- Email: [email protected]
A null ballistic method for comparing capacities
Sept., 1918.1
WESTINGHOUSE LABORATORY.
377
less by temperature changes. In the paper-impregnated samples }there is a maximum or a decided hump in the curve. In general a very decided change in the specific inductive capacities of materials is, seen to, acco~npany change of state of materials. The direation of the change seems to bear a definite relation to ,the composition. Further work would seem desirable along this line in order to determine ¢~,e cause of this behavior.
A N~LILL BALLISTIC METHOD FOR COMPARING CAPACITIES. By Dr. Phillips Thomas.
IN the course of some work on the dielectric properties of various chemical combinations, it was found that ,the preparation of more than a few cubic centimeters of the compound was undesirable until some preliminary knowledge of its properties had been obtained. Consequently, it became necessary ,to measttre with fair accuracy capacities of the order of lO-9 ,to lO-11 farads. It is possible to do this with considerable precision by any one of several methods, provided only that certan factors :which enter as correcticm terms in the results be accurately known. The troublesome corrections involved, together with ~he fact that values were particularly desired at equivaler~t frequencies lower than those convenient for use with standard methods of test, led the writer to devise a simpler method, embodying only easily obfainable or readily constructed apparatus, with which the corrections should be easily determined and simple to apply. The method arrived atl while not fundamentally new, has proved so serviceable and so surprisingly accurate that i~t is here offered for the consideration of others who may be confronted with similar problems. The idea of measuring the absolute capacity of a condenser by the discharge through a ballistic galvanometer of a single impulse from the condenser is well known. So also is the modification in which the single limpulse is replaced by a 'train of similar impulses produced by some form of commutator which permits only the charge or the discharge impulses, b u t not both, to reach the galvanometer. The capacity is given in terms of the current figure
378
VVESTINGHOUSE LABORATORY.
[J. F. I.
of merit or sensitivity of the galvanometer, the charging voltage and speed of ~the commutator. This method is applicable to very small capacities if thei. available aharging vol'tage is sufficiently large. There remains, qliowever, the difficulty of measuring accurately the commutator speed and the charging voltage, also the added and somewhat troublesome one of keeping bot.h speed and voltage constan.t; also 'the galvanometer must be checked at pro. i. ~ 4
3
I
~
,<
-
-
V
-
'
G
o,I
-
.I,
Jvw--
I
__.] F'-"
frequent intervals, and finally all the disadvantages of a deflectiort method are present. The present method is a modification of the above, in vd~ich the train of impulses f r o m one condenser is balanced by a second train from a different condenser, passing ~through the galvanometer in the opposite direction. The apparatus .and conn'eetions are shown in Fig. I,in which K is the eommu'ta~or, rotating counter clockwise as shown ; C1 is the condenser to be measured, C= a known condenser, usually a standard subdivided microfara.d, R1
Sept., 1 9 1 8 . 1
WESTINGHOUSE LABORATORY.
379
and R~ are standard dial rheostats. The commutator presents an 'insulating surface to the four sets of brushes, numbered I to 6, except for the solid segment, which latter is metal. The angle of the solid segment is 89 °, and the brushes are set in pairs on 9° ° lines, so that no two pairs of brushes are ever imerconnected. As the commutator revolves it connects together in sequence, brushes I and 2, 2 and3, 4 and 5, 5 and 6, and so on. If the switch S is .closed in position A, the following cycle is completed once every revolution of the commutator. Left hand numbers indicate b r ~ h e s connected. I , 2 ( I ) - C o n d e n s e r C~ is charged to a voltage V2, equal to VR~ RI+R2
2', 3 (2)-Condenser C,_, is discharged through the galvano,meter, G . • ° (3)-Condenser Ca is charged to ~ voltage V1, equal 4,5 VRa to RI+R2
5',6 (4)-Condenser,C1 is discharged through the galvanometer in ,the opposite direction to that of the discharge from C~. Since the directions of the discharge impulses from Ca and C2 through ~the galvanometer are opposite, there will be no permanent deflection of the galvanometer light spot if the discharge impulses are equal. This condition is realized when R1 and R2 are so adjusted that NC1V1 -- NC,V2
or, since V1 and V~ are proportional to R a and R_~, CI=R2 C2~R1
(1)
By throwing switch S into position B the galvanometer receives the charge impulses instead of those of discharge. In testing samples, a balance is obtained for both charge and discharge; a difference of more ethan one pe r cent. between the indicated capacities shows t h ~ the results are not accurate to one per cent., due to different relative amounts of absorption, leakage, or both, in Ca and C2. The result given by equation ( I ) is the comparative value of the total capacity in branches C1 and C2. T h i s value nmst be corrected for the small permanent capacity in the two branches, when the condenser Ca and C2 are disconnected. Tha use of a large condenser on the C2 side enables one to neglect this correction VOL. 186, No. III3--3o.
