Journal of Scientific & Industrial Research

Vol. 64, September 2005, pp. 666-673

Estimation of uncertainty in measurement in precision calibration of DC high

current source up to 100 A
Shiv Kumar Jaiswal*, V N Ojha and Ajeet Singh
JVS & DC Standards, National Physical Laboratory, Dr K S Krishnan Road, New Delhi 110 012
Received 03 March 2005; accepted 04 July 2005

In the present paper, calibration of precision DC high current source (> 2 A and up to 100 A), widely used in industries
and research, has been discussed as case studies of transconductance amplifier and precision power supply calibration. Since
transconductance amplifier requires boosting device for its operation, therefore effect of boosting device was added as an
additional component of uncertainty. The various sources of uncertainty in measurement and their estimation based on
Type A and Type B method as per ISO ‘GUM’ document are discussed in detail. The results are reported at coverage factor
k=2 for approx 95% confidence level. The standards used for calibration are traceable to the ‘National Standards’ and their
uncertainties were evaluated prior to use to avoid variation in results due to drift.
Keywords: Calibration, Uncertainty in measurement, Current shunt, Digital voltmeter
IPC Code: G01N37/00

Introduction amplifier or using precision power supply. As case

The direct quantum standard of current based on studies, the precision high current sources have been
single electron tunneling (SET) is at present in calibrated at 20 A and 100 A using the traceable
research stage1. Normally, current is realized from standard current shunt and high precision calibrated
voltage drop across the reference resistors using 8½ digit digital voltmeter. The various sources of the
Ohm’s law. This method is more precise than direct uncertainty in measurement and their estimation
measurement by digital multimeter because it offers based on Type A and Type B evaluation of standard
the choice of best artifact (i.e. standard current shunt) uncertainty as per ISO ‘GUM’ document is also
for current calibration. The standards of voltage and discussed6-8. Type A is evaluated by the statistical
resistance used for measurement are traceable to analysis of a series of observations, and Type B is
‘National Standards’ and in turn traceable to evaluated by other than statistical analysis of a series
Josephson Series Array Voltage Standard of NPL- of observations.
India2,3 and Quantum Hall Resistance Standard of
NPL-India4. Methodology of Calibration
High precision multifunction calibrator, a source of The transconductance amplifier is a precision high
DC voltage, DC current and DC resistance, is used as current source whose output current is proportional to
a working standard at NPL and is widely used for an input voltage. For the operation of the amplifier, it
calibration of industrial measuring instruments5. As a is energized by some DC voltage source such as
DC current source, it is normally used in the range of direct volt calibrator, multifunction calibrator etc.,
1 µA to 2.2 A, however for large value of current therefore the effect of this boosting device is also
used in various industrial sectors, this range can be taken in account for uncertainty evaluation. However,
further extended upto 100 A using transconductance if one is using high precision power supply, the
uncertainty due to effect of boosting device is not
__________ taken separately as it is in-built and included in its
*Author for correspondence calibration. The output current of the
Tel: +91-11-25742610–12/ Extn 2233 & 2273 transconductance amplifier/power supply is passed
Fax: +91-11-25726938, 25726952 through the standard current shunt and the ‘voltage
E-mail: [email protected] drop’ across the shunt is measured by using a high
 JAISWAL et al: CALIBRATION OF PRECISION DC HIGH CURRENT SOURCE UP TO 100 A 667

where, d= a constant; pi = a positive or negative

number; and xi = an input estimate.
Comparing Eq (1) with Eq (2) gives p1 = 1 and p2 =
-1. The voltage V has three components of uncertainty
as u1(V), u2(V) and u3(V). The resistance R has two
components of uncertainty as u4(R) and u5(R). Where,
u1(V) = standard uncertainty in voltage drop across
the current shunt; u2(V) = standard uncertainty of
digital voltmeter; u3(V) = standard uncertainty of
precision calibrator; u4(R) = standard uncertainty of
the current shunt; and u5(R) = standard uncertainty
due to temperature effect of the current shunt.
Uncertainty Equation
There are two methods for solving Eq (1):
(a) Absolute Method; and (b) Relative Method.
Absolute Method
For uncorrelated input quantities, the combined
standard uncertainty uc is given as follow:
N N
uc (y) = Σ [(∂f/∂xi) u (xi)] = Σ [ci2 u2(xi)],
2 2 2
…(3)
i=1 i=1
where, uc(y) = combined standard uncertainty;
u(xi)=components of uncertainty; ci=∂f/∂xi=sensitivity
coefficients; and N = number of uncertainty
components.
Fig 1 — Block diagram for calibration of transconductance The corresponding sensitivity coefficients
amplifier
[calculated from Eq (1)] are:

