of wire e The -29- o~ly problems with the readout circuitry have been of capacitors or diodes failures with 21 of the wires associated (after 10**7 pulses, about 10~ sparks per wire). ?he cylindrical cylinders chambers were made of concentric mylar with wires glued to their surfaces, shown in fig. 2.f. The cylinders are supported only by the ends. spacer rings at There were problems with the straightness of the cylinders and consequently with variations in their spacing. This led breakdown. to regions of It found was low efficiency that a and regions of large concentration of alcohol in the gas mixture allowed the chambers to operate over the wide range of electric field strength that occurred with the nonuniform gap spacing. The breakdown problem increased with use (possibly because of rou ghening of the surface of the wires, especially in the inner ch~mbers where the spark rate per wire was high). before tne chambers were installed, and one chamber Sever al wires vere cut because of breakdown. was shut off during the r unnin g because a breakdown in it was shunting current away fro m the ch~mber. adjacent The efficiency of these chambers ranged from 60 to 98,. There were two types of chambers in the cylind rical package. In one the wires were glued at a slight angle azimuthal angle varied along so that the the length of the cylinder. The wires on the two sides of the gap pitche d in opposite -30- (a) Plexiglass Rings Gaps Gas Window / ~~ Aluminum Frame ( b) Ax ial Cathode Anode ( not read out) Right Pitch Anode Cathode Figure 2.F: J• 26A 2 Cylindrical spark chamber: a) package b) detail of gap -31directions so that l~oking at both sides gave the lengthwise position as vell as the azimuthal angle. the wires read out. first vere parallel to the axis and o n ly one side vas The chamber package type In the other type, consisted of five alternating vith five of the second. of the The radii varied from 10 cm. to 60 cm., and the lengt h was 1 meter. rhe cylindrical spark chamber package was t he major piece of experimental apparatus for which the author v as responsible, and is described more fully in Appendix 1. electronics The supporting and gas system are described as ~ell as results of performance tests. The gas for all the spark chambers was standar d neon-helium, purified and recirculated. the Alcohol was added separately gas for the cylinders and the ~s chambers ran without alcohol. (plane) CD chambers. to The The high voltage pulse for all the chambers (300 nsec. long, about 4 kV depending on the cha mber) was formed by switching a charged coaxial cable by means of a thyratron tube. clearing field of about 40 volts to over from the spark. of 250 volts cylinder applied and remove understood. ions left to the MS and cyli nder chambers for The recowery CD chambers was 50 msec. produced an increased spark multiplicity. aot ~ he There was also a pulsed clearing field extra clearing after the sparking. the All chambers had a d.c. The time for Bunning faster ~his effect is experiment was run wit h a compromise -32of 25 msec. between events. desdti ■ e 2.G -- Cerenkov Counters Four Cerenkov counters were used were used for tagging in scattered e x periment. the beam particles. were large and complex, with large tag the particles. All angular were Two The other two acceptance, thres hold to counters, 5=cepting all angles of Cerenkov light. The beam Cerenkovs were simple pressurized pi pes in t h e beam line. C-pi was 5 m. long and 20 with hydrogen 25 psig. in dia meter, filled (Threshold ga mm a 37; 5.2 GeV/c C-k was 1 meter by 20 cm. diamet er filled f~r pions.) c1rbon at c m. dioxide at 35 psig. with ( Threshold ga mma 18; 9.1 GeV/c for kaons.) rhe multicell Cerenkov counter, Cl, was origi nally t~ identify pions using ~tmospheric pressure. a high for refracti v e index gas at In this ex p eriment, a i r had the right index for separating electrons and (threshold elec t rons, 20 hadrons MeV; for Because of the lo w refract i ve inde x of air amount of light was radiated. increased by moving upstream solenoid. and The designed the sealing to 5 GeV/c pion s, 5.8 Gev,. an insufficie nt Th e efficie n c y was therefore upstrea m it below the window 60 inner su rface cm. farther of the cell partitions were not extended since the -33:erenkoY cone was narrow (aarlmum 24 ailliradians) generally illuminated only one cell at a time). and The counter is drawn in fig. 2.G. The counter was innermost divided region was into four radial regions. split into two cells and the outer three annular regions were divided into t welv e The radial The cells each. distance from axis of the inner and outer walls of the cells increased with distance fro m the target so that stiff tracks tend to stay in one cell. ~t A thin mylar mirror the end of each cell (just past the end of the solenoid) reflected the light outward, through a long arm, it was focused by an ultraviolet-transmi tting fresnel where lens onto a 2 inch ph~tomultiplier tube. the extension arm, the tube required the extension magnetic field. Even at the end of heavy outer cells from The extension arms for the inner cells r~n out behind those of the oater cells so cells were longer. shie lding that the inner Not counting the upstr e am extension, the were 1.25 m. and the inner t vo 2 m. long. The bottom three cells of each of the outer three rings were not installed when the experiment was run. These 9 cells are counter with referred to as the "unequipped region". :erenkov C2 vas an eight segment several centi met ers pressurized of material in the path of tracks. vas filled with freon-12 at a pressure of 44 it useful only It psig., making for separating protons fros kaons. It vas -34- PHOTOTUBES AND SHIELDI NG SECONDARY MIRRORS FRESNEL LENS ' I' EXTENSION ARM I' ALUMI NU M SUPPORT SHELL I I ACLAR vVlNDOW PRIMARY 7 ·~-r--tt--:::,,,.~1.!RRORS -------- ----------, --- --- ---,, \ '\ EXTENSION 9- 78 Figure 2.G : \ ~ I SOLENOI D MIRROR RING 3~2 6 A 2 1 Cl Cherenkov counter, s_a~ i ' view 0 -35not studied or calibrated in this analysis. 2.B -- Shover Counters ?bree types of shower counters, experiments, were used. SLAC Group B, SLAC built earlier front face other They were named af·t .e r their owners, Group E, and Cal Tech. counters and three Cal Tech counter pairs the for we re Ten Group E mounted on of the dipole magnet as shown in fig. 2.H. Along with C1, these counters recognized the electrons. The Group the at the dovnstxeam experiment tagged some of the high energJ through B shover the counter dipole. (A end tracks of passing track straight down the solenoid center needed 5 Gev/c to hit this counter.) It has not been studied or used in this analysis. rhe Group E counters were designed output uniform light over the entire active area (18 by 25 inches). They consisted of 16 0.25 in. sheets of lucite give wavelength-shifter-doped with interspersed lead (t ype-metal) sheets, and five 2 inch phototubes along one tapered to to absorb less light acceptance was lo wer. edge. The lead sheets were energy near the far end where the The average pulse height (from . the five tubes summed) varied only 101 over the face. The Cal Tech counters consist of two 80 x 9 x 0.5 inch sheets interspersed lead with three 1.125 in. sheets of lucite. 9-78 E sea L __ I'sc1 I I I :sc6 I Figure 2.H: ·l..5.~ f g~., Lip . ,., SCIO SC9 50 cm Dipole-face shower counters L Dipole Aperture SC3 -----,------ __ _J sc2 1I I SCI: I I I J ,'26A10 E Group '~- SC5 SC4 jcal Tech r-- - - - - - - ----- --7 SCl3 SCl2 SCII I I Ol \J,I -37The light pipes at each of the long iach tubes vere meant ends, for five adapted to tvo inch tube s vith reflective aylar shaped into a truncated cone (shield ing tubes from the magnet was not practical). the large Two such counters vith 0.25 in. lead in front of the first a nd between the tvo g~ve acceptable energy resolution when the signals froa all four tubes were summed. yellow, In spite of the l u cite being very the light output varied less than 20% from the ends to the middle. Both types of scintillation counters counters differentiate electrons these were were preceded (called from by 0.25 ngam ma-veto " photons. counters) to The signals as follows: The four for CIT). types vere signal fro m t he last dynode vas inverted and su mmed f~r all t ubes in the c ounter GPB, from used in the trigger as well as i n the analysis. The phototube signals from each counter of bot.h treated in. (five for The sum was discri min ated for use in the trigger and its presence recorded. Th e were separately digitiz e d and recorded. In t he a nalysis the anode signals individual phototube pulse heights are weight ed according to the kno wn position of a hi t to get better e ner gy resolution. The Group Band CIT counters were c a librat ed at several p~sitions o ver t heir faces wi t h elect ro ns and pions at three be~m energies. This informatio n was used i n t he analysis to convert pulse from height to energy as a function of the -38p:>sition of a tract. Several counters were calibrated again, as a check, after the experiment vas dismantled(~,. 2.I -- Photomultipliers and Downstream Scinti llators c1 Those phot otu bes giving analog inf ormation.u in three types of and the shower counters, vere outfi tted with light emitting diodes for gain calibration. The calibration was d:>ne automatically before each run by pulsing the LED's 2048 times recording the pulse heights fro ~ th e tubes. and tube high voltage was manually ~hen the pulse When the sho r1er counters were height drifted too far (20%). recalibrated adjusted The the end of the experiment t h e light output at from the LED's was found to have drifted in some cases by as such as a factor of two. The response of the shower counters was therefore not constant over the exp~riment. attempt was made recorded at different efficie ncy used to to treat differently p~ lse height data times. correct The the to first order, electron detection result ~a s obtained from electron trigger events recorded over therefore, No took the these ~hole run, and va riations into account. The phototube high voltage was supplied resistor and gas tube voltage divider boxes~ e3.ch tabe ~r om adjustable The voltage to was monitored by a multiplexed d.i gital voltmeter which was read by the computer every 256 e vent s. -39- lll tubes were shielded by solenoid a from ■ agnetic the field twixt-region (C1, GpB and CIT coil" which The shower gamma-vetos) required another steel tube "bucking the layer of au-metal inside a thick steel tube extending well past the photocathode region. the of tubes in counters and and a so called carried a current adjusted to cancel the field. There were three areas where simple were used in the experiment: scintillation in the co11nters beam system; the gamma-veto counters; and the downstream arrays, HA h:>doscopes downstream and the lollipop counters are described below. s:>me of the triggers counters. These (although The arrays and were used HB in not in the main electron trigger). Hodoscope HA consisted of 42 rectangular inches, arranged without spaces. were only leave a 4 4 The center counters on the top 4 by was inch counter top make the 4 by by 33 hole inch and bottom They were raised and lowered to similar, surrounded by seven 6 8 by side, 21 above and below center, side inches wide. HB H:>doscope paddles for the undeflected having ten 4 counters on each inch beam. counters side. One and bottom just off center were displaced to 4 hole. These hodoscopes c:>rroborate the timing of downstream trackso were used to -40 - rn e lollipop c ounters were a s et of three concen t ric cir c ular counters of radii 8.6, 11.4, and 19.7 ca., The array was hung between the dipole and J B- DN centered on the beam spot. t1~ stacked They were used to define of the auxiliary triggers. Rafere nc es for chapter 2 1 • . The detector comp:>nents are described in more detail inc. L. Woody, A Study of Electron Pair Production in 16. 1 GeV/c pi-minus proton Collisions Using a Large Aperture Solen~id Spectrometer, Ph.D. Thesis, Johns Hopkins tJniversity (1978). 2. Design and Operati:>n of SLAC Beamlin~ 20-21-22, P. : . . Winkleman, SLAC Report 160 (1973). 3. Shapiro, s. L. et. al., IEEE Trans. Nucl .. Sci., NS-23, No. 1, 264 ( 1976) • o p . cit., c. L. W:>:>dy, Thesis, Appendix B. Shapiro, s. L. et. al., IEEE Trans. Nucl. Sci., NS-23, No. 1, 269 (1976). 4. -41CRAPTER 3 -- Electronics: Trigger and Data Acquisition 3.A -- Introduction Five separate triggers vere generated and or-ed together form the master trigger which caused the spark chambers to fire, the proportional chamber system to hold its the to preceeding 500 nsec., and l3tches to record the phototube ■ emory of the digitizers and binary information. The trigger palse also started the digital controllers which transferred the data into the memory of a PDP-11 computer. When the data signaled. next transfer completed, the computer vas The PDP-11 then reset the controllers to read the event, compressed the data, and initiated the transfer to tvo other computers. d3ta was on magnetic These computers recorded the event tape, and analyzed some of th~ events to produce visual displays (shoving a nd scintillator to monitor the )ne of the destination co mputers vas the SLAC triplex (tvo hits and reconstructed chamber tracks) and performance of some detectors. IBM 370/168s and computing for SLAC. a IBM 360/91) which did most of the IBM 18 00 which The other was an was used only to record data when the triplex vas not operating. The program on the triplex resided in the memo ry of one of the 168s along with several other jobs. The ntASS" job vas -42different in that it needed to be executed soon after data vere presented by the PDP-11 or the experiment The would stop. PDP-11 vas able to store two completed events and still read in a framework third. of a The LASS program vas built the program called BEALTIME which could demand execution with a higher priority than j~bs. program transferred The in REALTIME normal subtasks, for example from the subtask that computation data between received data frlm the PDP-11 to one that wrote it onto tape. 3.B -- PDP-11 System Almost no data processing functioned as (OMA). detector by the controllers PDP-11 by signals), pulse discriminators height memory CAMAC which and digitizers, complex fast digitizers, scalers and the multiplexed digital the cylindrical proportional chambers: IBM system; There were two D~A output to spark ports, one for messages system. The PDP-11 each detector OMA controller before each eYent an area in memory tc be filled. for voltmeter; 1800 and one which connected to an IB~ SYSTEM/7, a minicomputer extension of the triplex assigned the time the CD spark chambers (two DMA ports); and the MS chamber system. the chamber which direct There were DMA input ports for: read binary latches {phototube llgic done It vas connected to the other a controller. co mputers and the access was It also assigned coming in from the triplex computer. an area When it -43- finished formatting a data buffer it could i nstruct the SYS7 or 1800 D~A to transmit it directly from memory. The D!ls signaled the data the PDP-11 with transmission vas complete. a done an interrupt v hen The detector D~As each received signal from their respective device controllers and did not signal the PDP-11 until all were fi ·n ished. The data acquisition vas controlled using th e PDP-11 control It panel. consisted of information like a nd c~nfiguration of pushbuttons date many starting and and computers. logic levels entering used were There endi ng ru n.rs,, and indicators used to display the state of the PDP-11 produced for run nu mber. &~d for choosing a detectors for s witches prGg ram. It also to contro l t.f:.1.e fast logic and test devices for the detectors. It control1ed a "self scan" panel which displayed data logging infor mation : tape i.d •• file number. and run number. Besides administering data collection the P.DP- 11 the size and layout of detectors vere requested. messages da t a It the da ta logging confirmation. It read and volt meter data every 256 e ve nts su mmed intern a lly a buffers depe nding on which se n t to calculated begi n- r~.1 .n and end-run co mputers an d wa i ted for trans mitted th e scaler and run. It and be a m-count ing at e nd scaler of th a t vas reset every event. and it ran the LED phototube c al i br a tion at the beginning of each run. The LED test was s imilar to nor mal -44- d!lta taking vith the following excepti ons: The PDP-11 enabled the LED pulser which flashed the LED's 20 tiaes per Only the CAaAC D~A was enabled (it vas triggered by second. the LED pulse instead of a Yalues vere transmitted to intercomputer bea ■ summed over avoid the link. The particle). 64 The pulse height repititions . speed before limit ation being of the only data proces sing done by the POP-11 was the summing of the software scal e r (BTEZ) and the LED pulse height records. 3.C -- Fast Logic: Triggers, Gating and Sca lers The fast logic generated the master started the the recording of trigg e r event. signal The trigger was a coincide nce of a pion entering the target 6 a n enable called the event which signal g a te, a nd a trigger signal formed out of signals from Ceren kov C1, sho wer counters, an d scintillation counters. This trigger sign a l vas the logica l sum ("or") of five separate triggers, each selecting a part icular kind of event. ?he event gate was a coincidence of the run g ate , the PDP-11 )K flip flop, spark c ha mber deadtime, and c lea r i ng field OK. The 11 OK flip flop v as turned off by the master trigger and OB by the PDP-11 when it was ready for the ne xt event than 10 msec. except when data buffers bac ked up). spark chamber deadtime vas a si mple 25 msec. delay to (less The give -45the spark chambers time to clear out qld ions. field OK signal stopped the experime"&,t The clearing vhen the pulsed clearing field power supply for the !!S ch~,rabers latched up . . The event gated scalers counted only vhe!§. the experiment vas ready to be triggered. t3 the events The numbers they s·hoved corresponded actually recorded. Th~y were reset at the beginning of each run. Counters SE, XY, Bing, C- k, c-pi (BEAMTRIG). The logic is defined tke shown in figo l.C.1. segment XY counter produced signals IYGE1 vhen fired, and XYGE2 beam The four any segment when two were hit within 20 nsec. was used as a veto to avoid events too close in particles signal XYGE2 time. Two passing through the same segment ~ere not vetoed. The ring counter tagged particles which could hit the target wall or the support structure and produce a spla sh. A hit in it vetoed the beam signal if it occurred ¥ithin 150 nsec. (before or after). C-pi to fire as rhe final pion trigger required c-k and also for lta.ons, protons, and "jun Icons" (C-k fired vi th-cut c-pi). The ■ aster 500 trigger well. came Signals about were nsec. generated after the particle entered the target. Five kinds of triggers were and T4. generated, TO was the electron trigger. one of tne cells in C1 to fire and at named TO,T1,T2,T3, It required at least least one of the dipole face shover counters, plus its gamma-~eto counter, to 3 -77 XYBR XYTR XYBL XYT L R4 R3 R2 R1 SE B.S . IXY?:2 SCALERS AND ,., I X-Y?: I ":> Ck c.,,. VETO Beam l ogi c IXY?:2. SE •D<Y?: I •IR Fi gur e 3.C.1 : B.S. PROTON KAON PION D AND J l93A7 BEAMTRIG DISCRIMINATOR DOR C> I I cn .i:- -47- The fire. logic is shown in fig. 3.C.2. :1 corresponded to 1 count ADC above The threshold in threshold. in the shover counters. about 600 ADC counts above threshold. T1 was the total interaction absence of trigger. requi ring only the a signal from the 8 inch lollipop counter (LP3) in the undeflected beam path. The data were cross compared to a total from section this trigger bubble chamber experiment to check the normalization and the track efficiency. The probability of preventing electron data Tl C1 and were shower finding used also to estimate the counter ~cell sharing" identification in elect.r on pair events (ncell sharing" cuts are described in sect.ions s. B. 2 and one only one 5.B.3). T2 was the elastic trigger, requiring and element of HA and one and only one of BB fi rin g (undeflected beam went through the central hole). intended to check the accuracy The elastic data vere of momen tum and angle aeasurement. of the proton at large angle and of the pion in the downstream (dipole) system. (This check was not completed.) Tl was the beam straight through trigger. inch lollipop (LP1). requiring the 4 Tl was used online as a trigger for measuring proportional cha mber efficiencies oBly one hit should be seen in each. since one and The recorded data were 7 _/>------- ~ V I •1 _, Electron (TO) trigger ""' ~ L'.Ci- Inner To Buffer Strobes Figure 3.C.2: GV13 SC13 GV 1 (D~~~de) sc, c 1- c,2 c, -c, C1-B12 c,-s, C1-A12 C1-A1 r-..1 r(SC·GV) ~ I Beamtrig re,~, T0 7- 71 J◄l6All ~ . I I .i::- I 00 -49- used to align the cha ■ bers. r11, the gamma-veto trigger, required one or more counters to gam ■ a-Yeto These eTents vere topol.ogically similar fire. t~ the TO trigger, but accepted all charged tracks. intended to provide punch-through. It vas large angle tracks to estimate hadron Lacking computer time to process these data, th e Tl data sufficed for that purpose. rriggers T1-T4 occurred more frequently than TO . "scaled down" by counting each and pass ing it on to the master trigger only once in a preset number of They were ea c h divided by numbers vere not kept fixed) a were They factor occurrences. of 20 to 40 (these with the result that they made up about half the triggers. 3.D - - CAMAC and Chamber Readout The phototube signal latches digitizers (ADC's), the (buffer time strobes), digitizers scalers, and the multichannel digital voltme ter all controlled TDC's writing (and pulse also height counters fired as part branch read by CAM AC. one scaler), values of the and pulse (TDC's), the (DVM) vere There were two CAMAC The one that read the buffer c ontrollers. and and the strobes, ADC's, operated every event, bits event indicating record. controlled the scalers and DV5 modules. The which other It was read -soby the PDP-11 every 256 events. This branch vas also read bf a versatile scaler display system which connected to aany iientical display units vith thumb switches. rhe scaler c~unt corresponding to the channel number on the svitch c~ntinuously fetched and displayed on was each, so that any s=aler could be dialed in and displayed. rhe controllers for the proportional chambEr, the CD spark cb~mbers, and the cylindrical spark chambers were similar to ea=h other. Each represented hits in the chamber as an ~idress of the first wire on and a "width" giving the number of adjacent wires. This method reduced the number of data entries which were more than one wire wide. rhe for hits pr~portional chamber hits were grouped memory time slot. The regrouped the wires into clusters. system way program analysis clusters, grouped the time slots this for the for each expanded the same wire, and The proporticJal chamber was read out in an expandable chain so that chambers c~uld be added without rewiring. The cylindrical system had fixed "fiducial" simulated hits which gave immediate warning if it miscounted. The MS chamber system had a set of timers f~r each vand, which were stopped one at pulses a time were received, up to a total of 16 hits. hits were not recorded. as spark Additional -513.E -- Bealtime Network ?Be "realtime" program made it possible to use c,mputer system between various execution that controlled the of and tasks contained in the job. Tbe REALTIME job was composed of the following t~sk triplex for realtime data acquisition and control. It ran as a batch job and controlled the c,amunication the SYS7 "tasks": a data chan ne l, a task that wrote the data tapes, a task that contained the trackfinding program and analyzed a sample of the incoming events, a task that controlled the graphics display scope (located in the c,ntrol room), and several tasks concerned with the realtime m~nagement. Tasks tbe normal time which sharing could not wait for their turn in were sequence executed at high system in priority{l). 