0950 7671/31/12/307

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Home Search Collections Journals About Contact us My IOPscience

A 1-100 Ω build-up resistor for the calibration of standard resistors

This content has been downloaded from IOPscience. Please scroll down to see the full text.

1954 J. Sci. Instrum. 31 450

(http://iopscience.iop.org/0950-7671/31/12/307)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 134.129.120.3
This content was downloaded on 27/05/2015 at 00:41

Please note that terms and conditions apply.


A 1-100 a build-up resistor for the calibration of standard resistors
By B. V. HAMON,
B.Sc., B.E., National Standards Laboratory, Commonwealth Scientific and Industrial
Research Organization, Sydney, Australia
[Paperfirst received 15 June and in final form 23 August, 19541
The build-up resistor consists of eleven 10 R resistors permanently connected in series. Each
resistor is defined as a four-terminal resistor. Groups of resistors can be connected in parallel or
in series-parallel by separately connecting the appropriate current and potential terminals. The
resistances of the combinations so obtained are independent of the resistances of terminal blocks
and paralleling connectors to within 1 part in 10s. The ratio of the series to the parallel
resistance of ten coils is within 1 part in 108 of its nominal value (100) provided the individual
coils are adjusted to equality within 1 part in 104.
An example is given of the use of the build-up resistor to determine 10 and 100 fi standard
resistors in terms of a 1 R resistor to an accuracy of about 2 parts in 107. The build-up resistor
can also be used as a pair of bridg:: ratio arms in which the ratio, which is nominally 10, can be
determined accurately.
1. I N T R O D U C T I O N as in parallel. This cannot be done easily if separate resistors
are used. The difficulty can be overcome, however, by con-
In maintaining standards of electrical resistance in national necting the . resistors permanently in series, and providing
standardizing laboratories, it is usual to treat 1 Q resistors one current lead and one potential lead at each series junction.
as basic standards. Resistors of other values are then Fig. 2 shows the circuit of a build-up resistor designed in
measured in terms of the I n standards. The process of this way, and containing eleven equal resistors (RI to RI’).
measuring standards whose nominal values are 10, 100, Ten of these resistors can be connected in parallel by re-
lOOOa, etc., in terms of 1 0 is known as “building-up,”
movable connecting blocks. The requirements for the
and special “build-up” resistors have been developed for this
purpose.
The simplest build-up device(’) consists of ten nominally-
equal-resistors connected in series. Each resistor is measured
by substitution against a known standard of the same nominal
value; the ten in series can then be used to determine a
resistance that is nominally ten times the known standard.
The measurement of each of the ten resistors individually
can be avoided, with an appreciable saving of time, if it is
permissible to connect groups of resistors in parallel or in
series-paraIlel. Build-up resistors using this technique have
been described by Wenner@)and Rosa.(3) However, if the
individual resistors are less than about loon, it is difficult
to make accurate allowance for the resistance of connecting
blocks and paralleling links.
In the build-up resistor described in this paper, the effects
of lead and contact resistances are practically eliminated by Fig. 1. The parallel connexion of n separate four-
defining each resistor as a four-terminal resistor, and parallel- terminal resistors
ing the current and potential leads separately. In this way
it is possible to connect ten resistors in parallel without loss potential lead resistors, corresponding to equations (1) and
of accuracy, even when each resistor is as low as 10 Q. (2), are satisfied in this case if all potential lead resistors
have the same value (r), except the two end resistors, which
2. T H E O R Y A X D D E S I G N have the value 2r. The actual value of r is not critical; it
should not be high enough to affec; the sensitivity of bridges
Wenner(4) and Brooks(’) have discussed the parallel con- in which the resistor is connected, nor so low that adjustment
nexion of four-terminal resistors. Additional resistors must to equality is difficult or too dependent on the stability of
be included in the potential leads if the parallel four-terminal contacts. The accuracy with which the potential lead
conductance is to be equal to the s u m of the individual four- resistors should be adjusted to equality (or to a ratio of 2,
terminal conductances, and independent of the way in which for the end resistors) depends on the resistance of the current
the total current divides between them. Fig. 1 shows the leads in relation to the main resistors ( R I to R,J and on the
general case for the parallel connexion of n-four-terminal variations to be expected in the current lead resistance. For
resistors. The additional resistors in the potential leads are the case in which each main resistor is 10 a, an analysis of
r,, rz . . . rn;ri, r; . . . rln These resistors must be adjusted so the circuit showed that the potential lead resistors need only
that be adjusted to equality to about 1 % for a fractional error of
rlGl = r2G2 = . . . = r,G, (1) 1 part in 108 in the parallel combination.
‘;GI = r;G2 = . . . = r,G, (2) The junctions between adjacent resistors in Fig. 2 need
special design. Each junction can be regarded as a four-
where GI = l/R,, etc. terminal network,@)and will in general have associated with
It would be possible to make a build-up resistor using it a number of four-terminal resistance values defined by
separate four-terminal resistors, but it is a distinct advantage taking any pair of leads as current leads and the remaining
to be able to connect the individual resistors in series, as well pair as potential leads. If the build-up resistor is to operate
450 JOURNAL OF SCIENTIFIC INSTRUMENTS
A 1-100 build-up resistor for the calibration of standard resistors
so that the main resistors have the same effective values in found from these measurements be a = c1 - c,,~, and
the series and in the parallel connexions, the junctions must
be so designed that their four-terminal resistances are all
zero. This can be achieved if the leads are connected to a (4)
junction block in such a way that three of the leads are
symmetrical with respect to the fourth. Since exact sym-
metry cannot be obtained in practice, due to manufacturing It follows that e,, = cp + fi - a/lO (6)
Equation (6) gives the correction to the “unknown” 1On
standard in terms of cp, which is known in terns of the 1 fi
standard, and the measured differences a and 8.
In the measurements described above, it may be noted that
the corrections to the individual resistors need not be deter-
mined, with the exception of cJ in the 1 0 f i measurement.
Also, all measurements except one involve nine or ten of
the resistors, so that the power dissipaion is spread evenly.
The term a, which depends on the measurement of R, alone,
is divided by ten when it appears in the final equation
[equation (6)]. This tends to compensate any effect due to
the power being concentrated in one resistor in this
measurement.
For the measurement of low resistances, down to fi,
in terms of a 1 fi standard, a Kelvin double bridge with
accurately-known 10 : 1 ratio arms is most often used. By
providing eleven 10 n resistors instead of ten in the build-up
resistor, this known 10 : 1 ratio can be obtained easily. If
cJ1is the correction to the eleventh resistor, and if y=cII- c ~ , ~ ,
as determined by substitution, the correction in proportional
parts to the ratio of the resistance of coils 2 to 11 in series
to coil 1 is given by
p = y/lO - a (7)
Fig. 2. Circuit diagram of build-up resistor with
paralleling connexions 3. CONSTRUCTION

