Dielectric Recovery of Short A-C Arcs Between High-Boiling-Point Electrodes Citation Rader, Louis Telemacus (1938) Dielectric Recovery of Short A-C Arcs Between High-Boiling-Point Electrodes. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/BXYB-0G14. https://resolver.caltech.edu/CaltechETD:etd-06232004-143113 Abstract No Abstract. Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: Electrical Engineering Degree Grantor: California Institute of Technology Division: Engineering and Applied Science Major Option: Electrical Engineering Thesis Availability: Public (worldwide access) Research Advisor(s): Unknown, Unknown Thesis Committee: Unknown, Unknown Defense Date: 1 January 1938 Record Number: CaltechETD:etd-06232004-143113 Persistent URL: https://resolver.caltech.edu/CaltechETD:etd-06232004-143113 DOI: 10.7907/BXYB-0G14 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 2698 Collection: CaltechTHESIS Deposited By: Imported from ETD-db Deposited On: 24 Jun 2004 Last Modified: 10 Jan 2022 20:58 Thesis Files Preview PDF - Final Version See Usage Policy. 37MB Repository Staff Only: item control page

DIE'f:!ECTRIC RECOVERY OF SHORT A::.Q. ARCS Thesis by Louis T. Rader In Partia.l Fulfillment of the Requirements For the Degree of Doctor of Philosophy California. Institute of Technology Pasadena. California 1938 TABLE OF CONTENTS Summary. Pag_e I. Introduction II. Theory .. ~ •••••••••••• 4 III. Experimental Procedure ..•......................• 5 IV. Results 9 e 9 l ........ Iii> .................. Copper Electrod.es . . . . . . . . . . . . . . . . . . . . • . • 10 1. Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . • 10 2. Effect of Current.................. 11 3. Effect of Pressure .•..... , • . • • . . . • . 12 4. Effect of .A:rc Length............... 13 Steel Electrodes • • . . . . . • . • . • . • . . • • . . . . • • 16 1. Data............................... 1? 2. Effect of Current • . . . . . . . . . . . . . . . • . 17 3. Effect of Pressure . . . . . . . . . . . . • . . • . 18 4. Effect of Arc Length . • . .. . . . . . . • . . • 20 General - Temperature . . . . . . • . . . . . . . . . . • • 20 v. Conclusions ........ ., ........................... . 22 VI. A.cknowled@llent ...•••...........................• 24 Appendix A - Apparatus...................... 25 Appendix B - Experimental Technique • . . . • . . . . 28 Appendix C - Probable Accuracy of the Method 32 Appendix D - Temperature Calculation........ 34 A. l3. C. Appendix E - Equations for Spa.rking in a.n Ionized Gas • • • • • • • • • • • • • • • • • • • • 35 Appendix F - Reignition-Voltage .Time Equations 36 G - Arc Behavior.................... 38 Ap~endix SUivlMAii~ Continutng work done by an earlier investigator, the reignition characteristics of short arcs betvieen copper and iron electrod.es were studied. in detail. Current, pressure, Md arc length were varied to o bt1:d'.!1 curves of res triking vol ta.g"' ver!"us t.ime at curre•1 t 2er!). Data '?ere obtained by making use of an oscillating circuit, varylng the rate of rise of voltage a.cross the are termina.ls b~r varying the circuit constants, and. observing and record.1.n,g the voltage phenomena. by the use of 8. cathod.e ray oscilloscope. Results were obtained which check previous work on brass electrod.es as to the kind o:f voltage reco·1ery characteristic obtained for short arcs. The previously held theory that high mel ti11g point electrod'9s h.ei.ve a dielectr:i.c reco1rery directly proportional t.o a.re length for short times. is shown to be .false. An optimum a.re length for cu.electric strength for both copper a!ld steel electrodes is seen to exist between 1.0 E!.n:l 2. 05 mm.. Temper-· atures calcul9.ted for the arc space agree in ora.er of magni tucle with results for earlier determinations. The method used is held. to be p~.rticula.rly a~Jpropriate, becr;;nse a.vers.ge values were, obtained in all cases, s,n.d results obtained '·"i. th the i::pparatus vrere consistently checked over time intervals of more tha.n two years. The circu.i t u.sed was the same as used by Dr. T. E. :Bro~me, Jr.* The same method of obta.ini ng the de.ta ~as used and his the.sis should. be consulted for a more detailed analysis of the circuit. ·-------------------·-·---···------·* Browne. "Dielectric Recovery of Short A.-C Arcs between Low-BoilingPoint Electrodes 11 , Thesis for Ph.D. degree, C.I.T., (1936). I. INTRODUCTION In a rema.rkable paper published in 1892,1 Mr. A. J, Wu.rt?.: described his experimental discovery of the fact that metals fall into two ma.in cla.sses in their arc-suppor~ing che,r~cteristics. :Between electrod.es of a metal of one group, he found that an arc burned freely with considerable noise, light, and movement. Metals of this group, including carbon, tin, and lead; iron, cobalt, and nickel; and copper and aluminum, all having boiling points between 15QOoc. and 3600oc., he designated a.s 11 a.rcing11 meta.le.. Between electrodes of lower boiling point, such· as zinc, cadmium, and mercury, he found that an arc was maintained with difficulty, and that if it did burn, there r.as little noise or light. These, he designated as the "non-arcing" metals aw.l used their arc-quenching properties in a multi.gap type of lightning arrester. Mr. Wurtz was also the fin t to discover the almost paradoxical fact that as the electrode separation decreased, a higher voltage wa,s required to maintain an arc between zinc or antimony electrodes. Many workers have since studied this effect in more deta.il2,3, showing some dependence on electrode boiling point and electrode separation. However, a satisfactory a.nd complete theory of short- 1. Wurtz, "Lightning Arresters 11 • Tra.ns. A. I.E. E. 9, p. 102 (1892). 2. :Browne, "Extinction of Short A-0 Arcs", Trans. A. I.E.E., 50, p. 1461, (1931). 3. Slepian and Strom, 11 Arc in Low-Voltage A-0 Networks 11 , Trans. A.I.E.E., 50, p. 847, (1931). 2. .<.ire performance has not yet been proposed. As pointed 011t by Da.rrow4, the 'l"'Ork of many researchers in conduction in gases is of little or no use for purposes of correlation, because thetr data ware taken under such varying conditions, and results often given with the omission of some necessary pa.re.meter information, that it cannot be correlated into a theory of arc performance. For example, much d;:;ta have been taken on a.res between cylindrical electrodes a.s in the common carbon Rrc. In such cases the arc is usually bowed out a.t an edge so that no true figure ca.n be used for arc length and actnal separation of electrodes has li.ttle meaning. Browne5 with an exhaustive set of tests sta.rted a very systematic study on brass electrodes. Using the same apparatus built by Browne. the same set of tests was mad.e by the author on copper and steel, two high boiling point metals, in a.n attempt to obtain further data on the su.b,ject. Current, electrode sepe.rati.on, and pressure were varied to obtain their effect on the restriking voltage of an 9-rc. All results are given in the form of curves between a.re reignition voltage and time after current zero. When the current ~oes through zero in r:on A-C arc, i t does not immediately start flowing in the or,rposi te direction. Instead a qu.i.te consid.er- able voltage of the order of hundreds of volts is usually required to initiate current flow through the arc in the new direct:l.on. 4. Darrow, "Electrical Phenomena in Gases", (a book)______ Williams & Wilkins Co. {1932). 