Content uploaded by Cheng Liu
Author content
All content in this area was uploaded by Cheng Liu on Oct 07, 2014
Content may be subject to copyright.
Error Correction of Automatic Testing Systems for Hall Effect
Current Sensors
Cheng Liu, Ji-Gou Liu
ChenYang Technologies GmbH & Co. KG., Markt Schwabener Str. 8, 85464 Finsing, Germany
http://www.chenyang.de, Tel. +49-8121-2574100, Fax: +49-8121-2574101, cheng.liu@chenyang-ism.com
Abstract
In this paper an error correction method is proposed for improving the measuring accuracy of automatic testing systems
for Hall Effect current sensors. The errors of the original testing system are determined by a reference sensor or a refer-
ence resistor. These error values are saved as corrections into a data matrix in the testing system. The systematic errors
of the testing system are corrected by using the correction data matrix. In this way a conventional measuring system can
be enhanced into a precise measuring system. The proposed method can be used in all automatic testing systems.
Keywords: Error Correction, Measuring Accuracy, Accuracy Improvement, Sensor Test, Hall Effect Current Sensor,
Current Sensing, Automatic Testing System, Measuring System, Control System, Automation
1 Introduction
Current sensing is an important operation for many elec-
tric power, driving and communication systems. Tradi-
tionally, it was primarily intended for circuit protection
and control. With the technological advance, current sens-
ing has appeared as a method for monitoring and perfor-
mance enhancing. Therefore, current sensors are applied
to power systems, current and voltage regulators, linear
and switch-mode power supplies, inverters, rectifiers, mo-
tor drives, generators, automotive power electronics, elec-
tric powered locomotives, telecommunications, trans-
former substations, battery management systems, wind
turbines and photovoltaic equipment etc.
Hall Effect current sensors are preferred towards other
competitive technologies like shunt resistors, because
they provide many benefits such as wide measuring
range, good linearity, high accuracy, Galvanic isolation
between input and output, and wide variation of sensor
configurations etc. [1, 2].
In order to achieve reliable results and to ensure satisfying
quality, Hall Effect current sensors have to be tested be-
fore using. An automatic testing system can be very help-
ful to save the testing time. A simple automatic testing
system for Hall Effect current sensors consists of three
basic components: a digital multimeter (DMM), an
AC/DC current source and a PC system.
For a trustful quality control, it is important that the test
equipment should be more accurate than the sensor under
test. The measuring error of the testing system should be
lower than one-fourth of the error of the sensor [3]. The
most Hall Effect current sensors are defined with accura-
cy from ±1.0% to ±0.2%. So for testing all of these sen-
sors, the measuring deviation of testing systems must be
lower than ±0.05%. Nevertheless, this criterion cannot be
fulfilled for the most conventional current measuring sys-
tems. Therefore better and more accurate instruments are
needed for testing the Hall Effect current sensors. How-
ever such precise instruments are much more expensive.
It results in increasing testing costs of current sensors.
This paper proposes an error correction method of auto-
matic testing system for Hall Effect current sensors. This
method consists of error determination and error correc-
tion. By using the proposed method the accuracy of the
automatic testing system can be effectively improved and
controlled within ±0.03%. Instead of buying expensive
equipment, the error correction is a favorite solution for
testing systems. The principle can also be applied to
measurements of other electrical and physical quantities.
2 Test Equipment
Figure 1 shows one of our automatic testing systems. It
consists of a DMM (1), a DC current source (2), a PC sys-
tem (3) and a data acquisition device with analog and dig-
ital outputs (4).
The Agilent 34401A multimeter has a 6½ digit resolution.
It provides many measurement functions (AC/DC volt-
age, AC/DC current, 2- and 4-wire resistance, diode, con-
tinuity, frequency and period) and has a basic accuracy
0.0035% for DC measurements and 0.06% for AC meas-
urements [4]. The DC current source is an EA-PS 8080-
120. This device outputs a DC current of 0-120A. Ac-
cording to the data sheet, it has an accuracy ≤0.2% [5].
The NI USB-6008 is for data acquisition, but also offers
two analog outputs and 12 digital I/O pins [6]. So it can
be used for system controlling with software which runs
on the PC system. Due to the deviations of the DC current
source (±0.2%) and the DMM (±0.01%), the measuring
Sensoren und Messsysteme ∙ 03. – 04.06.2014 in Nürnberg
ISBN 978-3-8007-3622-5
1
© VDE VERLAG GMBH · Berlin · Offenbach
deviation of the whole testing system is higher than 0.2%,
which doesn’t fulfill the criteria for testing Hall Effect
current sensors.