83altime differed from a dedicated the greater speed availability of minicomputer and memory space of the IBM 370 and the peripheral devices (which would be too expensive to buy and maintain if they were not shared by all users at SLAC). iQch, four The high density tape drives (6250 bits per times managing the data. n~rmal high density) wer e Even vith the high dens ity, crucial to tapes were filled e very hour, and a total of 185 were writt en. Another ~ivantage vas the existing software for the t riplex: a good taxt editing and file storage system; an o ptimiz ing fortran c~wpiler: and a graphics display system. -52- ·r ile disadYantage vas that the triplex syste11 vas out of b~nds of the experimenters. The triplex often vent dovn, tile operating system being so complicated that its operation vas not understood. the detailed But for the cost of a tape drive, writing tape with the PDP-11 would have been ideal. References for chapter 3 1 • . Introduction to SLAC's Real-Time System, I. Denecke, Stanford Center fjr Information Processing at SLAC (SCSSCIP) (1974). -53:alPTER 4 -- Trackfinding Prograa 4.A -- Introduction rais chapter describes the computer program which r1w data tapes and produced data p1rameters of reconstructed tracks. the mathematical conventions, p1rametecization of the tr~cks summary Section as the tapes of the 4.8 coordinate read describes systems and the helices. Sections 4.C 4.E describe the steps by which the raw coordinates tnrough of hits in the cha ■ bers were candidates, into and match-points, aatch-points into checked plausibility and fitted to nearby coordinates. for track combined the rhe final chamber alignment procedure using iescribed in section 4.F. Section candidates the 4.G tracks is describes a comparison of tracks found in interaction trigger data tricks produced s1me beam. that in a with bubble chamber experiment using the Section 4.H describes a pre-track findin g filter quickly determined that a large nu mber of events could n~t cont~in an electrJn and removed them. Several other diff icult task. c~ts are described also. Miking a good trackfinding program rhe matnematical complexity was a of helical tracks, and of the irregularity of the chamber types conspired with low chamber efficiency, spurious "iirty" events hits, and a trigger that (i.e. containing electron sho wers). selected It was -54na=essary to include orier not to approxi ■ ately reject versus the number of track candidates. procedures described spurious the real tracks. trade-off was computer time ■ iximum 20~ below tracks in The most important examination of a Most of the complicated were intended eli ■ inate to implausible combinations at an early stage. 4.B -- Mathematical Conventions Two coordinate systems were used to points An orthonormal system took in the computer program. represent tracks z as the direction of the beam (increasing downbeam), and Y as and X to complete the system in the right-handed sense (North). The origin was taken at the center of the solenoid up, for X and Y, and the inside aGi-plate z. for face of the upbeam The other useful system was a cylindrical system (R,f, and Z) centered at X=O and Y=O when t, the solenoid cylindrical chambers, or, when halical tracks, concentric with the helix. referring referring to the The momentum of the track (discussed in Chapter 5) is expressed in spherical c~~rdinates, P, e and f.) and P were constant for the length of the helix. The magnetic field was treated p~callel to the (ap as constant, uniform, and z axis, in the region of z from Oto 285 cs. to Plug3/CD3). Beyona 285 cm., the path was calculated by numerically integrating the equations of motion, using a -ssrapresentation of the field interpolated between measured v~lues vith a polynomial. Inside the uniform field clia.racterized region, five by the track was a helix parameters: X0 and Y0 , the c:>:>rdinates of the axis; B0 , the pitch, 3t z position Z:,.) p:>sitive and t~king Z at a fixed beginning of CB and negative the ( Charge, CH, was +1 tracks place, it the inverse Cf' value, q>.,, or for -1 respectively.) Rather than vas denote used to also, either one determining of motion of the particle along the helix. the parameterization, the X and Y position at a given Z given by: <f the track, and recorded as a sixth quantity. were recorded direction B0 , and a starting B0 =d f/dZ=-CH•Rc/TAN (e): a given radius; = fo+Bo • (Z-Zc,) , I=Xo+Rc•COS (q>), Y=Y0 + R0 •SIN (<p) • The momentum was related to the helix by P=a • V ( R0 2 + 1/B O 2) , Pt. =a•Ro =P•SIN (6) , P£ =-a•CH/B 0 =P•COS(0), Px_ =-CH•2t_ •SIN (f), Py=CH•Pt •COS (<p), where a=(magnetic field)•0.000300=.006735 for 22.4 kG. the Using vas -564.; -- Match Point Making !3king "match-points" was the first naaber of track step candidates searched. in reducing the When N tracks hit a ch~mber that measures only X and Y coordinates. there are N2 X-Y pairs to choose from. reiundant By measuring the hits along a (linearly dependent) direction. and requiring the projecti on of the I-Y point in the extra direction to be hit. most false pairs are corroborated by eliminated. The corroborated pairs are called match-points. The proportional plane a measured chambers (3 plugs. and the 1.5 and 2.5 PiC's) measured x. Y, and "E" in separate gaps ca. CD plane chambers measured X and Yin the apart. The aboat 1 same gap, and the two redundant coordinates , "E" and "P", in a~other gap. about 5 cm. away. The in separation z nacessitated a large corroboration window. because the slope of the track cylindrical CD was ~ot known chambers, all at the small stage. In the L-R (left and right pitched wires) combinations were taken. ~f that This was possible becaase angle between the Land R directions. False combinations represented a point outside the limits unless the hits vere close of the chambers two physical Z together inf.) The raw data blocks from were converted into the lists chamber readou t controllers of physical coordinates, using previously determined alignment constants. Adjacent hits -57ware in Un the case of the CD planes combined into clusters. which the sansitive, cluster.) ~sually dita readout single The raw of every second wire wire gaps were data were occasionally less was permitted within garbled, a due to noise in the readout logic or a larger amount of than fitted in the fixed PDP-11 buffer space. Processing was terminated for such events. ~¼tch-points accepted. were Those reviewed after coordinates nev each used track vas in good quality tracks were "poisoned", that is, flagged and ignored in the process of subsequent trackfinding. In a region of 20 cm. radius in cnimbers 1 and 1.5, hits were sometimes coincident, so that · p~isoning vas not done in that region. In the CD planes, X-Y combinations were hit in corroborated tbe a either the E or P plane (or both) within 2 cm. of a p~int projected from the X-Y plane along middle by of the target. line from the These match-points were recorded at · x, Y, and z of the X-Y c~rroborated a chaaber hit. E-P pairs vere by an I or a Y coordinate, and recorded at the X and Y calculated from the E and P, and at the z of the E-P g~p. E-P match-points were not recorded when they coincided with an X-Y match-points. rbe proportional c~mplex. The chamber match-point corroboration distance criteria was were taken as aore the -58- of the discrepancies formed by projecting the three aini ■ u ■ c>~rdinates to ~arious a slopes z common along and directions. i~to a quality value (standard estimate of the chamber several lines, The distance vas converted error) dividing by resolutions. dropped. were by an ~atch-points which shired c~ordinates with more than 4 match-poi nts qu~lity having of better For good quality match-points, the number of shared coordinates was allowed to be higner, up to 7. _ Finally, all pairs of anased coordina t es were recorded ilso (as "tvo-vay",i.e. uncorroborated, match-points). rha match-points were stored in order of decreasing quality, with the tvo-vays last. This caused those most likely to be ra~l to be tried first for trac kfinding. ~a ny of the less likely points vere thereby poisoned before be in g tried. :ectain events had large numbers of hit s i ns id e small ia the l1rge areas chambers, leading to large number s of match-points, numbers execution of ti mes apparently good trial helices, up to 100 ti mes the avera ge. and An empirical t~st for such events was more than 50 mat c h-poi nts in two of the three CD chambers, ■ itch-points in Processing was or a product of th e number of the three plug chambers gr e ater than 1500. ter minated "lverflows " in section 5.G.3). for these e vents (called -594.0 -- rrackfinding Algorithm rrack candidates S)-called were "pivotal defined planes". by match-points If the in three intersections vith =nimbers were corroborated by nearby hits, the candidate vas fitted by minimizing the CHI-squared of the between the helix and the nearby chamber hits. agreement The program ll~ped over different sets of pivotal planes, starting those were that liKely tracks per unit time. bacause extra to ch!mbers other. minimized match -points dictated cylindrical produce the largest number of (This difference in the geometry that detector with were of all the computer time eliminated early.) The plane cylindrical pivotal surfaces) the and "planes" were of one (plane or type or the so-called "beam calls" consisted of only two pivotal planes, where the helix was constrained to intersect the all triplets of be!m track inside the target volume. For a given set of pivotal •~tch-points were examined. planes, Most because the change in~ divided by different were the rejected quickly change in z, vas between the first and second planes, than between the second and third. The change in~ was evaluated without cilc ulating the helix center by using ~ngle a between two points on the fact the circle mea sured from the center is twice the angle measured from the third the circu mference. that point on The ratio of df/dZ (between 1 and 2, and -60- 2 ind 3) vas required to be within a window around 1.0 whose viith increased empirically for points defining small radii. ?his increase allowed for the greater relative importance of resolution errors and scattering for small radii. c1se of the beam calls, the constancy of d /dZ 1~d used to beam track. calculate In the vas assnaed the point of intersection with the This point vas required to be not more than 10 ca. outside the target. For candidates passing refitted the initial cuts, the helix vas using the previously determined slop e to correctly project the pivotal c,ordinates to a common z position • . ?he corrected CD matc h-points were required to have a corroboration distance less than 5 mm. and the CHI-squared of the helix fit cut was above 25. If the candidate survived, the other chambers were examined for corroborating coordinates within the was chamber :oordinates a distance of 2.5 wire spacings (4.0 if not very corrobor~te the pivotal The poorly the trigger were track chambers nonetheless candidates. which resolved gainfully used to The corroborated coordinates (out of the possible the chambers). poisoned for match-point making were acceptable for corroboration. position between number number of where track intersected a chamber) was required to be greater than 651 when a large number of chambers were traversed, and ranging up coordina tes. to 1ooi for a small number of possible Corroborated tracks were passed to the fitting -61- subroutine. 4.E -- Fitting Procedure: Trac~ and Vertices This section describes the algorithm for halix fit to finding a set of chamber coordinates. a similar described here also. procedure in is therefore both cases was to of tracks minimize best The procedure for fitting a best vertex to and set the was quite The basic the value of CHI-squared by varying the parameters to be fitted. For track fitting, CHI-squared was the sum of the squares of the distances measured coordinates, resolution. expressed between the hypothetical normalized and the by the expected chamber the fin a l The uncertainty an "error matrix" giving the coefficients of in of helii the second degree terms in the Taylor series helix was expansion of the CHI-squared function centered at the best fit. In order to fit a vertex, the CHI-squared was the sum of the squares of the distances from the hypothetical vertex t~ each of the error matrices. point tracks, normalized according to the track When making a vertex for two tracks that were parallel at some point (as described in section 4.E.1), there were often two points of closest approach. When this occurred, the vertex was fitted at both points, and the with the lower CHI-squared was used. one -62- The minimization algorithm was Newton's t~ 5 dimensions (3 for vertex fitting). point gave a higher CHI-squared, h~lf as far away. corrections small.) a in method another If a new trial point was tried (This usuallv solved the problem of large cases when the gradient was anomalously After the first adjustment, coordinates which made contribution to the CHI-squared were dropped. The major procedure was iterated until the change in s~all, generalized or for a CHI-squared maximum number of steps~ was The confidence level of the fit was computed from the CHI-squared and the number of degrees of freedom. Pitted tracks rejected. 10-2 with confidence below levels 10-• were Tracks with less than 15 coordiaates were cut at because of clmbinations of the increa.sed coordinates. If probability a nev c,f track spurious had ~any c~ordinates in common with a previously accepted track# only the one with the higher confidence level was retained. A large number of candidates were produced and subjected v1rious tri a ls, ~any being rejected at each stage. to As elsewhere in nature, only the fitted survived~ ~.F -- Cha mber Alignment. rh2 approximate positi~n of the chambers was determined by -63The were optical surveying. positions fine-tuned ■ easuring straight (field off) tracks vith them. by The plane chambers were aligned in I and Y using beam tracks (?3 trigger) measured in the beam PWC's and the scintillation hodoscope. The Z position discrepancies of the plane chambers (plug, :o, and trigger) were checked by comparing the discrepancies in the X-Y position field-off tracks. vith the slope using wide angle (T4) The slope was determined assuming p~sitions of chambers 1.5 and 2.5. the Z The apparent position of the two sides of the CD chamber gaps coincided approximately at the middle of the physical gap. Discrepancies of several millimeters in the spacing between the all the proportional chambers x, Y, and E planes of were observed and have not been understood. Xhe cylindrical chambers were aligned package vas first made in self-consistent residual distributions for straight tracks. degrees of freedom to be fixed: Zand tvo steps. by centering the Thi s left radius of the chambers. is an error in the four (the longitudinal position and azimuthal rotation} of the package as a and possible shifts in Zand The whole, that varied linear ly with the The varying Z displacement appeared of the track, and the varying r~tation caused the track not to coincide wit h the beam -64- tr1ck at the center. A large set of straight tracks found im the cylindrical chambers transformed according to four parameters that mimicked these shifts a nd another four that ai ■ icked slope and These eight parameters were adjusted to minimize a p1ckage. displacement of axis the of the the CHI-squared of the agreement (summed over all tracks) of the tracks vith the beam track and the plane The four and alignment hits. _ Z parameters specified adjustments to the constants. displacement chamber of The effects axis the of were the slope and compensated at the the position trackfitting stage. In the cylindrical spark chambers, displaced about 4 mm. in the azimuthal solenoid was magnetized. The spark when direct ion direction of the the change corresponded to the electric field of th~?.? high voltage pulse r~ther than the reverse clearing fie ld* 1ppeared only at the anode end. were less than 1 mm .) Th e displacement (The cathode displacements The shifts were compensated in the alignment constants. ~.G -- Bubble Cha mber Comparison This section summarizes between interaction trigger tr1ckfinding program and the results {T 1) total even ts of a comparison(l) a nalyz ed by the cross section data from a bubble chamber experi1nent(2}, pi-minus proton interactions at -6516 GeV/c. The study checked trackfinding fanction of P and P efficiency as a. and the absolute normalization. The Tl trigger cross section correcting for empty were left. (Most of upstream interactions was 42.2 millibarns. After target interactions, 22.2 millibarns the empty where target no came fro ■ were found.) The events tracks bubble chamber cross section was 25.5 millibarns. Fig. 4.G.1 shows the multiplicity distributions. The bubble chamber events were corrected for the simulated stopping protons in the target. The T1 events were corrected for the empty target interactions. zero (The corrected cross section for track Tl events was negative 2 mb. Photon conversions were not rejected in the Tl triggers, and were iQ the bubble chamber data.) Although not counted the average multiplicity was 4.9 in both cases, the Tl events 1 greater number of contained events with an odd number of tracks, of indicating the existence of spurious tracks as vell as losses. rhe bubble chamber tracks were flagged as recons~ructible if they were compatible with the geometrical tha including trackfinding requirement assuming program, measured chamber requirements the reconstructible corroboration efficiencies. tracks (without empty target correction) are of compared T1 vith bubble chamber tracks in fig. 4.G.2 through ;:::,:) () \.) r,1- or- 5000 - >~ ==J-· X :X: :,;: :x: ..'", ✓= :{ I. r-1 •'II.: I • .. . . _: ·-.::_ '. --~1 .-- --· .-,~•.. •··-.;~ J ·; . -1. ;_,-1 ---(~ ·-- --, ,~ ~~ u J r.: PRONGS 15 (-., 00 H ~~ Fi[;'.\m~ 4.G.1: Cornp 21rison of multiplicity distl'ibutions: Tl (full rrdnus ernpt.y) and bubble chamber. 10 && ff- H Hist = BC data x T1 data -2500..._..,_~~-~~~-'--..__~_,___,___,__,____,c__..___.___.___._ ::.i_ ..D ('lj H ~ rf:! ?:500 10000 12500 H '> ►,_,1) ff I I en C) -~ -'~ <C ,S:1 0 2500 5000 e•tsoo r.r., 10000 12500 i5000 :x: :x: :x: :x: 0 I I _L_jL---L.-~- X :x: 1I 1 5 I u I .. I u PRONGS M I 15 .. I 1¥ , u 20 , .. , .. II Figure 4.G.2: Comparison of multiplicity distribu-Uons: T1 and "reconstructab]e" bubble charnber. 10 L_ ~r.:-1 -~ ~-ki :x: II Hist = BC data x T1 data II II 25 Ill I ----i I Cl :j_ ..0 ~ ~ ~ (f'J 0 1000 2000 3000 4000 0 :t.' GeV/c 4 6 Tl and "reconstructable" bubble chamber. Figure 4.G.3: Comparison of Pl distributions of tracks from 2 8 Hist = BC data x T1 data 10 5000 ~--,---,---,----.----.----.----.---,---,----,---,---,---,----..---,---,--,---,---,-;--;--...-...--.---.--...---. I I 00 0) 2000 j_ (I) 0 0.0 1000 0.5 1.0 ,c_ 1.5 2.0 Tl and "reconstructable" bubble chamber. Figure 4.G.4: Comparison of Pt distributions of tracks from GeV/c l.O I en ,.Q H I "' ,,, Hist = BC data X T1 data ~ rn 3000 4000 5000 ----~--~---.---------.--,-~-,------------------. -70- The momenta ■ distributions were in the multiplicity distributions g ood vere agreement, still quite different. In order to understand the Tl multiplicity distribution, the bubble chamber multiplicity distribution vas convoluted with a Poisson distribution simulating additional trackfinding loss each track, and the extra tracks. An was applie C i ndependently to normalization factor was allowed to float. The CHI-squared of t he difference between the (empty target corrected) Tl and the adjusted one and chamber distributions (for minimized a function of the track loss p robability, the as multiplicities bn bble above) vas extra track rate, and the normalization fac to r. fit track rate o f 0.42±0.2 per 97±5% per occurred event, a for track a spurious efficiency of (to 1.06±0.15 The t rack, increase best and t he n~rmali z ation of distribution). The multiplicity distrib u ti ons are shown a B.C. in fig. 4.G.5. ?he co mparison indicates ch!mber tracks that corresponded the trac kfin di n g progr~ m. installed in the Monte trackfin ding reconst ructible rather well to t he results of program for that electron tr!ckfin di ng efficiency determined in the believed to be correct bu bble The reconstructibility tests were Carlo efficie n cy the within 5%, estimated the pairs. The Monte Carlo vas accor ding to these £.1 ._, --:i.. crS :...., ~ rn 0 2000 .: 000 .t, 0 6000 • - L:- ·- 0 - 4 -~( MULTIPLICITY 6 8 10 smeared BC data reconstructable BC data = Tl data Figure 4 .G.5: Comparis on of mulUplicity distributions: T1 (full minus empty), and bnbb1e ch : :i.mber b efore ctnd after smearing. 2 ',/ 7<,;:--·• □ ◊ x 12 8000 r - - - - y - - - - r - ~ - - , - - - , - - - , - - - , . - - - r - - - , - - - , - - - - - I ....... I -...J -72re sults. ~. B -- Ea r ly Event Bejection A pr e-tr~ckfinding cu t "mask" on the pos sibility of i t rack finding satisf ying the electr on identifica tion criteria is des c ribed below. Other c uts which were filade in the of are trackfinding described also . process ihese cuts and the program flo ~ are sho wn in the left half of fig$ 5.!~1 (i n triggers for t he next chap t er). The "mask" selected 201 of the 5 The r e gistering C1 rejected cell and million TO events contained at least one one gamma-veto shover c ombination as re quired in the hardware trig g~ r. ~~nfig ur1ti on tc~ck . c ounter hut in a that could not have bee n produced by a single Th ese, and many that passed the mask, we~e due to separate spurious and uncorrelated counts in Cl and a sh ow er counter. in a The presence ~fa t rack regist ering both in Cl and showe~ counter (a "tr igg ering tractn1 was re qui re d again more l sa mple of tr ack s with momectum gre ater than 750 MoV /c (th~ l!terg both in t he trackfinding progra m. and rigorously in the data selection progra m. lowest acceptable elect~on momentnm) vas used to produce t he ~mask". The mask ~as an ar r ay with entries for ea ch pair of one of the 38 cells of Cl and one of the 13 shover cou nters. -73Entries were flagged "possible" if any track in hit of sa~ple Events were rejected if none both members of the pair. of the c ombinations the registerin g ci c el ls and shower chambers were c~unter s were possible. After all combinati~ns of s earched for tracks, the even t tracks satisfied the pivotal plane was rejected if none triggering track of the r.eguiremen t. The c~ mputer time for trackfinding in the cyli nders ~as ~void ed. Other cuts theceby made during the trackfinding process w~re the fatal unpa c k and mat ch-point overflo~ (~hich haw~ alrea dy been described ) ~ and the requ i rement of at least one trac k in the beam c ham bers . T~o cu t s made i mmediate l y after the trackfin1ing red uc ed the siz e of the DST's by a f a c tor o f two events 9 6250 BPI tapes): on ~ (to 120,000 second tr~ck id2ntified as an electron, in C1 but n~t necessacily in a showe r ~as required. in order Pina llye program counte r~ Eve nt s with multiple beam tracks wer e dr~ppe d to simplify a lat er cut involving the beam t r ~c k~ tracks ~ith momentum belov 750 MeV/c were extr apolated through the n~n-uniform field region as far as the end of C1. (This was do~e earlier for the high momentum tricks while searching fo r a triggering track.} -74References for chaptec 4 1. Comparison of E127 Total Cress Section Data with 16 GeV/c Pi- Minus Pr~ton Bubble Chamber Data, Co L. lioody. SLAC Gro up B Interna l Merao~ Jctober 5, 1979. 2. The bub ble chamber experiment was reported in J. Bartke et. al., Nuclear Physics, B107, p.93 (1976). -75- 5 :BAPTER Data Cuts, Processing: Background, and N,rmalization 5.A -- Introduction and Summary rhis chapter describes the cuts used sample of several hundred taat were reconstructed. w~ich the It to select the final events from the 120,000 events describes the procedure by background was esti mat ed and subtracted, and the pr~cedure by which the losses of signal were estimated and tbe final result was normalized. "L,w mass pair" will refer either to electron pairs produced by photon conversions and pi-zero with a mass iiantical. decays a or pairs 140 ~eV/cZ, these classes being almost "High mass pair" will al way s refer with pairs below Dalitz to electron mass above 140 MeV/cZ, wh ich ar e known to be almost free from pi-zero Dalitzes and photon conversions. T~e material is presented in sections 5.B through S.E in the s1me order as selection the program. corresponding cuts were made in the Section S.B is concerne d wi th electron iientification and the cuts requiring two electron tracks in s.c the event. Section qu1lity the electron tracks. ■ 1ss of describes tvo cuts made on the Section 5.D discusses low pairs which were erroneously reconstructed with large miss, and their removal by a cut on the association dist1nce -76- of the electron tracks with the beam track. dascribes the geometrical properties of low cut to remove events where the Section 5.E mass pairs~ a electrons originated as separate lov mass pairs, and two c uts on the position of the p~ir vertex. The flJw of the progra m and all the cuts described above are shown in the right half of fig . 