5. Browne, "Dielectric Recovery of Short; A-C Arcs bet'!1!'een LowBoiling-Point Electrodes", Thesis for Ph.D. degree, C.I.T., (1936). 3. This is the voltage known as the reigni t.ion res triking. or recovei:y vol ta.ge. The ri Ee of this vol te.ge across the arc spac"'"e is, of cou.rs e, a function of the circuit parameters and tirrie. ar<~ The ability of an space to wi thsta.nd breakdown; i.e., :its dielectric stren.gth, i $ a.lso a function of time so that "b;sr varying the circuit con st.ants, the time of breakdown of the arc space was vartea.. thus g:i.ving a measure of dielectric recovery with time. This subject is of vital importance in electrical engineering. Short arcs are found in all circuit breakers, such as knife and snap switches used on low voltages and in fuses. secondary netwo1·ks the fault must burn clear. In modern PrAsent day industrial welding is e, field intimately bound up with short-arc behavior. The deion circuit breaker and. the autovalve lightning a.rrester are excellent examples of the use of the extinction cha.racteristics of short arcs. L ~ c l T E cos 211'.SOt lT :c OSCILLOSCOPE ~ Fi~-- I. s. I ~ () I I I / I I I I ~ SIS ( I.() ~ ~ ~ c=L ARC I 230V .sol-60 J L ~ (;) l'f\ I ., /Ir. ,. I Sa T, s. I <\I sf 3KVA s. VOL TAG£ DlVIDER 53 Fiq. 4-. 4. A most complete summary of the theories underlying and explaining short arc phenomena is to be found in :Browne's thesis5. Re has made use of seven principal references on the su.bject from. 1923 to 1936 to give a comprehensive discussion and summary under the folloV<.•ing hea.ds: 1. Arc Reignition. 2. Space Charge Formation. 3. Dielectric Strength of S~a.ce Cha.rge Sheath. 4. Conditions .Affecting Dielectric Strength. 5. Rate of Dielectric Recovery. 6. Phenomena at the Anode. 7. Ionizing Activity. There ha.s been no departure in theory from those presented since 1936, so that it i.s felt unnecessar;y to reproduce this phase of the su.bject in this thesis. / I I .... \ \ I I I \ \ f I I I I I I --0 ...,.," F'ig. 2. I I '\ I ~, I I I I ' \ I \ I 'II E.. ' I I ' 11 -- --- TH"\ E. E .., 0 .... .+-I ~..., +' " ,, , Fig. 3. ll.I!._~J.;PERI MENT.A.T, PROO EDURE The apparatus used for the taking of data was the same as that used by :Bro1"rne5. As shown in Fig. l, current from a 690 '.rolt circ-i.li t wa.s limited by almost 9ure rea.ctr:>.nce, 1, l)laced in set'i.es with the arc. All readings were taken iri air on continuously- bil.rning a.res bet•rleen plane pa.rallel metal electrodes. The met~'ils were copper and st.eel; the currents 12.5, 25, and 50 amps.; the electrode separations 0.45, 1.0, .2.05. an:l 4.13 millimeters, and the pressures 1/4, 1/2, ano. l atmosphere. Condensance, C, (J!'ics. 4) was placed a.cross the a.re to very the nat1.1ra.l freouency of the circu.i t, e,nd so give varying rates of vol ta,ge rise across the arc. A cathode r"t:'t. y oscilloscope of the :Bra,un type (R.C. 104) with fluor- escent screen 'Fas conr:,ected acro!?s the arc e.nd ,cr,ave a. visual picture o:f the voltage wave at tl1e time of current zero from l':'h~.ch ~VEP.AQ:~ ampli tud.e readings were taken. The time relationships existing between supply voltage e 9 , of ma.xj m·(lJTI value E. arc voltage ea, end a.re mlrren t i a. are shewn i.n Fig. 2. Due to the very expanded scale of the figure, (the interval t = o to t = ts, representi.n,g: a maximum time of about 250 micro-seconds) the applied voltage \vh:i.ch is at its maximum, appears as a horizontal stra:i.ght line. Except near current zero, the current 1!'rave is a i;rn.re sine v.:r:.;.,'!e laggi'.1.g the applied V()lt1:1.ge by ninety degrees. As the arc current, approaching zero sinusoid!'lJ.ly, 6. reaches a small value of the ora.er of 0.1 amps .• it suddenly fails, falling immedia.tely to zero, where i t remains until A. reignition voltage is reached so tha.t at t = ts it starts to flow in the opposite sense. The current in the inductance cannot dro~ instantly when the arc current fails and it surges into the capacity in the circuit, the ree.ctors 1 distributed ca.paci ty which is in para,llel with the capacity across the arc. Du.e to oscillations, then, the voltage across the inductance and capacity in parallel, rises from the value it had at t = o, to em at t = tm and es at t = ts.· The boundary conditions obtaining at t = o are the values of E1 the arc voltage and I 1 the arc current. The former is quite constant but the latter is subject to fairly large random variations with consequent variations in em. At t = ts reignition occurs, the voltage across the arc falls to the glow voltage. and a glow discharge· characterized by low current density and high cathode drop occurs. As the current increases, the glow turns into an arc the current then follows a sine wave and the arc voltage remains substantially constant for the next half cycle. l!'rom t = o to t = ts the arc voltage can be computed quite simply by the circuit equations since the current is zero. has been done by Attwood, Dow and Krausnick6. 6. Attwood, Dow and Kra:usnick, This See Appendix A. Reign.i tion of Metallic A-C Arcs in Air", Trans. A.I.E.E .• 50. p. 949, (1931). 11 ?. ea = E - Em cos (w t + Q) where ea. = E the arc voltage during the period of t = 0 to t = t1. = maximum v-a.lue of the applied A-C wa.ve. Em = E - em= amplitude of the voltage oscillation a.bout E. '!/\' = 2TT _l_·_ = 2n times the natural frequency of the /LC' circuit. em = the negative maximum of the oscillation. Q = tan-1 !J~WC = the phase angle by which the negative E - E1 maximum la.gs behind the time t = o. The above equation can be solved for ts by subs ti tu ting es for e and obtain:ts =JLG cos-1 E - 9§ E - em + tan··l j E- El - 1 ~ E - With L in henries, C in microfarads, t is given in micro- seconds. Two kinds of negative maxima occur; viz! the negative glow, and the no-glow cases. The former is shown in Fig. 2 and occurs when the current fa.ils earlier in the cycle and so drops from a larger negative value to zero. The negative electrode voltage falls very sharply to a sufficient val-ue to produce a negative glow which lasts for a few microseconds. The same effect occurs for low val-Iles of C. as may be seen from the equations: and s. so that en increase in I l or a decrease in C both tend. to increase Em .!md Sm· For this negative glow case, the initie,l voltage El is identical with the negative maximum em. so th.at the inverse tane;ent term disappears in the expression for ts and ts = / LC :El - es E - em For each value of arc current and ga.p spacing, readings <"ere taken, with values of capacity e,cross the arc from !?.ero to the highest possible at which the arc was stable. For ea.ch of these :readings ts was ca.lcula.ted and plotted against es giving e. curve of rei.gni tion voltage versus time after current zero. The method ·1.1sed for obtaining the data plotted on the accompanying curves was to ad.just the gap spacing and current to the desired values. varied. The value of the condenser, c. (Fig. 4) was then This change in the value of the condenser changed the time required for the vol ti:ige to reach its reign.i ti on potentia.1. Th:i.s is the ti.me t 3 • the inductance, Knowi.ng the value of the condenser and of the time t 8 was calcu.lated. from the eque.tion alone. 