6 %\. 31334 K23»- Digitized by the Internet Archive in 2017 with funding from University of Illinois Urbana-Champaign Alternates https://archive.org/details/sirwilliamthomsoOOkelv For Private Circulation,] 9 SIK WILLIAM THOMSON’S REPORT TO MESSRS. SIEMENS BROTHERS ON TESTS OF DIRECT UNITED STATES CABLE, TAKEN AT BALLINSKELLIGS BAY STATION, September 16 and 17, 1875. LONDON : PRINTED BY NICHOLS AND SONS, 25, Parliament Street, S.W. 1875. For Private Circulation.] SIR WILLIAM THOMSON’S REPORT TO MESSRS. SIEMENS BROTHERS ON TESTS OF IIBECT UNITED STATES CABLE, TAKEN AT BALLINSKELLIGS BAY STATION, September 16 and 17, 1875. LONDON : PRINTED’ BY NICHOLS AND SONS, 25, Parliament Street, S.W. 1875. % I XjlA. . ' . *?0 6 / f\jLc- dt io a i 3. \°i "*>4 K a 4 'f' SIR WILLIAM THOMSON’S REPORT. On Wednesday evening, September 15th, I arrived at Water- ville, and proceeded thence to the cable station at Ballinskelligs Bay. There your electrician, Mr. Ebel, met me, and showed me the instruments which he was ready to put at my disposal for the tests ; and Mr. Gavey, the Company’s superintendent, obligingly lent me a number of additional condensers which I desired for measuring the electrostatic capacity of the line. Having made preliminary arrangements, and learned that, by orders from London, the line was to be at my disposal from 7 till 10 next morning, I returned to Waterville for the night, appointing to meet Mr. Ebel at the station in the morning at 7 o’clock. About 8 a.m. on the 16th, after some preliminary trials of the instruments in connection with the cable, which showed strong earth-currents, I commenced testing for insulation with a battery of 20 cells, having its two poles constantly joined through a resistance of 20,000 Siemens units, but found so great disturbance by the earth-currents, that it was impossible to get a result. I first applied the battery for a quarter of an hour, zinc to line, then for seven minutes line to earth, then twenty minutes copper to line, and lastly, thirteen minutes line to earth. During the whole time a galvanometer, in circuit with the cable, showed strong currents alternately in the two directions, and varying from extreme positive to extreme nega- tive with great rapidity. To keep the readings within a conve- nient range I was obliged to shunt the coil so powerfully as to I 4 reduce the deflection to about —th of what it would have been with a degree of sensibility proper for measuring the insulation resistance. The deflection was read off and written down every ten seconds during nearly all the time. A careful examination of these recorded results shows no sensible preponderance of current in the direction due to the battery, whether with copper to line or zinc to line ; and no perceptible difference in the currents when the line was put to earth directly instead of through the battery in either direction. The currents observed were frequently ten times, and sometimes more than sixteen times, as strong as what I afterwards found to be the true leakage current, and the extremes were about as often in one direction as the other. The strongest current of all chanced to be in one of the periods when the line was simply to earth without battery. The sums of the readings taken during successive minutes show that the insulation resistance, whether with zinc or copper to line, cannot have been less than a megohm, and that it was probably not less than two megohms. I next measured the electrostatic capacity by the method which I described in a communication to the Society of Tele- graph Engineers, published in the Proceedings for 1873. I used three boxes of your resistance coils with 20,000 units on one side of the point put to earth and an adjusted resistance made on a box of 10,000 on the other side. The current from a well-insulated battery of 80 cells was kept flowing through these coils. With the condensers lent by Mr. Gavey, in addi- tion to your own, I had 80 microfarads in all. By means of the battery aud resistance coils arranged in the manner de- scribed, the cable and condensers were charged oppositely in measured proportions, then connected together, kept so for about ten seconds,* and then discharged through the galvanometer to earth. Mr. Ebel’s insulation galvanometer with its ordinary magnetic adjustment was used, but its coil was strongly shunted to avoid over sensibility. Thus, after repeated trials, I found * Five seconds would have been quite long enough. that when the cable and condensers were charged to oj^posite potentials in the proportion of 1,600 to 20,000, and then put in connection with one another, the charge in the cable was over- borne by that in the condensers ; but with the proportion 1,630 to 20,000 the charge in the cable preponderated. Hence, about 1,615 to 20,000 would have given zero; and, therefore, the capacity of the cable was 20,000 QA . , , — x oO microfarads, 1,615 that is, 991 microfarads (or -409 of a microfarad per knot, the length