5.A.1. rhe various 1irect background electron pair processes vhich contaminated signal are discussed in section 5.F~ Most of these processes produced the same where both opposi te electrons charges~ bad The the same same number of one estimate Misidentified of hadrons the charge pairs contamination pairs were pairs charge as those with vere treated iien t ically to the opposite, so that the former were for the not in a one the latter. sy mmetric in t his sanse~ and ~er e treated separ~tely. Se=tion the SoG describes t he esti mation of t he normalization of the final result. acceptance The geometrical a==ept an ce {for elect=on identification in the Cerenkov shJ wer couaters) applied to tha trackfinding iientification and was estimated using track pairs generated by a Monte Carlo program. vas and The usual tr ackfindi ng progr am simulated chamber measure~er.ts to estimate efficiency. procedure Losses in the , . e~ectro n and the beam associ ation di st a nce =at were ceproduced in the ~on t e Carlo using ~alues me asur e d with re~l tr a cks. Losses are tabulat ed and used to EXAC'l'l,Y 1 I! OR MORE ELECTRONS IN Cl I I 11,472 415,681 242,440 ~ I 11 a,100 (Three/More Electron DST) I C> ..,~,u "t:1Dr\OT'1""1.1 ~ > Comblnation < < llAKE ELECTRON PAIR VERTh'X < < < I Charge I Target Wall Beam near !.lbs~ belo-ir I HO l,fov/o" RE1'RAC!EO OST ELECTRON IDENTIFICATION IN Cl AND SHOWER CTR3 Figure 5.A. 1: Program Flow and Event Selection (Rot.racked DST) I COl!PU'f'E I.OIi' IJOi-f. TffKS THROUGH 1-W IXT r!El.D ~ >...___,r-,___.,,. 10,033 Z Outside 122,133 Selected DST) lKV~ To.rgol Cannot Mako Vert.ox 120,307 No Low Ma•~ .Ila~ Cut Beam A.esoc . Distdnca Flag Matrix Level Conlldonce Identi!led In CI 1.7:il More Than 2 Identified In Cl Leoo Than 2 1147,110 < I No Trlgggfflng Traci< 27.01' 962,666 4,18?,7ll6 I fl ,160,4 10 SELECTION PROGRAM r-- 1'.60-fx10• 46,'162 I I ~ To No Triggering Tro.dc Tra0k No Beam Match Point Ovorflow TRACKFINDJNG PROGRAJ,I Fato.l Unpack TO Trlgg61'3 B<>o.m Pi o~ 179 117 7,015 1,646 13? 7 ,146 12,691 " 626 1,300 41 ,774 38,875 11 8,100 I ....... ....... -78noraalize the final result. The final differential cross section as a function of mass (valid for ■ asses 140 MeV/c 2 ) is presented in fig. S.G.6. The total nu ■ ber of greater than events with opposite charge pairs minus the number charge pairs hadron sa ■ e (background esti mate), vith mass greater than 140 MeV/c 2 vas 62±17. the of After subtracting punch-through t he component of which vas not c harge-co mbination symmetric and correcting for the loss of real pairs due to the removal of low mass pairs, the result was 57t29 events • . 5.B -- Electron Identification rhis section describes the electron identif i cation procedure using the Cl Cerenkov counter and the sho ve r counters. electron identification misidentifying hadrons efficiency is The and t he probability of estimated. Th e rejection of events with more than two electrons is discu ssed also. All events presented to the selection pro gram, i.e. the "candidate events" passing the Cl-shover co unter correlation m~sk, track". had been selected to have a so-ca lled "triggering A triggering track must have traver sed at least one C1 cell which registered light, and hit wbich (and wa s counted in the had a corresponding large pulse height "gamma-veto" counter). a shower The counter triggering electron identification requirement in the p rogram was stringent, a s described below. ■ ore The program also required a -79second track to be identified as an electron in C1. s.e.1 -- Shower counters If a track traversed ~ny of the dipole-face shover counters. with tne pulse height of that shover counter (or of the one greater path length, if two were traversed) vas used to datermine if it were an electron. The pulse height sum from have 111 tubes on the shower counter was required to fired tne "buffer strobe" discriminatoc, corresponding to a sum of pedestal-subtracted 600. ADC counts greater than approximately A count was required in the corresponding gamma-veto counter in order to reject tracks that could not produce the hardware trigger. The identifying counter was required not to have been hit by any other track with momentum above M9V/c. (To test for lover momentum 750 tracks hitting the counter would have required a large extra computer effort.) T~e pulse height cut ~f 600 accepted 50% of electron with an energy of 750 MeV, determined before the run by placing the counters in a test beam. verified This measurement was later using "test tracks" from recorded TO events. A=ceptance versus pulse height cut for various energy tr1ck electrons and pions is shown in fig. 5.B.1. trick tracks electrons are described test (The test immediately below, and the pions are described in section 5.B.4.) u 0.0 0.1 0.2 0.3 <'-1~ ~ ~ ►-< 0.4 0.5 0.6 0.7 0 E-... z <.t; u u ~:r.::i o. . Cl Lr.J r:-, 0.8 0.9 1.0 @) 0 , l ·\ \ I { r\ \ \ T ~ I I I• I 4 , • l \ I' \ \ 2000 PULSE HEJGHT CUT I . I .'. I " ," ".:?- hat b 3000 i 4000 I Lines inchcate grouping o:f points only. @) n ."15<p<10 GeV/c © e 3<p<4 GeV/c (De 1<p<1.5 GeV/c @ e 1.5<p<2 GeV/c @ e 2<p<3 GeV/c Figure 5.B.1 a: Accepted fraction as a function of pulse height cut in GpE counters. 1000 ' rs- ~ i- " , " 11 \ ·kJ~.L T I 1i \ \I ~ l Zr \ 11 ~T\ 1\ CD @\ rsi ii i ~11 ~N • \T"- :~ \ i \\ _;~°"l \ 5000 I I 00 0 ~ ~ ~ <r, [-< 0 >--1 •-~ ,._:::.~1 ~~ o. . f:.:.l u u E--' w 0 0 0.0 0.1 T\ I I 33! -,, 2000 3000 PULSE HEIGHT CUT ---J: - I --;r,==I f:;:;J-=-.. ._... -'!'~--· h_ "1,-.l\ 4000 Lines indicate grouping of points only. ® n ."l5<p<10 GeV/c @ e 1.5<p<2 GeV/c G) e 2<p<3 GeV/c @) e 3<p<4 GeV/c ® e 1<p<l.5 GeV/c (De .75<p<1 GeV/c Fieurc f1.B.1 b: Accepted fraction as a :function of pulse height cut in CIT counters. 1000 I ' "·f--• .=::t:±J--.. --·-r ,~ -i... \l \ l t- \I t ~tl i \ l\ l1 r\ r\ ti \ \ 'l \ . f1l \ (l 1 \ 11 1 .{ 0.2 0 :f { 1 J, I \l \ ® (±)\f( ! \ t ll ~ ® I =K \ \ \ \ TI'tl'i • \L ~ ~" ~ "i 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 5000 I I ~ 00 -82Tast track electrons vere to determine the counter identification efficiency. which electron used shover necessary to correct for the loss of electron pairs iiantification f,rmed low electrons of with MeV/c 2 ) pairs identified opposite charge. The test tracks were free with selected iientified mate satisfying the hardware trigger. tracks the (m<SO from trigger bias since the events were of in These were tracks in TO events which step. mass vas the The number making low mass pairs with electrons of the same c~1rge was one per hundred opposite charge pairs. This is t1ken as an estimate of the probability of chance pairing of 1 hadron h1iron with the identified electron and therefore of the contamination in the test tracks. The further requirement that the test track be identified by C1 reduced the contamination effectively to zero. Ine acceptance was found with a Monte Carlo program assuming ! constant average identification probability for all tracks !ssociated with a shower counter. i~=luded spaces between the The area part of the shower escaped. 1lso included a 4 boundaries to misalignment. cm. allow band for The Monte association shover counter enclosure and ictive element. and low efficiency regions where of Carlo the edges The area of association beyond track near the physical counter uncertainties and counter identification probability was taken as the identification efficiency averaged over the whole area of acceptance. The average was obtained by -83t~king all test tracks associated with a shower counter and computing the fraction which registered in the buffer-strobe discrimi nator and the gamma-veto efficie ncy of the counters vas only test tracks 2 cm. counter. measured The separately. true using inside the area of the counters. Pig. 5.B.1 shows the accepted fraction as a function of pulse height cut where the test tracks were 2 cm. inside the counters. Fig. 5~B.2 sho~s efficiency for buffer - strobe and gimma- veto for tracks 2 c m. inside the counters. 5.8.3 shows the efficiency fo~ buff er-strobe and Fig 9 gamma-veto for all associ ated tracks. rhe validity of averaging over a ~rea fairly la~ge depends on the similarity of the angul ~r distributions of the test tracks and the electron tracks had the aLgalar candida t es~ parent pnotons ver e iatecactio n , virtual photon. both it test d~s tributed li ke Since the pi-z ~ cos and vi rtua l prod uced is The distribution of t : e ~i-zero3 that produced the m and the real electrons were the insensitive in the re~ sonable to hadronic assu ~ g the angular of electron 1istributions ~ere not ~adic a lly different . rne identific ation efficiency ~~mentum is as a fu nctio~ shown in fig. 5.B.4 in the form it was used in the ~onte Carlo. The "O"s in the figure represent t he esti mated identificatio~ probability at each zom entum value. T~a "X"s represent the one stand ar d deviation li mits~ ~bi c h ~ ~ <r: 0 f--1 0 E-< z 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 TI □ Ph>Thrsh and GV counter fire s Pulse height above threshold 1 MOMENTUM GeV/c 3 ; 4 + Figure 5.B.2a: Fraction producing hardware trigger inside active area of GpE counters. 2 + +++++ + TI e Ph>Thrsh and GV counter fires o X e Pulse height above threshold ~ 5 6 I -&:I 00 ~ µ..i <~ 0 0 >--< I:-< z 0.0 0.1 0.2 0.3 0.4 0.5 0 1 MOMENTUM GeV/c 3 ' 4 f Figure 5.B.2b: Fraction producing hardware trigger inside active area of CIT c ounters . 2 -M++t+it++ □ 5 Ph>Thrsh and GV counter fires 0.6 1r e Ph>Thrsh and GV counter fires X 7T Pulse height above threshold o :x: e Pulse height a bove threshold 0.7 0.8 0.9 1.0 6 I I 00 \J1 ~ ~ ~ 0 0 1--f E-• z 0 1--f .... >---< '-·· ~ 0 l-7 E-< ~ 1--f 0 CJ w ~ 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 1 J. 3 4 ~ MOMENTUM GeV/c J_ Figure 5 .B .3a: Fiducial area efficiency in GpE counters. 2 (f 5 6 -=i I I en 00 ~ ....-:r: µ.4 (Y. 0.0 0.2 0 - - (.) t- ~"\ 0 r--: -- 0,✓ ,1 - .. ':!: - - 0.6 - 0.8 - r :;.:, C) >-I > 1---< CJ Z, I:--< (Y. '-' C) GJ 0 1.0 1.2 1 ¥ 2 3 MOMENTUM GeV /c 4 Figure 5.B.3b: Fiducial area efficiency in CIT c ounters . 1 1➔., -1 l +t t(W Jt1 T t r'I~ 5 6 7 I I "-I 00 [::., ~:1.l fa.• t•- 1 u ,. ....., zC."Li u ~ 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 0 I X 1 / 0 X X I 0 X 0 ,/o X/ X 0 D x/ 0 x_,,.-;; 0 / 0 X ? fv X ;-a X X X O X 0 0 X X o_ X X .::.- X )( X X X X 0 ..2-0-0 >C X X X i,-0 3 MOMr~NTUM GeV / C O O X o-x X X X 0 X X 0 X X D 4 X D X 0 X D X 0 X D X 0 X 0 xxxxxx><}< 5 Figure 5.BA,a: Electron identification probability in GpE counters, one std . lovrer and upper limits, and polynomial fit. X x/ 0 X X X X X xxoo)(xx 6 I I 00 00 w r.:r.... G:.. w ....... u ....... :Z, u ::.-,... t I l I 0.0 O 0.2 0.4 0.6 0.8 1.0 1.2 X /o I 1 X D X x/0 X X 0 X ,,S )( D X X X D ✓ X X 0 )V D X 2 X X ,,a' a 0 D - 0 X o X - 0 X 0 X X 3 .,..,.-x-,;xxx "x D X 0 . -X D " X X X _,, 0-0 X MOMENTUM GeV /c -x X X DD X )( )( O X O X )( D )( D )( X 0 X X 0 )( 0 X X 0 X X 0 X X )( X 4 ;..--~--- O X )( 5 Figure 5.B.4b: Elect ron identification probability in CIT counters, one sld. lower and upper limits, and polynomial fit. X D I X X X / X X x" o X )( X 0 X 6 I I 00 lO -90vere used to estimate the nncertainty of the acceptance factor calculated by the Monte Carlo. final The dashed line is a smooth 8 parameter fit to the measurement (of the form: f(p)=C•p + 1/Q( 1/p ), vhere Q(1./p) was c a a sixth-degree polynomial). ~nly to determine binning ranges as would was have constant and This fit vas used described below. (It been more str aightforward to use the fit as the estimate rather than the Os. The final result would have been the same.) The o•s and x•s were calculated at a large number of momentum values using test tracks within a variable range of mom enta. These ranges generally efficiencies measured at nearby statistically on depending overlapped, momentum values due to was the statistical The bin were not The size of each bin was chosen the change rate of fairly constant. range in uncertainty size that independent. of the parameterized efficiency in order to sum over a momentum range efficiency so momentum where the The estimated uncertainty was combined with the to produce an overall uncertainty. was chosen to uncertainty. The o•s are the fraction identified in the resulting bin and the X's correspond overall uncertainty. 5.B.2 -- C1 Cerenkov Counter minimize to this the overall minimized -91The basic electron identification criterion in C1 the track produce light in not counted. "s hared".) (Such that one of the cells it cross ed. :ells which were simulta n eously wece was crossed cells will by another tract be referred to as Every track produced by the trac k finding program w~s used to invalidate shared cells (including tracks which c omprised about 20 % of the total.} Ide n tificati on efficiency was th1t made measured using test tracks lo w mass pairs with "triggering" electron tracks of the opposite charge (i.eof identified in Cl and a cou nter). As one perc ent. c ounter menti~ned above, the contami nat ion was about (This vas verified by making a cut pulse height middle e xcluded. or in inner The cells were ring, g rou ped into the unequipped cells bei ng The t wo central "beam" cells we re not measured~ Above pedestal, the pulse c ?ntaining tracks electron height distribution for m~y be because no_se cells and cells ~ithout any apparent tr~cks overlapped to a large de~eev as shown in fig. (rhis shower for the subsample of the test tracks hitting the sho wer counters.) outer, sho~er and electron c~r responded to single photo-electrons.) 5.B.5~ tracks both Th e rate for pulse height abo ve zero was 601 for electrons and 0$5% fer empty cells. (The efficiency of c ells with electrons appears bacause the sample low vas do minated by lov momentum tracts.) rhe empty cell rate ~as measured using interaction trigger z en ~ ::> r£l ~ rI I 0 50 100 150 200 - 0 250 H-- I 300 -<J vnrn ELECTRON 400 (66280 cases total) x EMPTY CELL ( 1180 cases iotal) D CELL Figure 5.B.5: C1 pul se height for cells traversed by an electron an d for cells traversed by no reconstructed track. PULSE HEIGHT ABOVE PEDESTAL 200 =490 X =66000 □ 600 J I I N lO -93events and cells not trav e rsed by any reconstructed track. ! pulse height cut just above the pedestal value for each C1 call gave the best electron-hadron separation. Cells 15 and 33 were deadened in their spurious rate count the program because of high (counting when no track hit the cell) and c ell 28 because it was inoperative (due to a detached mirror). The detecti~n effi c iency depended on the position and of the tra ck within the cell. angle Once an electron becomes extre me l y relativistic at about 100 MeV/c, light angle are approximately radia t ed c, nstant. the and the Cerenkov The momentum had th erefore no efficie n cy .. cell However* the the amount of direct effect on light collection eff iciency of the cell was low if the direction of the t r ack w~s not parallel to the cell axis. with the axis The cells were pointing toward the target, and so had good efficiency for high momentum tracks which followed a stc aight angles the cell {1', effici encies. the axes, angle* and therefore had from given by GeV /c) /1)0) loijer a track made with the cell axis may be approxi mated by the change in the angle track fairly Low momentum and large 0 tracks made larger path. with built of the where it left the target to where it enters C1* COS(1.J,')= The 1 SIN2(9)eCOS(~lr}, constants 2 GeV/c, is wh ere Af = (2 related to the magnetic field strength and the distance between the target -943.nd C 1. For small "Y, The efficienc y was momentum and outer ring. these "4J was appro rim a tely TAN (8} • (2 GeV /c) /P.,} therefore whether measured as a function the cell was in the inner, middle or , The ring selected a roug h range o f variables of deter mined '"fo) so that Measuring efficien cy as a function of r in g wa s also i mpor tant because the cells of different rings differed in length. Trac ks someti mes ~ent through several cells o in any Pulse height of the cells was con sidered sufficient for electron identifica tion (providing identificatio n ·:rh e tr;.ck) • s:> metimes hig her cells. The the t ha n individual the cell contained probability efficiency cell of efficiency no other therefore vas the individual was lo ver when a track went throug h several cells be c ause the light f r o fil soch tcac ks wa s directed a t the inter-cell partitions. measure men t s were ma de fo r tracks hitting only one c ell and f o e t ho se hit ting mo re than one . deter mi ne Separate The momentum bins used to t he efficien ci e s ver e co mbined in varyi ng numb e rs as described f or the s hover c ounterso Fig~ 5 ~B~6 sho~s the oue upper t he sta nd ard effic iencies deviation and lower and li mit s of the parameterization {of the same form as was used for the shcver counters). rhe identification efficiency for a track (computed 5 ante Ca rlo) vas ca l cul ated cell(s) it tr avers ed and a from the cons tant in the efficiency of t he r epresenting the -~ r.::=1 f:c l h.; 1-' "' ·1 C. . ) ►- GJ Z, u >-4 Cl.. i:."Lj CY: u t::,J ....:i ....:i [.I:l C.1 C::.j u ........ ....... z w u :>-< 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 x ><ox 0 X ,x j;<ooxx D:ic MOMENTUM GeV/c 3 4 5 TRACK CROSSED CELL BOUNDARY(S) 1$}__.x ootr.) o oa-o- o ouo-o o O"'O'O o 0--:,1,0 o -ucroo cnn, 0015'1)' o oOOo o 6?>0 o t:5'trn"o o'trm:J o 01ftio 07'"-0 oo tfl5b od1fb o cffi'tf11o oa-o-o o OOU" o O'OO"'D o o-oo x o 0 xxx xxx><:.<><xxxx xxx.xxxxxx>< xxxxxxxxxx xxxxxxxxxx xxxx>< xxxxx:xxxxx xxxxxxxxxx x><xxx xx><><>< xxxxxxxxxx xx><>< XO o, fa X 2 X)( X Xx>: >< X XX XX X XX XX X ;..: x XX X XX XX X XX XX X XX XX~ XX XX X )000( X XX XX X XX XX X XX XX X XX XX X >< X XX A Y. X XX X )( X XX X XX XX X XX XX 1 TRACK ENTIRELY IN SINGLE CELL Figure 5.B.Ca: Elecl:ron idenLificaUon probability in the inner ring of C1, o ne std . lower and upper limits, and polynomfrtl fit. 0 / X X OX I o, )/ox X Io• Jo X )(Xx><x XX><><x xxxxx xxxxx xxxxx xxxxx xxxxx X)(XXX xxxxx xxxxx xxxxx xxxxx XXXXXXX)O(X xxxxx )()()()()( xxxxx xxxxx xxxx ~e~~~~~~?:~ ~~~~ ~t~~~ ~~~~-~e~f~~~~~~ ~~~ ~-~ ~ ~~~~~~~~~~~~~~~~~~~~ ~~ 6 I v, LO G:l L:,:.: r·:,_, :-.-~-, (.) 1~-1 GJ Z; u ;r, C:L.. r.:i l p::~ v ril ,_J ...J [:r_.; [I., l.:,:·1 u_, w ........ >-a u z, 0.0 ().~2 0 .4 O.oD J \...,• Op 1.0 1.2 0.0 O. qr.:.J 0 )()0()( :oc xx:-.: xx XX~ X )00()( )().:>(X X )000( X X > - > O ( X )00(X X 4 )()( )O(X ~>l)O(~ XX XXX )(XX)()( 5 xx >c;.(),(: )(ki-.,t TR.ACK CROSSED CELL BOUNDARY(S) r, S:. ~ -. _. -;--·•, tH.t~oO!."-t'J·~o i, tJ-n OO.•S;00 t.t.t.LD,00 K?.P.Jlort_Mr,051.!J.oO~tJot!.2.._!? on_t;!!.0 opooCJ CLJ2..J70QJLJ)onr.u,ot.1,l:).Q..OO.ot)."4) ~&\+o .. ;,;,._,, µ ?-( :.(.~ ;.: :,,; ,<,,~;~ x X ;< ,-,;..1,X :<1~ )(X X )t ;.,_ )(,t X ~ XXXX XX >O( X >< XXX XX >O<X X )( XX}O( X XXX XX X XX X XX XXX XX XXX XX XXX XXX)( X MOMENTUM GeV /c 3 TRACK ENTIRELY IN SINGLE CELL ~~** R~~~-\riiH@ iffr~ ~rimt1617® ~ R19~ITTrn~ffli~!lITTt~rntrmm-:r~~~~lR~tirH~} Figure 5.B.6b: El e i-:t.ron identification probability in the middle ring of C1, one std. lower and upper limits, and polynom:ial frl. X X X X 0 XO O-< 'D 0 X-'S)< '.,<; 2 ~X K i,; ){ , >-)( x X t i U o C,t.1 x x :,,' ;..'.~ X~x -~-/ ' ; " ...: ,<a~·:::>' 0 x 1 X~Koxx>< XX , ::. ~ , Q*Q ~~i)iR !'!m* lo,,~;~!/Q~ ,~( 11 ~~!i X' I J 0.6 0.4 I 0.8 l.O 1.2 6 I m I LO µ:J ., [i.., [:. ...... u w ......... z u :,.... 0... w ~ 0.0 0.2 0.4 0.6 0.8 u w 1.0 1.2 0.0 0.2 0.4 0.6 0.8 ....:i ....:i w G:... [:r.., :,.... u z w ...... u ._. 1.0 1.2 0 0 o x " X " X XX 2 4 5 TRACK CROSSED CELL BOUNDARY(S) 3 TRACK ENTIRELY JN SINGLE CELL MOMENTUM GeV/c X XX XX X x x>< xxx xxxxx ><X>o<xxxxxx ><xxxxxxxxx xxxxx xxxxxxxx:<x xxxxxxxxxx xxxxxxxxxx xxxxxxxxxx xxxx x>< x)t ><xx>< O O"O'Ux><~x~ o o.o..QOOC!..Q_ooQJl..D o ~ o 0.0.000 Q.Q.DOO..Q . .flOOa..Jl..DOci.Jl..DOC0..0.OO..00..0 00..000 Q.0..QOO..O.OC0-0-0O0-C-0-OO.CO..O Qg.OQ X XX)(oODooOx x><XXX XX><lx>< XX XXX XXX XX XXXXX><Xxf<x XXXXX XXXXXXX XXX XX4,(XXXXXX XXXXXXXXXX X><><fx><XXXXX XXXX xx xx>< 1 O /4 • xxxx o/ 0 / O x :---Bo oo.c..n 00 51r00 e1J.E oo:~ o.o..Qoo 0-Q.000,.0..0OO.ruz.ooa.o...ooon..o.0040-0 00.000 0-0-0og..o.oo~ooe-o-O0'0tTD om:n, o ~ 00 xx><>< xxxxx xx><xxxxxxx xxxx>< xxxxxxxxxx xxxxxxxxxx ><xxxxxxxxx xxxxx xxxxx ><><xx x x Xx xx x>/<XX x><xxx xx><xX )(XXXX xx><xx x xx><>< xxx><x xxxxx xxxxx xxxxx xxxxx X><XXX xxxxx xxxxx xxxxx xxxx x x Figure 5.B.6c: Electron identification probability in th e outer ring of C1, one std. lower and upper limits, and polynomial fit. / ,x xx / o{ X x 6 I I lO ......, -9 8 fcactio n of the events in ~hich that cell was shared. fraction ~as measure d for interaction two or more tracks~ t he electron electron tciggec (TO) pair and abnormally events ra ther events, because the latter of t en co ntained electromagnetic showers, conversi ons, vith events The interaction trigger events ~ere chosen as a model for direct than trigger This do~nstream photon large multiplicities. I t has been ass umed that the interaction trigg er events repr€sented th e direc t pair events in terms of angular distribution multi pli city. and The noshare pcobabilities {per cell) ~ere as outer ring, Oe905; middle ring, Oe864: inner ring 6 follows: 0.819. The efficiency meas ured for each cell on the fi q. 5. 3.6 or zero f or unequipped or track dead (as in cells) was multi pli ed by the probability t~t the cell was not shared. If the tr ack hit sever~l cells, th esa products ~ere co ~bined t3 get t he probability that at least o n e cell identified the as an The electron. efficienc y calculated using the nosha re probabilities {for Cerenkov measured identification efficiencies and electro ns in the range 0.75 to Se ~ GeV/c from high ma ss pairs generated by the Monte Carlo} aver aged 57 % per tracko The lowe r ~omentum li mi t for electrons vas .c hosen to be MeV/c because the efficie ncy drJpped sharply there. in Ci and 750 shower counter Pions with momenta abov e 5.8 GeW/c -99produced pion Cerenkov light. Pig. 5.B.7 shows the number of tracks making light in test probability radiated The detection very lov below 7 GeV/c since less light is was near C1. the threshold. However, the momentu m aeasurement errors were large (about 1 GeV/c) above 6 GeV / c, s~ there were pions registering in Cl as lov as 5.8. upper momentum limit for electron identification vas at GeV/c, allowing a safety margin. 5.0 the pion test tracks is section 5.B.4 described pertaining to The chos en (The selection of below in the part of hadron punch-through in C1~ l The group in figo 5.B.7 at the higher momenta corresponds t o pions which produced Cerenkov light. momenta corresponds to a much The group at the lo wer smaller fraction of pio ns which were misidentified for other reasons, probably beca use a photon converted punch-through in tracks the same cell. The number of below threshold is however no gr ea te r than the estimated electron contamination as described in section 5.B.4. 5.B.3 -- Electron Detection Efficiency In the Monte efficiency Carlo, the shower read from a 5.B.4 was corresponding to pc~bability that ~eV/c. fig. counter identificat i on table of efficiencie s and multiplied the by t he it was not hit by another track above 75 0 As with Cl, the nosbare probability v as taken as t he frequency that the counter was not hit by a track in the z ~ r.:.:..7 (.D lo:'... ,.;-2:. ::.) 0::: o 1 0 2 ,_ v 0 4 1::.., ,, Ul_j.J..LJ_LJ...L.Ll.J_[ , 4 0 n r r r=t::·· GeV/c , t - : l = b I d 8 &I • , ,,, 10 , Figur.,) r, .B.'i': Pi on t e~,t tracks in C1 celJ.s: numbe r registering light, and loto.l number traversing, as a function of momentum. I . . L . L I J _ j_ _ _ PIONS FIRING C1 - - - 1000s OF PIONS PER GeV /c THROUGH C1 ,, , 1 , , , 12 , 5 --~--~~-.----.-----,...-..---.--,--......--.--....-...--.---.---,----,..--..--,--....--...--.--.---,---.---, I I 0 1--' 0 -101interaction trigger data. The noshare probabilities are hit shower given in table TS.B.1 below. inen both electron tracks counters, the identification efficiency was calculated to reflect the fact th~t the event identification would failed have for been both tracks. the efficiency was the probabilit y regardless shover rejected of only if For these events, identifying either, of whet her one or both actually registered. counter identification probability described (for Mon te Carlo high mass pair calculated events) 751 per event. Table T5.B.1: PROBABILITY SHOWER COU NTER WA S NOT SHARED WI TH AN OTHE B TRACK Group E SC 1 SC2 SC3 SC4 SC5 .889 SC6 SC7 • 917 .905 • 928 .973 sea • 974 SC9 .975 SC10 .983 .982 Cal Tech SC11 .9 0 6 SC12 .851 SC13 .740 .977 the The as averaged -102- rne total electron identification efficiency vas the product Cerenkov for efficiency electron for an event of the shover counter probability and the each of the two electron tracks. This was evaluated in the Monte Carlo to estimate the identification correction factor. The total electron identification efficiency for Monte Carlo generated direct electron pairs averaged 22%. (This number includes the effects of cutting out shared cells.) 