Fig. 3 is a drawing of a build-up resistor containing eleven


tolerances, the four-terminal resistances of the junctions will 10 fi coils, together with the links used to connect the first
differ slightly from zero. The four-terminal resistances of ten coils in parallel according to the circuit of Fig. 2. One
all the junctions used were measured and found to be less of the symmetrical four-lead junction blocks is shown at A .
than 5 x 10-8n. It was estimated that this would not It consists of a short copper cylinder on the end of a copper
cause an error greater than 1 part in 108 in any application stem, the latter serving as one lead. The other three leads
of the build-up resistor. are copper wires soldered into shallow holes drilled into the
The use of a build-up resistor of the type shown in Fig. 2 cylinder at 120” spacing. The current and potential leads
for the calibration of 10 and 100Q standard resistors in from the junction blocks are taken to mercury cups B and C,
terms of a 1 fi resistor will now be discussed. respectively. The current leads are connected in parallel by
Consider the first ten coils ( R , to Rio) of the build-up copper blocks D, which have amalgamated projections on
resistor shown in Fig. 2, the nominal value of each coil being their lower surfaces. These blocks also connect to the main
10 Q. k t cq be the correction, in proportional parts, to the current terminals E by means of amalgamated surfaces F.
qth resistor, so that its true value is lO(1 Icq) n. The Terminals E are of the same design and spacing as the ter-
correction (again in proportional parts) to ten resistors in minals of standard resistors, so that the two units are physically
series is then interchangeable.
10
e, = 0.1 cq
The potential leads are connected in parallel by means of
(3) the block G, which contains the potential lead resistors. The
y=l
block can be clamped in place by means of the spring-loaded
But if cq < the correction cp to the parallel resistance clamps H. The lead resistors are self-supporting bare
of the ten coils is also given(’) by equation (3), to an accuracy manganin resistors of 0.2 fi each, the two end resistors being
of at least 1 part in 108. cp, and hence c, also, can be 0.4n. They were adjusted to within a few parts in 104.
found by comparison of the parallel resistance against a The surfaces to be amalgamated were machined plane after
1 R standard by substitution in a bridge circuit. Comparison assembly, so that only a small amount of mercury is necessary
of the series connexion with an “unknown” 100 fi resistor, to ensure contact. The cups are formed by close-fitting
again by substitution, gives the value of the “unknown” sleeves of stainless steel. Connexions other than the parallel
directly. connexion shown in Fig. 3 are obtained by using blocks of
To determine the value of a 10 fi resistor, a bridge is set suitable design in place of D and G.
up in which 10 resistors can be compared by substitution. The 10 n resistors (E, Fig. 4) are supported on aluminium
Balances are obtained for resistor R, (correction c1), resistors plates A insulated with a thin sheet of polythene B. The
R, to R,,inclusive in series-parallel (correction and the plates are separated by aluminium spacing bars D. Clearance
“unknown” 10 standard (correction cy). Let the differences was provided between the resistors, the plates and the sides
VOL. 31, DECEMBER 1954 45 1 ****
B. V. Hamon
of the grooves in the Perspex spacers C, so that the resistors constant of about 10 &, and that a total power dissipation
are not constrained. of 0.025 W in the ten coils did not raise their temperature
The resistors are in good thermal contact with the alu- more than 0.01" C above the bath temperature.
minium plates and with the oil of the bath in which the unit The annealing and adjustment of the resistors followed
the general principles laid down in recent years for standards
using bare wire annealed at a high temperature.(7-9) 24 s.w.g.
manganin was used, and was annealed at 550" C for 1) hours
on steel plates immersed in boric acid. After removing the
boric acid with hot water, copper leads were attached in
such a way that the lead lengths were equal and as small as
possible, so that the temperature coefficients would be
disturbed as little as possible. Initial adjustment was by
etching using the solution recommended by Schulze(lo) but
reduced to one-tenth strength. Final adjustment was made
on the completely assembled instrument by scraping the
coils where they protrude at the bottom of the former assembly.
All coils were adjusted to within 2 1 part in IO5 of nominal.