precision digital voltmeter (Fig. 1). The value of c1 = c2 = ∂Ix /∂V = 1/R; c3 = 1 (because amplifier’s
current is calculated from the ratio of voltage drop transconductance is 1 S); and c4 = c5 = ∂Ix /R = -V/ R2
across the current shunt to the resistance of the The combined standard uncertainty is given as:
current shunt as per Ohm’s law (I=V/R). As case
studies, the calibration of 20 A is discussed in detail uc2(Ix) = (c1)2 u12(V) + (c2)2 u22(V) + (c3)2 u32(V) +
using transconductance amplifier and for 100 A, the
(c4)2 u42(R) + (c5)2 u52(R) …(4)
precision power supply is used.
Relative Method
Mathematical Model Used for the Evaluation of
For uncorrelated input quantities, the relative
Uncertainty
The following mathematical model is used to combined standard uncertainty uc(y)/y is given as
estimate the uncertainty in measurement. follows:

Ix = f (V, R) = V/R = V.R-1 …(1) N

[uc(y)/y] = Σ [{ pi.u(xi)/xi}2]
2
…(5)
where, Ix = current output of transconductance i=1
amplifier/precision power supply; V = voltage drop
across standard current shunt; and R = resistance of where, uc(y)/y = relative combined standard
standard current shunt. uncertainty; u(xi)/xi = components of uncertainty in
relative form; pi = a positive or negative number; and
Eq (1) is in the form of y = d.Πxipi …(2) N = number of uncertainty components.
668 J SCI IND RES VOL 64 SEPTEMBER 2005

Table 1 — Experimental observation Average value V = (ΣVi)/ n = 1.000380 V.

Input voltage to transconductance amplifier: 20.000000 V
Nominal output current of transconductance amplifier: 20.00000 A Experimental standard deviation s (V) = 8.30×10-6 V.
No. of Measured voltage drop
observation across current shunt Standard deviation of mean u (V ) = s (V )
V
= s(V)/√n = 2.62×10-6 V.
1 1.000390 V
2 1.000388 V So, Type A uncertainty u1 (V) = 2.62×10-6 V.
3 1.000387 V
Relative type A uncertainty u1 (V)/V = 2.62×10-6 V/
4 1.000386 V
5 1.000384 V 1.000380 V = 2.62×10-6
6 1.000380 V
and degree of freedom ν1 = n-1 = 10-1 = 9.
7 1.000376 V
8 1.000373 V The measured current is: Ix = V/R =1.000380/
9 1.000369 V 0.0500237 = 19.99810 A
10 1.000367 V
V = 1.000380 V Type B Evaluation of Standard Uncertainty
There are four components of uncertainty for Type
B evaluation as follows:
The sensitivity co-efficient for Eq. (1) is given by
(a) Standard Precision Calibrator
Sensitivity co-efficient = pi (y/xi) …(6) Prior to use, the precision calibrator was calibrated
using DC voltage reference standard, voltage divider
The relative combined standard uncertainty is
and high sensitive digital voltmeter/nanovoltmeter to
given as:
avoid variation in results due to drift. The following
[uc(Ix)/Ix]2 = (p1)2 [u1(V)/V]2 + (p2)2 [u2(V)/V]2 + (p3)2 mathematical model was used to evaluate the
uncerainty5:
[u3(V)/V]2 + (p4)2 [u4(R)/R]2 + (p5)2 [u5(R)/R]2 …(7)
where, u1(V)/V = relative standard uncertainty in VX = (VZ + δVDVM − δVD) r …(8)
voltage drop across the current shunt; u2(V)/V =
relative standard uncertainty of digital voltmeter; where, VX = voltage applied by the calibrator;
u3(V)/V = relative standard uncertainty of precision r = ratio of the voltage divider; VZ = average voltage
calibrator; u4(R)/R = relative standard uncertainty of of in-house DC voltage reference standard;
the current shunt; and u5(R)/R = relative standard δVD = difference in voltage between the in-house DC
uncertainty due to temperature effect of the current voltage reference standard and divided voltage of the
shunt. calibrator; and δVDVM = non-linearity and instability
From Eq. (1), p1 = p2 = p3 =1 and p4 = p5 = -1. of sensitive DVM/Nanovoltmeter.
Eq. (4) and Eq. (7) give standard uncertainty of
Eq. (1). Two methods are same except their different The uncertainty of standard precision calibrator at
ways of evaluation. In relative method, it is easy to 20 V was estimated as ± 4.0×10-5 V at k = 2 for
handle the equation having large numbers of approx 95 % confidence level. Since the probability
components in product/quotient form. distribution is normal, therefore the relative standard
Evaluation of Uncertainty uncertainty u2(V)/V in precision calibrator is:
Type A Evaluation of Standard Uncertainty
U2(V)/V = 4.0 ×10–5 V/(2×20 V) =1.0 ×10–6 at k = 1
For evaluation of Type A uncertainty, 10 numbers
of observations were taken (Table 1). and degree of freedom ν2 = ∞.
 JAISWAL et al: CALIBRATION OF PRECISION DC HIGH CURRENT SOURCE UP TO 100 A 669