5.B.4 -- Hadron Punch-Through rhe probability of calling a hadron an electron to was needed estimate the number of false electron pairs reaching the final sample. It vas measured directly in the shover counters and inferred in Cl. Interaction trigger events were used as a source of hadrons. A sample of test tracks was chosen requiring through equipped within the electrons). to giTe active cells Cl area (to and hadrons for they go hit a shower counter vell maximize The sample was further cut as test that the rejection described of below the shower counters or for Cl. Tracks were not used if they made a low mass pair with any track was tract of opposite charge. To measure shower counter response, each test required to give no light while staying inside a single cell -103- of C1, vell starting away from a partition. used only if no pair of opposite charge tracks in C1. had a mass SS of the tracks in the se events gave less than 150 MeV/c 2 • light Events were Since C1 was about 90i efficient under these conditions, cutting out the tracks that made light left a maximum of o.si electron contamination. The contribut ion to the shower counter punch-through rate from a photon hitting the counter simultaneously was assumed tj be the same as c~unter without a depending the frequency track. This of a phot on hitting a varied from 0.2% to on the position of the counter and its area. results are shown in fig. 5.B.8. probability wa s 13.3% The average 2% The punch-through (for buffer-strobe and gamma-veto). rae dependence on the pulse height cut is shown earlier in subsample of fig. S.B.1. The punch-through in C1 was measured using test tracks smaller than subsample wa s with 150. a pulse The calculated height electron by in a the shower counter contamination comparing the in the pulse height distribution in the shower counters (before the pulse height cut) vith the distribution for known hadrons. ~f pulse heights above 1000 was tracks being pure c~ntamination at the fcaction consistent The frequency with the test hadrons and with an upper limit of 1.3i two standard deviation level. The of the contamina ting electrons surviving the pulse -104- ~ J. (.) "> + ~ ()) 0 C,') i.-.:::-! ,.c.:; - ~.-:> f-1 r zr:.~.:i 1 ~l I ~ 0 "":cl ~ I N T ~ -t 1- 1 c.o CQ -..r., l ~ l - r-b "='-l I I l() N 0 I I I I 0 02 0 0 I ' ! 0 0 L{) 0 0 0 0 0 0 -105- haight cut of 150 vas estimated from height as The contamination after the distribution 1.41. the electron palse cut is esti mated therefore as 0.021. The measurement was done for tracks in the middle and ring of since C1 shower counter. For momenta tracks in the inner ring seldom hit a The result is shown in below 5.0 GeV/c outer the fig. 5.B.7 (above). probability averaged O.J7%±.06% (7 tracks made light in the outer ring, 3 in the middle). T!ble TS.B.2 below gives the frequency of each light without any observed track. whole outer, cell The average rates for the middle, and inner rings (excl udi ng unequipped Table TS.B.2: PROBABILITY OF SPU R IO OS COUNT IN C1 CELLS Counts per 10* *4 events OUT ER RING MIDDLE RING I NNER RI NG Cl-01 Cl-02 Cl-03 Cl-0 4 Cl-05 Cl-06 Cl-07 Cl-08 Cl-09 Cl-13 31. 9 Cl- 14 36.4 Cl-15 37.0* Cl-16 13.7 Cl-17 u Cl-18 tJ Cl-25 Cl-26 Cl-27 Cl-28 Cl-29 Cl-30 Cl-31 Cl-32 Cl-33 Cl-3 4 Cl-35 Cl-36 82.8 Cl-37 90. 1 Cl-38 87 .7 12.0• Cl- 10 hawing 1 o. 3 6.0 7. 1 2. 2 u u tJ 7.6 2. 2 9.2 Cl-1 1 21. 1 Cl-12 13. 0 Cl-19 Cl-20 Cl-21 Cl-22 Cl-23 Cl-24 u 19.3 19 .9 27.9 34. 1 25.6 U=Unequipped Cell *=Not Used (Deadened) u tJ tJ 90.7 291. * 64.2 7 5. 1 9.9 BE AM CELLS 183. * 359. * -106and deadened cells) respectively. For were the cells 0.091, 0.271, that registered h~drons below 5 GeV/c, the expected rate of outer and middle ring) and 0.891 light for (weighted average for registering without tracks would be 0.14%, as coapared with 0.17 measured with The probability of tracks. a hadron registering in the absence of n9utrals, undetected tracks, and phototube noise is taken as the difference in the rates, 0.03%±.06%. rate using isolated pions cell in a test beam as 0.021. and The punch-through was measured on a prototype C1 This rate was taken as zero, the overall punch-through rates were therefore taken as the counting rates for trackless cells given above. 5.B.5 -- Two Electron Cut After the detailed electron identification described above, mJre than half the original candidate events were left with less than two following types: it electrons. (These e vents were of the events where a track was not found because too slow or for other reasons, where a track failed vas to go through the equipped part of Cl, and where failed These events were to register or were shared.) Cl cells rejected. Of the remaining 31651 events with two more electrons, 1388 (four percent), had three or more electrons. The three 3nd aore electron or events were rejected in the interest of -107simplicity. been would In most cases, the extra electrons removed later by have the cuts designed to reject trac~s wnich came from photon conversions. The loss of final pair events due to the cut was taken as 4% on the assumption that ~n extra electron was equally probable in the final pair events as any pair event. s.c -- Track Quality Cuts This section c~nfidence presents level cut two cuts on track quality. The removed poorly meas ured tracks which c~used poor mass resolution photon conversion pairs. The "m~trix flag" cut re~3ved tracks which could not be properly fitted. 5.:. 1 -- Confidence Level Cut The confidence level cut removed most false tracks and those ra~l ones which were fitted with ~ad therefore an inccrrectly rn easared. incorrect The cut was chosen at the ll confidence level in order to ~ake a s ma ll th~ sig nal. A cut at 10% match-point confiden ce level irupact on failed to in~rease the ratio of sig na l to background. Fig. s.c.1 shows the c~nfidence level distribution of tracks i1entified as electrons in C1. If the res o l ution errors in -1080 ...-I 0) 0 co u5 0 ~ C) ca .....r-. i:: l"- 0 ...,r-. 0 u a.> ,......, (l.) t:O 0 0 ~ µ:j > ~ ~ l[) 0 w. u z~-;J ~ ,-...-, 'tj< rt-~ 0 f:1 0 •M -,..) ~ ,..0 .,-, r-. .,.:, ff) ...... -c, (l) > C) ...... Q) t.l /_;, iL-: r;:~ 0 (l) u -0 .,. . ....... r.:; 0 C0 I l (.:) ~ r.... 0 .....c.o ,,...; J 0 0 0 ~➔ 0 0 0 C0 0 0 0 C\2 S)I:) V?:!1 i!O tl38HDN 0 0 0 ~- j 0 t,) ('J 0 0 0 0 LD .. ..-1 1.() } \ 0 0 0 ,., ~ ~ -109the chambers were distributed as a Gaussian vith the ~idth c~rrectl y specified in c~nfide nce level the would be program, flat, the and ce moved exactly 1% of the real tracks. di s tributi on of the cut ~ould have The excess of tra c ks wi th c onfidence level near one indicates that the resolution estimate was too reflect s non-Gaussian the distribution. large. Th e tail small of increase nea r ze ro the measurement error The cut re moved 2.08 % of the pai rs (1 .0 4% of the trac ks} . 5.C.2 -- "Matrix Flag" rhe chi- squared function us ed to fit the track parameters was uns tible for high momentum tracks, low angle tracks, and in cas es where the ma ~ch-points did not fit a helix well . rhe "matrix flag" indicated when the fit failed to o~ whe n the converge approximate derivatives of chi-squared of the f it ga ve a non-positiYe-definite error matriK$ rba c ut eli~inated the unfitted tracks and those with a err or mitrix. T~e e=ror bad mat~ices were n eeded to fit the var tex pair, and the fit was needed to de t e r ~i ne the mass. Si nce el ectron tracks needed an angle of 30 milliradians to -110- hit the inner ring of C1, and were limited to a momentum below 5~ 0 GeV/c, most of the bad tracks were not Only 6 events in 71409 events electrons . (6 o ut of 1428 18 electron tr~cks) ~e re cu t . 5.D -- Pair Vertex Position and the Beam Association Cut Below i t i s shown bel~w how a measu re ment erro L on one of the tra c ks from a low mass pair, which ~ere numerous because of photon conversions, could cause the r econstructed mas s to be several hundred MeV/cZ. mos t The beam asso cia tion cut removed poor. ly measured and er-ron eou s tracks., thereby r.eraov ing t ha t so urce cf false si gnal~ The loss cf signal due to the c ut is es timated. 5. D.1 -- Mas s, Geometry, and Conversion Pairs Neg lecti ng the electr~n mass 3 t he parent mass of a give n b y m2 =2GP 1 <11tP:i. 0 { 1-C OS {"4>)) mom enta and~ the ~riginat ed. helices the "f depended tracks is and P;t are the track is the angle b~tween them. ~ac e de scri bed by helices, .11:in g where E\ pair were Since the tracks upon the believed position to have The point of origin~ the pair vertex, vas taken ~s the point giving the lowest combined chi-s qu ared for both tra cks, more or less the point of closest t r~ cks. J~t approach of the For photon conversions, in ~hi ch the tracks started pa ral le l, the veLtex position was espec iall y uncert~in~ -111- ?his is analyzed results: The in section with 5.E.1, the following typical uncertainty along the beam direction of the reconstructed vertex for photons was 20 cm. The typical for low mass pair of well-measured tracks was 25 MeV/c2 with many as high as 80. The reconstructed mass a reconstructed mass of photon conversion pairs where the parameters of one track were inaccurate by several centimeters could have masses as high as 300 MeV/cz. Photon conversions of primary which vertex were could produced not be far do wnstream the separated fro m direct pairs because of the uncertainty on the vertex position of low aass pairs. s.0.2 -- Photon Conversion Pairs at High ~ass Fig. 5.D.1 is a scatter plot showing the electron passing pair quality vertex cuts, any other opposite (c~nsisting of (for charge opposite charge trac ks, track) and a primary to The vertical Direct pairs comprise the The horizontal band at the botto m. pairs, shows several The displace men t towards the right (pair of photon vertex conversion downstream main vertex) photons converted downstream The between three or more hadron tracks). n1rrov vertical band. effects: distance and not making a low mass pair with scale sh~vs the mass of the pair. corresponding the points to the left corresponds vertex to the fact that of the pri mary interaction. show the extent of the vertex I ++ \ + ♦ ♦ ♦ +, '' + ·~-♦ -- .,. -.r...; .. t+ ♦ + ~ t ~ ♦ t" t ♦ + +'\~ ♦ ♦ +#-t.+ :1 ♦ .. ♦ ~ + (· ... ♦ •"' ++po- ... ♦i,. ♦ + ♦• ff • + ♦ ++ + • .. ♦ .. : + + • .,. + \ <t:. .......♦ ~ ~ ~ ................. I+•. .♦•. ~ . ~--♦· ♦ . : *•·· ♦ •♦ 0 50 ~ t• + .... + ♦ 100 ~ + + + te,Mt) :~♦.i• ~ + +;.. .... ~ + Distance from main to pair vertex .• . t+ DARKNESS i2JF P0INTS C:NHANCEO AB0VE 1'JO MeV /c ;:::~g~t::.r!2L~~~~~J1i~z!~~~1r~~l2'.:?:~z~:i~ ♦ ♦ 1..' ·;♦ ♦ ' ~ ..... .~\. '• ,: .... .<~· r ... ♦••... ,t t ♦ Figure 5.D.l: Separation of primary and electron pair vertex versus mass of pair, all opposite charge electron pairs. 0o0 -5 0 0ol 002 0o3 j ']04 VI c., i 0o5 t) 1= 006 r 007 008 0o9 LO ♦•• ♦ 150 N I .....I ..... -113position resolution error. The increase in mass to the left shows how the apparent mass increased with the error on the vertex position. rhe occasional points primary vertex at have mass and away from the indicate the problem which necessitated the beam association cut. must large crossed Above 140 ~eV/c2 in mass, the at an angle large enough tracks that the resolution of the pair vertex position was better than 5 cm. Since no known long lived particle decays into electron p~irs, one of the tracks must have been very badly measured or entirely fictitious. rhe latter fictitious case is relatively provided the tracks were distribut ed randoml y in charge. The subtraction of the same-charge When both tracks benign, pairs cancel s the effect. corresponded to real electrons, the pair was almost certainly a p hoton conversi on pair with a resolved poorly vertex, and would have appeared a mo n g the opposite ch!rge p~irs only. Even a small leakage of t he photon into the high mass region would be a proportion a lly large effect. In fig. 5.D.2 the plot for sa me charge pairs corresponding t, fig. 5.D.1 is !bsent. pairs shown. The photon conversion band is (The narrow band of apparently di r ect same charge an example is of the background processes and charge c~ntributed to both opposite discussed below in section 5.F.} same that pairs, There wer e considerably ~ UJ U) C >al 0 ...._ 50 100 ( CIK.) 150 .::- ... ♦+ ,.., :,. ·!t ~ ' · ++ :+ t1"" + + ..... ++ ....-,... +-1"\r. l+ + ,....+ 1, ... )- ♦ ·•· "+ .... ,:++ ♦ ., -+ + + •• ♦ ♦ !:..-+ ~ ♦ ♦ Distance from main to pair vertex 1 + +...( . • "4 ♦ .J + ♦ .....I ..... . i • ~ .. •*++ •+ ~ + +' ~ ++ .,. + + + + .. I + + Figure 5.D.2: Separation of primary and electron pair vertex versus mass of pair, all same charge electron pairs. -5n O"O 0.1 0.2 J.3 0.4 0.5 0.6 0.7 0.8 0.9 LO -115faver same charge high aass pairs far from the main vertex tbsn opposite charge, indicating that many photon conversion psirs were aasses. present in the opposite-charge ?he beam association distance cut data at high described below was designed to remove these spurious high mass pairs. 5.0.3 -- Beam Association Distance cut l cut requiring the pair vertex was rejected w,ald miss downstream appeared to be vertex for to be near the following reasons. conversions which near the primary vertex. primary vertex vould have to be discarded. pair the primary This cut coincidentally Events without a The same charge subtraction would no longer be valid for the following reasons. Opposite charge tracks bend in opposite directions i~ the field and so c~arge tracks diverge more quickly than vhich tend to follow each other. the same The vertex p~sition resolution in the case of small opening angles worse for same imYolving a charge fit to pairs the Any cut than for opposite. pair vertex would va s therefore c,aplicate the matter of background subtraction. Imstead, cuts involving the c~nsidered, such fr~• the Tertex, ~r the cG saon ■ ain tracks individually were as distance from the beam track, distance sum point along the beam track. of the distances to a The cut vas made on the distance between the bea a track and either of the electron -116tcacks at closest approach. complicated. am. The The others were considered too This distance vas required to be less cut removed 101 of a of sa ■ ple than 4 test tracks (described below) chosen to resemble direct electrons. Fig. 5.D.3 sho ws again the pl~tted reconstructed against the s~tisfy the 4 mm. cut. the main distance to the pair vertex mass where both tracks There were 2 faraway {>5 cm. from vertex) pairs above 140 MeV/c2 out of a sample of 8453 events in which the hadronic vertex was accepted. background above 140 MeV/c2 due to pairs inside the (The 5 cm. limit had to come from pairs whose "real" mass vas above 120 MeV/cZ since the maximum error on reconstructed mass was 4 MeV/c 2 per cm. error on the position. That background is treated in section 5.E.3.) N~t necessarily all faraway high mass pairs shoveJ There 5.D.3. and possibly cases of a do wns tream conversion the main upper cases of downstream fig. were cases in which no main vertex was made, reconstructed at high mass and near limit in for photon fig. 5.D.3, because the number of conversion it was failing wrongly vertex. An high mass a to be recognized in reconstructed near the main vertex, was fo un d by obser ving that less than 101 of the low mass pairs vertex. high were reconstructed within 5 cm. of the main Using 10% as an estimate of the probability that ■ ass a faraway pair was too close to be detected in the 0,6 ~ Ul Ill 0 <I) :> • ♦ .;..* ~t~ .~ ... ♦ .. C . ♦ ♦ ♦ + 100 Distance from main to pair vertex (~Ill\~ 50 ~.::~~1?1, L~~,~~1i:WJ~~~!~t~~~~~~~i1J:~4~fJ}/i>- ~-4 . _.; ~♦ ,.+ ♦ ,,--t+. ♦ ~~ ' .. . .,, . ~ 11: ;,. , .;: ~; ♦ I • t" ~~ t ... I· \t .,\ + DAR K~JESS 0F P0 I NTS l: NHA NCED AB0 VE l !JO MeV/c Figure 5.D.3: Separa tion of primary and electron pair vertex versus mass of pair, opposit e. charge tracks passing be am association cut. -50 0,0 0a 1 0 a2 0, 3 0,4 -- 0,5 (J + f. 0 ,7 •- • .. ,i 0, 8 0 ,9 LO 15 0 I ~ I ........ -1188453 events, the additional contamination was 0.2 events • . Another 0.7 false events are expected to have occurred vhere no main vertex was made (assuming the same rate). st1ndard deviation upper limit on the events from low mass pairs with is 2.0. The background expected The two 0.9 p~orly resolved verti ces is estimated therefore events {quoting half the two std. as extra 2.9t0.5 upper limit), before correction for acceptance. The loss of real pairs due to the cut was studied using s~mple of test tracks. interaction triggers. the test tracks were taken from The They were required to go througn a (Hadrons vere single equipped C1 cell without making light. expected to have geometricalproperties similar to the direct electrons, while most observed downstream conversions, were extrapolated back.) electrons, farther the spark measuring they traversed. removed most artifici~lly of large from beam when the The test tracks were required to have cegistered in 80 % of stations from coming the and fictitious proportional The last requirement tracks which beam association distances. sh~vs the beam association chamber produced Fig. 5.D.4 distance distribution for the aoi require ment did not resolution (and thus passing cut). test tracks. It is necessary to show that the significantly sharpen artificially increase the the track fraction the A z ~ "'-"""' ~ 0 0.0 500 ~ 1000 m 0.5 1.0 cm. (SUM ABOVE 2.0) ~ Figure 5.D.4: Beam association distance of test tracks. 1.5 2.0 I lO 0 ..... .....I [;:r... E- ~ ~ 1500 ~ U) 2000 2500 -----------------..--------,---.---,,--..---,--,--,--,----.--,---.------ -120- subsample of the test tracks with moderately large angles and momenta was compared with and without the 80% coordinate requirement. There were few false tracks in this region phase space. 91.71 of the unrestricted tracks passed, while of quality the good subsample, 93.4% passed. known how much of this increase was due to of fictitious resolution. tracks, and how much the due It is not elimination to sharpened The difference was treated as negligible. The test tracks were separated into a tvo dimensional were momentum and e.) rectangular groups order to significance as described for the efficiencies in section 5.E.1. The additional of bins due to of in in Bins array combined into increase the statistical shover counter uncertainty using data where the fraction of the tracks removed was different than at directly the nominal values, wag estimated from the data (rather than using~ fitted function as was done for C1 and the shower counters). The additional uncertainty vas the sum of the magnitude of the between the efficiency for each of the four subsections and that for the whole rectangle. ~inimize The rectangle was chosen to the overall uncertainty (statistical plus that due to the spread). table difference of This procedure produced a dimensional the fraction passing the cut and the uncertainty, as a function of momentum and angle. dimensional two cross sections of this Fig. 5.D.5 table. probability of passing the beam association cut shows one The average (using the ~ Jl'OR P-0.75 TO O.Bt ~ e e ra. ~ ~ z ~ ~ I.) 4 6 6 L> !:; ii?. ra. 0.2 I 3 0.2 0.4 0.6 0.8 2 MOMENTUM GeV/o 3 FOR ~-0.27 TO 0.32 ~~<}0~~ ~ ~ ~ ~ ~ Figure 5.D.5: Result of beam association distance cut as a function of momentum and angle. MOMEN'l'UM GeV/o I II II I I 2 r.. ii?. z ~ ~ FOR 8-0.030 TO 0.035 ~ 0.4 ~ ~20 0.25 I I 0.30 4 5 6 1.0 J#~~4HH444 ii~ i i z ~15 THETA (RADlANS) ~10 1.0 0.6 0.. 0.05 ~ 1.2 ,-,-...,...,....,.......__,.....,.....,-,---,-,.......-,-.,.....,,_,_-,-..,......._--,-.....,._........,.....,._........ ij ~ I P'OR P-5.5 TO 6.0 ~ 1.2 ,.-,,--,--.-..-,---.-....-.,....,........-.--,-.....,.--,--,-.....,.-,--,-.,.....,---,--,-......,,.........,......-, 0.8 L> , 0.00 0.0 I 0.2 0.4 0.6 0.8 I) (HORIZONTAL BINNING RANGE IS GENERALLY LARGER THAN INDICATED BY SPACING BETWEEN POINTS) TIIBTA (RADIANS) 0. 0 ........._._...,_.._._,__._.._,'--'-_._-'-L-l.-'--'-'-L-'--'-.L-1-'--'--'-''--'-................, 0.00 0.05 0.10 0.15 0.20 0.26 0.30 0.2 0.4 0.6 0.8 e !:; la. ~ E 0 z 0.. ~ ~ (.) ~ 1.0 e 1.0 ~ 1.2 .....,.-,--.-.,.....,-,..-,-.,.....,---.--.--,-,............-...--.......- . - - - - - - - - - 1.2 .....,.--,--.-.,.....,-,--,-.......,,...........--,-.....,.--,--,-.....,.--,--.-.,.....,---.-....-............................. I I ...... N ...... -122aoaenta and angles of !onte Carlo generated high aass lov electron electron pair tracks) v~s 931 per track. S.E -- Low Mass Pairs ?his section analyzes geometrically the p3irs: photon S)nrces of conversions and pi-zero Dalitz decays. background reconstructed aass at are discussed: lov mass Two pairs high mass and single electron tracks from phlton conversions (referred to as "half-pairs"). The "low miss mate" cut. designed to remove half-pairs. is described. Tbe contamination of the high mass signal with photon conversions and electr~n pairs from pi-zero Dalitz decays is estimated. 5.E.1 -- Geometry and Mass Resolution. rhe mass of the electr::>n pair (neglecting aZ=P1 •P2. •2• (1-COS (f)) is the angle between where P, them. masses) For low mass pairs. The opening angle SIH(191 )•SIN(~2 )•COS(~,-~,)• where TAN(en = Plj /Px-) Along .the helical track, the 11omentum changed. components direction and is and TAN(f) = P-t./Pi. Px_ the ♦ COS(f)=COS(G, )•COS{e4) The is and P-z... are the momenta and "f 1pproximation • 11= Y(P, •P~ •"f, is used. given by rest P-:, of the changed ac::ording to dcjl/dZ =a/Pi • while P..t, Pi.• and t:9 remained fixed. For the 22.5 kG. field, a was 0.006735 (GeV/c)/cm. -123- PhJton conversions and pi-zero Dalitz decays produced with mass close to zero. t:> each other at the The helices were therefore tangent (recurring helices closer 0 1 and <9-i. vere point of creation. ag:aal everywhere and <-?L - ~1. was zero origin only at their at isolated faraway points). a small neasureaent large shift in the position ipproach of the tracks (the reconstructed in a view end-on error) of closest vertex). Pig. The' drawing on the left shows 5.E.1 illustrates this point. in common Moving the (representing resulted pairs of a photon conversion. The track helices !ppear as circles. The slope perpendicular to the 1/?AN() of Z distance per unit arc-length. in units page driving on the right shows the result of moving one balices closer point moves by from a V small to v• The distance e. by a tricks. the rhe of the The intersection distance h, given by Z positions of the tracks at v• ~re almost identical for the following reason: s~ that the is z change per arc-length is the e, equals e~ same for both The arc-lengths from the beginning to v• are almost same, because they are b~th approximately straight line segments. The fitted measurement error in vertex the is therefore near opposite direction did not have this effect.) rne apparent mass calculated as a function of the distance e r2 Figu re 5 .E.1.: Effect of a small shift oi the center of a helix on t he verbx position. h= ✓2~~ e ,it r1,Jl:r2/(r1 +r 2) 0 g==e I + N I-' -125c:11., •=0. 116• Vee• (Pc.t position is 1 The + Pt.'2.) ) • h/TAN(9). error This on comes 1 z=17.2•'/(e•(P1 +P:i)/SIN(0)) /f in cm. and GeV/c. = (P, +P:2.) IVCF, •P1 ) the vertex out as The number f depends only on the ratio of the momenta and lies between 2 and 2.97 for momenta in the range 0.75 to s.o. The rate of change of the distance is dm/dz=c•f•SIN(8). with mass vertex error (This does not require a zero The angle 9 was 'limited to the range 0.1 to opening angle.) 0.2 by the shower counter requirement for one of the tracks. (Fig. 5.E.2 shows the angular distribution of triggering dm/dz therefore is expected to be in the range 1.3 tracks.) to 4.0 (MeV/c2)/cm. It is histogrammed in fig. 5.E.3 e taken from opposite charge electron values of P, , P~, and p!irs. It fell between 1.1 and 3.5 in 991 of the cases. The momenta were mostly distribution the of for low, averaging reconstructed 2.2 mass The GeV/c. is calculated, assuming the distance e was the same as the beam association distance of the individual tracks in fig. 5.D.4. ~ssociation distance is well fit by The beam the function dN/db proportional to EXP(-b/(.125 cm.)). These values predict a distribution dN/d (m 2 ) = constant • histogram of for the EXP(-(m/(34 MeV/c2)) 2 ) . measured low mass mass of Fig. 5.E.4 shows opposite-charge pairs constant • , m•EXP(-(m/(34 Mev/c 2 ) 2 ) . of the pair masses at low result of masses a fit by dN/dm = The downward distortion is believed to be the the coordinates of each track being close enough -1 26, - ---,--- - - . - - - - - . - - , - - - , - - - - - , - - r - - . - - - - r - - - . - - - , - - - , - - - - - , - - , - - - , - - ~ - - . - - ~ - - . ---. -tj'< 0 C0 0 r.-, ..-a v r-J h .,-., l:.O C1 ·c U) 1:0 C,i) ·....., ~ ,.-, lfl 7 C\1 ~ '-"'-> 0 • :--•• : '"C~ <'C I-< ,.