4. T E S T RESULTS

Since no independent ratio device of comparable precision


was available, it was possible to test the build-up resistor
only by examining the consistency of the results obtained
when the resistor was used in different ways. To do this, a
100 d resistor Z and a 10 d resistor Y were compared with
a 1 51 resistor by two different methods. Method 1 was the
normal method as described in Section 2, and involyed only
substitution measurements in Wheatstone bridges. Lead
resistances were eliminated by taking two balances with a
galvanometer lead and a battery lead in different positions.(')
Method 2 was an adaptation of the method described briefly
in Section 2 for the measurement of low resistances. Coil 1
and coils 2 to 11 in series were used as two a m of a Kelvin
double bridge, in which the 10 n resistor was fist measured
in terms of the 1 d,and then the 100 d measured in terms of
the 10 51. Lead resistances were balanced out in this case
by the use of an adjustable mercury rheostat.(ll) All
Fig. 3. The 1 to 100 P resistor, with Parallelkg measurements were camed out with the resistors immersed
connectors in an oil bath whose temperature was controlled to ~0.005"C.
The sensitivity was such that the different bridge circuits
could be balanced to i O . 1 parts in 106 or better, although
to achieve this sensitivity the bridge currents had to be
increased to the point where a small correction was required
for the heating effect of the current.(?) This correction did
not at any time exceed 0.4 parts in 106.
The table shows the results obtained in four sets of measure-
ments. The figures given are the differences, in parts in
106, between the values found by the two different methods
C for the 10 and 1 0 0 resistors.
~ The results show that the
build-up resistor can be used for the intercomparison of
D 10 and 100 d resistors with a 1 R resistor to an accuracy of a
few parts in 107. This accuracy is limited at present by the
A sensitivity obtainable in the different bridge circuits. As
discussed above, the accuracy of the build-up resistor as a
ratio device should be at least 1 part in 108, or an order of

v/ B
E '
magnitude better than is apparent from the table.

Differences between resistances measured by two methods


Diffrrcnce(parls it, 106)
Fig. 4. Mounting of 10 d resistors in the build-up Da!e Ion(Y) 100 n (Z)
resistor 29.6.53 +0.1 0.0