380
\VESTINGHOUSE LABORATORY.
[J F.I.
on .that side ; the correctior~ fo.r Ca is easily applied by disconnecting Ca, obtaining a balance, and subtracting the. indicated capacity from that found when Ca is connected. Thus, CI+CI=
hence
C~R2 Ri
C1 connected
e 1= C~r2 rl
C1 disconnected
C1 C~Ru
C2r2
If C2 equals co, and R1 equals rt, we have C, (R2-r~) c1 =K,
(2)
which enables the correction to be made by simply subtracting the value of R2 found with C1 disconnected from that found with C1 connected, and using the remainder inl equation ( I ). Obviously, the commutator speed for any capacity must not be high enough to prevent full charge from enltering the. condensers during one charging ir~terval. Fortunately, satisfying this condkior~ for C1 also does so for C2, because the resistance in the charging circuits are R~ and R2, and the two time con.~ants, R1C~ and R2C2 are equal by equation ( i ), when balance has been established. This condi,tion, however, sets an upper limit to the product N C R ; in practice the maximum permissible speed corresponding to the desired product RC is taken from a curve kepi~ with the apparatus. T h e upper limit in capacity or voltage is set by the appearance of strong forced vibrations of the galvanometer moving system, due to the fact ,that the charging or disctmrging impulses, while equal, do not reach the galvanometer at the same instar~t, but are spaced a half revolution apart. T h e accuracy obtainable is of course decreased when the ligh!t spot is thus broadened out into a band of light. The lower limit of accuracy is set by the relative values of the residual or permanent capacity in the C1 branch, and t~he capacity being measured. T h e method is very easy to use, and its sensitiveness is limi~ted only by the current sensitivity of the galvanometer. Small frictional voltages induced by the rotation of the commutator prevent the use of the most sensitive galvanometers obtain'able, but a galvanometer of sensitivky lO-9, current, is good in this'respect and
Sept., 1918.]
v'~rESTINGHOUSE LABORATORY.
381
sufficiently sensitive to measure IO-10 :farad of capacity to one percent. on a few volts. The fact that neither voltage nor speed enters into the equation giving the results, makes it possible to obtain curves of variations of capacity with voltage and with equivalent frequency, since in most cases these coefficients of change are small, and one need not employ means for either closely controlling or very accurately reading either speed or voltage. Neither the sensitiveness nor the accuracy is affected by the relative magnitude of C a and C2, assuming that R 1 and R 2 are accurate over any range desired. Hence very small cor~densers may be compared directly with a large standard subdivided microfarad, in order either to calibrate the small or th:e large condenser. This is a great advantage and is possessed by none of the standard methods of testing. The writer uses, commonly, fixed values for C2 and R~, of 0.05 microfarad and 5000 ohms respectively, resulting in ~he convenient equation C, = I O - "
(R~ - - r2)
Rheos!~at R 2 has dials giving IO x o. I and 9 x (I+IO+IOO+IOOO). With Ca disconnected the galvanometer is sensitive.to one division on the o. i ohm dial with an e.m.f. ( V ) of IO volts, and the balance gives C~ =' 5-4 X IO- ~ farads
as the Ca residual. Hence if one per cent. accuracy must be attained, IO-a° farad is the lower limi,t of capacity which can be measured. A new commutator is now being built, 'having two segments each 45 ° long instead of one of 9 o°. With this will be used brushes set in/pairs at 45 ° instead of 9 o°. This will cause Q1 and Q_~ to reach the galvanometer at the same time, and will thus do away entirely with the forced vibrations of the moving system. It has been found that with certain modifications, this apparatus may be used to measure the insulation resistance of samples as well as their capacity. It seems probable also that the effects of dielectric polarization and absorption on b~th resistance and capacity may at leas.t be estimated, if not measured quantitatively. This work is now in progress, and the results will be embodied in a later paper.