___., 0 .....0 Q) ,,_ ~J i::; (ti <D C-2 t:::il V':l Q.) r... ,•-' .:::P r:,:_, 0 0 0 "'1' 0 0 C0 0 0 C\? 0 0 ..-I n ~\I S)I8VH1 iIO ''.:Irrq,\-t- L'.l.cd'i l _1 0 0 - z ::::> ~ m i::r.::i 0::: 0 G:.. P... <t; ~ lfl 0 50 100 150 0 t- 200 • - 250 2 3 dmass/ dZ in MeV / c per cm. 2 4 Figure 5.E.3: Distribution of dmass/dZ for electron pairs, where drnass/dZ=c j.tSIN(0)*(P1 +P2)/vPl*P2. 1 5 7 300 r-r--r--,--·,--r---r--,--r-----.----.---r--r-r--r~--,,--r---r---r--r--r--r--.--.--. I -...J N .....I -1280 Ir~ N ,-.._ N 1 (.) '----... >Q) o ~ 11 0 ~ -tj< C'1') ...-1 J '----- s ~ ...__,, l (I) ,,I ~ X a s '6 .,.--..,_ 0 aJ I! '---"' ~x1~ oC'\"-' 4-1 r<) "' I I ,'.) > (0 ~ 0 ( \] ,---l_) ,............-......;----' 0 0 ( •') 0 0 C\2 I -------.... i 0 0 ...-1 . -0 -129- in t og ether that they are combined into one point, of th e pl anar chambers. seYe r al Othervise th.e agreement is El e ctron tracks fro m photo n conversions would often mimic produced by two s tr i king . 5 .E .2 - - Low ~ass Mate Cut h igh mass pairs separate photo n s. when the tracks ~ere The "maten fer each (coming from the same ph)ton as the first electron and vith which the pa ir mlss ~3. s low} was often not identified as an electron because its mome ntu m wa s belo~ the c ut-off or because it did not hit an equi pped Cl cell. tha ma ss e s Such events were rejected by c alculating of all pairs consisting of o.e of the electrons ~nd an o pposite charge track not identif ied as an ind th en m3ss . requiring that electron. non e was less than 150 ~eV/c 2 in Pairs with tr a c k s with bad matrix flags were not angle Cf) tas ted b ecause the vertex could not be fitted. Fig. 5.E .5 shows a scatter plot of :lg 3. i nst the mean momentum ( ~ } ~ppo sit e c ha r ge electr~n pairs. for a test sa~ple of One electron above, ~~rr espo nded to the ca ndidate ele ctron tracks. w~s all req uired cuts opening described pas s the the vas required so that The it other only to register in C1 in ord er to iJclude as f1r as v~s possible mates ~uich fell outside the mo~entu m~ c onf iden c e l evel u and beam association distance limits. 'l'he ~ I..:;:~ r::r:i <t; :zl "'"'-~ ~-...-4 0 2 JJ '1 M M MMMFl 2 3 3 1 3 1 C 1 \\ 1 2 4 2 1 4 3 2 3 I I ' I 2 2 1 1 --.....,_ 1 1 ---...._ 1 I - 1 ~ 1 -- . 1 1 ~- ' ___L - 0. 10 OPENING ANGLE (radian s) I I 1 '--1-..~ 1 1 -------- 1 '2._ "-- ------- 1------- --...._______ 1 2 1 0.05 i ' 6 2 1z ? I 5 3 B 6 .M 1 1 :\1 2 1 1 :~ '\1 111 1,1,, 2 \. :---....._ "' 1 \ ~ 1 13 '77 2 41 1 2 1 I 0. 15 j I I Figu re 5.E.5: Scatterplot of opening angle versus mean m omen tum of electron lc s l pairs , vi.r:ith curves of const ant :invariant mass. 1 4 '32 -·1 6 3 -1 ).JMMMM},I --M M M M M 7 M?JM 7 M7 -M M 8 M M 0 4 3 9 6 6 M 4 6 J.JrJ 5-··i: M M 0 0 .00 1 4 f3 7 M 1J 9 M M 5 t"'-~_:'I M 9 5 5 \,,.;,- -4 6 6 7 5 ·o 4 E-< z r:,~ 7 2 7 1 3 I~ ! 6 2 1 :1 ::::> 3 1 2 2 • 2 ..-:~ '-s-' Q) "'c.,> Q 4 5 1 0 .20 ~ 1 I c:, \J-l I I--' -131- :1 requirement cut out most very lov momentum ■ ates, but vas necessary to eliminate electron-hadron pairs. Tile solid curves correspond to 150 MeV/c 2 • U(P1 •P:i,_)'•i' = mass = 50, 100, The dense region of conversion pairs covered a larger region of at lower momenta. c~ntained curve representing a mass of 100 MeV/cz. The the by dashed curve is an example The dense region of l\f•((P, •P2 )**1/4) constant, which is derived from the formula for the ingle of zero mass pairs due to "fz=2•e•c•f •TAN (tt) / \J("P, •P2. f, taking a for is = opening helix mismeasurement: constant upper limit the separation, e, and the momentum partition, f. This curve does not seem to fit as well as the mass cutoff. Fig. 5.E.6 shows the mass distribution described above. accidental The combinations separate conversions). as well. of pairs at electrons the test pairs high masses were (coming They appeared as same from charge two pairs A drop occurred in the distribution at both 80 and 150 HeV/c 2 • many test of more A cutoff of 100 MeV/cZ (rather than 150} passed same sign pairs, a measure of electrons co~ing from separate conversions. significant number ~f This indicates that there was electron a tracks having mates which formed masses between 100 and 150 MeV/c 2 • The mass cutoff therefore was chosen as 150 MeV/cz. The loss of direct high mass electron pair events due to z ,4 ..,,ro. :::> ~ µ:J 0 G... P... <r: ~ - lf) bl 0 25 0 50 ,_ 75 100 125 100 MASS MeV/c 300 2 400 Figure 5.E.6: Invariant mass of electron test. pairs. 200 500 600 l 150 ~--,--.--.--.---r-...--.--,...-,----,----,--,-......-,r----r---r--,--.---,-___,r----r--,--.--,---,----.-- I N I.N .....I -133chance lov mass combinations with the other tracks in the event was compensated on an event by event basis by giving each event a weight equal to the reciprocal of the estimated probability of both probability of avoiding a chance low mass pairing with one of the opposite dimensional surviving charge tracks as function table aomentum and electrons a angle, and the was the read cut. from The a four of the electron charge. number of opposite charge tracks in the event. The table was computed using the events and T1 (interaction artificial electron tracks. The probability of an electron with a given momentum and e making pair with one trigger) a low of the tracks of the event was taken as the fraction of the circle of c:p values over which the less than 150 MeV/c 2 • of mass vas For each T1 event. the probabilites of surviving the cut were calculated for range mass electrons over a 32 momentum values, 32 e values and both charges. and recorded as a function of these quantities and of the multiplicity of the opposite charge tracks against which it was tested. (The probability of making a low mass pair with none of the tracks was calculated probabilities were assuming For independent.) the individual each bin, the mean probability and the uncertainty on the mean were t~ produce the table. single electron calculated Fig. 5.E.7 shows the probability of a passing the cut for opposite charge track aultiplicities of 1 through 8 as a function of electron ~ ~ 1 4 5 6 0.10 0.15 0.20 THETA (radians) 0.05 8 0.25 P= 1.75 vi -l=1 .....I 0.30 I IDJ I 11111111111 Figure 5.E.7: Probability of passing ]ow mass pair cut as a function of momentum and angle (for 1 to 8 opposite charge track multiplicities). MOMENTUM GeV/c 0.00 o.o 3 0.0 O 2 0.1 0.2 0.2 0.1 0.3 ~ ~ § 0.4 ~ ~ 0.5 ...... 0.3 m <t m o.4 0 THETA= 0.155 0.6 0.6 f-1 0.7 0.7 - 0.6 0.6 ::,... 0.9 0.9 ~ ...... 0.5 f-1 ::,... 1.0,,, 11111111111 1.0 -135momentum at a fixed angle. and of angle at a fixed momentum. The probability (The values sho wn are for negative electron tracks. tr3.cks. rnose was one when there vere no opposite charge for positive tracks were only a fe~ percent different.) The uncertainty of the probabi lity estimate was recor de d as a dapended mostly on the number of cases proces s ed which fell single value for as the multiplicity increased). were an each multip licity The recor ded {it uncertainties upper limit of the uncertainties for all values of ~omentum, e, and charge. The distribution of the event weights (the reciprocal of the probabil ity for both electrons pa ss ing the cu t) is sho~n for the high mass electron pairs in fig. 54 E. 8. When binning the weighted event added to a events, a single weight bin co ntaining only s~all weight events greatly reduces the statistical sig nif icance. a large For example# bin containing 9 events of veight 1 has a significanc e of 9±3. Adding an event of weight 4, i.e. 4±4, gives less significant indication of a signal. such a loss of significance, a ceiling ~as event weights. 13±5, a In order to avoid i mposed on the A weig ht of 2 wa s assigned to all events in which th2 probability of passing the low mass pair cut vas less t han 0.5. The cutoff affected 49 out of a total of 325 events mass (with greater than 120 MeV/c 2 } . reduc ed the average weight from 1.592 to 1.458 . The cutof f There ~as - z :::> ~ en i::r:i ~ r.:r.., 0 ~ <t:: ~ if1 0 10 20 30 40 0 t 50 - 1 ~ 2 4 WEIGHT 5 6 7 8 final sample of high mass electron pairs. 9 - ; - - - - - - - 1 " - -- - - - - - - - - - - -- Figure 5.E.8: Low mass mate cut compensating weight for 3 - - - - . -- - - , - - - - . - - - , - -- 10 j .....I I m \J,I -137no apparent correlation between weights abo ve 2.0 and the •~ss of the pair (scatter plot factor of 1.092 was shown included in in f ig . S. E. 9). the f inal acceptance A correction to give the correct average weig ht. S.E.3 -- Pair Vertex Related Cuts Three cuts were concerned with vertex fitting dropped. procedure the X-Y ver t e x : If the failed to conve rge t h e ev ent was Events were also dropped when the the fit was outside the target. the pair Z po s it ion of Finally. e vents were cu t i f z o f t he f it was position of the beam track at the near the aluminum target jacket. rhe vertex fit failed for 27 out of 9793 pair c an d idate s. All were opposite charge and the difference in 9 f or the two tracks was always less than 14 milliradianso that they were low mass pairs. This indicates It is ass um~d that the c ut had no effect on the high mass signal. Fig. 5.E.10 shows the z position of all t he liquid hydrogen e xtended from 23 to 114 cm. Plug1/CD1 The material in (Not all low ma ss events seen the Plug1 peak were created there. one The extended from 121 to 131 cm .• a nd that of the 1.5 chambers from 165 to 167. than pai r s . wire spacing chamber coordinates. were This in Tracks separated by less measured distorted to the have identical fitted helices ~ <-.t: U) U) 0 <l) :>- - 0.1 ~ 0.2 f - 0.3 f - 0.4 " 0.5 0 0.6 0.7 f - 0.8 f - 0.9 I 1.0 1 1 1 1 1 2 1 1 I I I It I 2.0 1 I lI I 1I WEIGHT I 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2.5 1 1 1 1 1 versus the mass of the pair. Figure 5.E.9: Scatterplot of the low mass mate cut compensating weight 1.5 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 1 1 1 1 1 2 3 2 1 1 1 1 1 1 1 1 1 1 2 1 3 1 1 2 1 1 2 2 1 2 2 2 2 1 1 1 1 1 2 1 1 2 2 2 1 2 1 1 5 1 2 3 1 1 1 1 1 1 3 2 2 2 1 2 2 1 2 2 1 2 1 1 3 1 3 2 2 1 4 1 2 1 1 2 1 1 1 3 3 2 1 1 2 1 1 3 1 1 1 2 6 1 1 1 1 1 2 1 2 1 3 1 1 2 1 1 1 3 2 1 1 1 1 2 1 1 1 4 1 1 3 1 2 3 1 1 1 4 2 1 1 4 2 1 3 1 3 1 1 4 2 1 1 1 1 1 3.0 l I I 00 ~ ~ - 139 0 0 C'-1 0 L0 L ....... ,,--..._ Q) o +-> r'\ G.5 __,. .--..-1 / -:-i 0... '"CJ ~ ~ ~ 0 ~ ~-, 0 ID <.TJ Q) r ,..., -;-' '-\ U) ,< ,,., ~ lj ts: 0 :-c ...__,, ~i (.) --i:· 0 0 a:> 0 0 CD 0 0 C\2 0 -140enough that the verte x fitted best at the position of the chambers.) The linear ly rising slope =ontaining the low pai rs va s mas s cf the histogram a co Th sequence of the increasing number of p hotons available for conversion at the downstream end of the ta rget. The z position distribu tion for high mass pairs is shown fig. S.E.11. in The his t o gram is level, in(icating that most high mass pairs came fr om primary interactio n s. rhe z position was req ui r ed to be no mo re than 5 cm. the ends of the t a r g e t~ This insured that beyor.1d all pairs produced in the target we re ac c epted even if sBeared by vertex resolution. Wi th 5 c m. smearing, pairs produced in material outside the t a rget could not fall within with the following e xc eption. of Plugl could Bowever, no be the s me ared as the cut, Pairs produced in the front far upstream as 116 e ra . pairs we Le ob se rved bet~een 116 and the cuto ff ~f 119 cm., and thus backgro und from ma ter ial beyond the target was not signif icant. Pig. 5.E.12 s h o ws the distance fr o m the c ente= of the target to the bea m trac k at t he cut z position of the p~ir . This was at 2.3 cm., 3 mm. fro m the aluminum jacket, in or der to remove possible inter a c tio ns in that materia l . OB the beam flux was negli gib le. The One event was cut. effect ,i:::..i ,--,. ::=) ,.~ f::l ,,C"J .. ., r1,., • ~IO LJl □I ~ I I L_i___,, L . _ J _ L . - . . J . _ . _ _ _ _ . _I..LLJ_--l.,___ [Li] 2 cm. 3 Figure 5.K1::i:'.: D1s tanee from target. axis to high mass pair vertices. 0 L i _ ~ l . _ _ L _ _ j _ _I_ _ j , - ' - - 0 1 4 I .....I l--; 1~.f:" N 10 20 C) fa., (.1..., <.f, er~, ....... U) 30 . - - - . - - ~ - - , ~ - r - - r - - - ; - - - r - - , - - - - , - ~ - - - , - - - i - - - T - ~ - - r - - . - - - . - - - - - . . - - - . - 143 s. e . 4 -- Lo w Mass Pai c s Appearing at High Mass The n umbe r o f low mass 9airs appearing at high mass due to ve ry l a rge e rrors in the reconstructed pair vertex position was es timat e d di scusses in the section 5.D.3. The present section possibility of pairs from pho ton conversions ind pi-z eco Dalitz decays ~ith a vertex reco nstructed within 5 cm. o f the c orrec t ma s s s ignal. p~sition, being an d are t h e as the high The number of lov mass pairs is taken dir~ctly fro m th e final mass spectrum~ p1irs counted photon o nl y The relative number of Dalitz conversions is calculated assuming pions significant source of photons.. The mass position is res o l ution e xcluding the error on verte% cal culated. The number of these pairs at high ~ass is e stiraated to be less than one. Th e approxi ma te ratio ~f photon conversion to pi-zero Dalitz d e c ay pairs was obtaiaed by computing the probability photon interact in g wi th the of material in t he target. a The ~mount of material in the ~a th of the photon depended on its a n gle , increasing as i/S IN{0). The angle oft e photon the daughter electrons , and so an same as that for the a v erage value of the electron angle, 0.16, as shown in was fig. s. E. 2 was used. (Photons origi m.ting in the last 18% of the t~ ~get through the mylar end ~indow instead of the exitted a l uminum outer jacket. The estimate errs by ign o ring It errs by abou~ 10% in the oppasite that effect. ab out 15% by -1 44directi on by ignoring the material t he in CD1/Plu g1 chambers. ) Tracks perpendicular to the target hydr ogen and penetrated 0.0 7 1 c m. of aluminum. 2.6 cm. of This constituted 0.9 % of a radiation l engthfi), or .7% probability of producin g an electron pair (0.2% fro m the aluminu m) . hydrogen and 0. 5 % from the Correcting for the increased ptth le~gth because of the angle v and c ounting both pho to ns from the decay, each pi-zero produced 0.09 elec tron pairs on the average. The pi-zero Dalitz branching ~atio is 0~0115 (2) ~ leading to predict ion About of 8 conversion pairs for each Dalitz. 7 815 pairs were dete c ted belo w 140 Mev;c2. a An estimat ed 870 therefo re c ame fro m pi - zero Dalitz decays. The mass error on low mass pairs had t hree t he co rrect vertex modified ¼ngles posit ion, opening the compnnents~ lt mismeasu r e~ent of the trac· angle of the Mismeasure~ent of the momenta changed the which the opening angle. Mismeasnrement of the to reconstructed multiplied t rac k positio n led an in c orrect factor pairo ft"pl IP p'2. )) vertex position the::-eby changing th e opening angle. ?h e contribut ion re construction question was to the less (ver-tex error mass erro r from the vertex than 20 MeV/ c 2 , for the events in < 5 c m.). -145- rhe error on the track m~mentum iacreasing The momentum. in matrices, this error was not = shape. with V{ P1 • P-i f, by divided Fig. 5.,E.13 1 dV{ P,,1)P.._) /!)(P 1 eP:z}• correlated with the opening angle, dmass/mass (for the momentum s~me was order to better characterize it. sh.ows the histogram of dLOG(P 1 •E\)/2 Assuming increased error on the guan ti ty calculated from the tract error Y{P 1 cP,_), generally errors o nly) had the Because of its proportional nature, this error w1s small for loH masses. Tbe mass error due to mismeasurement of w~s the opening angle estimated by measuring the difference in the measured 0 angle of the elect rons in opposite charge pairs~ The error one was independent ~f the vertex position 4 A small openi ng angle.- >!f, ma.y be resolved a p ?.:-oximat e ly into compone nts parallel and perpendicular to the angle from z axis to the tracks. co mpo ne~ts These the r~spectively to tbe two ter~s in the correspond approxi r:i a.tion, ::: '£his + geometrically in fig . 5.E.14 u-here a reg i on of the unit sphere is approximated by a plane.. "f is the hypotenuse of a right triangle vbose legs are the two ter • s mentioned aboveo Assuming that the directio~ of the direction ~as rando~q the distribution error of on the tr ack cecon~tructed f~r zero opening angle pairs can be inferr ed~ -146L.0 C'.J. 0 0 C\1 0 H ..., ;J ..-, .... ....().) s 0 s il) ClO ttS 1-, Q,) I.() ?- """'" >c::: 0 P-. Cl :'j "--._ > -<>: µ., 'O p~ 0 0 >-'-i .-=1 <C >-= __r ?-, \_ !:'"""1 0 1 ---' ~? 0 0 L_L .-:.:;i 0 l() C\1 0 0 C\1 l.() 0 0 0 ..-! ..--l 0 LO 0 0 0 Q) N b.., -;I'x. , a r2; 0 ' .,_, H;> (.) ~ :~ ,-, ri1 1_() iJ.) ;... <,,..,. ~ ~ .S-P i::r... <"::1 ~ ,.., µ., F .. \ ~ i () h 0 ;... h •rl ~ n ·" ..-; i::.,.:i 0 0 ;'.=-, Figure 5.E.14: Illustrating the decomposition of the angle ,I, into components parallel and perpendicular lo the angle of the pair relative to the Z axis. I 0 X ( q> 1- q> z) '\ (®1-®2) ~'\ '¥ I I " ~ ~ -148If the distribution of the Is, - ~2 1, difference, independent, the t? t.l, is !)a <P= SI!i (9) •I~. - the and Adding exponents, dN/d!l equals EXP (- ('l'/K) 2 ) , where cill. is dtle•dA Cf, 2•1't'•SIN(f) •d'l.) is dN/d'o/ therefore equals and proportional to (replacing SIN (1? by"Y}. "f'•EXP (- ("f/K) 2), ?his procedure was applied to the quantity Ve~ •P-i ) were constant 1 desired same distribution as a functl.1~11 of both is the product of the tvo similar distributions. If is i.e. dn/d~e = EXP(-(h8/K)2), and the distribution G~ussian, of the A fi= difference, vould result •~ss•EXP{-(mass/K)Z). and dN/dm 0 be dN/dmass For 7 IR 13 = V{P1 •P-t) on£) Gaussian, the proportional to wer,e real data with war ying momenta, the distribution of m6 might be the sum of ~a ny Gaussia ns. It th e sum of two vas, in fact, fitted very well G~ussians, as shown in fig. 5.E.15. by The prei'H cted curve for the measured mass of photon conversions is also sho wn. 10000 zero mass conversions, less than a t e nth of an event is expected above 40 appropriate only ~eV/c 2 • The above for pairs of zero mass. d istribution mass is large, distributed as dN/dAM = V3.lue. That EXP(-(Aa/K) 2 ) dis tr ibu tion falls aroamd fas te r the the overestimated. (The number similarity at to central dN/dmass = estimate for masses was distribution of hiqh er the the ·than Using the latter as am increase, When mass>> K, the mass resolution, AM, is a3.ss•EXP(-(mas s/~Z). mass is r t overestimates the mass error if the pair mass is greater t h an zero. the Por ~ I--·.;,- ,. _, :::) . ...::::-, m__. ... r:r.:i 0 250 500 ?'50 1000 1250 10 J. -'• ..,t ~. ) \·.~ V 20 C) -, 1i,;, ,:::..'I/ C ,~ JV1, 1-'," S' --.: 30 Figure 5.E.15: Ma~::.: n1::::c1 t-1lbon, rn, excluding vertex position error, as predicte d from :mt, vihfr:h co:rre:::;poncfo d to lhe 0 component of t.he opening angle (l ' /-5 A.,,,,,- Prediction for n1=·✓p 1xP z<'l' ocrnx ( e -(m/.006B)z+ .43 xe -(m/,012)') FIT o< ( e-(mt/.0068)2 + .43xe-(mt/,012)2) Histogram of mt= ✓P 1 xPz<l0 1 -0 2 I 40 1500 ~---~----,~----,---.---.--~--.--.-----.---.--,---,---.--------,..-- I -1=" lD I ...... -150dl/d•e aentioned above is due to ■ asses, the the that at large component of the angular error parallel to the opening angle changes the mass, co ■ ponent fact does not, just whereas t he perpendicular as~ was ignored in calculating • If•) The theoretical mass distribution of . the electron pi-zero Dalitz angular in decays vas generated accor din g to the paper bf Kroll and Wada(3). the pairs This distribution was convoluted with resolution mass increase distribution. The result vas smeared according to the resolution. is shown as the integrated The final result fraction expected above a St~rting given momentum proporifi:i.o nal iu mass, fig. 5.E.16. with 870, the background contributi on from pi-zero D~litz decays is expected to be less than 0. 3 (less than one with 95% confidencet above 20 possible additional aeV/c 2 120 Counting MeV/c2. from the the vertex position uncertainty, the li mit was 140 MeV/c2, and the background at hiqh mass from lov aass pairs that had 5 cm. or less vertex resolution was less than O. 3 events (unveig ilrt ed). TBe mass of electron pairs produced (Bethe-Heitler process) is in generally masses since the q2 of the virtual photon tBe charges conversions ph oton on lJ a few electron th at couples to in the matter is approximatel y proportional to tBe mass of the resulting pair. (qZ is approx imately 2•m•k, vBere t i s the incoming ph~ton momentum and m is the pair 0.007 l 0.000 ~ 0.001 ~ <r: o 0.002 0 .,._. E- z 0.003 ~ .,._. ~ 0.004 ~ ~ ~ 0.005 U) ~ 0.006 0 > ~ 8 0.008 0.009 0.010 70 80 110 m (MeV/c 2) 100 120 130 Figure 5.E.16: n° Dalitz decay contamination to high mass including resolution errors 90 Momentum and angular resolution added Angular resolution added ,,.o Dalitz 140 150 I I V1 .... .... - 152- Because uuss.) were there about 10• conversions, even a s ma ll fr action at large ma sses could be a serious backgro u nd problem. then The high mass tail of the dis tribu tion pr e dic ted propagator by the 1/g 4 because s ~all q2 of larger the photon interactions are reduced by the screeni ng effect of th e atomic egaluati~g suppre ssion is electro ns . The tools for the mass distribution in different materi a ls are given in the paper of attempted because of the complexity. it Tsai(4), but th e author has not The a ssu mption is s~ie that i ncl uding resolution effects, less than one i n 10 • o~curred above 140 ~e V/cZ, which was the case if the true mas s did not excee d 100 Me V/c2. 5.F -- Bac kgr ou nd Processes This sec t io n discusses the vario~s their sources rates, and their subtraction9 of backgr~und, ~uch o f the bac kground w3.s remo ved by subtracting the same c h:i.rge pair events. The 11 d itio n1l cor rections are discus sed below. rhe ~ass o f the final op po site charge pairs is in p·g . fig . 5.F.2 hi s togrammed shows the histogram with the wei ghti nJ that compensated for the low mass mate cu te wit h th e ~eighted s~ rn e c harge his togr a .umed in fig. 5 . P., 3. "direct " electron mes ons other than pairs the pair s. along The d"fference is This difference ?.ias co mpos ed of (counting t he pions "direct") as Dalitz decays of and pairs of -1530 0 0 ~ 0 0 0) 0 0 a) 0 0 l:'- 0 0 CD C\) 0 0 L() 0 ~ "-. Q_) ~ U2 0 0 ,......._ _J .-! 0 r - - - - - - _j - -~I C) l ~, I lj r-:::.J ' ·- ~ C0 0•·~ ( ., 7 C\2 lI '=:J 0 _J 0 .,.....; 0 0 0 i.() ~' ~ ~ 0 0 0 0 0 C\2 {) } <, 0 0 0 0 0 I.[) 0 -~· IJ) ..,.. &) 0 ~ (2, 0 5' ,. . , !1} ~ 0 ~ ..,:, (1) µJ ..-~ lr.l ,_ z 0 Q I I 2 1 4 6 8 10 E--< l. '-..J U2 GJ P... ·t ' ) 14 I I l 100 I __u_ _( () [L_L_l 0:~ ~--l 0 ~ \r.::~4 r:..., r) >~ 4 (,.) "----.. N 6 8 12 10 ,__ 14 ~I 200 I 500 600 CHARGE SAME I I I I I I I I- 1 I I I I I 1-=1 1- _[hn I Glilli_li_ Jl_j__ 700 800 900 1000 -1 I I I I I I I Rf11cc, llG' 111vf.c1,) A c-• ,.::, r< M 1 _e ,7; 1 c 2 40 0 II Figlu-8 f>.fi' .f: : Mw,s of pairs - ·- weighte d for the low :rn nss mate cut. ·J I__ rio· ,:)! I , II I I full ~ruu~ 11 11 OPPOSITE CHARGE I t-' I -r- v, (.) [:r.J > zGJ E--1 U) P-. [:r.J -' ~, C-,..• 0 ~:.'1 ~::.··-~ (1) •·- "'.,..- N r 0 200 n 800 MASS MeV /c 600 2 1000 n ~ ~ 1200 1400 1600 _,__L--~-...L---I nn Figure 5.F.3a: Mass disLrihution r,fter same charge subtraction. 400 ~ - - ~ - ~ ~ - - " - - - - - - ' · - - ' - - - - ' - -__.__ 11 I 1 ~ ml u ~~j ~ J , n r fin l r ~ II . ~ - 1 0 ~~ -~ -5 0 5 10 15 () i:i:l > zw ~ U2 0.. w ~ L/") 0 ~ (1) > '--..._ N -20 o 20 40 0 200 800 2 1000 1200 r ·-·rh-~. MASS MeV/c 600 _l Figure 5.F.3b: Mass d istribution after same charge subtraction. 400 I L lW, l j JJrfl 1 1400 rh 1600 J I I m u, I-' - 157Ris identified hadrons (which are discussed in sec tion 5.F. 4) . 5. F.1 -- Sources of Ba c kground Jp posite c harge pair backgroQ~d Opposi te c harge pair s of low ~b ove 140 MeV/c 2 • arose as reconstructed Opposite charge and conYersions, misidentified pair s were roughly equally an d could be t~o ways: pair background two uncorrelated electron-identified tracks ("half- pairs" from el e c trons, in There was no corresponding contribution t~ th e same charge pairs. ~l s o mass occ urred same- charge pairs, Co mpt on had rons). distributed so that a atomic scattered The uncorrelated as large opposite-charge part of this ba c kground was re mov ed by the same-c harg e subtraction. The backgro und of the first type, pi-zero Dalitz and c onversions 1 was treated above. es timated (in sections The number of such pairs is 5.D.3 and 5.E.4) as 2.9 ±0 .5 due to pJo r vertex recons t ructiont and 0±0.l due t~ account for the low mas s mate cut, giving of to all other The estima te mu st be increased by a factor of 1.6 e ffects. rest photon the background discussion 4.6±1e0. The co ncern s second-order c orrections to t he uncorrelated pair background due differences its distribution into op posite and s ame c h:1.rge pairs. in to -158 - 5. F . 