(b) Standard Digital Voltmeter (DVM) u4 (R)/R=1.0×10-7Ω/(2×0.05Ω)=1.0×10-6 at k = 1

Prior to use, the precision 8½ digit digital
and degree of freedom ν4 = ∞.
voltmeter was calibrated by using precision calibrator
to avoid variation in results due to drift. The (d) Temperature Effect of Standard Current Shunt
following mathematical model was used to evaluate The temperature affects the resistance value of the
the uncertainty8: current shunt. For determination of temperature co-
efficient of the current shunt, the current shunt was
Vx = Vstd + ∆Vx …(9) calibrated using automatic DCC bridge at three
temperatures9, 23, 25 and 27 °C. The temperature co-
where, Vx is the voltage indicated on digital efficient factors (α & β) were calculated from the
voltmeter under calibration; Vstd is the voltage applied following widely used working formula10:
from the standard precision calibrator; and ∆Vx is
R23 = R25 [1+ α (t23 – t25) + β (t23 – t25)2] …(11)
resolution of digital voltmeter under calibration.
The uncertainty of standard DVM at 1 V was and
estimated as ± 3.0 ×10-6 V at k=2 for approx 95 % R27 = R25 [1+ α (t27 – t25) + β (t27 – t25)2] …(12)
confidence level. Since digital voltmeter was
calibrated at the point of its use, so non-linearity where, R23 = resistance of the current shunt at
component of uncertainty does not come into picture. 23°C; R25 = resistance of the current shunt at 25 °C;
Assuming probability distribution as normal, the R27 = resistance of the current shunt at 27 °C; α =
relative standard uncertainty u3 (V)/V in voltage drop slope of the curve (ppm/°C) at 25°C; and β = rate of
measured by DVM is: change of slope of the curve (ppm/°C2).
The calculated values of α and β are –1.17 ppm/°C
U3(V)/V = 3.0 ×10-6 V/ (2×1 V) = 1.5 ×10–6 at k = 1 and –0.069 ppm/ °C2 respectively. From the values of
α and β, the current shunt is characterized for the
and degree of freedom ν3 = ∞.
other temperature using the following formula:
(c) Standard Current Shunt (SCS)
Rt = R25 [1+ α (t – t25) + β (t – t25)2] …(13)
The SCS was calibrated using automatic direct
current comparator (DCC) bridge. The following The resistance of current shunt with respect to
mathematical model was used to evaluate the temperature (Fig. 2) is decreasing, as the values of
uncertainty9: temperature co-efficient factors (α & β) are negative
for this shunt. For working temperature of 25 ± 1°C,
RX = (RS + δ RD + cS δ tS) r.rl – cX δ tX …(10) temperature co-efficient of the standard current shunt
was taken as –1.51 ppm/°C ≈ –2 ppm/°C as the
where, RX = measured value of the unknown maximum value (Table 2). Since upper and lower
resistor; RS = value of the reference resistor; δRD = limit of this uncertainty component is given,
drift in the value of reference resistor since its last therefore, distribution was assumed as rectangular.
calibration; δ tS = temperature deviation of the For this distribution, relative standard uncertainty,
reference resistor; r = the ratio of the unknown u5(R)/R, due to temperature effect of standard current
resistance to reference resistance (i.e. RX /RS); rl = shunt is:
non-linearity and instability of the high resistance
bridge; δ tX = temperature deviation of the unknown u5(R)/R=(2 ppm/°C×1°C)/√3=2.0×10-6/√3=1.15×10–6
resistor; and cS & cX = temperature co-efficient of at k=1 and degree of freedom ν5 = ∞.
reference resistor and unknown resistor respectively.
Relative Combined Standard Uncertainty
The actual resistance value of the nominal 0.05 Ω The relative combined standard uncertainty is
SCS from its calibration is 0.0500237 Ω ± 0.0000001 calculated as follows:
Ω at k = 2 for approx. 95 % confidence level. [uc(Ix)/Ix]2=(p1)2[u1(V)/V]2+(p2)2[u2(V)/V]2+(p3)2[u3(V)/V]2
Assuming normal distribution, the relative standard
uncertainty u4(R)/R in SCS is: +(p4)2[u4(R)/R]2+(p5)2[u5(R)/R]2uc(Ix)/Ix = 3.527×10–6
670 J SCI IND RES VOL 64 SEPTEMBER 2005