9. IV. RESULTS. Results a.re given for copper electrodes under the following Figure numbers:COPPER ELECTRODES Effect of Current Separation Pressure in Atmos. Fig. Nos. 4.13 mm. 1 1/2 1/4 5,6,7 2.05 mm. 1 1/2 1/4 8,9,10 mm. 1 1/2 1/4 11,12.13 0.45 mm. 1 1/2 1/4 14,15,16 1.0 Current Separation Effect of Pressure 4.13 mm. 50, 25, 12.5 17,18,19 2.05 mm. 50, 25, 12.5 20,21,22 1. 00 mm. 50, 25, 12.5 23,24,25 0.45 mm. 50, 25, 12.5 26,27,28 Pressure Effect of Arc Length Fig. Nos. l atmosphere Fig. Nos. Current 50, 25, 12.5 29,30,31 1/2 u 50, 25, 12.5 32,33,34 1/4 II 50, 25, 12.5 35,36,37 Figures 38 and 39 are representative curves which are given to show the correlation between the actual data and the curves drawn through them. 10. A. Coouer Electrodes · .h__:Qata Figures 5 to 38 show 34 sets of curves. In all, 589 rea.dines were taken on copper, - about 15 points making up one curve. About 90 of these readings were made to re-check data in order to make certai:!l that the ap:9aratus gave consistent results and to check probable mistakes. Da.ta were ta,ken for all possible combinations of 50, 25, a.nd 12.5 amps.; 4.13, 2.05, 1.0, and 0.45 mm.; a.nd pressures of 1, 1/2, and 1/4 atmosphere. res~l ts It was impossible to get consistent with 50 amps. at 1 f:ltmos. and 0. 45 :nm. separation, though a uortion of the curve is shown. The number of curves is too large to include them all on separate sheets, so that the results are given under three main heads:(1) Figures 5 to 16 show the effect of current on dielectric recovery; (2) Figures 17 to 28 show the effect of pressure, and; (3) Figures 29 to 37 show the effect of arc length. Figures 38 and 39 are submitted to show how, in general, calculated points agree with the final curve drawn through them. They are representative examples. In all cases the curves drawn follow the data points very closely. • • • . • • -· -t ~ .- ···' .· ! •I . -+ ·- ·- ·: i ··. . ... ••.:. . .. - . ·1' . . . 1' .! . • • • ----· • -. -··. .... ~·-···--· . -. . - . . __:___ - ~ --.. . I. . L-: --- . : :. f:.. . I I . ~ .--..,..--- . -· -: . l I • . :_ ~ l . I I. . I • __,; • : .· . : - I - . • \ . I ---- I-~ ' ' • I :• .. !: - - ' - t:__ -·- I I : . i ' ' _J.·: ' I I' ) ' ' . __L_ __ ' I -· - . _. ! - .: --···. I • . . •1 ) ' . · __ . -.. . . --. --~ I --- -- . . l - I ' .. - I ·. \ ;· ' - . - ' --.l·-· ' ' I . . . I . . . • I • • I '• . 'I ·. . I l ~L .. , f __ ;_ -- • .-~- -;-. ')--i-:' "T r._l: · ";1'.. I. ci . : . . ~. ..:!~ ~ ..J,~ { i !:.· : 't .. , . . ' ., . ....' ...'. . . .. . ... - -. .. . -. -- -~ • • • .: . .. . . : : I. . . -·. . -.. . . . . - .l .. . . .i.. • l + .. ,, . .. . ,., • • - -1 .... ,._ . . . . . . . . t •• • ·;: . . ' - · :--. :· : ' . · - •j --- ~} i" . l -:: . : . ::L; . . .. ~ I . --L I f• • ~ 'I . . .:. . !1·..• ---~---.4 • • • •' 'I_. ••• • -- • -·.- : ~ ..: • , : r• : '"- •' -· i_-..:. - ........ ....- :: --·, .... ~~1~ft ::~~ ~~!: :~i...1:1· ~ ~ ;:1-~1fi ~ ~~~ ~~ ;~ : .. :i : • ! ! .. .._~ :J ... ·-, .. .3 ....··i··.. • t • ~... -~ I ' :--: ::~: ~;~~ ; - . : :-:.: ~ .. : r.~i -· . . :. . :: : •. :· :_j:.... . . '.'t) - -. . . r·· .... . . . -· .. . ..: ~ . . : : ... ~ __: ~· -- . . . ::-:-1 -: -- .. ·-· ----,. __.- I ~ • I . . ! .~ -·-·-t-~· ·--·-r--:- .. T • -- : : ~ . 1 --- ·_.__: . - l . -! ' . . . :f:i -·-¥. .. . - .. - ~ t • .. . . __, :> ~ ·· ... - •" •• i ~- •• I • ' • . , - -!-. . - I ' ' I 'i . ,. .~~----J . I ' ! . : f - . - . ~- :'!' .: . ·:. ::-+- ··:l··· . . I . . . '!' ··t • ·• • ·- ...:4·I:_. _ _:__~ . •.:_ :·. :· .. : -·1:·:: ... , . I. ·- . .:· ···b : . .. - 1.. . . . . .... I .. ! : -- ... .. . .:.. - - . . ··-- ____ [. .. . I. . . I! . .. ~ . ,· .· . . . ID . ---- . "f . ; 1 I ! • . ! . :. -+ - -- + ; . ' . ~ • . I . I . - ---i• . .•• ' . !" - .- 1l t ...: . ___ j_ ' . -~ .t:. ... ; : -=~~t - I• --- ' - -- ' ' - --t--::;; ; 6 .· :: . : :;~·~: . ~~-;; - - . . . ..: :. . -:_ ---- . . -·-' ~· .. :J .. .·._t -~:· : --_,__ , I.:;. :. : . 1:· ·- ~ .......... . : l-·. :· :- : .. ~ ..... 't ... ' '. '. - - . . .$ .w~ ...... •• ~. • 1. :;;·r :....· ~~·-h::: :.-::·~~~- .·:: _, ___ ,...·----. ---... rt+--... . ' ...-~ . ._ . :'.; . ..: ~---. :-~-~ ··-·1 .. ,.._. .- .. . ' •-" -. .-- .. ·- • 1•- - .. --· -- --·.... -• ·I • · :.. .. i - '1 . .. -. . .. I. ·-·~; ' t· • --- ·r. • I . l . I : . . . --r.-""I . . .' . . - -~ ' . I C' :.......: ;._~ ' .I i.-.i ,,. f'i . '.· : :- l . : : ;-· 1 : ': ·ID· -·zt:-- -. - . ; : :..i i :: ·-; . :l~' .. .!:.......... . . . i'. ... . . - ~ ..---J. ·I·._, . : .1:; '. ·1 :. :. T ~ . . ,. I ... . 1·... ,. . . -.....+:--· .•. _.: . . T .. ::::1 · • t • • - ••• . .. _;-_-:"1-:-: I .. .. . ... . . ' - .- . --- -· . I: - ;::. : •.· - • ' ' ~ J -• • :_:_ T • :~~<- ;~ . .. . . ... . .....- . . . .... ··~ . -- J --1--- --- ~ .: .. \ - • ._. ·••• + . . l,,. ------. . :. . ' . ' • I, '' t I • I . ---r- . . .. ·r ... I • : . . -t •• _. ·-. • • -· r ~~ ~ l ~ : . . .. . . .. : . • ; : 1· -· ••• ·-:.. ·:-:.::.-; # . . .:·~:1 ~_;-= .:~~: .. : : ---· ...... . ~ • - • ' • ... ! I ~t··-- ! . ! :' . . I l 1 --i-- i 1 . ! • :I . ~l· - t. . .... I .. . l :. . ----·r. .. .' -· • +-_,.' ~ t I . . . ·-. !.. : : ·- I , .. ' . .. I :'~7:. .. : .... .. ! .. . k . :1 .. 'j···~ ----. ~f:--: -; -- . If- ... . ' .... :.. I ,. -·- · 1' .·· .. I --' ·I· ..• t • .. . ,_ : __1...: . . ·- - I 1 • • . . 1.:.: ,, • :.:. 11. 2. Effect of Current - (Copper) Figures 5 through 11, show the definite effect that increasing the current lowers the initial recovery voltage and also the subsequent sparking voltage. However, at 1/2 atmos. and 1 mm. (Fig. 12), the dielectric recovery curve for 50 amps. is higher than that for 12.5. Figu.re 13 (1/4 atmos. and l mm.) and Figure 14 ( l atmos. and 0.45 mm.) show this same effect. The curves shown on Figures 15 and 16 show the anomalous relation that the dielectric recovery curve for 12.5 amps is higher than that for 25 amps. but less than that for 50 amps. The initial recovery rate, however, is still smaller for the higher currents. It should be noted that with the exception of Figure 12, these latter results all occur at 0.45 mm. separation. Assuming that current density is invariant in the arc, neglecting lateral diffusion from the ioni2ed core, and treating diffusion and cooling as one-dimensional processes, then the reignition curves for various currents should all coincide. The results obtained show then the variations in these assumptions. Lateral diffusion must be appreciable for the longer arcs - 4.13, 2.05, and 1.0 mm. For, still assuming uniform current density, the ratio of core perimeter to core area is greater for 12.5 amps. than for a greater current, so that diffusion is relatively more important for the smaller currents. 12. The effect occurring in Figures 13 to 16. that of a higher recovery curve for 50 amps. than for 12.5 amps., can be explained by the fact that it was very difficult to maintain an a.re e.t 0.45 mm. and 1/2 or 1/4 atmos. with 12.5 or 25 amps. Almost immediately after initia.tion the arc appeared as a. diffuse discharge and quickly (in less than a second) changed to a glo1v discharge. For these cases then, the relations existing were obviously very different from those holding at the other spacings and there could be no assumption of uniform current density. 