is immersed when in use. From tests with copper coils in 30.6.53 -0.2 -0.1
place of the mangain ones, it was found that the temperature 15.7.53 j0.4 f0.2
of the resistors followed that of the oil in the bath with a time 15.7.53 -0.2 +os1
452 JOURNAL OF SCIENTIFIC INSTRUMENTS
A I-IO0 C2 build-up resistor for the calibration of standard resistors
ACKNOWLEDGEMENT (.5.) BROOKS,H. B. Trans Amer. Inst. Elect. Engrs, 39,
p. 549 (1920).
The assistance of MI. H. C. Collins, Division of Electro- (6) SEARLE,G. F. C. Electrician, 66,p. 999 (1911).
technology, C.S.I.R.O., in the construction, adjustment and (7) THOMAS,J. L. J. Res. Nat. Bur. Stand., 5, p. 295 (1930).
testing of the instrument is gratefully acknowledged. (8) THOMAS, J. L. J. Res. Nut. Bur. Stand., 36, p. 104
(1946).
(9) BARBER, C. R., GRIDLEY, A., and HALL,J. A. J. Sci.
REFERENCES
Instrum., 29, p. 65 (1952).
(1) GLAZEBROOK, R. Dictionary of Applied Physics, Vol. 2, (10) SCHULZE,A. Metullische Elektrische Widerstanh-
pp. 717,718 (London: Macmillan and Co. Ltd., 1922). werkstofe, p. 86 (Berlin: Dr. Georg Liittke Verlag,
(2) WENNER, F. J. Res. Nut. Bur. Stand., 25, p. 229 (1940). 1941).
(3) ROSA,G. L' Elettrotecnicu, 36,p. 50 (1949). (11) WENNER, F., and WEIBEL,E.: Bull. Not. Bur. Stand., 11,
(4) WENNER, F. J. Res. Nut. Bur. Stand., 8, p. 559 (1912). p. 65 (1914).

The transient response of direct current amplifier systems


M.A., D.PhiI., A.Inst.P., The Clarendon Laboratory, University of Oxford
By J. H. SANDERS,
[Paper received 30 July, 19541
Direct current amplifiers of the d.c.-ax. conversion type have an upper frequency response
limited by the detector circuit, and when negative feedback is used there is a large overall gain
for signals due to charges induced instantaneously on the input terminals compared with the
gain for steady signals. The form of the transient response and methods of reducing its magnitude
are discussed.
IDEAL SYSTEM Now consider a simple type of transient input to this
system, a charge - Q induced or placed instantaneously on
A typical direct current amplifier system(',*)for measuring the capacity C,, so that VI instantaneously assumes the value
a current I is shown in Fig. 1. Here a fraction ,6 of the output
Q/C,.The output voltage will rise instantaneously to the
voltage Vo is fed back in opposition to the potential drop value A ( Q / C , )and will subsequently decay exponentially to
IR, across the input resistor RI. In piactice a capacity C, zero with a time constant Q. The initial value of Vo should
must be considered to exist across the input terminals of the be compared with the expression for the steady state output
voltage. Here, if we regard the signal due to the current as
ZR,,which is the signal input to the amplifier in the absence
+
of feedback, the voltage gain of the system is A/(1 A,@, a
quantity which is generally much smaller than A. In the
case of the transient signal due to the induced charge, how-
ever, the whole gain of the amplifier is effective and the
output signal is A times the input voltage Q/C,. Generally
speaking transient signals appearing at the input to a d.c.
1 amplifier are spurious and may be due, for example, to
Fig. 1. Ideal negative feedback d.c. amplifier system fluctuations in the voltage of the h.t. electrode of an ionization
chamber, movement of the electrodes, or pick-up from
external sources of interference. We have, therefore, the
amplifier, this capacity being partly the capacity of the asso- undesirable state of affairs in which unwanted signals may
ciated circuit (e.g. the capacity to earth of the collecting
be amplified far more than the wanted signal.
electrode of an ionization chamber) and partly the input
capacity of the amplifier itself. The behaviour of such a
system, provided that R, > ,6R3, may be summarized as PRACTICAL SYSTEM
follows: In the above analysis it has been assumed that the d.c.
A
output voltage v0 = - (-)IR~ amplifier is ideal, i.e. that the output voltage is - A times the
1-A8 input voltage no matter how rapidly the latter varies. In
input resistance 7 = (- 1
1 -AB
) R~
some types of amplifier, the gain necessarily falls at a com-
paratively low frequency. Examples of this type are the
contact modulator, capacitance modulator and magnetic
1 modulator amplifiers,(3)in which a steady input voltage is
time constant To = (-)R]c, converted to an alternating voltage, amplified by an ax.
1 -AB
amplifier, then rectified to produce a steady output voltage,
The last quantity, the time constant of the system, refers part of which may be used for negative feedback. The
to the exponential decay of the output voltage when the response of such a system to signals which vary with time is
current I is instantaneously stopped, or to the exponential restricted by the time constant of the rectifying circuit,
rise of Vo when the current I is instantaneously started from which necessarily has a value of at least, say, ten times the
zero provided the value of I is independent of the voltage drop period of the modulation frequency.
V, across the input terminals of the system. The circuit to be analysed in this case may be represented
VOL. 31, DECEMBER 1954 453

You might also like