Fig 2 — Resistance versus temperature curve of 0.05 Ω current shunt

Table 2 — Temperature characterization chart for 0.05 Ω current Truncating to the lower side to get the maximum
shunt value of the coverage factor (k) and thereby the
Alpha: –1.17 ppm/°C practical expanded uncertainty.
Beta: –0.069 ppm/°C2
Temperature Resistance Deviation from value at 25 °C Expanded Uncertainty (U)
°C ohm ppm
The relative expanded uncertainty is given by
22.0 0.05002419 + 4.11
U(y)/y = k × [uc(y)/y].
23.0 0.05002405 + 2.61
where k is a coverage factor. From student’s
24.0 0.05002399 + 1.37 t-distribution6, the value of coverage factor k for
25.0 0.05002392 + 0.00 (Ref. temperature) approx 95 % confidence level and for νeff=29 is 2.11.
26.0 0.05002384 -1.51
U(Ix)/Ix = 2.11×3.527×10–6 = 7.441×10–6
27.0 0.05002376 -3.17
U(Ix) = 7.441×10–6 × Ix =
28.0 0.05002367 -4.96
7.441×10–6 ×19.99810 A = 1.49×10–4 A
The complete uncertainty budget is summarized in
Effective Degree of Freedom (νeff) Table 3.
Effective degree of freedom of relative standard
uncertainty uc(y)/y associated with the output estimate Results
ui(y) is given by Welch-Satterthwaite formula6: The measured value of current of nominal 20 A
current source alongwith expanded uncertainty is
[uc(y)/y]4 19.99810±0.00015 A (Table 4). The reported expanded
νeff = ——————— uncertainty of measurement is stated as the combined
N standard uncertainty multiplied by the coverage factor
{∑ [ui(y)/y]4/νi } k=2.11 which for a normal distribution corresponds to
i=1 coverage probability of approx 95 %. In the similar
νeff = 29.54 ≈ 29 way, uncertainty estimation of the precision high
 JAISWAL et al: CALIBRATION OF PRECISION DC HIGH CURRENT SOURCE UP TO 100 A 671

Table 3 — Statement of the uncertainty budget in transconductance amplifier calibration at 20 A