3. Effect of Pressure - (Copp~r) ]'igures 17 through 28 show the effect of gas pressure on dielectric recovery. all other quantities being kept constant for any set of curves. The Figures show that in general the curves cross at some point, showing that the effect of pressure is different for the first and the latter part of the curves. Equation (5), Appendix (E), shows that the dielectric strength of the space charge sheath sho·uld vary directly 'l'l.rith the :pressure (since pressure varies inversely as temperature) and inversely as the ion density. ]ecause of the relation existing between the ion density and coefficient of diffusion, Equation (5) shows that for deionization by diffusion the early recovery rate should be A.L~ost independent of pressure. After deionization of the space. the only effect of pressure is upon the density of the gas between the electrodes. For the same temperature the dielectric strength should be higjler for the higher pressure and directly proportional to it. The fact tha.t this condition does not hold for later recovery times, shows that the temperatures are different, which is easily appreciated since the cooling rate is higher for a lower pressure. These relations of lo"'' dielectric strength for short reignition times and of high dielectric strength for later times with increasing p1·essure are well shown in the curves of this set. l!,igures l.8, 21, 24, and 27 agree identically in form with Browne's Figures 45, 44, 43. and 42, respectively, for b~ass, differing only somewhat in magnitude. 4. Effect of Arc Length - (Copper) Figures 29 through 3? show the effect of arc length on dielectric recovery. With only the exceptions noted below, the shorter arcs are seen to have a higher short-time voltage strength than the longer ones. Inasm11ch as this property has hitherto been held to hold only for the low-boiling-point electrodes, this is a very significant fact. Considering first the 50 amp. curves 9,t l, 1/2. ~md 1/4 atmosphere (Figures 29, 32, and 35) as being of most practical interest, it is seen that the 4.13 mm. arc has the lowest restrikine; voltage. In Figure 35 (50 amps. 1/4 atmos.) deionization takes place for all spacings at 30 microseconds. Both in these c-iu-ves (Figure 35) 14. and in those of Figure 32, the 4.13 mm. arc has a slightly higher recovery voltage than the 2.05 mm. arc but the difference is "Pi thin. the range of possible eX!)erimental error. In Figures 29 and 32, complete deionization does not take place for the 4.13 mm. spacing within the time range shown. It was impossible to continue this curve further. because of the explosive violence with which the arc behaved. For as the arc shunting capacity was increased to obtain longer reignition times, the arc movement became ver~ rapid often reaching the electrode edges and bowing oa.t into an arc length of a foot or more. It is ·u.nfortunate that readings were not taken on spacings between 2.05 mm. and 4.13 mm., for there appears to be a rather significant large variation in dielectric recovery 'between these curves (Figures 29 and 32). Figures 29. 30, and 32 show that as the time gets longer the dielectric strength of. the 2.05 mm. space becomes greater than th.~t of the l mm. or 0.45 mm. a.res. At still later times i t is to be expected, as shown in subsequent curves. that the 4.13 mm. arc strength will rise to a value greater than any of the others shown. The 25 a.mp. ~arves at l, l/2, and 1/4 atmosphere (Figures 30, 33, and 36) are also very interesting. They show again that for short times the breakdown voltage varies inversely as the arc length, and then as diffusion becomes more important the long arc s9aces reassert their higher voltage strength. an excellent presentation of these phenomena. Figure 36 is indeed 15. Figures 31, 34, and. 37 for the 12. 5 a.mp. arcs show another effect which is also evident on Figure 35. It is that even for short recovery times the l mm. arc has s higher dielectric strength than the 0.45 mm . . This shows very definitely that for 12.5 amps. at lea.st, there is a separation between 1.0 mm. and 2.05 mm. at which the dielectric strength of a copper arc is a maximum. Figures 30, 31. e~d 33 of this thesis agree very closely with data taken under the same conditions for brass5-(Figures 39. 40, and 41, respectively). These Figures 40 and 41 also show that an optimum spacing exists between l mm. and 2.05 mm. These results show that short time dielectric recovery is not dependent on boiling points, as far as copper and brass are concerned, unless indeed their boiling points are not sufficiently different, - which is a rather untenable hypothesis since the boiling point of zinc is only 93000. and that of copper is 2310oc. The extremely close relation between Figure 31 for copper and Figure 40 for l:.lrass is a good check on the consistency of the experimental apparatus, for data, for these curves were taken exactly eighteen months apart. 16. Results are given for steel electrodes under the following Figure numbers:- STEEL ELECTRODES Separation Effect of Current Pressure Atmos. 4.13 mm. 1 1/2 1/4 40,41,42 2.05 mm. l 1/2 1/4 43,44,45 LO mm. l 1/2 1/4 46,47,48 Separation Effect of Pressure Current Fig. Nos. 4.13 mm. 50, 25, 12.5 49,50,51 2.05 mm. 50, 25, 12.5 52,53,54 mm. 50, 25, 12.5 55,56,57 1.0 Fig. Nos. Current Pressure Effect of A.re Length Fig. Nos. 1 a.t:nosphere 50, 25, 12.5 58,59,60 1/2 It 50, 25, 12.5 61,62,63 1/4 " 50, 25, 12.5 64,65,66 . ·I • ... . _J __~ - __ i_:. v ... t -- ... - ~ -. .- ... ~: : I>.. ·····-· . . .~ -. ..... - - -- ;_: ii:t'. . - -_ :~ ! . --- -----1- ·:. -:-.~ ·. ~ ~ :. j - -: - ;~ 17. B. Steel Electrodes l. Data Figures 40 through 46 show 27 sets of curves. were made up from 472 readings. These curves As for copper, a.11 combinations of 50. 25, and 12.5 amps; 4.13, 2.05, and l mm.; and pressures of 1, 1/2, and 1/4 atmosphere were used. These Figures are divided into three main groups:- (!) Figures 40 to 48 show the effect of current on dielectric recovery; (2) Figures 49 to 57 show the effect of uressure on dielectric recovery; (3) Figures 58 to 66 show the effect of arc length on dielectric recovery. 2. Effect of Current - (Steel) The nine sheets, Figures 40 to 48, show definitely that increasing the current does not have much effect on the initial voltage-time recovery rate but that after a short time, of the order of 30 microseconds, the recovery voltage rises very much more rapidly for the 12.5 amp. arc. In none of the curves, except perhaps in Figure 41 (1/2 atmos. 4.13 m.~.). did complete deionization take place for the 50 amp. arc up to 130 microseconds. All 26 curves, if projected back to cut the zero time axis, show an initial arc reignition voltage . 1· ' .. . :.L..:..:. - I . . ' t 1 . . . : • . •• ..I . . . ~ _L . I . '- ·:·· - 3 ~'ot.l.1 i;; ... , • ' I -~ ... :-•· .. --· ' r . :+ r··· • 1. . .- . •• ' >•I -_ . ~--:I • ~ ..•-- - •• ~ ·-. .. . - . ... . • • • ' . -· :~: . ,.. ~-("- ___ ---- ,,4. ~ - . ·. ·-· . -. ::_ -=·_: . =--r p - __ .. ;·:: -· - --· ..... ; . _:::~·; --r .... . • ·::~ -. : :t ... : ··. .. ' -- . : ·i .. - . ., J . ·j . : . .. ~ ! . ·.