of the cable being 2,420 knots). I could have come much more closely to the exact proportion of the charges required to give the zero had the cable been at my disposal for half-an-hour longer, or had I perceived in time that one of the commutators which Mr. Ebel had left at my disposal could, in a few minutes, have been arranged to make the requisite connections by a simple manipulation instead of a somewhat cumbrous arrangement which I had extemporised. The quickness of this process, even with the cumbrous arrange- ment which I used, is such that it is much less disturbed by earth-currents than the ordinary tests for insulation or copper resistance. It is, in fact, disturbed only by whatever change there may be in the terrestrial potential along the line of the cable during the ten (or five) seconds between the insulated wire of the cable and the battery, and discharging the con- nected wire and condenser to earth through the galvanometer. The last quarter-hour of the time for which the cable was at my disposal was spent in a somewhat hurried measurement of the copper-resistance. The line was found to be still greatly disturbed by earth-currents. A battery of 20 cells was applied during 12 minutes, first zinc to line and copper applied to earth, and then suddenly reversed, and kept so till 10 o’clock (the time arranged for the conclusion of my tests), when signalling from the remote end commenced. During the whole 12 minutes there was a difference of potential between the Irish 6 and Nova- Scotian earths, varying rapidly in amount from a least minimum of 5 cells to a greatest maximum of 18 cells, but always in the same direction : — the Irish earth positive relatively to the Nova- Scotian earth. The ordinary bridge method could have given no result at all in so disturbed a con- dition of the line ; but the simple method of deflection (the only proper method for measuring copper-resistance in a submerged cable), in which is observed the difference of the readings im- mediately before and quickly after reversal of the battery, gave an approximate result, which in round numbers I took as 7,300 Siemens units. The cable being offered to me again from midnight till 2 a.m. on the 17th, I made another series of tests at that time, for the main object of measuring the insulation-resistance. I found the line in a much less disturbed state, and was able to make a per- fectly satisfactory insulation test by the ordinary galvanometer method. I applied, however, also a new method which (no electrometer being available) I had planned to meet the contin- gency of the line being disturbed by earth-currents so much as to render the ordinary test unsatisfactory, but not so much as to vitiate an electrometer-test. This method, which I think may be found generally useful for testing submerged cables when an electrometer is not available, is as follows : — 1. Apply the ordinary test by battery and galvanometer for a certain time. 2. Insulate the cable for a certain time and then shunt the galvanometer to prepare for No. 3 (unless you have conve- niently available a second galvanometer suitable for discharges). 3. Instantaneously re-apply the battery, through the insula- tion galvanometer properly shunted ("or a special discharge gal- vanometer), to the cable, and observe the maximum of the sudden deflection produced. 4. Go on repeating Nos. 1,2, and 3 as long as you think proper, according to circumstances. 5. To determine the proper ballistic constant of the galvano- meter for utilizing the observed result of No. 3, find the maxi- mum of the sudden deflection which takes place when a sudden change of electrification is produced by instantaneously chang- ing by a small measured difference the potential of one electrode of the galvanometer, the other electrode being in connection with the cable. 6. The change of potential which, in the operation of No. 5, would give the same deflection as that observed in No. 3, is equal to the change of potential which the conductor of the cable has experienced during the time when it was left insulated according to No. 2. Hence calculate the insulation-resistance in ohms or megohms as in the ordinary electrometer method, when the electrostatic capacity of the cable is known. At 12h. 2m. on the morning of the 17th, the 20-cell insulation battery (with its poles again, as on previous occasions, joined through 20,000 Siemens units) was applied, zinc to cable, through the insulation galvanometer with a shunt of 5,000 Siemens units on it. Then, commencing at 12h. 2m. 50s. the galvanometer indication was read and recorded every ten seconds till 12h. 6m., when the cable was insulated during a minute, according to No. 2 of the directions above, and a shunt of 30 substituted for the 5,000. At 12h. 7m. the battery was instantaneously re-applied, the throw of the galvanometer observed according to No. 3, and the shunt of 30 removed, and 5,000 re-applied. The battery was kept on till 12h. 8m., when