(Relative Method)
Quantity Estimates Relative Probability Relative standard Relative uncertainty Degree of
Xi xi Limits distribution uncertainty contribution freedom
±∆xi/xi (Type A or B) u(xi)/xi ui(y)/y=pi u(x)/xi νi
(±) (±)
Standard 1V 3.0×10-6 Normal 1.5×10-6 3.0×10-6 ∞
DVM Type B
Standard 20 V 2.0×10-6 Normal 1.0×10-6 1.0×10-6 ∞
calibrator Type B
Standard 0.05 Ω 2.0×10-6 Normal 1.0×10-6 1.0×10-6 ∞
current shunt Type B
Temp. effect of — 2.0×10-6 Rectangular 1.15×10-6 1.15×10-6 ∞
current shunt Type B
Repeatability 1.000380 V — Normal 2.62×10-6 2.62×10-6 9
Type A
Ix 19.99810 A — — — 3.527×10-6 29
-6
U(Ix)/Ix — — — coverage factor k=2.11 7.44×10 29
-4
U(Ix) — — — coverage factor k=2.11 1.49×10 A 29
Table 4 — Statement of the uncertainty budget in transconductance amplifier calibration at 20 A
(Absolute Method)
Quantity Estimates Limits ± ∆xi Probability Standard Sensitivity Uncertainty Degree of
Xi xi distribution uncertainty coefficient contribution freedom
(Type A or B) u(xi) ci ui(y) = ci u(xi) νi
(±) (±)
Standard 1V 3.0×10-6 V Normal 1.5×10-6 V 19.99 S 3.0×10-5 A ∞
DVM Type B
Standard 20 V 4.0×10-5 V Normal 2.0×10-5 V 1S 2.0×10-5 A ∞
calibrator Type B
Standard 0.05 Ω 1.0×10-7 Ω Normal 5.0×10-8 Ω -399.78 2.0×1-4 A ∞
current shunt Type B A/Ω
Temp. effect of — 1.0×10-7Ω Rectangular 5.77×10-8Ω -399.78 2.31×10-5 A ∞
current shunt Type B A/Ω
Repeatability 1.000380 V — Normal 2.62×10-6 V 19.99 S 5.25×10-5 A 9
Type A
Ix 19.99810 A — — — — 7.06×10-4 A 29
-4
Expanded — — — coverage factor k=2.11 1.49×10 A 29
uncertainty U(Ix)

current source at 100 A was done using high Conclusions

precision power supply. Since the precision current In the present work, the calibration of DC high
source has inherent boosting device, therefore, effect current source, widely used in the industrial and
of boosting device in the uncertainty budget is not research areas, is discussed in detail. As case studies,
warranted. The measured value of current of nominal 20 A and 100 A current sources using trans-
100 A current source alongwith expanded uncertainty conductance amplifier and precision power supply
is 99.6435 ± 0.0176 A (Tables 5 & 6) as evaluated by were calibrated. In case of transconductance
the two different methods. amplifier, the effect of boosting device is included in
672 J SCI IND RES VOL 64 SEPTEMBER 2005

Table 5 — Statement of the uncertainty budget in precision high current source calibration at 100 A
(Relative Method)
Quantity Estimates Relative Probability distribution Relative standard Relative uncertainty Degree of
Xi xi limits (Type A or B) uncertainty contribution freedom
±∆xi/xi u(xi)/xi ui(y)/y=pi u(xi)/xi νi
(±) (±)
Standard 10 mV 2.0×10-5 Normal Type B 1.0×10-5 1.0×10-5 ∞
DVM

Standard 0.0001 Ω 3.0×10-6 Normal Type B 1.5×10-6 1.5×10-6 ∞

current shunt

Temp. effect of — 6.0×10-6 Rectangular Type B 3.46×10-6 3.46×10-6 ∞

current shunt

Repeatability 9.96475 mV — Normal Type A 7.54×10-5 7.54×10-5 9

Ix 99.6435 A — — — 7.62×10-5 9.37

-4
U(Ix)/Ix — — — coverage factor k=2.32 1.77×10 9.37

U(Ix) — — — coverage factor k=2.32 1.76×10-2 A 9.37

Table 6 — Statement of the uncertainty budget in precision high current source calibration at 100 A
(Absolute method)
Quantity Estimates Limits Probability Standard Sensitivity Uncertainty Degree of
Xi xi ± ∆xi distribution uncertainty coefficient contribution Freedom
(Type A or B) u(xi) ci ui(y)=ci u(xi) νi
(±) (±)
Standard DVM 10 mV 1.99×10-7 V Normal 9.96×10-8 V 9999.6 S 9.96×10-4 A ∞
Type B