·;,~=H·_;· - -· -· ~ . - .- - 18. between 400 and 500 volts. This value is of the order of the minimum sparking potential for air and undoubtedly represents that value for iron electrodes in air. These variations of vol ta.ge with current show definitely that no simple one-dimensional proce,ss of cooling or diffusion is adequate to explain recovery rates. Ho'l'Tever, the observations made for the case of the copper. electrodes apply 'll"i th equal force to this case. 3. Effect of Pressure (Steel) Figures 49 through 57 show the effect of pressure on dielectric recovery for the 50, 25, and 12. 5 amp. arc at 4.13, 2. 05,· and 1. 0 mm. separations. Looking first at Figures 49, 52, and 55 (50 amps.; 4.13, 2.05 and 1.0 mm.), it is seen that the dielectric recovery curves are practically hori7.ontal. The dielectric strength is greater for 1/4 atmos. than for 1/2 or 1 atmos., but the difference is indeed small, being principally the difference existing at earliest restriking. For this case it is true that the i~itial striking voltage should be slightly higher for the lower pressure, since cooling takes place in the arc space immediately after current zero at a rate depending on the mean free path which is inversely proportional to the pressure. Complete deionization does not occur for any of the 50 amp. curves up to 130 microseconds. It was impossible to obtain points beyond this time, because of the very unstable arc which then existed. i ----=~·Rf+-· . . '· . ~ :· - • - I· - -:_:J-·. ·:.! • 1 -J_~ _:-_Ui~ i . - - . . ' . - -- . ' -I. I - - , ·- ----1-- OJ\i Olll..1 ' . I ; --~ -~ -~~_:_· . -· iI . ' I I • __ :.__ ---L. - . n1- 'V, . ·-:-T . ' ' ' ;-:_-:_j- _-• '": ~= J. •:' --· .- .,.·-.·- -............... . ',.._ • .: r • - ~ : !:~1 =-:_ii : ~-~i ~ ~ ~· -- ·~ ~-==.:~ ·-· -· .: :=:r=-~: .:..:.; .-:... I . . . -+ ·-I ')~~ .... • • • .. I I . ·-- . •+: 1. ... !. : ---r. . •. --~ · I . · . I : -- ' . I .. -1 . ---·---. i. . . ~ ! . . . ; . . !. I -.II ·1 I ·--r-. • I : . ' ---j ' ' . - - -!. . I I -- ''- -!' ' '' - ·-l- : . . :_ ... I ·: . ' --.. :: .... -·· , . I . . .. ; _·- . :: . ... I _ . - . ,.. - . \:: -· l-.. ,.. .. . . .......... --·- . - . . ... .. . .... . ... . .. . ·- . ~-~ : ~· ~~ J. .! : ~ -~· . ·- f i ' -;-·-~ : • - • : ..• I -. ..j ·--• ! • .. ...... _ ~- - • • .:~ ... .. - . - -. ,. . -·~- -_ • . i • ...__T • ~ •• -~ • - ' . ..,. •··. . ··- ... <- . . •: f - 1 • .. . -:--r --. . . ••• -r • • ·; •• : - ~ : . ___,_ · I ·· • ' I I . " ; I . . . : r I I • -· ~i- . • .. . '. ; ' -· . -----' --·-- --:-1-· ' . . '·-; : : !. . _ _l__ - .. t • _··; .-:- . I :·:]: ·1 ·:· I ~ . .4 • : l : i I -- - : - . ----J-- D i - c:1 t ) 19. Proceeding now to a consicleration of Figures 50, 53, and 56 ( (25 amps.; 4.13, 2.05 and 1 mm.), we find that the 4.13 mm. arc sho11•s complete deionization at about 80 microseconds. The 2.05 mm. gap shows a somewhat greater gain in di~ectric strength for the 1/2 atrnos. case, deionization appee,ring to take place very quickly. For 1 atmos. pressure, however. as for 50 amps., deionization is not completed in 130 microseconds. The minimum breakdown voltage, which we should expect to be independent of current, is very nearly the same for 50 as for 25 amps •• Figures 51, 54, and 57, showing the effect of 12.5 amps. at the various spacings, give results which are ver~ similar in form to those obtained for copper. After deionization takes place, the dielectric strength of the air space should be directly proportional to the density; i.e., to the pressure; the temperature being assumed constant for all pressures. A significant fact to note is that for the 12.5 amp. arcs the initial breakdown voltage is about 100 volts higher on the average for similar spacings than that existing for the 50 or 25 amp. ca.se. This lends weight to the theory that the air is partially replaced by vapor from the electrode material, which has a much lower recombination rate, thus lowering the initial breakdown voltage. The 50 amperage arc would, of course, produce a much greater evolution of vapor than an arc one-qua.rter its magnitude. 20. 4. Effect of Arc Length - (Steel) Figures 58 to 66 show the effect of arc length on reie,,ni tio!'l. voltage versus time. Fieures 58, 61, and 64, (a,t 50 amns., and 1, 1/2 and 1/4 atmos.), show tha.t arc length has very little effect for this value of current either on initial restriking voltage or subsequent rate of rise ,,vhich is very slow. As is expected, for later times the longer arcs sho~ slightly higher breakdown strengths. Examining the curves for 25 and 12. 5 amps., the sa.me tYPe of voltage recovery takes place R.s was discussed for the copper electrodes. Here again we find that the short arc lengths have a higher dielectric strength for short times after current zero than do the longer arcs. 62, 63, 64, and 66. This is clearly shown in Figures 60, 61, Again a preferred arc lengtr. exists somewhere between 1 mm. and 2.05 mm. for the 12.5 amp. case at which shorttime reignition voltage is a maximum. Again there is a remarkable similarity between Browne's Figure 40 for brass, Figure 60 for steel, and. Jl'igure 31 for copper. These curves show conclusively that the dielectric strength of short arcs is independent of boiling point. C. Results - General Although it is not the purpose of this thesis to correlate data obtained with that found for brass, still results are so analogous, especially for the case of copper electrodes, that .,, " i .. : • . I .. --~ -·· .•. ,_L.,_ .:. • I I .. ' . ; , ' . .- . . ' ·1· - ' .... .. . . . _.:._:..: :....:...:_ ..:.: ..:::-.:... .. : : ; : r ~: .. •••• . .. f ·:· • • • : .....,._ .. .. ;. " ' ~ - :. •: ! • . : . . ... .... . .. : .' · • 1 .. . ~.: L:;: .... :: :1-~! ... .::'!·; .. . . . ~ '- . ......- -. ~ - l - ' "t~· : .. --- •..:..:l:! • - ••• :... ::-·· ..... ...····p· --:: --.·.-·.:+:·:... 1 . .. . . :' .l . . . . - .1··_ ._.,. . _i_:..:... ··r . ' : • . . I . . . .. . ••• .,-... . .. .. ........ ' ' '· , ...... ·-··· . • - • • ·- .. ..... •t • ..__ • ~ ' t--4 ....._ • 6I-::.:; :4;· ~-: :: . ! : - . l "';:" -:~ .. II . .. • -·· .. . _,. .. . ·: -j . :... :. ! ·-·. .! : ! ". .. -I I ! .. I ~ ~ . i : ~ .. . . -· . ... . . -· .. . ·~:... -~-: • • • -· t :: . : . :: : I .. . . . . . I · · - . •l • . . t :. 21. results obtained there can be applied directly to this case. It will be noticed tha.t the curves almost without exception show a definite transition point between two distinct regions: the first, a concave upward curve; and, the second, a nearly straight line. The first concave-upward portion represents dielectric recovery due to space charge growth associated with deioniza.tion, as 11 given by Equation 5, Appendix E, and as discussed earlier under Resultsn. The second, usually a straight-line recovery, repre- sents increase in dielectric strength due to the effect of cooling on the gas density. Assuming the latter to be true, the 'tempera- tures indicated by the striking voltage in air in this second region were calculated for three cases, (Appendix D). For steel at 1 atmosphere and 1 mm •• the extrapola.ted temperature curve gives for the a.re temperature a value of 27500 K. ; for 2. 