2 -- So u r ces of Uncorrelated Electron Tracks El e c trons from "half-pairs" {photon Dil itz low mass were e qually often pairs conversion or pi-zero where only one electron vas found) positive and negative~ fev p e rce nt. Compton scattered electrons, ho~ever, vere always excess of by only misidentified overall differed and ne gative ne g a tive , giving an hadrons Positive negative a electron trac ks. In th e data the final corr e c ting number of high mass for the l~w mass mate cut was: pairs, opposite charge, 250.2 ; d:::iuble n-egative, 10600; double positive, e xc ess of double negative pairs after implies The 7004 . the ratio of ne g ative to positive electron candidates vas 1.23±0.17$ Th i s cha=ge asymmetry caused a mu ch smaller asymmetry in the s am e to opposite rati~ as follo~s: Assuming no correlation, ~nd a negative to positive electron ratio of s ~me to opposite ratio i t was 1.021 wece {a 2 +b2} to 2@aeb. {-0.019 +0.036} ~ believed weights Yas were subtraction. section 5. P. 3., to The by to b, the In this case Since the same charge originate i ncreased a events in an uncorrelated way~ their a factor of L 021 before negative excess is analyzed in detail in - 159There was one more c orre latio n between r e al electr on s with c onservati on in difference between pairs. the each the and to charge Although th e sources of co mpto ns) interaction were totally introduced a ra t es fo r same and opposite c harge example, an positive and negative trac ks has only due other and vith the ha drons, cha rge hadronic the For co r rection, hadrons. (half-pairs uncorrela ted hadron necessar y event nine with three each opposite-charge and six sa me-c harg e pair co mbinations , all of vhi ch hav e a c hance of being mi si d e nt ified as electron pairs. The number of pairs of misidentified hadrons using the progra m measured pu nc h-t hrough r at e was estimate d a n d a Monte Carlo up for had ron tracks. misid entified hadron and on e rea l e lectron wer e completely {Pairs mad e remo ved by th e subtra ction of the sa me-ch a rge prediction of pairs~ } one !he was 42 opposite-ch arg e and 26 sa=e-char ge hadron p~irs. The 16 pair difference was the estimated beyond the c o rrection back~o und by sa me- charge subtractio n# and ~as subtr a cted fro m the opposite-c harge result as a function The of hadron pair calculation is described in Section 5.P. 4 . 5.F.3 -- Sources of Negative Elect rons :o mpton scattering, high electrons bea m we re in the energy knock-on investigated electrons, as and sources of -160negative elec trons. scattering was Of these possible sources, only Compton significant. with a momentum g r eat er than 750 MeV/c were much less likely to be produced A than Knock-on Compton electrons vith the same momentum. knoc k-o n of momentu greater produced a with by electrons pion than 750 MeV/c 750 MeV/c. be a momentum less than 4 .0 GeV/c. such pio~s were rare co mpare d t o photons which only cannot Electron conta minat ion needed have in the beam was esti mated as o ne per million, far too s mall to compete with the other processes. The foll owing paragraph describ~ a roug h estimate for Com pton el ectro ns relative of r ate to half pairs in order to show that Compton scattering ~as a pl ausible sourc e for the excess was negativ e calc ulated for electronso First~ the probability a photon conv ersion to produce a half p~irv i.e. for one t r ack not to be found and the other to momentum greater than 750 MeY/Co have a (Geometric acceptance and identification efficiency of C1 are assumed to have an e qual effect in the two processes.) Photons we re generated ~ccording to a momentum distribution calculated in :arlo study distribu tion of of pi-zero the momentum electron ~a s taken as flat(5). track was aonte production and decay. The fraction going into each The probability of losing a assumed t~ be 5% for momenta above 100 MeV/c and 50 % for those =orrect a below. (The assumed o n ly within a factor of two. efficiencies may be The goal was an order -161- of magnitude estimate only.) production cross section{6) ■ icrobarns for gave Multiplying a by the pair of 530 cross section a photon producing a half pair in hydrogen. Osing photons vith the same momentum distribution the section for a Compton ~bove 750 MeV/c was different Z• (Z+l) z scatter(7) 170 of the for pair production) (8) to in hydrogen aluminum with electron momentum microbarns. dependence Accounting frequently than for the processes (Z for Compton, and the ratio known Co mpton to occur scatters, times ten i.e. negative electrons vere 20% more frequent than positive, assuming dilution by hadrons. of the path of a photon exiting the target wall, half pairs are predicted 11:>re cross no This is in agreement with the measured ratio. 5.F.4 -- Two Hadron Punch-Through This section describes the estimation of misidentified hadron pairs using the punch-th:cough probabilities. The predicted punch-th:cough pairs, opposite charge minus electron pair subtracted from the the excess number of measured Cl excess of same, vas to give the produced by the background corrected result. The input hadron events came from trackfinding program starting a DST with coordinates generated according to tracks artificial measured chamber in the -162bubble chamber experiment (Bartke e t . al.) (9). 3200 events vere used. Pairs of tracks ~ere selected where one of the tracks ~ent through a sen sitive C1 cell and hit a shover counter and the other we nt throug h a se n sitive C1 cell. Ab o ut hal f a pair per e vent wa s used. Ther e were about charge sa me. The pair vas assign e d a weight pa i rs than 50% mo re oppo site equ al to the product ~f the punch -t hrough probabilities of the was tracks, traversed. wh ich 0~ 13 wa s depended on which ring of C1 us ed the pro bability of s h ower as cou nter punch-t hrough . rbe pair punch - through was determined in n~ minal described in sec ti on value s of C1 punch-through t~o c a s es . The 5.B.4 gave the nomi na l hadron pair punch-through. se=on d devi aticn uppe~ limit was case, t he two s tandard calc ulated by incre2sing the single ~~te according cell Ci In the pun c h-thr ou gh to the uncertainty due to the poss i bi lit y that the presence of a hadron punc h-t hrough in a cell increased over the cou nting rate for empty cells. increase was measured as co nsistent with zero having t he (The an U? pe r limit of 0~15% ~s described in section 5.Be4o) rhe expecte d number of punch-through pairs using the nomin~l v~lues for the C1 misidentification rate is 5.F.4 for opposite ~nd same c ha rge shown in fig. combinations. The 0 > G.J zG.J (-.; rD µ_, r-, [ ,] ex: 0 ID ..i:.:~ 14.~~.4 (l) ~;> ~ N 0 c_-,(0 1 2 3 4 5 G .....{, , __ 100 200 500 MASS 1v1eV/c 400 2 (300 "100 800 Figure 5.F.4: Mass of two misidentified hadrons predicted from bubble chzi:mb er data . 300 ➔~-7 x SAME CHARGE o OPPOSITE CHARGE 900 1000 ~ -=::z{- 8 .-----.---,---.....----,r-----r---.--....---.---.---r----r----,----.---,.--r----;----.--.....---.--. I 0) v,I I I-' -164difference, opposite minus same, is histogrammed in fi g . 5.F.5 for the no minal and the upper-limit :1 punch-t hrough. MeV/c 2 was of The total difference for masses above 140 16 in the nominal case~ and Q2 assuming the t wo standard deviation upp er limit. error probabilities The one standard was taken as half the difference: deviation The background was taken as 16±1 3 e vents. 5. F.5 -- Empty Target Backgro und The c uts on the position of the pair vertex were designed to remove all i nt eractions in hydrogen {and the matter mylar other windows). correcti on ~as taken as the ratio of than the liq uid empty target interaction rate The the for the 0.026 cma of mylar comp are d to the hydrogen~ The analysis of empty target runs ~ith an 0~~%~ integ~ated be am f lux equal to 2% of the normal runs, yielded two pairs, both less than 30 ~eV/c 2 • The null Lesult for high mass pairs implies a t~o standard devia tion ~p posite u pper ch a rge limit of with 150 mass high-mass pairs in the normal runs, a useless result. 5.f.6 -- Background Subtracted Result rhe final weighted Jpposite c harge electron pair.s, and pai~ sample ~as 250o2z19a2 176.3±16.7 same charge pairs. () > µ] z µ] E-< (/) 0... µ] P:': 0 LO ~ >(!) - '--.__ N -2 -1 0 1 2 3 4 5 6 7 8 0 200 J I 300 -x ~ I I X- I 500 2 -X-' MASS MeV/c 400 L I -x 7 I 7 -x 7 600 ~ L.- ~ 700 -¥ 7 900 ~ ~ -x...1 rX- 800 r~ NOMINAL PUNCH THROUGH RATE x MAXIMUM PUNCH THROUGH RATE ◊ L >(- i I Figure 5.F.5: Predicted hadron punch through, opposite minus same charge. 100 ~x~ u I r~ r x- 1000 - -x - I V, en I f-J -166Subtract ion of type one backgro~d (poorly resolved conversions) events. Subtraction left 24506±19.3 photon of s~ me-ch arge pairs ~eighted by 1.021 left 72.9±25.5. re movi ng of the same c h arge the hadron pairs in excess correc tion, the estimated number of with masses deviation direct above 140 Me V/cZ vas si gna l. The empty target After electron pairs 57±29 6 a two standard correction was negligible. rhe cor rected mass spectrum {not corrected for the 4.6 event expe c ted background due to poorly resolved low whos e mass distribution was unknown) mass p~ir sq is presen ted in fig~ 5.F.6. 5 .G -- Monte Carlo and Acceptance Corre ction Thi s s ection geomet ric describes acceptance, ije ntif icati~n the Monte Carle trackfinding losses~ and the calculation efficiency. loss du€ to of electron the beam ~ss oci ation cut, as a f un ction of the mass of the pair . The back gro und sub t racted ~ass spec t ru m was the corr ected for losses listed in table TS.Ge 1 . The normalized mass spectru m {di ffer ential for cross section pi ele ctro n pair+ anything) is presented. 5.G.1 -- Monte Carlo Track Generation minus + proton--> -167- 0 0 a) ~ 0 .,.., ..., t> d 1-, ....., -! ,.D :::J I iD 0 0 CD ~ :1 :1 0 .... . ~· (\] ~ 7 0 ~ 0 CY) 0 C\2 0 ....... '.:) ,>i 2 JI/',1.~' J 'i 0 0 ,,...; og ct:2 d S1N~L\3 0 ~ lo ,- .... -...,/ (\2 C0 I ! -168The I and P + distribution for pi minus+ proton--> rho zero anything at 16 GeV/c measured parameterized and used by Bartke electron momenta given iavariant mass were vith (Distri bution s according respect to center the was to generate moment~m values for an electron pair center of mass system(11)e mass, et. al. (10) to of In the center of generated corresponding t o a an isotropic 1+SINZ (0) mass dist rib ution. and i+cosz (0) 1J1ith direction vere momenttH investig ated also, for pairs with th e mass of the rho only.} The electron momenta Yere transformed t o the lab fra me, and used to calculate track trajectories with a startin g distribu ted af randomly along th e length :hamber in t ersections were generated at dropped the point the target. nearest wire, accordi ng to the measured cha mbe r efficiencies, and passed on to the standard trac kfin di ng progra~ ~ Only ev ents wher e both tracks ~ent. th rough equippe d Ci cells and o ne hit ~ shower co unte rg vere recons tructed . e ve nts The fractiJn of was called the geometric acceptance. 6% to 10% depending on the mass of the pair ~ such r anged fro m rt as sho ~rn in f r a cti on of fig • . 5.G .1.. Tha trac kfinding efficiency was taken these eve nt s r ec onstr uct ed . the pair (96% where both as the tracks succes sfully It ~as about 92% independent of t he mass of per tra c k) • The difference bet~een the reconstruc ted a nd the initial mass vas a measure of the mass resolution. Thi s is s hown for several masses in fig. 5.G.2. C) ~ ~ <G 0 t--' 0 ,__. z 0.00 0.05 ~ 0.10 r:il 0 n... i:£1 E-< 0.15 0.20 0 I I 200 I 400 600 MASS MeV/c 2 I I 800 I I 1000 Figure 5.G.1: Geometric acceptance from Monte Carlo. I I I I 1200 ~ I ~ I lO en r:n z :::> ~ m i::r.:I - O:'; 0 G:.. ~ ~ ~ O:'; 0.0 0 .1 0.2 0.3 0.4 0.5 + \ 0 II ~ I+ I 200 I+ 600 2 MASS MeV/c ~ n 800 A Figure 5.G.2: Pair mass resolution in Monte Carlo. 400 I 1000 + INDICATES GENERATED MASS 1200 I I t-' I -..J 0 - 17 1- 5.G.2 Elect r on Ident if i c ation and Beam Asso c i a tion Accep tan ce The Cere nkov and shower counter identification probabi l ities we r e c al c ulated f or each track in the Monte Carlo ac c ording to its mom e ntum and which cell(s) and shower coun ter it hit, usi ng t he 5.B.2 an d efficien c ies and 5 .B.3. probability a3s oc iati o n Th e distan c e cut and e of the track using 5.D.3. methods described in sections of passing t he beam was calculated from th e momentum the table described in sec tion It ~as a bout 93.5% per tra c k, almost ind e pendent of the mass of t he pair. These acc eptan c es co mbined were together to forra f~~ the pairs This is shc~n as a function of the ruas s of the pair& in overall of probabilitiess on the cc mbined acc eptan ce fig. Uncertainties on th e measured efficie n cies led to an 20% a a c ceptance probability for e lectron ide ntification a nd beam association 5. G. 3 multiplied uncertainty The uncer tainties shown in the figures are s tatistical only. Th e product of : ~rlo, all geo me tric, acceptances track determined by the fi nding, electron identification, and beam association is sho ~n as a f unc ti o n of mas s in 5 . G. 4 . of The 20 % syste matic uncertainty applies. the · rho, the Monte acceptance was 18% fig& At the mass l over for a 1+COS 2 ~~, µ..., ~ <t, 0 0 ,._, t~'"" 0.1 8 0.20 0 <C .-'7' 0.21,.,, ~~ l ) 0 f - - '-- - · µJ u - --- f;::'. 0.24 C..iJ Q 0.26 0.28 I T 400 I I 600 MASS MeV/c 2 I I I 800 I 1000 I Figure 5 .G. J: Eleclro;-1 identification and beam a ssociation acceptance calcuhted in the Monte Carlo fo r differ ent m usses . 200 I 1200 j -I N I I ...... ......, µ., ~ u<G [-, 0 !-t .L., 0 .00 0.01 ~ 0.02 0 µ::J Q l-i.l E--D... 0.03 0.04 I 0 I - - iI I 40 0 MASS MeV/c 600 I 2 I 800 I I 1000 I Figure 5.G.-1: Accephmce clue to all los~;es calculated in Monte Carlo. 200 I I I 1200 LJ, ~ I IJ,j --..J I ...... -174- distribution, and 121 higher for a l+SINZ distribution. 5.G.3 -- Final Acceptance correction and Normalization rable 5.G.1 lists all the stages where data could have lDst along with the estimated fraction kept. depended mass. on t he Where the loss pair mass, the value shown is for the rho The loss due to match-point overflow is taken as fraction the of overflow events out of the total number of pion interactions, on hardware been the assumption that they produced the trigger and were not correlated vit h direct pairs. (rhey are believed to be electron field large lines, producing showers, numbers of Table T5.G.1: FRACTION OF SIGNAL PASSED Geometric Acceptance Mask Correlation Test Unpack Ok ay One or More Bea m Trac ks Ho Match Point Overflo w Trackfindi ng Efficie ncy Exactly One Beam Track Electron Identification Two Electron Cut Confidence Le ve l Cu t Matrix Flag c ut Beam Associ a tion Cut Low Mass Mate Cut Wt Ceiling Vertex Cuts 0.0830* 1.000 0.953 0.982 0.997 0.920* 0.918 0.253* 0.960 0.979 1.000 0.874• 0.916 1.000 -----------------------=0.0121 Overall Acceptance •=~ass Dependent -- Value Shown is for Rho ~ass following the hits the in -175chambers, and triggering in C1 and a shover counter.) This gave a factor of 0.9974, one minus the fraction of overflows per interacting pion. (If the overflows vere not responsible for the triggers, they disguised the fraction of direct electron pairs normally present at that stage of the selection, and the fraction lost of the direct pairs vas the fraction of overflows in the data presented trackfinding program, making the factor 0.63.) related losses, garbled data from the to Other the event detectors (fatal track, vere The factor shown for the low mass mate cut compensated only unpack), no beam track, and more than one beam assumed to be uncorrelated with the signal. for the event weight ceiling, most of the loss (it averaged 0.67) vas corrected by weighting the individual events. seen As in the table, the acceptance at the rho mass was about 1 in 80. The corrected mass is spectrum shown in fig. 5.G.5. 9.604•10 8 pions entered the target during the sensitive time of the trigger. attenuation With a 91 cm. target, along the target, and the correcting sensitivity was 3.JqJ events per nanobarn. The final cross section for "direct" mass D~litz, pairs etc. differential with included) cross cross section in the greater was section then 1.14±0.68 in rho-omega shown mass for 140 producing BeV/c 2 (eta microbarns. The in fig. 5.G.6. The region, 700 to 860 -176o - - ..---~-,-----.---.-,----,- ,-.-.--r----.~----,-----y,-r-.----r--r111117 l I I J °0 ..-< s ;::! h +> t.l r Q) P. f.l) 0 0 {/j co !/J I'S s ~ s ~ "O ..,..,Q.) u a:l ..,.,h ..0 ::s Lr:J 0 0 © -r• s .. I '.::l q ~ 0 h t\l () '---., I 0 0 'tj< Qj) ,..,c, 0 r,j >(j) ,.a ;E C; CCi U) U) <r. ~ ';j ~ ...., Q.) () !l.) ,... h 0 0 Q) tJ ;.:; CJ -<-' p., 0 (..l 0 () 0 C'1 c:r; l.0 0 ;__() Q.) a QI) -~• µ., I ! 0 0 0 (') I I I I I 0 I I 0 0 0 0 0 0 (\ ) ....; 0 0 0 0 ..--l 2 JI / f\8 :!ii ·~ 09 d .!:1 d - N I '\ ...,. r-f L 0 0 0 - S.LN3i\3 - 177o r-7;--i-----,--.-r-r-,---,---r--r--.-.-.--r-;~-----,----,--.~..,.,--,---,----.-....-------,--~. - - ~ 0 , O 1 ....-i 0 0 s co L 'u '-...._ t, "Cl Cl 0 :;; t.) (l) Cl) :,~ rJ) 0 0 © () '------ > <:i) ~ 0 -~ + 0 h I c.;, C (\} 0 ?<: U) if) <t; ~ -+ w :;; + C: ~ c:; C /\ t.;.- 4 ~1 "O ,...d (',.) N ...... ....., d P, + ,.. s "'0 ~ 0 h z ''" (,') D ,r., f:: ::J .~P j.":;;., -1 I ! I I I 0 0 I.() r-- 0 !.() I.() C-J r-i 0 0 z~ / f, 8 J\I 09 0 J I I I 0 0 0 • ~t--p~ '.,.., --l rl, - - TI.Ip .18d s"'U 1 j l 0 L() (1.1 0 0 0 I I I .Dp ;D -178~eV/c 2 • was 0.16±0. 13 microbarns. Th e normalization of the f i nal level sample is uncertain to a of 20 % due to electron identificat i on efficiencyg 10% due to the uncertainty of the center of mass decay angular distri bution~ and possibly 30% or more due to the difference between the real X distribution of the events and that of the rho whi ch was assumed The I · distribution in acceptance calculation. is in the un certain ty treated the next chapte r. References for c hapter 5 1. Radi.3.tio;:i length in alu mitrnm and hydrogen taken fr.c>~ Review of Pa rtic le Properties, Phys . Letters 508, N~ . 1, (1 974) whi ch was c-evi sed from CE RH NP I nterna l R&p o r:t 7 Li-1 • Pair production c~nst itu ting 7/9 of the "radiation le ng thn inter act ion cross section is r,2ported in E .. B. Rossi, High Energy Particles,, Prentice-Hall, (1 952). 2. Branchi ng ratio reported in Review of Particle Properties, op . cit. 1 as taken from Budagov et. al3, J .E.T.P. 11, 755 121, 275 {1961}. 3. and Saruios et. al.,, Phys . Bev. N. Kroll and w. W~da w Ic tern1 l Pair Pro duction Associated with the E~ission of High-Energy Gamma R~ys, Phys. 4. (1960) Rev. 98, 1355 (1 955) . Y.-s. Ts~i, Pair Produc:ion and Bremsstrahlung of Ch a rged Leptons, Rev . Hod. Phy s.., 46 , No. 5. The distribution is nearly flat. op. cit. 6. Rossi, op. cit. 4 {197 4 ),. It i s given in Rossi~ -179- D. Bjorken ands. D. Drell, Belativistic Quantum Mechanics, 5 c Graw - Hill (1964) o 7. Jo 8. Rossi, op. cit. 9. Bart ke et. al., Nu cl. Phy s. B107, 93 {1976}. 10. Ibid. 11. D. L. Blockus, Johns Hopkins Dniversity, Ph.D . thesis, to be published. - 180CHAPTER 6 -- Results 6.A -- Introduction Tkis chapter discusses the direct electron the previou s chapter. The pair res ult of limited region of geometrical ~=ce ptance is discussed first s ince assumptions were made in order to extr a~olate beyond that region. the known The signal due cross sec tion for the decay of the rho and omega aesons deca ying i s used to check the results~ rho/omega to cross section is c onsistent measured The with both the prediction and ~ith zero. The r e sults of thi s experi ment are compared with exp eriments at si mila r energies: a tvo bubble other chamber exp eriment with full solid angle electron detection: hybrid strea mer ch amber direction~ The is co ns istent with section masses above 200 ~ev;c2 . a experiment vhich identified muons ~nly in the forward cc css and overall these lepton pair experiments for There is evidence for a source of pairs othec than vector mesons and Dalitz decays. The signal between the pi-zero and rho ma ss is c ompared with the prediction of a pion annihilation model f eat~res of several earlier models. c onsistent with the model above 200 ~eV/c 2 • which The includes results The data are point for 150 - 200 MeV/c 2 is not in agreement with the steep rise -181predicted by the model. 6.B -- Geometrical Acceptance and Rho-Omega Signal In order to obtain the cross section in presented the previous chapter, it vas necessary to estimate the number of p~irs outside the region of acceptance. The final cross section therefore is quite model dependent. factor correction vas computed in the Monte Carlo assuming the Feynman X and Pe distributions measured for rho with The and omega mesons, the result that pairs originating as these mesons were correctly normalized. mechanism is not At lower masses, where the production known, the calculated different from the actual cross section. result could be For the particular model discussed below, a larger fraction of these pairs were produced in the region of acceptance than those due to the rhos, in which case the cross section was overestimated by a factor of 1.34. In this section, the observed distributions Carlo predictions is Monte Since the signal in the rho is weat, a mass plot showing a strong pi-zero signal presented electron the for production and acceptance of the rho in I and Pt are presented. region and as evidence identification the final sample of that worked. events is the reconstruction and Finally, the smallness of reconciled predicted before the experiment vas run. vith the size -182- The X and P dis t ributions we re ge nerat ed in the ~onte Carlo according to the rho-zer os produced interactio ns ( l ) . distributed measure ment s of Bartke et. al. for inclusively in 16 GeV/c pi-m inus proton have {Omegas similar ly in been the to obse rved be one co nstraint ch annels and ~ssumed to have the sa me be havior in th e multi-neutral final states (2}.) since Pt.•) The Pt. was generated first~ ,an d then the x, wa s distributed differently in diffe re nt ranges of X At lo ~ Pt, X peaked near one, centered aro und zero . ~hereas :...t high Pi. it vas The projected distributions are shown for X and Pt. in figs. 6.B.1 and 6.B.2 res pectively. The geo metric al acc e ptance was daughter The requiring acceptance of X and P~ in table T6.B. bet ween X and Pi. , L, e 4 prod uct by both tracks to pass through the equippe.d part of C1 and ~ne to hit a shower co unter. bins defined is shown There was lit tle interplay the acceptance ldas approximately of f unctions of each separately. the The acceptance as a fu ncti on of X integrated over Pl, and as a function of - integrated over X is in figsa 6~B .3 and P~ 6 ~B.4 res pe cti vel y. As fall in g to zero below fu nctio n it is a window cen tered at X=0 .27 and about of X a shown in function of P, it is almost uniform, 80 and 0.2 in width$ These acceptance using ~ith pairs above 1400 functions the mass of the rho. ~eV/c. vere c alculated Th ~ acceptance was fai rly constant as a functio n of mass 1 but at lo~ masses the -1 8 30 ,,-I 0 ,_, l() h (1'5 u ll.l ..,.., 0 !'.::1 0 ~ s 0 h in -~ ...... 0 Q 0 0 0 µ-, >< ......., ::l ~ ;.., .,_, ~:2 'tJ :x:: .. ....... co ::::, L·' Q) ! 1_() H ,.., ,., 0 µ:~ I 0 0 0 CX) 0 0 C\2 0 0 (0 t.I38PnN o a.u •rS i - 184LO ..--1 0 ~ (l:S u ...,tP ,.., ;-1 0 0 ~ ...-1 80 ......t.. {.,1 ~Q. 0 ..., ;J..; !=: . ;:, .0..., ,.D ·c ..., r.? .,.., 'O .., i:,., L.0 0 (\2 a:i CD ~ ~ QD ii: I 0 0 0 ..--1 0 0 co 0 0 © I I 0 0 "tj" H38:V .ON 0 0 C\l ' C:.:) 0 0 <G u (_) ~-il 0... ~ <G 7 ......., w u L 0.0 - 1.0 0.1 0.2 0.3 0.4 I I I I I I F X0 .0 -w. A'- l I w:w lw: w:::s I I 1 • I I Figure 6 .B.3: Acceptance as a function of Feynman X -0 .5 I I 1 1 1 1 fT •• 1..0 I I-' I u, 00 <t'. u u p_. r:::.:l w u z < £---4 i-- 0.0 0.00 0.05 - - 0.10 I - 0.15 I I Pt (GeV /c) 1 1 1.0 1I ,, / ' Vi.gDre G.B.-1: /\ccepta n ce as a fu.nction of l ro.nseverse JTiornentum 0 ,vr, 1 I I I 1.5 - I I 00 01 f--1 - 187 - p~ ir nee ded more Pt for a n electron to hit a s ho v e r c o un tero ? b e X an d P~ di s tributions of the obs er ved ev ent s ~ith g r ea ter than 140 Me V/c 2 are s hown in f igs. 6.B .5 an d 6.B o6 . s a me ch a r g e pai r s we r e subtrac t ed v nn:l.e fo r Pc The mass but no c or rection cha rg e -combi n a t i c,n c o rre lat ed ha dron punc h- t hro ugh e distri b ution is quite c o n s i s t e nt wi th t ha t o f th e Ta ble T6$B : AC CEPTANCE AS A FUN CTI ON OP Xf AN D Pt I N PER CENT -.1 Pt t was <-- Xf - -> o. o •1 •2 . 3 I I I I .4 I .5 I "6 .7 I o. ♦--- - - + -- ---+- --- -+ - ---- ♦ - -- - +- - - - - +-- ---+-- -- - + I 0±8 I 0 ±6 111±8 I 17±8 i 13±7 I 3 ±4 I 0±3 I 0±2 I . 1 +---- -+- - --- +- - - --+--- -- + - ---- +---- -+ ----- +----- + I 0±3 11 4± 5 ! 2 1±5 133:c6 ! 33±6 l 5±3 I 0±1 ! 