WESTINGHOUSE LABORATORY.
377
less by temperature changes. In the paper-impregnated samples }there is a maximum or a decided hump in the curve. In general a very decided change in the specific inductive capacities of materials is, seen to, acco~npany change of state of materials. The direation of the change seems to bear a definite relation to ,the composition. Further work would seem desirable along this line in order to determine ¢~,e cause of this behavior.
A N~LILL BALLISTIC METHOD FOR COMPARING CAPACITIES. By Dr. Phillips Thomas.
IN the course of some work on the dielectric properties of various chemical combinations, it was found that ,the preparation of more than a few cubic centimeters of the compound was undesirable until some preliminary knowledge of its properties had been obtained. Consequently, it became necessary ,to measttre with fair accuracy capacities of the order of lO-9 ,to lO-11 farads. It is possible to do this with considerable precision by any one of several methods, provided only that certan factors :which enter as correcticm terms in the results be accurately known. The troublesome corrections involved, together with ~he fact that values were particularly desired at equivaler~t frequencies lower than those convenient for use with standard methods of test, led the writer to devise a simpler method, embodying only easily obfainable or readily constructed apparatus, with which the corrections should be easily determined and simple to apply. The method arrived atl while not fundamentally new, has proved so serviceable and so surprisingly accurate that i~t is here offered for the consideration of others who may be confronted with similar problems. The idea of measuring the absolute capacity of a condenser by the discharge through a ballistic galvanometer of a single impulse from the condenser is well known. So also is the modification in which the single limpulse is replaced by a 'train of similar impulses produced by some form of commutator which permits only the charge or the discharge impulses, b u t not both, to reach the galvanometer. The capacity is given in terms of the current figure
378
VVESTINGHOUSE LABORATORY.
[J. F. I.
of merit or sensitivity of the galvanometer, the charging voltage and speed of ~the commutator. This method is applicable to very small capacities if thei. available aharging vol'tage is sufficiently large. There remains, qliowever, the difficulty of measuring accurately the commutator speed and the charging voltage, also the added and somewhat troublesome one of keeping bot.h speed and voltage constan.t; also 'the galvanometer must be checked at pro. i. ~ 4
3
I
~
,<
-
-
V
-
'
G
o,I
-
.I,
Jvw--
I
__.] F'-"
frequent intervals, and finally all the disadvantages of a deflectiort method are present. The present method is a modification of the above, in vd~ich the train of impulses f r o m one condenser is balanced by a second train from a different condenser, passing ~through the galvanometer in the opposite direction. The apparatus .and conn'eetions are shown in Fig. I,in which K is the eommu'ta~or, rotating counter clockwise as shown ; C1 is the condenser to be measured, C= a known condenser, usually a standard subdivided microfara.d, R1
Sept., 1 9 1 8 . 1
WESTINGHOUSE LABORATORY.
379
and R~ are standard dial rheostats. The commutator presents an 'insulating surface to the four sets of brushes, numbered I to 6, except for the solid segment, which latter is metal. The angle of the solid segment is 89 °, and the brushes are set in pairs on 9° ° lines, so that no two pairs of brushes are ever imerconnected. As the commutator revolves it connects together in sequence, brushes I and 2, 2 and3, 4 and 5, 5 and 6, and so on. If the switch S is .closed in position A, the following cycle is completed once every revolution of the commutator. Left hand numbers indicate b r ~ h e s connected. I , 2 ( I ) - C o n d e n s e r C~ is charged to a voltage V2, equal to VR~ RI+R2
2', 3 (2)-Condenser C,_, is discharged through the galvano,meter, G . • ° (3)-Condenser Ca is charged to ~ voltage V1, equal 4,5 VRa to RI+R2
5',6 (4)-Condenser,C1 is discharged through the galvanometer in ,the opposite direction to that of the discharge from C~. Since the directions of the discharge impulses from Ca and C2 through ~the galvanometer are opposite, there will be no permanent deflection of the galvanometer light spot if the discharge impulses are equal. This condition is realized when R1 and R2 are so adjusted that NC1V1 -- NC,V2
or, since V1 and V~ are proportional to R a and R_~, CI=R2 C2~R1
(1)
By throwing switch S into position B the galvanometer receives the charge impulses instead of those of discharge. In testing samples, a balance is obtained for both charge and discharge; a difference of more ethan one pe r cent. between the indicated capacities shows t h ~ the results are not accurate to one per cent., due to different relative amounts of absorption, leakage, or both, in Ca and C2. The result given by equation ( I ) is the comparative value of the total capacity in branches C1 and C2. T h i s value nmst be corrected for the small permanent capacity in the two branches, when the condenser Ca and C2 are disconnected. Tha use of a large condenser on the C2 side enables one to neglect this correction VOL. 186, No. III3--3o.