05 mm. , a va.lue of 450Qo K. i~ obtained. Both these values are for 12.5 a~ps •. A calcul.a.ted point for copper at 1 mm., 1 atmos. and 50 amps. shovrs an arc temperature of 35000 K. Brass electrodes under the same cond.i tions yielded 38QOo K., - close enough agreement. For a D-C arc between copper electrodes, Sui ts o·btained a value of tempera.ture of 40500 K. 22. V. CONCLUSIONS The following conclusions may be d.rawn :(1) Dielectric recovery by the arc space after current zero is due to the formation of a space charge sheath, as predicted by Slepia.n. (2) The chief deionizing process for short arcs is not recombination. but is diffusion. ( 3) The initial recovery vol ta.t-;e is lowe1·ed by increasing current. (4) An optimum spacing exists between 1.0 and 2.05 mm. for copper electrodes at which the reignition voltage is highest for short times after current zero. (5) Short arcs in copper have a higher dielectric strength than long arcs over the range shown by the curves; i.e., over the same range at which brass exhibits this property. This definitely disproves the theory that this property holds only for low boiling point electrodes. (6) For steel electrodes the very short time breakdown voltage is a\)oui; 100 vol ts higher for 12. 5 amps. than for either 25 or 50 amps., supporting the theory that the a.ir is partially replaced by electrode vapor. 23. ( 7) She rt arcs in steel have a. higher breakdown strength than the longer ones. This is very definitely shown for a..11 c-.u·rents and pressures used, except a.t l atmosphere and 25 and 50 amps., where the curves are too closely grouped. for any ane,lysis. (8) A preferred arc length between 1.0 and 2.05 mm. exists for steel electrodes also, for which the breakdown voltage is a maximum. (9) Temperatures are calculated which agree in order of magnitude (40000 - 6000o K.) with recent determinations. There still remains considerable work to more thorougb.ly analyse the curves contained herein. It is hoped that further correlation can be made in the near future and the results published. 24. VI . ACKNOWLEDGMENT The writer wishes to ex:press his grei,ti tude to the Electric.al Engineering Faculty of the Californie. Institute of Technology for thAir coopere,tion, to Dr. T. E. Browne for his excelle.nt ir1structi.on and inspiration in this work; to Mr. T. Fahrner for aid in obtaining data; and t.o Miss Esther Gilbert for her invaluable help in preparing this thesiiL To the National Youth Administration a. special acknowledg-, ment is due, - for 1,117i thout the aid of the two assistants whom it su.ppli ed for six months, the very laborious and long machine work and calculati.ons wo11ld have very much curtailed the sc<:rpe of this investigation. 2..," . !tl'ENDIX A. APPARATUS Fi@.ure 4 shows in more detail the actual circuit used. Three, single-phase, l K. V • .A.., transformers in parallel, together with a · 7. 5 K. V. A.. transformer in series. provided the power su:~:rply at 690 vol ts. The reactors, air-core. three in number, were designed for minimum distributed capacitance and consisted of 19 layers, ea.ch of 19 turns of No. 7 wire. Ea.ch coil ha,d a resistance of 0.63 ohms, an inductance of 0.0334 henry, and capaci.ty of 18 micro-microfarads. The inductance of the circuit was varied by taps on the reactors and also by varying the coupling between the coils. The electrodes were squa.re blocks, 5 cm. per side initially 1.27 cm. thick. The lower electrode was drilled with a No. 59 hole (0.9 mm. diameter) through which a tungsten wire was projected by means of an extensible bellows (Silfon) seal to make contact with the upper electrode and then, falling back, to s ta.rt the e,re. The steel electrodes 1vere shaped after each reading, polished smooth and plane with carborundum paste on a. fla.t disc .?.nd. then cleaned. With the copper electrodes it was possible to cbtai1'.! clean surfaces by- sandpa.pering; whe,n they became at all pitted, they were reshaped also. The electrode sup:porting structure was mounted on a drilled glass base. ground to make good contact with 26. e. 12 11 belljar ~nd the whole connected to a ve.cuum pump and a closed tube mercury manometer. The voltage divider was composed of s:i.x identical resistor- capaci tance sections in series. Each section consisted of a 600,000 ohm carbon resistor in parallel with a small, 50 to 100 micro-microfarad "trimming" condenser. The resistors were adjusted. by fili.ng to give equal D-C resistances, and the condensers by insertion into a tuned circuit to give equal voltage ratios at 50, 000 cycles. For small displacements on the oscilloscope a vol tage-di.vider ra.tio of one was used by connecting the l~ads directly across the arc, and for larger deflections a ratio of 2 or 3 was used. The divider was symmetrically arranged above ground to elimin.a.te errors due to ground currents flowin.g through non-symmetrical capacitances to ground of other parts of the circuit. The rest of the circuit. a.s much as possible, ~as also balanced ~ith respect to ground to eliminate distortion in the electro- st~itic deflecting field caused by high frequencies if one plate is closer to ground potential than the other. The oscilloscope had only one pair of electro-atatic plates 'll'Thich was connected across the arc through the voltage divid.er. Deflection for the sweep was obtained by magnetic field coils connected through a phase-shifting transformer to the same su:!;)ply as the arc. The coils were so designed tha.t deflection of the spot across the tube screen took ple,ce onlji~ when the :field current. 2?. was nea.r zero, end so a linear sw~ep was o·btained. Condensers in series with the coils made the circu.i t a pure resistance. The reignition phenomena to 'be observed mere centered on the screen by means of the phase shifting transformer. By applying a voltage of the same freq_u.ency as the arc supply. but displaced from it by 900, the spot was cut off on its ret-arn path so that the reigni- tion phenomena recurred for one polarity only. The concensers used for shunting the arc were of two ty:pes: the smaller, mica insulated in molded composition; and the larger, ordinBry "telephone" condensers of rolled paper. All were accurately measured at frequencies from 10 to 50 kilocycles, the smaller in a 1;,;med circuit, and the larger with a condenser br;.dge. The arc and a.l.l measuring equipment were about 15' from, and on a line per:;iendicular to the axis of the reactors, to avoid the latters'magnetic field. The transformer voltage was read by means of a standard 20-1 potential transfonner. A 200 volt D-C supply with voltmeter was also available for calibrating small deflections on the oscillo- scope. 28. APPJi;NDIX :B. EXP]::R.!MENT.AL TECHNIQUE Data were taken in such a manner as to obtain curves of reignition volts versus time after current zero. The current pressure, and electrode separation were kept fixed for each set of readings and the arc shunt ca.pa.ci tance was va.ried to act as a parameter for the required curve. The a.re current was taken to be the short circuit current flowing when the system "ll"as shorted on itself. This current was checked regularly and. its value altered by changing the ta.ps or the coupling of the reactors. The pressure was adjusted and controll!