the cable was again insulated for a minute, the galvanometer shunted with 50 (instead of the 30 used the first time), and the operation of No. 3 repeated. This process of re-applying the battery and re-insulating the cable, in alternate minutes, was continued till 12h. 26m. Then an interval of five minutes was spent in determining, according to No. 5, the proper ballistic constant of the galvanometer, by applying alternately full power and f g- of full power of the insulation battery : the change from one power to the other being made in each case as instan- taneously as possible. Lastly, the shunt of 5,000 was re-applied 8 at 12h. 31m. for insulation-test, and one more period of the alternating process performed from 1 2h. 32m. to 12h. 34m., when the cable was put to earth to prepare for insulation-test with cop- per to line. Either three or four, generally four, galvanometer readings for ordinary insulation-test were taken at intervals of ten seconds in the second half of each minute during which the battery was applied. Twelve galvanometer readings taken at ten seconds intervals during the second and third minutes of the electrification gave for mean deflection 127, and the readings taken during the second halves of the fourth, eighth, tenth, twelfth, fourteenth, sixteenth, eighteenth, twentieth, twenty- second, and twenty-fourth minutes gave for mean deflection 82 T. The sensibility of the galvanometer in the condition in which it was used for these readings was such that a deflection of 290 would have been given by the actual battery, with a resistance of 1 0 6 Siemens units. Hence, the insulation resistances proved by the mean observed deflections were as follows : — Mean deflection. Insulation resistance. 127 (2nd and 3rd minutes) 2*28 x 10 6 Siemens 82*1 (4th, 6th, . . . 24th ) 3*54 x 10 6 units. The new method described above gave the following ballistic deflections or u throws — End of 5th minute . . 70 divisions. >> 7th . 102 >> 9th . 102 11th . 109 >> 13 th ,, . 57 15th . 52 >> 17 th . 92 19 th J? . 102 >> j? 21st ?? . 110 23rd ?? . 102 Mean . . 89-8 Say . 90 9 The ballistic deflection due to instantaneously changing the potential by T V of that of the insulation battery, in accordance with the rule of No. 5 above, was found to be 112 divisions. This is 1J time the preceding mean throw, which therefore showed a change of potential equal to ^ of that of the battery. Hence the mean of the falls of potential in the ten alternate minutes during which the line was insulated was ^ of the potential at the beginning of each of them, or (nearly enough) ^ of the mean potential during the minute. Therefore, the loss was at the rate of ^ per minute, or 3000 P er second. Now, I had found the electrostatic capacity of the cable to be 991 microfarads. Hence the insulation resistance proved by this mean result is or 3*027 megohms, or 3.170.000 Siemens units This agrees quite as nearly as could be expected with the 3.540.000 Siemens units deduced from the means of the galvanomer deflections during the alternate minutes when the battery was in action, as described above. At the conclusion of the u zinc to cable ” test, the cable was put to earth and kept so for two minutes, till 12 36, when a similar series of tests with copper to cable was commenced, always with the same battery of 20 cells. These tests were much disturbed by another storm of earth -currents (not quite so severe however as that of the preceding morning) which came on very suddenly about the fourth minute of electrifica- tion, and nearly stopped the leakage current (giving an apparent insulation-resistance of 7,000,000). The new ballistic test was then applied and continued in alternate minutes as before. The first ballistic deflection gave an apparent resistance of 5,000,000, and the second actually showed an increase of potential (rela- tively of course to the Ballinskelligs earth) during the minute of insulation. Then came ten minutes of comparative tranquillity, till 12h. 52^m. when the leakage current from the battery was 10 rapidly reduced to zero and reversed, and a quick succession of violent pulsations supervened for about a minute and a half, throwing the spot of light alternately off scale to right and off scale to left, at irregular intervals of ten or fifteen seconds. About 12*56, there being still great deflections, but not so rapid pulsations, a shunt was applied which allowed the amounts of the deflections to be observed, and showed them to range fre- quently from ten to forty times the proper leakage current, some in the contrary direction to it, and some in its own direction. The disturbance diminished gradually till 111. 