0±1 i .2 ♦ --- -- ♦ - - - - - ♦ ---- - +----- + ---- + ---- -+ - - - - - +- ---- + .3 +- -- - -+ - ---- +-- --- + - -- - - + ---- • ----- ♦ --- -- ♦-----♦ t 0± 2 t 3± 2 i 3 5±5 !3 4 ±5 i 19±4 I 4±2 I 0 ± 1 I 0±1 l I O± 2 l 4 ± 2 I 2 5 ± 5 l 3 3 ± 5 i 29 ± 5 l 1 ± 1 I O± i ! O± 1 i • 4 +- -- - -+----- -i ----- - +--- - - +- -- -+ - ----+--·- - -+ --- --+ I 0±1 l 2±1 123±4 129±5 ! 21 ± 4 1 2± 2 I 1±1 i Ot 2 I . 5 +- --- - +----- +----- +- - - - - + - - - - +---- - +-----+ --- - - + I 0 ± 2 I 6 ±2 t 17±4 ! 3 4±6 ! 21±4 f 6:t3 I 0 ± 2 i 0±3 ! .6 ♦-- --- ♦ -- - - - +-- - -- + - -- -- + -- --+- ---- ~ -- --- ~ -- - - - + .1 ♦ -- ---+----- + - - --- + - - ---+-- --+ -- --- + --- -- + - --- -+ .8 ♦ - - - -- + ---- -+-----+ - --- - + - - - - + --- - -+----- + ----- + I O± 3 I I 0± 3 t 1 ± 2 l 2 0 ± 6 I 3 9 :t 7 I 26 ± 6 ! 6 ± 3 I O± 2 1 0 ± 4 I 0± 2 118±6 t46±9 1 18±6 111±51 2±31 0±8 t I 0±3 I 0±2 t '12± 5 125 ±8 f 15±10 ! 3±4 I 0± 6 i 0 ±: 21 ! . 9 +- ---- +- - - - - +-- - - - +- - - -- + - --- +-----+-----+-----+ t 0 ± 5 I 0 ±3 I 7 ± 5 120 ±7 1 28± 12 118 ± 13 ! 0± 131 -- - f 1 .0 +-- --- +--- - - +--- - -+ -- ---~--- - +----- +--- -- +- - ---+ I 0±6 t 0 ± 4 I 5 ± 5 !3 5±12 1 16 ±15! - - - t - - - t --- I 1 . 1 +-- ---+- - --- +- ---- +-- - ·- -+ - - - - +- ---- +----·- +- ----+ I 0 ±1 5\ 0±6 f 17±12110±11! - - - t - - - I --- I -- - t 1 .2 ♦ - - - - - +----- +- ---- + ----- +-- - - - + ---- -+ - ---- + - - --- + I 0 ± 2 1t 0±10! 0±23! - - - l - -- 1 -- - I - - - 1 --- i 1 .3 +- ---- + - - ~--+---- -+ -----+-- - -+ -----+-----+ ----- + -188co 0 c-. 0 ,-.... '"O CD 0 ...0, C) (ti .....,I-< ..a LC) ::::i l'/.l 0 ll) Q.l) h ltl ~ :~ 0 !l) (.) s (!;, (,) .._.,, C0 !,') .,..;, 0 ~ i.) :;, ll) j:::., C\2 0 >< ...... 0 ,.., ;..., .... ...., 0 ::; ,-! ~r:.• 0 ..... . .,.., I[) -c; X II I 0 ,...., .._, t_() co 0 1; ..-; 0 i C\2 0 I I 0 co I I 0 CD 0 -.:::r' I I I 0 C\2 H38J\JnN i ! CY) I 00 0 C\2 I I ·;3 .so C.-, -189I L0 I ,--., 'v n) ..,.., v ;.j ....,l-< ,.0 ;::l ti} tl) 0 r:i.o r-, (1'.S ~ .i:: t) !1) 8ltl r,, '-"' "' -,.) ...... IJ-i i:: Q) :> Q.) .....0 -...,..., 0 .,., B ·c LO 0 . ~) ~1 ;,a ..., 0-, ,:) C'..) cc 0 /., I ~ I 0 CD I I 0 0 N I c,_o r;.: I I I :;:; I I I 0 C\1 I \F;rn ~~-rn 0.1.....1.tl H N 1i ! 1 -190r~~- Tha X distributi~n snows the gcaater cross possibili ty of a much section for pairs vith X b~tween 0.1 and 0.2 The s tatistical accuracy of the data does tlnn O. 2 and O. 4. n~t however allo~ a definitive statement. 4.4t3.6 events occurred in the r ho-omega re gion, 700 to MaV/cz, after bac kground c~crespo uding to these subtraction. events vas Th e cross section 0.1ft0.13 Although correctl y noc~~lized in ter ms of signal uncer ta in is by 20 1 due to x microbarn. .a nd polarizati~n uncertai nty of the vector me sons . cc~ss section was measured as l'he electron to the and the ratio exclusively into branching o. oinito. oosi is pairs of t he The rho total millibarns 4.8±0e4 tnis Pt, electron identificati~n efficiency and 10% mo re due possible 860 and 0.076%±0.017 % ras pective ly{4 ). All the o~ega a nd 63% of t he r ho s ig~al ~re e xpected to occur Mith in expectatio n of 0 . 43 ±0. 09 o.i1 ± 0 .1 9 , the mass range. microbarns . The Th i3 yields an di ffer enc e was i.4 st~ nd ard deviat ions frog t~e pred i ctio n@ !ni 1.2 st an d a rd deviati ons fr om ze ro. ~ince the observ a tion ~f direct rho and o mega electron pai rs Wis questio nableq the fu nctioning of the reco nst ruction ~nd electron by identi fi c ation re=o nst ru cting belo w 50 was verified pi-zeros from two photons which ~ece seen as c~n version pai r s. m~ss procedures Each oppositely charged track MeV /c Z, in Yhi c h at pair .ith least one track ~as -191- ilentif ied as an electron, was ~terpreted as a photon whose ~osentum wa s the sum Jf the momenta of the tracks vertex position. The pn,tons is sho~n in at their invariant mass of all pairs of s uch fig. 6.B.7. Pi-zeros ve re clearly present near the correct mass of 135 aeV/cZ~ A sig nal of 3500 pairs ~ith negligible hadron H!S expected wh en do~ n ti me of 501) . the e Kperiment was plLnned, The fluK was reduced a a=~eptance wa s do~n dJ ~nstream shower a by counter of factor used . Electron identificati~n tcacks and five. Fi na lly, the measured cross sectlon because the count e r were not inefficiency for The anticipated o hadron wa s the t wo making the rate for pairs ~ith iie ntified track. a thousand ooe five times vas poncb- ~hr~ugh un1erestimated by a factor of ten in C1 ane inticipated two the low mass ~ate ~t yielded another factor of than cou nters, of The geo me t rica.l t~o Cerenkov 3Dd (assuming a factor of spark cha nb er latency problems. b3=ause punch - through doubly in the show er ti mes highe~ th~ D one and As a result~ the punch-t hrougn si ngly caused a ba=kgcou nd subtraction of about 201& 6.: -- c o~pa rison with Other Experiments rwo other experiments production by pi-minus have near measured 16 direct GeV/c: lep ton pa ir a bubble ch~mber 0 µ::j :> (I) i::1 ...> rn 0-i S---4 (I) ~ 0 (I) ":g> N 0 10 20 30 40 50 60 70 80 90 100 110 120 0 100 500 2 Mass (MeV/c ) 400 600 700 Figure 6.B.7: Invariant mass of two photons 300 II I I I II I I I II I I I II I I I II I 200 j "lI'~ 800 900 1000 I I N t.J tO -193experiaent(S) with which detected electron pairs at all angles a final sensitivity one ninth that of this experiaent; and a hybrid streamer chamber experiment(6} vh ich identified auon pairs in the forward direction wit h a f .i nal sensitivity twenty-four times that of this experi aent. Both these experiments ran also with an equal amount of pi-plus beam. In the bubble chamber, identified, using production The data in 90% of t he shover methods plates installed i n.sid e the chamber. of pairs identified. (No where correction efficiency contamination Dalitz pi-zero i nc luding both variation of identification from vere electrons several metal consisted positively about e lectrons was with were made for the angle.) The deca ys vas believed to occur completely below 130 MeV/cZ in mass e In the mass region between 130 and 1000 were observed pi-plus. where in pi- minus interact ion s an d 22 events only 7 in ~ost of this difference occurred below 250 ~eV/c2, there were 15 in pi-minus and 3 i n pi - plus. m3ss enhancement was not seen in the three MeV/c2, of the LASS d ata. This low All but lo w mass events occurred with an X less than 0.15, below the LASS region of accept a nce . The results are compared in three T6.C. region The enhancement. 140-300 Me V/c2 mass reg ions c on ta in s The 300-700 region contai ns the in table the lov mass rest of the -194- non-resonant cross section~ and the 700-860 region dominated by decays of the r ho and omega vector mesons . combined rho and omega muons). (The c hamber higher data are presented both as and a s wo uld pass th e LASS acceptance (normalized accordin g t o the rho). change some~hat o bserved c ross section was 160±130 nb.) The pi-minas bubble observed, The cross sect ion for making electron pairs is expected to be 430 nanobarns (and f or is the cross This acceptance restriction did not s e c tion; however, it re mov ed most of the events so that the disagreement was no l onger statistically significant. Table T6,C: CO MPARISON OP EXPERI~ENTS Cro ss sec ti ons in nanobarns. Mass Range (~eV/c2} I 140- 300 ! 300-700 t 700- 860 t - +-------- +--------- +--------~·+ -+-- ------ +--------- +---------+ This experiment (L ASS) I 40±540 a ccept. corr. assuming rho I l i1 00±400t 360t130 i ! I i - + - - - - ----+---·- - - - - - + - - - - - - - + Bubble chamber (pi-plus) 490 (2) 490 (2) 250 ( 1} I j Bubble cha mbe r {pi-minus} Bubble chamber (pi-minus) passi ng LASS a c ceptance -+--------+---------+---------+ 1 3400 (16) t 1300 (6) 0 f i i 3500 (3) f I -•--------+--------- +--------+ 0 -+--------•---------+---------+ Strea mer chamber (pi-minus) I 124±14 I t I accept. corr. assuming rho 620 {4) 445±27 I 250±20 I -~ ---------+- -------- +- - ------+ Numb ars i n parentheses indicate the number of events. (Af ter ~eighting by acceptance, events do not contribute equall y, and the statistical uncertainty corresponds to the number of events in parer.theses.) -195- The auth3rs of the bubble c hamber paper estimated the p~ssible contamination fro~ the Dalitz decays of the eta and omega mesons as less than 750 nanobarns in the 140 -3 00 region and less than 100 nanobarns in rhis estimate is applicable to the the 300-700 LASS region. data with the c~ution that Dalitz pairs occurred at lover X than the rho ind were therefore overestimated. In the strea mer chamber experima!t, both muons were required to pass through ten interaction lengths of absorber . The minimum muon momentura was therefore 2.4 GeV/c~ ~hich limited the acceptance to greater than 0.3 for the X of the pair. The acceptance was fairly uniform above 0.3 in X and at masseso The pi-plus and pi-minus mass distributions vere si milar in shape and the pi-plus t~o-thirds all cross section as large ~s pi -ninus overall~ was about The rho and omega cont ribution was about one-fourth of the total and was about hilf as large as expected from the X dis tribut ion, cross section and branching ratio of the rho and omega. The regions of acceptance of LASS ch amber overlapped only slightly. eve nts in t he rho region in and of the stre! mer The larger proportion of the streamer chamber is consistent ~ith the expectation that the rho and omega are produced at higher X than t h e non-resonant pairs and Dalitz decays. The streamer cha mber cross sections in table T6.c -196- h~ve been normalized assumi ng that 55% of the cross section h~s X greater than 0.3 as is the case f or the rho. 6.D -- A Mo del for lepton Pair Production In th is section, the pion Duong-va n, and model annir~lation because it is nu merically of Gold~an, Blankenbecler{~ is described and evaluated for the conditio ns of this ezperiment. section of the right e~plicit The model uas c hosen and pre d icts cr oss a and because it involves magnitude, interes ting physics. The orig inal and most basic model for lepton pair production by the i ntera c tion of constituents of collid ing the Drell-Yan model(BJ which treats the hadrons case of is the innihilation of a free quark and anti-quark co~tained in the incoming hadrons. The weakness of the t~e the treatment of constituent Drell-Yan quarks a3 neglect ing the interaction between the bicding the processs electro magnetic annihilation processes that are ignored are: model is free, i.e~ forces and I mportant hadronic interact :o n of the annihilating quarks, hadronic bre~sstrahlung, large raomentu~ tcansfer with interactions spectator quarks ~ and 1nnihilations involvi~g quarks produced in the collision (9}Q for diagrams 6.D.1 . The predicted cross section is too lo~ by up to orders of ~agnitude, these and processes too are shovn in fi g~ Pey n~an two few pairs a re predicted at -197- ., P===~~~~~ p ~~~~:()~ a} Free quark annihilation b) Pion-pion annihilation Drell-Yan model GDB model p ======~~~ p ============(£ c} Hadronic interaction of annihilating quarks d) Annihilation of quarks produced in the interaction e) Electromagnetic bremsstrahlung by interacting quark f) hadronic bremsstrahlung by annih ilating quark 11' - e+ e p 11' - - e+ p e - g) Large momentum transfer interactions with spectat or quarks Figure 8 .D. l: Feynman diagrams for lepton pair production -198- l~rge transverse momenta. The GOB model is an analog of the Drell-Tan model ~here meson rather than the h~drons annihilate. quark cons:titue.1 1ts of the incoming Meson - meson annih ilations are t~ be more important th an bare quar below 1 GeV/cZ. ~nnihilating mesons is included f~rm-factor measured in t he electron-p ositron anni hilations at ma sses alo ne colliding is considered . coupling strong-interaction of believed Charged pion an nihilation is the largest c3ntrib ution, and th at proce ss effect by it does not the t he pion 11::::i n g for ma& i on of pion pairs in beams.,, Th 4 .s for m-factor coup l e to t vo p!0n s} involving a third pa rticle, such as hadr o nic ind The between includes the effect of the rho reson an ce~ t a t not the (because the 2 o~s ga Effects bremsstrah l ung large momentum tra n sfer to spectators. := e not i n cl uded becau s e the fo rm - factor refers only to exc~~ -~ire pion pair production. The lepton p~ir cross .Pif(z) is t he section is g iven distribution functio n for a Dion of mo mentum fractio n z occurring in the i n c o~iog pion and is z <z 0 , for a pion of whe re zO =0• 4 an d Cii= 2 • 8 a Pp (y) ~~s given momentum fraction y occurring in the inco~i~g pro ton and by is -199given ye Pp {y) by y <y 0 , vhece y 0 =0.3 and Cp=0.4. not independe nt ly measured. =r oss c~ and Cp we re Th e const001 t s They were ad just ed to g ive the for the rho measured in the streaner c hamber section experi ment, and the correct ratio for X b.e,l ow an d above 0.3. The pion form-factor was parametriz -e,t; as P-wave a Breit-Wign er function measured at Or.say (10} : p:;11' f .t-1)/"-= ( where to tr1f If f /[ ( ~r-Y,'\-i.f" + y}i? rt·(11\ir h·,4 )1 (r /Po} t] Mf=0$775 GeV/cZ~ 17 =0,. 15 GeV/c, and p is the c.m. mom entum Ge'l/cZ !l F0 =5. 83, p0 =0., 36 of ::;;i ons .. (p tha ;ga,s taken as O for pair masses below two pion ~asses.) The differential function in fig. cross 60D.2. the full momentum ::.s The large co nt ribution nea r delta shown as a off co uld of mass be X of 1 co mes function for a pion constitu e nt hav i ng the the diffractiV"e pr:odi.::.ction. pealt du--/d.md! of X at the tvo pion thres hold atd at the rho mass from 3.llo t1ed section, shell smeac-ed incomi ng pion, co rresp o nd ing to I f the exchang~~d fc' : :::: r:i o mentum wer0. in the space-lii:.f, Jir ection~ this downwards observed X distribution of the rho t.o correspo nd (fig ~ 6,, s .. 1) . to the The high X ~omponeut is lcrger at highe~ pair mas ses~ rhe model was evaluated using the acceptance fo r experiment )6 that of (0. 36;aEXP {- (X-0. 27) 2 / (2~ (0~ l 1) 2 i t he LASS and using the streamer cha mber experiment { zero for X belo~ O. 3 and 30% for X greater than O. 3i ~ The results of the (.) ""O "b '"O 8 ""O :>< r:x.. I -1.0 Q I XX 1 .XX ~~ I I ! ! ' ' t::::W::4-aH I XX I - ·0 .8 -0 .6 - 0 .2 0. 0 XF 0.2 0 .4 0.6 Figur e 6.D.2: Predicted X distrution in GDB model -0.4 0.8 1.0 I :::t 0 XX I XX I N 0 ,.Q 2 3 ◊ m=280 MeV / c 2 2 0.. (I) ~ .. 0 (I) "> N 4 X m =780 MeV / c 5 - - - - - - - - - - - - - - - - - - - -- - - - - - ~ -201 model are compared with the strea~er chamber mea s ureme n t in fig. 6.D .3, an d th e LASS measu r ement in fig. 6.D.4. In the muon cas eg th e cross section has been multiplied by a factor {1 - (2 e~~ / m) 2 ) t o account for the lepton pair phas e space . The cur v e has been nor malized to the rho peak {excluding the omega), and th e agreement c or resp onding prediction normalizat io n a bove increased by a factor overesti mation of is good at lower masses. The fo r LASS used same Belo~ 9 the production is 70 0MeV/c 2 • of 1.34 to th e 11eV/cZ. The agreeme nt experimental upper 200 L4 eV /c 2 150 with the possibl e the measured result due to assuming an X d ist ribu tion lL·e that of the r ho. within reflect the well is The prediction l imit for ;na sses above 2 00 data point wa s not in The the predicted ste ep rise of the modeL 100 - 15 0 ~eV/c ~ data point contains pi- zero Dalitz decay An overall cross sect i on for pi-minus - proton going to events and is therefore not valid. 6.E -- S uillma ry electron pair with ma ss greater than 1QO MeV/c 2 in cl us i ve was meas ured as 1.1 ± OD7 microbarns# sing le rat io l epton to au pion of 2.6 co rresponding x 10 - 5 • a The cross sec tion in t h e ~ ho ma ss r egion is co ns istent bot h ~ith and the known cr oss section for rho and o mega . to zer o ~ :> Q) ~ ....> lfl 0 -1 (].> $.....1 LO C\l -2. (1) ...,,., "-> 0 C'J 200 0 10 300 600 2 Mass (MeV / c ) 500 700 800 A \ll Curve = predicted Figure G.D.3: Prediction for :.:;treamer chamb er - GDB m odel 400 2 0 - ~ ~ l fl 30 •- · 40·- Hist. --= measured 900 50 , - - - - - . - - - - - r - - - - r - - - - - - r - - - - - , - - - - - r - - - - - - r - -- 1000 ---, I I N 0 l'-l f.£1 ..-· ..__ z [£1 E-< U) et: Ll p.., 0 L0 (i) > :~ ~ () [\J ll _ . L_ _ J L _ _ . ! _ _ J __ _ - ½ . . . <$'~ f . GDB model 800 1000 ~ - - - L - - - ' - - - ' - - - ' - - ' - - - - ' - - . . , _ __ , 400 600 2 MASS MeV/c ___L_ T = m.easured Curve = predicted o Figure 6.D .4: Prcdicl:ion for l.J'.\SS experiment - _ L_ ---- -30L _ _ j_ _L_._J__ _j__~ _ _ ~300 0 - 20 - -10 ,___ 0 10 ,__ 20 30 ,_ 40 I I \J-1 N 0 -2 04 The signal in the 200 - 700 MeV/c 2 mass range 11 not due to entirely Dalitz decay s a nd vector mesons~ is consistent ~itb the two e xperiments of s imilar co mplements the strea me r This energies~ chamber b y experiment observi ng a different region of I and masses below the tvo muon threshold 4 It bas be tter statistics than the bubble chamber in limited reg ion of experiment x. The lowest mass data point is net co nsistent vith the rise at low mass predicted by the model. obser ved in the bubble chamber in steep Such a ris e vas pi- minus but not pi-pl us. References for chapter 6 1• 2 • 3. 4. J. B 3. r t k. e et . J Ba r t G k a L , Nu c 1. Ph ys • B i O7 ., 9 3 {1 9 7 6 } • e et . a 1 • s N uc L P h ys • B 1 rn ~ 3 6 O ( 1 9 7 7 ) • Ibi d. and op. ci t. , refs 1 and 2. Branching ratio of rho and omega i nto ele ctr ons f rom Reviev of Pa r t icle P_oper ties , Phys. Lett. SOB~ No . 6. a 1 (1974) and refe:.:ences thereina R. E. Cassell, Ph.D. Th esis 6 Vanderb ilt Universit1v {197 9) , also K. Bunnell et . al.# PRL 40, 136 7. T. Gold u, an et. al., 8. S. Drell aP-d T.-Me Yan, PRL 25, 316 (i".; 70 ). P h y s ~ Rev . D20,, 619 (1978). (1970). in -2059. For example: J. Bjorlcen and H. ieisberg, P.R. D13, 1405 (1976); c. Sachrajda and R. Blankenbecler, P.R. D12. (1975); N. Cragie and D. Schildknecht, ( 1977) • Nucl. 3642 Phys. B118, 311 10. D. Benaksas et. al., Phys. Lett. 39B, 289 (1922). -206- lPPESDIX 1 -- Cyliadrical Spark Chamber System 11.A -- Introduction rhe layout of the solenoid and its physical and detectors dictated the electrical characteristics of the cylindrical sp1rk chamber package: good resolution of the curvature l1tge good detection efficiency for several angle tracks; tracks in a small area; low mass minimize scattering; and compact solenoid able to operate in the w1s necessary in the dovnbeam electronics field. inside the resolution in the PHI direction than in the z direction used to calculate whereas only a slope was calculated from the curvature to Better since the PHI angles vere The end of a curvature z measurements. vas to be measured for tracks with Pl in the r1nga from about 100 to 1500 ~eV/c. 1 maximum excursion of only 30 100 MeV/c Pl tracks had cm., while 1.5 GeV/c Pt tr~cks deviated only 1.5 cm. from a straight line inside the 50 cm. radius of the package. The following sections describe how the constraints were met and the problems It is assumed that that the description in chapter 2 section P has been read. 11.B -- Physical Description r~e amount of matter past the beginning of minimized by utilizing the ~rength the target vas of styrofoam and of -207c ylindri c al tubes of 0.125 mm. diae The cylindric~l co ncentri c mylar ring s at each ende cbi mber inside fit cyl inde~$ S~ me occurred These sheets ~ere rolled into bat wee u rings at each end. Jf spark bet~een co pper c lad - alu~inum ~ires glued onto th in sheet s of mylar . tQbe s. The mylar. shape cylindrical was maintained by cl amp ing The cyliader package ~as a s et c ylinders, between conc ent r ic s upp or t Th e t arget and c ylindrical the hole proportional formed by the inner gas - bag The upbeam su pp~rt rings ~ere 0 ~25 inch luci te. downb eam r ings ue re styrofoam wrapped with Saran ir~p. ot hers were hollow lucite. A tubular aluminum fr ame surro unded the pac kag ef supporting each end hy the outer most rin g. Tbe mylar tubes were quite ci gid once the ends were fixed; the nacrou ones ~ere ~ble to support many ti mes their ~eight hung from the end e The outer onese vhose diameters were c~ ~parable to the len g th, had a tendency to buck le ; and be t) thic ke r~ A problem due to the glued se~~s requir e d much work and was never co mp letely ware overl appe d ep~zy. l~ss ~~ th~ 0.25 near the mi ddle of the t ube ~ It fi eld vas operating voltage. edg es inches and glued with Scotchcast 221 a fla ~ spot and a variation in the gap electr i c The eliminated~ T~e extra ri g :di ty of the seam caused curvature had st ~ength theref~re was not difficult a region of This res ul t ed spac ing so th at unifo ~m over th e t o find a suitab le -208- l first set of chambers {the p~irs) coul d inner 3 of the 5 eventual be made to operate only in PHI-bands, roughly b~lf of t he area would not operate at voltages that did c~nse spontaneous breakdowns elsewhere. w~s made stiffer with thicker mylar. not A new, final, set This set (except for one cyli nder) was made reasonably efficient by using a large a~ount of alcohol in the gas to ~iden the operating voltage pl~teau. Two type s of chambers made up the package. In the first type the wires vere glued onto the mylar at an angle so that tne PHI angle cha nged with z. t~a gap pitched left The ~ires on the inside of with incce3sing The PHI difference [countercloctvise distance}, these outside pitched eight. measu red at the upbeam end gave t he z position of the sp~rk. The wires in the s9c o nd type ~ent parallel to the axis~ only the ou ter ~ires were read out. The sparks in the 31ial c him bers vere used to eliminate ghosts in the pitched vire unread inner mat ch-po ints. I~ the original surface inner three ch am bers s~eet . sucfaces the of the azial ~ire c harn~cs was the 0.5 mil aluminum l1yer of aluminized nylaro ~ire chambers, were glued The close proximity electrically The oater wires of the pitched ~n the inside of the same mylar necessitated (vhich connecting thes e ma de the high volta ge system -2 09- more co mpli c ated) o For the final inner cha mber pairs_ the a~wnbea m s uppor t ri ngs eere c hanged from styrofoa m to hollo~ lu~it e. Th e unread pla ne vas c hang ed to glued v ires (the alum inum developed visible pits), ve~e enlar ged {rhe outer to t &o and outer cha mb e~s sepa r at e th e previously shared surf~ces4 ch amber pair s wer e separat i~n al though using wires ~lu mi nu Bf so these surface s elec tric al ly the that co nne cted.) built without the on beth sides instead of still had to be Th e separation allowed the seams tJ be st ~gge re d so th!t the trouble areas did not coi nci de in PHI . s inc e the axi1l c hamb ers we re then too far fro m t he pitc hBd c hambers to use for c orroboration of match-points, the y were treated a s sep a rate chambers. r~ble TA1.B lis ts the ch amber surfa c es . The spack gap was 1 c m. in all cha mbers o V3 S wire vas The !ctive len gth 91 e ra ~ Each attac hed to an individual readouc c i rcu it . rhese were loc ated on a p ri nted ci~cuit bo~rd bent in th8 s hap e of a hoop. The p.c. bo2rds s upport ri ng s the on were glued outer cha mber s. between the upb ea m Tho se on t he inner c h~ mb er s wer e plugged into a connector so that piec~s ba ce moved for repai~ . could !7en so, t he six inch space betwe en the ch aaber and the steel plate to ~hi c h it wa s bolted was fille d so co mpletely that replacing a diode would have t~ken h~urs. ia No s u c h re pai rs Ne re done except to disconnect wi~es small regions of tvo c hambers vbere breakdown int er fered with the operation of the whole c hamber. -210- 'I'able TA1eB: CYLINDRI:AL SPARK CHAMBER P~Rl\tiETEBS CYL. radius name cm. t ::>f wires spacing mm . pitch 320 2. 2. 2. L 352 384 416 2. 15. 28 16.30 17.32 480 512 2. L 2. R 544 2. B3 18. 33 576 2. C1 C2 20. 37 21..39 22.41 23.43 640 672 704 736 2. 2. 2. 2. L R 2. L pol- side arity of gap + inner outer inner gas bag Al 12 13 B1 B2 C3 10. 19 11. 20 12.22 13. 24 R + inner outer inner + + outer inner outer inner + outer + inner outer int er ii! ediate gas bag D1 D2 30c 56 31. 58 31.58 03 32.59 E1 E2 56.02 57.04 57.04 58.06 E3 outer gas bag 960 2., 992 R -------solid- -----1024 2. 