=!d by the vacuum pump and manometer. The separa.tion was obtained by the use of gauges of the correct thickness. The lower electrode was first clamped into its supporting structure with the ini tie.ting tungsten wire near the bottom of the centrally drilled hole. The top electrode was then laid on the lower and its movable hold.er brought down to it and clamped. The top holder and elP.ctrode were then ba.cked off and clamped firmly. using the gauges as separators. By this means it was unnecessary to do any compli- cated levelling to make sure the faces were parallel. The tungsten was then pushed through the lovrer electrode to make contact with the ·u.pper one, a q_uick return being obtained by a fa.irly strong spring. Since the oscillograph sensi tivi ty varied somewhat, depending on the width and intensity of line ussd, settings on the focusing 29. or accelera.tion anodes, a.nd line voltage, it was calibrated for every rea.ding. All deflections were read by means of two glass slides ~:hich moved vertically across the whole face of the tu.be. A millimeter scale capable of small la.tera.l adjustment 11'1as fastened horizontally below and above the screen so that positive and ne(5:ative distances from the zero mark could be read. Before every readi~g the zero on the scale was adjusted to correspond to the position of the electron beam. The line voltage was applied to the plates a.11d the deflections measured in both positive and nega.ti ve directions. The applied voltage ras read through a standard 20-1 potential transformer and a vol tme.ter. This deflection was usually 33 mm. in the positive direction, and 35 mm. in the o~posite direction, using a voltage divider ratio of 2 and peak a.pplied voltage of 1000. A 200 volt D-0 supply was used for reading directly the voltage corresµonding to any deflection up to 15 mm. (V.D.R. = 1). Between these two :90ints a constant mm. per volt sensitivity existed. As soon as the arc wa.s started, one slider was moved to read e 6 , the average reignition voltage Emd the other to read average em. The a.re circuit wa.s broken end restarted several times to check these va.lues and also to obtain a reading of E1. a. voltage divider ratio of one was used.. Wherever !)Ossi ble, With steel electrodes at other than 1/4 atmosphere, the initiating hole was usually blocked '111i th molten metal just after the a.re was struck so tha.t all three 30. readings e 8 , em. and El, had to "be obtained while the arc .r;as burning the first time. W'nen it was felt that these rea,dings were not sufficiently definite or reliable, a new set of electrode surfaces was used and the reading repeated under the same conditions. Several precautions were observed. In some cases, particu- larly at the lower values of shunt capacitance, the arc did not In these cases, it was necessary to obtain the reading of move. es very quickly, ·before it fell as it invariably did when the a.re stayed on burned metal. Often by extinguishing and restriking the arc several times, it was possi"ble to make it move around on clean metal, in which case, the values on the screen were very uni form and well defined. For the higher values of capacity, the arc usually moved with great rapidity toward the edges where- it bowed outward.s. This condition of burning co-i.lld be noticed on the oscilloscope screen and no readings were taken when this condition persisted. The greatest variation in vah1es appeared in -em. the negative maximum. Often two well-defined and consistently repeated values a.pp eared si:mul taneously, either when two no glow traces appeared, or when the no gloi11 and glow case appeared, which occurred near the point where the transition took place from no glow to glow. Readings were taken and recorded of' all such consistently appearing values. 31. For each reading the following data. were recorded:Open circuit volts. Short circuit volts. Deflection in mm. corresponding to open. circuit volts in both directions. s.c. amps. Shunt capacity. Separation in mm. Pressure. Reignition volts, e 9 • Negative maximum, em. Initial voltage, Ei. Glow frequency. Behavior of arc, - kind of movement. etc. 32. APPElIDIX C. PROBABLE ACCURACY OF THE METHOD A very complete and searchine analysis of the accuracy of the experimental method followed, is given by :Browne5 and will only be summarized here. Readings of deflections were probably accurate to within five percent on the la.rge deflections and ten percent on the small deflections. Errors due to non-linearity in oscilloscope deflection ~!)'ere elimina.ted by calibrating as much of the scale as possible for each set of readings, and then plotting a curve of voltage versus deflection for the screen. The maximum probable error was perhaps ten percent. Voltmeters and ammeters were checked against standards and were accurate to within one percent. L=-~ is thus accurate to at 1 ea.st two 2 n f I The acvaal current in the arc was less than the these readings percent. Inductance calculated from measured short circuit current by about 5 percent, due to the neglect of the arc voltage. The capacita.nce of the condensers u.sed was measured to within two to five percent. The distributed capacitance of the circuit, found from the inductance and observed resonant frequencies of the circuit, was ta.ken as the mean of several rea.di.ngs to be 150 micro- mic rofarads. 33. !>lieasurement of electrode separa.tion was easily accure.t.e to five ~ercent, and pressure to one percent. Browne investigated fully tbe error occurring due to calculation of time by means of the simpHfied. equations (X) and (Y). which neglect series resistance and conductance in pe.rallel vri th the arc. - [ Using the full equation, including series resistance::z.t.. --.:&--c[ where El) cos w t (E w = + _l_ wC RC 2 L (E - ElJ sin~~ R2 4 L2 and. calculating with typical constants and R = 5 ohms, the effect of res is ta.nee of this magnitude was found to be quite negligible. From an analysis of several oscillograms. l3rowne found tba.t leakage conductance rose only when the a.re voltage was above 600 or 800 volts. Since the greater portions of the normal voltage transients preceding reignition for the cases tested are belo~ this figure, the evidence showed. tha.t leakage conductance could ·be entirely neglected without much error. The total overall accuracy was estimated to be a.bout five percent for vol tags and ten percent in the calculated values of time. 34 • .APPENDIX D. Tlll\'lP .EP.ATURE C.ALCULATI OM Slepian and Mason. give a. curve of sparking voltage versus p,£, for plBt..ne pa,ra.llel electrodes in air, (Figure 70). Knowing a. sparking voltage then, the corresponding value of d can be found. The apparent pressure is then:- Then sl.nce p Po 1irhere = d do is gas density at pressure Po and temperature To Then since T = Todo d :: 2?3 -1lo_ p in o X. The assumptions included in this determination are:(1) That Slepian 1 s and Mason's curve can be extrapolated back as shown dotted. (2) That To can be taken as 2730 K. (3) Tbat the perfect gas law holds even though the arc space was not in thermal equilibrium. (4) That the gas temperature curves on Figures 6?, 68, and 69 have been projected back with the correct slope. 