7m. A.M., when the cable was discharged to prepare for measurements of capacity and copper resistance. Notwithstanding the very disturbed state of the line the means of the regular observations taken between 12*39 and 12 53 gave satisfactory results in respect to insulation resistance. Thus the mean of 24 deflections observed in the ordinary galvanometer test during that interval was 112*2, which gave for insulation resistance 2,585,000 Siemens units, this being for the time from the end of the fourth minute to the end of the sixteenth minute of electrification. The ballistic deflec- tions* observed were : — End of 5th minute + 52 divisions „ 7 th >> . - 10 >> „ 9th V . + 80 „ 11th . +160 „ 13th . + 92 „ 15th >> . +102 >> „ 17th . + 62 538-=-7 Mean . . 77 This with the ballistic constant, determined in the manner ex- plained above, gives 3*52 megohms, or 3,690,000 Siemens units, for the insulation resistance. 11 The measurement of electrostatic capacity described above was next repeated, and the previous result confirmed, but there was not time to attain to more minute accuracy in the ad- justment. Lastly, the copper resistance was measured by the simple gal- vanometer method. The insulation-galvanometer, quickened three or four fold by a magnetic adjustment (which I had used also in the insulation-tests), and with a shunt of 20 Siemens units on its coil, was put in circuit between line, battery, and earth, and the deflection was observed and recorded every ten seconds during the whole time of the test, which was from lh. 36|m. to ill. 58m. As was to be expected, large and rapid variations of the deflection were continually taking place on account of earth- currents. The direction of the earth-current was from east to west the whole time, as was shown by the u copper ’’-current being always greater and the u zinc ’’-current less than the true mean concluded from the observations. It increased gradually (but with some slight backward pulsation) from the beginning (lh. 26 Jm.), when its amount was that due to a difference of potentials between the Ballinskelligs and Torbay earths, equal to 1*7 of a cell (one cell and seven-tenths), till the end (lh. 58m.), when it was more than five times as strong, and corresponded to nine cells ; the Irish earth positive relatively to the Nova-Scotian earth the whole time. To measure the copper resistance, a time of comparative tran- quillity was chosen, a reading taken, and then as quickly as possible the galvanometer short-circuited, the battery reversed, the galvanometer circuit re-opened, and a fresh reading taken. Half the space travelled by the spot of light from the first reading to the second is taken,* as being the deflection which would * Supposing there to be no instrumental error, the sole error in this process is that depending on the change of earth-current between the first and second read- ing, hence the importance of quickness, and the value of the “ dead-beat ” galvanometer for such observations. So far as the cable is concerned five 12 be produced by the battery applied in either direction were there no earth-currents. This was done seven times, and the half-ranges found were as follows : — 235 231 229| 2344 231 235 230 Mean . . 232*3 At 2h. 2m. I found that the same battery applied in the two directions through the galvanometer and 7,300 Siemens units gave 232 divisions on one side of zero and 233 on the other — mean 232*5. Hence the copper resistance to be inferred from the observations is 7,300 x 7^ ^ or 7,306 Siemens units. £ o seconds (as I find from my mathematical theory) is amply sufficient from the instant of reversal to the second reading to secure that there he no sensible error on account of the current not having become perfectly uniform from end to end. But far more than five seconds is required to get the second reading, on account of the swinging of the galvanometer needle when the customary “ astatic mirror ” is used. With the dead-beat galvanometer the second reading is easily taken within five seconds of the first. Summary of Tests of Sept. 16 and Sept. 17. Insulation Resistance of whole cable of 2420 knots in 2nd and 3rd minutes . . 2J millions Siemens units, or 5,445 millions per knot, from 4th to 24th minutes . . 3^ millions, or 8,470 millions per knot. Copper Resistance of whole line .... 7,300 Siemens units, or 3 02 „ „ per knot. Electrostatic Capacity of whole line .... 991 microfarads, or *4,095, say *41 per knot. Throughout the preparations and observations re- quired for the tests which I have now described I received skilful and efficient assistance from your electrician Mr. Ebel and his assistant ; and I desire to take this opportunity of expressing through you my thanks for the patience and care with which they went through the somewhat irksome and tedious series of operations which I had to ask of them. In conclusion, I am glad to be able to say that my tests proved the cable to be in perfect condition as to 14 insulation, and showed its electrostatic capacity and copper resistance to be so small as to give it a power of transmitting messages, which, for a transatlantic cable of so great length, is a very remarkable as well as valuable achievement. (Signed) WILLIAM THOMSON. The University , Glasgow , Sept. 23 , 1875 .