880 896 Q.. l 4. R - ------solid--- - --912 4. + + + + i::ner outer inner outer inner outer inner outer -211- rae high voltage pulse was hr.ought in on RG58 coax cable, wb~se conductors attached to a bus on the p.c~ boa rds (or to ! copper strip in t h e case of the unread surface}. c~bles supplied dep endin g each cha~ber on its size. in parallel, A lo~ voltage signal indicating the meter 8 (spectra-strip), using wires weighed t wo hundred upbeam end H~s o ne conductor pounds). c ables per spark vire (the Gas entered on It left through the outer cyli nder and the outer gas bag. led through the entire volume by ~ires the at punching alter n~t ing cylin ders. An intermediate gas bag vas outlet the sa me end as the inlet. on p.c. the through the lucite spacer betveen the inner gas the bat~e e n the ribbon unshielded b~g and the first spark cylinder. between of signals uere co nnected to a r a ck outside the by solenoid nu~ber the pre sence of a spark v1s formed at the far end These Several e nd s The gas small holes 4 • co nsecu~ive of added r ing to put t~ e Gas tubing, c oax and signal cabl es ~ere threaded through holes in the iron end induced by cap. In the origi na l ultraviolet spurio us inner chambers~ photoionization spark occurred ~ere sparks o bser ved nea r in an adjacent c hambe r. where a This was re medied i n the final inner c hambers by pa inting the back of the mylar sheets black. -212A description of the co ns tr uc tion and testing of the se ctions for original inner three c ham bers has be e n published(i} . A1 . C -- Electric al Description The syste m is divided the follo~ ing high voltage pulser sys tem; the cleari ng description: t he field system; vo ltag e into the chambers , capacitor diode circuitry ; the shift tb.e auto mati c electronics tester.. the s hi ft (ground) re gister and anode syste~s (high th e re gister attached sy stem : and 'l'he chamber circuitry an d asso cia ted v o ltage electric a lly s eparate . There levels vhic h acted iijfloating as with vith the cathode aJices pulsed ) vere a nu mber of volta ge for different groundsv~ These levels are given na mes to r educe references systems . betwe en t he syste ms . GHD refers to real lab ground~ rbe cl earing field was a co ntinuous field used t o sweep and coll ect The ions f rom ol d tr acks and previous sparks . p~larity was mad e opposite to the pulsed voltage so tha t the slo ~ beginning rise of the high voltage pulse 1nd did not cl ear out the fresh tracks. - ijOY (negative on the anode surface ), it - 250 V f~r one the spark pulse . ce anants vas vas c anceled Th e voltage was increase d to millise c ond one beginning millis e co nd after This reduced the time needed to cl ea r the of the spark~ increasing the allowable pulse rate. rt was dis cove r ed that a field above 80V/cill ~ ·,ould sust ain -213- the arc started by the spark. lt 40v. tracks remained cipable of producing sparks for 1 microsec. l "rectangular" high voltage pulse of about 4 kV was produced by switching coax cables charged to 8 kV across the ch~sbers with thyratron tubes. with terminating resistors to The chambers were equipped minimize reflections. The pulse vas delivered from the thyratrons through 50 ohm coax. The cathode wires were connected to the coax shields and the anode wires to the center conductor. The center conductors v~ich came from the thyratron cathodes were through connected an inductor to the clearing field voltage level, so that while the anode voltage averaged to the it could be pulsed for the spark. the c.f. voltage, The clearing voltage vas thereby applied through the HV cables from Since also the thyratrons. terminating resistors were also connected across the gap, series diodes were inserted to prevent shorting Figs. 11.c.1 through out the clearing voltage. them from Al.C.4 are schematic diagrams of the chamber electronics. The cathode wire read~ut system anode system is described was slightly more complex. c~pacitor diode readout system vas the . c1pacitor, R1, C1, and diode first. The The basis for the resistor diode and shown in fig. Al.C.4. The a1-c1 time constant (5 nsec.) was so short that the voltage a=ross C1 was R1 essentially tracked the chamber current. charged during the spark. D1 prevented it from discharging 11 - 79 LL.J LL LL w (_) - z (_) >- o' ..__,. 0 ..--... 0 r F i g u re Al. G. 3: 0 00 c:.. . 40 60 80 100 DEGREES 180 270 Efficiency as a function off f o r cy lind er s urface B3. 90 J720A.'i 360 7 I +:- I N I-' I l-·79 1 " 1 T (~.-.., ,,.._, 0 -r IOµH CFt' . I , G I @ ) 3720A I0 Substitutable Resistors for Adjusting Size of HV Pulse J HV Pulse to Chamb ers 6 or 8 per, 1 ., I 34 Total Vl~ r;- ~ J High voltage pulse generation s ystem: I I I charge storage cables , thyratron, voltage adjustment resistors. Figure A1.C . 2a : Cho rge Storage Cables 300 feet 6 or 8 per Thyratron ·- ,-7 -c' Volt age Supply BkV High ~-,/ _4.,...o,,,.,K_n @ c - -0 ,-.\J ,~,,;,' ~ l c:~l G2 G2 0 to 15.0. THYRATRON PULSE GENERATOR ( 5 toto I) I i L I watt I {34 total) Cf isol at ion diodes and pulse monitor. High voltage pulse ter'minator: H.V. t errn inator m I N f-' 3720A9 to p~lse monitor time s 3'0 (60 fl imped ance ) ·,___,,,__ _ _ q..r K ~~ -0-- T f...., i t_·i ·... i-· Fi gure A1. C. 2b: 10-79 chamber from 1.8 kn -217- I Ofl Anti R i ng ing CF to Chamber s 34 Cab le s CF Io Thyratrcns 5 Ca bles I / s1;t1.tched I (" by Tes ter I I I I I 5 Volt bias I I I .J_ y I I I VT 5 Volts 40 Amp~ VCAN I I Tester I GND 1, , - - - - - - - - - - - - 4 - - - - - - - - - - - - J . . - C.F. Vo l tage Genera tor + 5 Volt bia s 5 Volt s 80 Amps 0 ~ -- - - - - - - - --4- - - - - - - - - - --""- - l I - 19 Figure A1 . C. 3 : Low voltage s upp l ies: SV power to IC 1 s , b ias voltage , cle aring field , and tester . VCCAT G~JDCAT - -- Electf.'onics : .~lK ,,4w / / R4 47K I bil / l/4W 47K R ___ _ _ Jl __ /J,F T .0047 C4 4 I 1bit shift of register 74_165 HIii• GNDCAT - - - - VCCAT Ind ica tes bus ring on chamber electronics board connecting to al I circuits ,I 6 CARDS VBCAT ' ~ 10 114 ~ 1/4W PRlhlTED CIRCUIT fLF chamber-mounted, and di s criminator/ shift-reg ister cards. / -- CA BL E ELEC TR ONICS - .0047 _ h' 2 ,,_F_ _,,,,,______ lN.J070 'a -:2 G,aond Bos R;ng / /,. !/ 2 W R3 _ •• 004/ 1c4 VBAN VCAN of ~ t-----'Vv\,------...._--~ 7'1165 !OK sh ift l/4W register 1-- GNDAN 10047 µ.F // lc2 8m 47K l /2W / HV Side Ground Bus Ring ( to CF) CH AM BER k:"~ 1N30 70 L·--k]>-------R2 w.. Fi gu re A1. C. Lf: / / le, l/2W 101µ.F !~A r ID, i L [ l C1 Ca th ode Wfre CHAM BE R SPARK Anode Wire /// HV Pulse Bus Ring I 00 I-' I N -219through R1 afterwards. problems vith resistor wires, the The resistor and diode solved two system using limited the voltage that the spark so a capacitor only: The difference between adjacent did not spread. The diode prevented inductive ringing from discharging the capacitor. rhe rest of the circuitry converted the charge in C1 into voltage compatible with a the 74165 TTL integrated circuit. ?his IC discriminated the voltage into hit or no-hit, stored this information, and acted as a shift register for t~ the computer. Before readout the spark pulse a small current flowed from the bias voltage bus, VBCAT, through R4, 83, R2, and the diode. VBCAT was about 3.Sv positive of the fl~ating ground, GNDCAT, which was connected to real ground, GND, except when the tester operated~ Therefore the input to the shift register chip sat at about 2.0v relative GNDCAT. wire, C1 was When Iamediately a spark afterward, hit the the junction about -40V relative to GND. v~ltage to this charged. C1 and R2 was at Current in R2, B3 caused this decay and the voltage at the input of the shift register to fall. k~pt of to A diode inside the shift voltage from falling belov -.7v. Cl decayed with a characteristic time of 250 register chip The voltage on microsec. and the voltage on C4 reached a minimum at about this time. The shift registers were strobed 250 microsec. after the HV pulse, causing a bit to be set if the voltage vas less the threshold, about 1.4V. than The shift register input drew - 2 20cur rent only ~hen the strobe was enabled, and C4 • ~mentar y the voltage. n e c essary to split R2 and R3 to damp out ringing t he c urrent spectra from c hanging strip and ground cables. kept this It was in~olving (Ringing fro m the HV p u l se had been rectified~ and charged C 1 as a spark ~ould.) Jn the anode side, the capacitor diode circuit was connected t ~ HV , delivered through the center conductor w~ltage feed cables. of the hig h The floating ground for t h e part of t he circuit in the outside rack was GNDAH (co nn ected to c learing bus except when Therefore these busses uere at dur ing field and immediately the the tester same the operated}. voltage except after the spark pu l se (when HV ~as about 4 kV more positive}. During this time 0.1 amp through This a mounted to a bout 300 a mps R2 for each wire~ tJtal, rising in tens of destroyed the s hift nsec. Th i s regi s ters if r each the m. The pur pose of C2 c urrent isolate and and current flowed wou l d hav e it bad be e n a llowed t o B3 ~as to sh un t the se n sitiwe electroni cs, the C2 pass e d t~is current i n to t h e CP bus, a nd if no spark occurr e d, was c harged 6 V positive in the process. The extra RC filter on the anode side peak at C4 to 450 microsec. incr e as ed t he eguilibr iu m point. the a f er t he spark . w~ltage on VB AN was s e t lo ver ~ ab o u t :2 delaye d volta ge Th e bi as 2.0 v , since th e 6Y on v ~l tage a t C4 to 1 Volt above th e sta tic The vol t age at C4 va s agai n ab out 2. 0V -221providing the high voltage palse came and no spark occurred. If a spark occurred, C1 was charged to -40V and C4 hit a low 450 microsec. after the pulse, the time of the anode side strobe. Most of the exceptions busses were were HV common to all chambers. The and the bias voltage busses (VBCAT and YBAN). It vas desirable to fine tune the £ize of the pulse for each chamber. spark Better sensitivity vas achieved by adjusting the biases independently since different HV pulses produced different no-hit voltages at the shift registers. Al~D -- Electronics Fabrication and Testing The 10000 diodes and individually tested sh3rt, correct open, 1000 shift and rated. labeling purchased so that IC's were The diodes were tested for of according to the reverse breakdown vere register polarity, voltage. ~nd sorted Extra diodes only those able to withstand 250V were used. The IC tester vas selected interfaced to a PDP-8 with whi ch any of the eight inputs, applied a voltage, sensed tae input current and read the shift register r:•s computer extreme values bits. Those of threshold voltage or a large input leakage current were not used. -222- The chaaber p.c. boards were tested by conditions similar to chamber the computer operation. under A high voltage pulser system capable of delivering 300 nsec. pulses of Oto 100 aaps vas connected to each of the syste• of register. 100 oh ■ rise the p.c. board, A 10 kV pulse fro ■ 64 circuits spectra-strip, of and the shift a thyratron in series resistor simulated the spark. the with a This, having a longer time, did not stress the components as much as a real spark. The test consisted of determining the current at which each circuit registered a hit, pulsing each 1000 times at maximum while checking that none of the others current registered, and rechecking the threshold. This test uncovered many bad connections, solder bridges, and arc-over problems, well as as causing early failures in several diodes. When the inner chambers were rebuilt, the p.c. waveform to the spectra-strip voltage on an component values were After end and oscilloscope. easily displaying current Faults or changes in noticeable in the pattern. 2000 sparks per wire average, approximately 1% of the circuits had numbers were They were tested by appl ying a 10V triangular refurbished. versus boards gone bad. These were approximately of shorted diodes and leaky capacitors Cl. equal In many •~re cases the capacitor vas only leaky enough to reduce the sensitivity. resistance About 101 of the Cl capacitors had a of less than 50 k ohms and were replaced at that -223ti ■ e. The Cl capacitors vere s ■ ~ll size chosen for their extremely even though they vere rated at SOV maximum. retrospect, a larger operating voltage at the In expense of size would have been chosen. Digital Readout 11.E The shift register IC's were chained to form l~ng shift registers. two The IC's for the cathode readouts of all chambers were chained together amounting to The anode side IC's 7000 bits. formed a shift register of 3500 bits (only pitched vire ch~mbers being read). shift extremely An extra (8 bit) register was inserted in the chain between the end of one chamber and the beginning of •fiducials" no-hit. were hard wired to the next. read These no-bit, They allowed the computer to check that 8 bit six hits, the shift registers counted correctly. These long shift registers besides the floating needed power only supplies: three connections the LOAD {strobe) pulse: the CLOCK {shift} pulse: and the DATA output from the first stage. t~ the These signals, two coming from and one going digital "control box" passed through level shifters which electrically isolated the Using clamped optical a schottky output from the isolator, impedance matching, inverting, and buffering, input. even they with delayed the signal by only 125 nsec., allowing a 4 MHz shift rate. -224- Th e cont r ol box received a trigger coincident p ulse to the chambers. with the It produced the LOAD pulse for the c athode side after 250 microsec. and for the anode 450 tai c rosec. 3500 CLOCK Follo wing the second LOAD pul s es to side trans mitted an to did the p ulse~ the number c l ::> c kidse-most li k ewise for pulses. 70 00 It PDP-11 o n ly vh en it e nco unter e d a hi t q of contiguous hits "last" and "first" it: downstr ,e a m} Sparks hittin g wir e s of the c yl i hder, o f co u rser were not co mbined by t he co~ trol bo x . transferred along following (facing t he electc-onics en d" i. e~ e dge and width mi Lus o n e fo r each clu =t e r~ the it Then it s Hitc h ed to the giving the number c o rrespon di ng to t h e first wire hit with at the anode side shift registe r r eading the DATA sig n al after eacha cathode side It pcoduc e d the si gn al whic h dro v e the 250 V c learing field p ulse. s ant HV Thi s i nf orm at io c was to t he c ompu t e r wh e n th e end of a cluster was datecte ds Th e n 0 %t CL OC K transfer ~as finish e d~ inserted in t h e long pu l s e de J. a.7ed until the Fo u r one bit shift re g isters wer e CLOC K and c ontrol roo m g n d the cha mbe rs. ~~s Mas DATA cabl e s bet~een the The c a bl e d el a y bet ween th e~ less t han 12 5 ns e c .u s o that no data ver e l o s t when t he =~ntrol bo~ pa u s ed . Th e s hift rat e was becau s e pr ~ble ms of t i ming 2 MHz t h at we re ne ve r l oc ated $ The ces ul ti nq sca n t i me of 5 mill is ec . millise c. t:> cle ar. r edu c e d \!as a cceptable to si ne e 2S deadt i me wa s requi c ed t o al lo~ the s par k c h ra bers -225- Al.E -- Automatic Tester T~e tester applied voltages to the floatin g groun ds in a va y as to test all of the circuits in c ontrolled by the au tomatic. all PDP-11: 1'he C.D~ syste m ~ere tested e lectric al elect r onics s imply by a .t· alleL and running such It was switching vas t.he shift register a program on t he PDP-11. ~hen enabl ed v the tester c ycled through eight las ting about 0.5 s e c., states, each c oming to rest again in state Oo Sep arate tests were performed in states 1 through 5. States 6 and 7 were du mm ies~ identical to state O~ s tates 1-5, CF was switched from the clearing fie l d generator to GNDCAT and g aoerated Du r ing GNDAN were by t h e tester. t he rela ys had settled~ switched to a vol t2,ge Half way through e ac h state, ifte~ the control box Bas trigger e di l~ ading and rea ding in the shift registerso In st ate 1v VT vas +20 , relative to GHD (i.e~ to in the spark chamber)~ open. wac e zero. In state wires curre nt through t h~ dioa~. e 2. and RJ produced a low input to the 74165 , unless w!l s the one 2s VT sas -2 0V and t~ All shift c egister inputs ~ere C2, or the diode leaked current. 0 of these bias volt~ges Lo w unless C1, -226In state 3 VT was at -20V vitb the bias s~me time the At the vas triggered, an SCH pulled vr to as If readout zeco, making condition s for back normal on. o pe ration. a ciccuit worked properly~ C1 ~as charged at -20V which caused a low input to the shift registers after the proper delay of 250 or 450 microsec$, when t he LOAD pulses camee In states 4 and 5 VT was zero. off In 4 the bias was switched to test the ability of the shift re gister to r ead hits. In 5, it vas back on, testing the ability to read no-hit s . ?he res ults of all five tests were stored the co~puter memo r y. were also collate d in The loc at ion and nature of inco~r.ect resp onses were printed. fi ducials and The hard vired verified, pa~terns localizin g fro m breaks the in the c hain. l display s hoved ~bether the relays fail e d to re c onnect tnings properly after the test. Al.F -- Gas System Neon (~ith 10~ heliuo ) gas Yas circulated to al l ch ambers throu gh an LBL ga s cart whic h t he spa~k used activated charcoal and a liq ui d nitro ge n bath to re mov e alcohol~ water v~por, air and ot he r i ro puri _ies. 11 e thyl alcohol ~as added by bubbling the gas throug h li q uid at -3 deg rees c. (The -227teaperature range of the syste• was capable of varying the concentration from 0.051 to SI.) Good regulation of the . .. . concentration vas achieved with the following techniques: All the gas vent through alcohol at lov temperature. than rather trying to divert a fixed proportion through alcohol at high temperature. The bubbles aquarium using an gas was aerator precooled in the same bath that temperature broken into many small stone. The the alcohol. cooled gas was The the bath was controlled electronically to a of fr~ction of a degree. The gas was circulated at about 10 cu. feet this rate, the pressure per the necessary pressure at the the considered vas The outlet of would collapse pressure drop inside the chamber vas quite small. The inlet and outlet pressures were desired beginning unsafe for the chamber, and the negative pressure at the end of the it. At drop in the tubing was about 0,5 inch of vater: inlet hour. flow adjusted rate vith about 0.15 inch (water) the chamber. If the chamber pressure outflow was interrupted exceeded 0.25 the inflow bubblers kept the pressure electrical to by an was give the pressure at fell below 0.1 the electric valve. If it interrupted. within ± 0.5 Oil filled inch if the system failed, The gas flow was interrupted if the alcohol cooler failed in concentration alcohol from of order to prevent a large entering the chamber. The pressure regulation system was inadequate in that there was -228no sensitive backup protection. (The oil filled bubblers operated only at relatively large pressures.) 11.G -- Cosmic Ray Tests After the experience with the first second set vas tested at Caltech determine whether the chambers eBtire area. set were chambers, with cosmic sensitive chamber package. 1 opposite sides triggered coincidence the occurred about once per second.) rays to over their with words 28~ the center hole in between either set of spark and Events were readout. read of core and a disk. (This into whether a track attributable to the track. recorded. lf z and vent through the region, whether there vas a hit near the track, and whether there vere not a The PDP-11 program divided the cylinders into regions of PHI and counted the Four 1 meter by 6 inch scintillation counters vere arranged as a square tube inside PDP-11/45 of hits The width of the sparks was Tvo thirds of the sparks covered only one wire. the rest Most hit two, the number falling by a factor of four for each additional wire. The program had two basic pieces: read in a foreground piece which events (either online or from tape), found tracks, and accumulated counts in the region histograms; and a background piece which accepted and executed commands from a terminal. Commands controlled the definition of regions: -229displayed statisticsw single events, histogramsv and tables; and controlled rea ding, ~ritiug, or updat ·n g disk files containi ng the accumulated counts. For this section 9 follo ws: the cha mbers will be as innermost crossed wire ca thode surface - A1; anode surface - A2; next chamber (axial wire} A3; designated si~ilarly through E1, E2, cathode surf~ce and E3 for the other fou r cha mber pairs. One c hamber pair vas designated to find tracks, and was (Another In practice the outer {E) pair wcs used~ tested. not cun using the D pair determined th e efficiency pair.} atch-points in the PRI-Z Tracks vere defined as two approximately at cha raber. opposite corro borated in the PHI only cyli nder. PHI for the angles, E both Tracks were required to intecsect the outer c ylinder within the active z r ange . Fig. 11 .Ga 1 shows inner mos t the axial-wire PHI distribution cha mb er (13). of since h i ts in the tracks ~ere regaired to go through the center 9 this is reflection of the cosmic rsy distribution ~hich goes appr.o~i mately like c3sin e sq uared of the angle from t he vertica le The four dips are the s hadows of two aluminum pol~ that ran the length of the ch~mher at ± 45 degrees. Also visible are regions of inefficie~cy an d narrow dips in the efficiency due to dead vices bins are or readouts. Someti mes the adjacen t Figure A1. G.1: 10- 79 0 <f) .., • 90 ii!.i iC._. o~ t • !!Q)() '-· V • I u .._ ; ' ---~ B "'.;s' ' "%·A ..1,-..!r., ~ ~!{..... ·~ 1:~ I, - 270 ~ @" h C', -~Jl-~ - "1 ''' J 720A'• 360 . '\J ( , I ! H h, ~1 c~ , \ti ~ ~{.<-i4: r H~ _. 4 ht\· . !""',n-• _,..~GRr-- r• t•..... c,-, I ____,i_ ' . ~_f\;t \ ,: /1 S ~I I~ f0[t. I ~.,,! i j~ I' _~_;{'i I ] tH tJ Ir!foi ! , __ distribution of hits in cylinder A3 ::: hoHing cosmic ray angular distribution. ~--- ~i, ~ ~.~~.J--.. . h~~ :1\ ~~ - _ ;~1~-~~~ • lf '"h'''l ~, ,~ . I . t~ Jt .i .fd ~fl ; I "" C) I N -231- abnormally high 6 indicating the spark vas pulledg perhaps because the prope r voltage was not present at the dead spot. A typical event is shown plotted in fig. A1. G. 2" The hi ts are at the PHI of the vice at the front of the chamber. Hits in pitched vice cylinders appear to be rotated by a dista nce proportional to the distance from the fro nt end. Fig. AlQGel and A1oGo4 shows plots of efficiency for Z all as a function of PHI for the best and wor-st cha mber. surfaces (bins of about 1.5 cm. of efficiency for all chambers were limited circumference). was 90%. All The average inefficiencies to regions rather than being u n iform over the c ylindera The consistently louer efficiency (oc surfaces was a result of the lover sensitivity of "2") of the anode the anode readout and the need to ignore ce rtain wires permanently regi ste red hitsb that (The fact that sparks ha ve t~o ends leads to the expectation of equal efficiency in the " 1" aud "2~ surfaces.) Efficiency versus PHI for an in ef ficient cbsmber is shown in 20 cm . Z slices i~ fig G A1oGe5~ was Horst in the middle. One dip The other w~s wor st at the back. Cl bad a brea kdo Yn regio n which bec ame vers e over the weeks as the chamber was sparked. The deterioration is thought to be o~ due to the roughe ning electron er1.i s si on in the f5.e ld., i~ t his region every tLig g 0 r~ the spark ~ires lea ding to Even'· ually a_ spark occur. red It b e c am e so bad that the -232- / f Downst ream Z Shows as cp Di splacement . / {. -Tri gger Scin tillators 10-79 3720AI Figux'e A1. G. 2: A cosmic l'ay event: letters indicate q> position of upstream end of wir,es. ~ --- 11-79 w LL LL w u- z 0 >- 0 ........., 0 20 40 60 80 100 0 Figu:r-e A:1 , G. 3 : r I 1---,-4 I DEGf-~EES 180 Efficiency as a function off ~'"'--' ari L..r--1 270 L~ for for cylin der surfaces B3 . Lr 3720A5 360 7 I N I 1..,-l 1..,-l ---~ 11 -79 w LL LL w 0 - (_) >z 0 ...._., 1 0 0 I Figure Al. G. 4: 20 40 ~ 60 80 100 ,_ r.p 270 for cyli.nder surface B2 . DEGREES 180 Efficiency as a functlon of 90 l,) 3720A6 360 7 I I +"" N \.N 10 - 7 9 w LL LL >u z w u ---- 0 ~ ............. 60 80 100 0 Figur e A1 . G. 5 : 20<Z<40 -..z ..) 180 360 0 J 720 A 7 180 360 BO<Z<IOO 180 360 0 60<Z<80 cp i n bins of Z for . surfa ce B2. DECREES 60 Q J7 40<Z<60 Ef fic ie11cy as a f i. . n ction of l'C1._, r~<n l_ . ·- j ~ 180 360 0 u7-' O<Z<20 I N I V, \.N -236breakdo un sparks pre vented the voltage on the chamber from reaching the value nec essa ry to produce sparks elsewhere chamber., tilis The wires in this di sconnected allowing the rest of the Bre akdown occurred frequently 2 cm. ch amber ~is to were op erate . at the edges of this region which beca me worse until again the chamber was wires uere cu t$ region dead. Mor e The breakdo~n began again after the package installed at SLAC. During the e ~peri me nt, this c ham ber w~s run at lo ~e r vol tage givi ng about 50% efficiency. A1.G.6 sho~s in a plot of c hamber after t h e surgery. Fig. spurious sparks vs. PHI for this A s2all region ~as cut in also, with sa ti sfac to cy resultse Reference for. a ppendix 1 1. H.B. J ensen et . al., A New Single ~i~e Readout Sy s tem .. for Sp ark Ch aIDbers" Nucl., Inste /-! ,2 th. 12 5, 25 (19 75) . E3 11-79 0 0 ::> z f~ L ~- 0 C I ':l,.,c' Figure Al. G. 6: 0 20 40 I 60 80 100 ~ 90 1r=- L,.,e:' Ll DEGREES l80 LJ J 720A8 360 -,.D-i1MJ Two bins are truncated. 270 LC: ~ ~ Spurious sparks in C3 gap sho wing breakdowTl . - r I I I I --.J 1,,-.\ N