35. APPENDIX E. SPARKING IN IONIZED GAS A commonly used value for spa.rking voltage iu air at ordinary pressure and temperature is:V = 30,000 d volts (1) In more general form, applying Paschen•s Law:- v where T0 T = 30,000 d • _o_ (2) T = 2730 K. Now d and A = A = v2/3 (7) ( 3) . (~3 T . n (4) nl/3 (136) (...JLJ l/ 3 c Eliminating d, v = ~6.8)(1019~ k l c Taking into account changes in pressure by the use of (2) above, instead of (1), then:- v = (6.8)(1019) !!a. Co where 7. -n 1 • (- ~o J 3 p5/2 volts k0 = c0 = a.vera.ge thermal velocity at l atmos. and 0° C. p = k = ionic mobility. T = absolute temperatu.re of a.re gas. d = n = density of ionization. ( 5) mobility at 1 atmos. and oo C. pressure in atmospheres. thicJmess of space charge sheath. Slepian, "Electrical Propertie;-ofGase~(a.book) _________ Westinghouse Elec. & Mfg. Co. (1933). A,.~~ndi:s~ .. F ·· R.ei-an;f,on VoltQ9e' -Tiw.• Equaiions. = edl.. +- .$... £= al c. ta....::. o - d.t'1.. -- dt .. L <fl:.~ J,J- - .!/.. .::. 0 = Ldtc c d L {:$_ + !J_ d.1-,. c. A n~W :::: + !l c 0 f + O~wt '""' al- t;:. o or , l..1. :: - Jr::1q --ti cJ.f = -t'c: ,_,.A~t:.Jt -c;Ul5~"'->f ....,, -q_ A= F,.om ,, CD L = -wB = £ -.!L c B -= c,_( £-£,) '71'( (-1,.. pc. nol eQ.. = E - F' •• (t. r\ t-'d) L{ w"'-C{ £-£,) ~w-1: - ((.£-£,)-a+ (~~c~ )~ = E = •• c. . .: ""' wI,~ w-1:} ~ (AJi + ·~-~=I'==- t..»C(E-E,) E - £,.,, ~ ((.c)t t- ti) whcu•e Em ~ ()"' ~ -'··=·=h;.-1- 4>C{E-EJ f<;:zy t:!c)~ + I, ::: w C (£,:. ~ (E-E;)l. £=~·<='1t1.. E- em '4) t = Cffe _, E-er/JL - e £-tf!!m .r-~ = ::t11 pJ..C ) 38. APPENDIX G. ARC Jl~JiAVIOR Due to the large variety of conditions under which the arc behavior was recorded. the data are given in note form below. with an explanatory summary following. In the notation used the first figure denotes current, the second separation, and the third the pressure. COPPER 50 - 4.13 - l Rapid increase in gas pressure: violent flames when arc lee.ves the electrode edge. 50-1-1 For low shunting capacity, a soft sounding arc; as .Q increases, the arc becomes staccato, completely covering the electrode with a spotty trace; arc bowed out into length of 8 inches. 50 - .45 - l Arc self-starting. 50 - 2.05 - 1/2 Continuous spider-like trace to edge. 50 - l - l/4 Large burnt spots, no depth of burning. 25 - l - 1/2 Arc goes out after 2-3 seconds. 25 - 4.13 - l Continuous trace up to Bl seconds. then goes into explosive movement. 25 - 2.05 - 1/2 e 8 drops by 50% when the arc reaches the edge. 25 - l - 1 Hard to keep arc going when Q is increased. 25 - 4.13 - 1/2 Great increase in pressure. 39. 25 - 4.13 - 1/4 Arc travels to the side very quickly but vi•i th no preferred direction; es drops by 30% and em increases; no violent flames as with 1 atmosphere; soft fuzzy electrode traces. 25 - .45 - 1/4 Self-starting; after surface is covered, arc t\1rns to pu.rple glow. 12.5 - 4.13 - 1 Start of sta.ccato trace at 138 sec. 12.5 - 2.05 - l Start of staccato trace at 75 sec. 12.5 - 1 - 1/2 At high Q very diffuse arc; after electrode covered arc changes to diffuse -purple glow; develops great heat. 12.5 - 1 - 1/2 Complete diffuse coverage of all electrode surface. 12.5 - 2.05 - 1/2 When oxidized around center hole, arc could not be restarted. 12. 5 - . 45 - l/ 4 Arc changes to glow in 6-8 seconds. 12.5 - 1 - 1/4 Trace consisted of black spots mostly near edge. STEEL 12.5 - .45 - 1 No readings could be taken at .45 mm. because arc was continuously shorted by thin spires of metal soon after arc started. 25 - l - l Two main craters; reading had to be taken quickly before char~wteris tics changed. 40. 25 - 4.13 ·- l Initiating hole weldecl over. Rapid movement only at high v~11es of Q. 12. 5 - 1 - 1 Very rapid movement, maintained only for 2-3 cycles. 12.5 - 4.13 - 1 Difficult to maintain arc. 12. 5 - 2. 05 - 1 Difficult to maintain arc. 50 - 2.05 - 1/2 Da.rk,fine :path: good movement. Arc someti:nes shorted out by spire of metal. 50 - 1 - 1/4 Soft arc. 25 - 1 - 1/2 Good coverage - spotted trace. 25 - 1 - 1/4 Arc self-starting; electrode roueh as though arc had passed but not stopped; some small molten spots also. 25 - 4.13 - 1/2 At 101J1.T values of Q. total movement 'IJ'Yi thin 1/2" diameter circle. For higher values more movement. SUM.W.i.ARY For low values of shunting capacity, the arc tended to burn near the center o~ the electrodes. arc movement became more rapid. As the capacity increased, the When it reached the edge for copper electrodes at 50 and 25 amps., the arc bowed out in violent fl~~es. The manometer indicated a rapii increase in pressure, ofter1 great enough to send the mercury in the manometer tube surging from its 41. position recording a l/2 or 1/4 atmospheric difference, to striking the top of the tube with considerable violence. This effect is well known and has ·been described in early li terature8. The different kinds of traces are shown in Figures 71 to 81. All the conditions holding at the time they were ta.ken are noted under them. Figure 81 shows an enlaxged view of the center part of Figure 80. The hole shown is the initiating hole of diameter .9 mm. Measurements indicate a current density of apµroximately 10,000 amps. per square cm. for this case. Rougher calculations for Figure 79 sho1.11.r a current density of the same order of magnitude. This agrees with Langmuir's criteria for the high field theory of cathode spot. In general es wes much lower when the arc did not move, so that in such cases the circuit was broken and the arc restarted until it did move on new metal. It was also noticed (see 12.5 - 2.05 - 1/2) above, that it was harder to start the arc i f the metal around the initiating hole had been previously burned, showing that the oxides and other products of burning when cold, support the arc with more difficulty than the parent metal. Another point which is of interest is the fact that the axe ceased its continuous trace and broke into a fast spotty motion at roughly the same time when transition between the two kinds of dielectric recovery took place. 8. Thompson, J. J. and G. P., "Conduction of Electricity through Gases", (a book), Vol. II, page 566. Cambridge University Press, (1933). 42 . The motion of the arc a.cross the electrodes has been discussed in literatures, and its cause is fairly obvioQs. The addition of 107 ergs of energy to 1 c. c. of ga.s would i l"lCrea.se the pressure by 6.5 atmospheres. Consider this same energy 1 c. c., a venr large over-estimate of the volume 100 of a short arc, then the pressure rise of 660 atmospheres is experi- applied to enc ed. This high pressure would spread as a wave giving the 11 bomb" effect descrH>ed above, but would also cause the ionized gases in the core to move. The lowest dielectric path then usually occurs at a position slightly different from the previous arc -position, and the arc then reignites at a new point. Copper ~o- , 4-~- I e ::, ::: '14 ~ t-s = I 3j U 2 C•p p(i! (' S'o-1 ·I C!:. : S'S-4-:. I 8-l::. t:. Stee:/ volh.. c 0,, per Fag. 'l4 Sb - :;:i ·OS - ~ e:. ... 8 43 -t~ .:: 4- 1 ~ - ;a.os- e~ .:: ~C\ 4 t :. =- n r-s I .LtCJ. TS j C.of:>f"et' e~ 50-1-I 1 i io .:::. t ~ .:. 14-0 0 ~ ~-1-i C ofop-.r· e~ .... t~ :. 6 S 4~1.s C.ofc>pc.t> "S'D-. 4S"- I C o pper ~s- ~.o ~- I es= ,.. ~s es::::. t:~ 3G.. I t~ .. '-5' s = 30 :t"S' - , - -;_ . e~ = 4-<o~ ~""'""="~~-==-~t-~ : b·~~~-==~====-~--!~==..:;.:::;:::=~~~ , • "Ste e. I e~ =- + ~'f ts : 11. 1