Standard current 0.0001 Ω 3.0×10-10 Ω Normal 1.50×10-10 Ω -996395 A/Ω 1.49×10-4 A ∞

shunt Type B

Temp. effect of — 6.0×10-10Ω Rectangular 3.46×10-10Ω -996395 A/Ω 3.45×10-4 A ∞

current shunt Type B

Repeatability 9.96475 mV — Normal 7.52×10-7 V 9999.6 S 7.51×10-3 A 9

Type A

Ix 99.6435 A — — — — 7.59×10-3 A 9.37

Expanded — — — coverage factor k=2.32 1.76×10-2 A 9.37

uncertainty U(Ix)

the uncertainty budget. However, in the calibration of the resistance value of the standard current shunt and
100 A current source using high precision power its evaluation was highlighted. For this purpose, the
supply, the effect of boosting device, which is in- temperature co-efficient of standard current shunt was
built, does not come into picture. Uncertainty also determined (Table 2) using recently established
calculations can be done by using Relative Method high precision automatic DCC bridge and its effect on
(Table 3 for 20A & Table 5 for 100A) and Absolute uncertainty in measurement was incorporated.
Method (Table 4 for 20A & Table 6 for 100A)
resulting into same results. However, it gives the user Acknowledgements
an opportunity to use either depending on the ease of Authors thank Dr Vikram Kumar, Director and
use and his understanding of the sensitivity Dr P C Kothari, Head, Electrical and Electronic
co-efficients. Further, the effect of the temperature on Standards of National Physical Laboratory, New
 JAISWAL et al: CALIBRATION OF PRECISION DC HIGH CURRENT SOURCE UP TO 100 A 673

Delhi for their constant encouragement and support. 5 Ojha V N, Singh A, Jaiswal S K & Sharma S K, Automatic
Authors also thank Dr Sudhir Kumar Sharma for and manual calibration of high precision multifunction
calibrator, MAPAN, J Metrol Soc India, 18 (2003) 43-48.
useful discussions.
6 Guide to the Expression of Uncertainty in Measurements, I
References edn (International Organisation for Standardization,
1 Keller M W, Standards of current and capacitance based on Switzerland), 1995.
single-electron tunneling devices, Proc Fermi School CXL 7 Ojha V N, Evaluation and expression of uncertainty in
VI: “Recent Advances in Metrology and Fundamental measurement, MAPAN, J Metrol Soc India, 13 (1998) 71-
Constants” (IOS Press, Amsterdam) 2000. 84.
2 Ojha V N & Gupta A K, Josephson voltage standard at 1 8 Guidelines for estimation and statement of overall
volt level at NPL India, in CPEM-Digest (USA), edited by T uncertainty in measurement results, NABL Document No.
L Nelson (Washington DC, USA) 1998, 558-559. 141 (Department of Science and Technology, Govt. of India,
3 Sharma S K, Ojha V N, Singh A, Jaiswal S K & Sharma R, New Delhi) 2000.
Traceability and calibration-A case study, MAPAN, J Metrol 9 Singh A & Ojha V N, Calibration of national standard of
Soc India, 18 (2003) 57-61. resistance using newly established automatic direct current
4 Ojha V N & Singh A, Traceability of DC resistance standard comparator resistance bridge, MAPAN, J Metrol Soc India,
at NPLI, 5th International Conference on Advances in 18 (2003) 49-55.
Metrology, Admet-2005 (NPL, New Delhi) February 23-25, 10 Resistance Standards Instruction Manual, 742A series
2005, 78. (Fluke Corporation, Everett, USA) 1989, 1-9.