380
\VESTINGHOUSE LABORATORY.
[J F.I.
on .that side ; the correctior~ fo.r Ca is easily applied by disconnecting Ca, obtaining a balance, and subtracting the. indicated capacity from that found when Ca is connected. Thus, CI+CI=
hence
C~R2 Ri
C1 connected
e 1= C~r2 rl
C1 disconnected
C1 C~Ru
C2r2
If C2 equals co, and R1 equals rt, we have C, (R2-r~) c1 =K,
(2)
which enables the correction to be made by simply subtracting the value of R2 found with C1 disconnected from that found with C1 connected, and using the remainder inl equation ( I ). Obviously, the commutator speed for any capacity must not be high enough to prevent full charge from enltering the. condensers during one charging ir~terval. Fortunately, satisfying this condkior~ for C1 also does so for C2, because the resistance in the charging circuits are R~ and R2, and the two time con.~ants, R1C~ and R2C2 are equal by equation ( i ), when balance has been established. This condi,tion, however, sets an upper limit to the product N C R ; in practice the maximum permissible speed corresponding to the desired product RC is taken from a curve kepi~ with the apparatus. T h e upper limit in capacity or voltage is set by the appearance of strong forced vibrations of the galvanometer moving system, due to the fact ,that the charging or disctmrging impulses, while equal, do not reach the galvanometer at the same instar~t, but are spaced a half revolution apart. T h e accuracy obtainable is of course decreased when the ligh!t spot is thus broadened out into a band of light. The lower limit of accuracy is set by the relative values of the residual or permanent capacity in the C1 branch, and t~he capacity being measured. T h e method is very easy to use, and its sensitiveness is limi~ted only by the current sensitivity of the galvanometer. Small frictional voltages induced by the rotation of the commutator prevent the use of the most sensitive galvanometers obtain'able, but a galvanometer of sensitivky lO-9, current, is good in this'respect and
Sept., 1918.]
v'~rESTINGHOUSE LABORATORY.
381
sufficiently sensitive to measure IO-10 :farad of capacity to one percent. on a few volts. The fact that neither voltage nor speed enters into the equation giving the results, makes it possible to obtain curves of variations of capacity with voltage and with equivalent frequency, since in most cases these coefficients of change are small, and one need not employ means for either closely controlling or very accurately reading either speed or voltage. Neither the sensitiveness nor the accuracy is affected by the relative magnitude of C a and C2, assuming that R 1 and R 2 are accurate over any range desired. Hence very small cor~densers may be compared directly with a large standard subdivided microfarad, in order either to calibrate the small or th:e large condenser. This is a great advantage and is possessed by none of the standard methods of testing. The writer uses, commonly, fixed values for C2 and R~, of 0.05 microfarad and 5000 ohms respectively, resulting in ~he convenient equation C, = I O - "
(R~ - - r2)
Rheos!~at R 2 has dials giving IO x o. I and 9 x (I+IO+IOO+IOOO). With Ca disconnected the galvanometer is sensitive.to one division on the o. i ohm dial with an e.m.f. ( V ) of IO volts, and the balance gives C~ =' 5-4 X IO- ~ farads
as the Ca residual. Hence if one per cent. accuracy must be attained, IO-a° farad is the lower limi,t of capacity which can be measured. A new commutator is now being built, 'having two segments each 45 ° long instead of one of 9 o°. With this will be used brushes set in/pairs at 45 ° instead of 9 o°. This will cause Q1 and Q_~ to reach the galvanometer at the same time, and will thus do away entirely with the forced vibrations of the moving system. It has been found that with certain modifications, this apparatus may be used to measure the insulation resistance of samples as well as their capacity. It seems probable also that the effects of dielectric polarization and absorption on b~th resistance and capacity may at leas.t be estimated, if not measured quantitatively. This work is now in progress, and the results will be embodied in a later paper.