Figure 1 Components of Automatic Testing System
3 Error Correction Algorithm
For improving measuring accuracy of measurement sys-
tem, error compensation and correction methods can be
used with the help of modern data processing methods [7-
14]. Firstly, the errors of the measuring system have to be
determined. There exist systematic and random errors
[15]. The random errors can be reduced by signal pro-
cessing. Systematic errors are caused by the inaccuracy of
the system and can be determined and corrected. There
are two methods for error determination of the testing sys-
tem.
The first method is using a reference current sensor (see
figure 2). The output voltage Vref of this reference sensor
is measured by a high precision measurement system with
accuracy of ±0.01% in advance.
Figure 2 Error Determination with Reference Sensor
For calibration, this reference sensor has been measured
by testing system under correction again in order to get
the output voltage Vsys for the same input current values.
Thus, the absolute voltage deviation can be calculated by:
(1)
According to the given ratio k between output voltage and
input current, the errors of the testing system ∆I can be
determined by
(2)
The second method needs a reference resistance with high
accuracy (±0.01%) and low temperature drift (< 10 ppm)
(see figure 3). The error detection procedure in this case is
comparable with the method above. The resistance has a
value of 0.001Ω. The voltage Vm which can be measured
is directly proportional to the input current (see figure 4).
This voltage can be easily converted to the input current
Im:
(3)
The error of the testing system ∆I can be determined by
the difference between measured current Im and the output
current Io from the DC current source:
(4)
Figure 3 High Precision Reference Resistor
Figure 4 Error Determination with Reference Resistor
Sensoren und Messsysteme ∙ 03. – 04.06.2014 in Nürnberg
ISBN 978-3-8007-3622-5
2
© VDE VERLAG GMBH · Berlin · Offenbach
In both cases the measured errors ∆I are saved as devia-
tion data matrix in the PC system. The principle of the
calibration is shown in the following figure 5.
Figure 5 Determination of Deviation Data Matrix
The deviation data matrix contains only the absolute devi-
ations. By using this matrix, error correction algorithm
can be performed for a testing system. Figure 6 shows the
principle of a testing system with error correction algo-
rithm.
Figure 6 Testing System with Error Correction
The error correction uses the deviation data matrix and
linear interpolation for eliminating the systematic errors
of the testing system. Concretely, the digital error correc-
tion is made by (5)
where represents the current output before correction,
the current deviation and the current output after er-
ror correction.
Another way is to manipulate the control voltage of the
current source if available. The control voltage V is pro-
portional to the desired output of the current source,
which can be set by the user under using a set current
(ideally ):
(6)
where is a conversion factor. In order to perform the er-
ror correction, the absolute deviation can be simply mul-
tiplied with the linear factor and added to the original
control voltage for getting a new control voltage :
(7)
By using , the current source is corrected and can
give out a more accurate current output . The testing
system can directly measure the current sensor output. In
both cases, the measured values are saved with the current
output () on hard disk of the testing system for further
processing.
For calculating deviation values which are not listed in
the data matrix, a linear interpolation can be used. Linear
interpolation is processed in order to create new values
between two known neighbor values. This method allows
the use of less real measurement values and therefore less
data space.
4 Results
The rest errors after correction are mostly dominated by
the random errors, because the majority of the systematic
errors are compensated by the error correction algorithm.
Moreover, with both error detection methods, the results
are nearly the same. An accuracy of about ±0.03% can be
reached. Table 1 shows the relative deviation of the test-
ing system with the error correction algorithm in the
range [0A, 120A]. Figure 7 visualizes the measuring re-
sults.
Furthermore, it is also possible to make even small cur-
rents more accurate. Table 2 and figure 8 show the results
in the range [0A, 5A].
Based on the above results, the testing system under using
error correction can be defined with an accuracy of
±0.05% at least. It fulfills the criterion for testing and cal-
ibrating Hall Effect current sensors.
Figure 7 Relative rest deviation in range [0A, 120A] of
testing system after Error correction
Sensoren und Messsysteme ∙ 03. – 04.06.2014 in Nürnberg
ISBN 978-3-8007-3622-5
3
© VDE VERLAG GMBH · Berlin · Offenbach
Current under test (A)
Relative deviation (%)
0
-0,027
5
-0,009
10
-0,005
15
-0,021
20
0,002
25
-0,011
30
0,002
35
-0,015
40
0,003
45
0,015
50
-0,007
55
0,013
60
-0,011
65
-0,028
70
-0,003
75
0,004
80
-0,013
85
0,004
90
-0,013
95
0,019
100
-0,014
105
0,001
110
-0,024
115
-0,005
120
0,013
Table 1 Relative rest deviation in range [0A, 120A]
Figure 8 Relative rest deviation in range [0A, 5A]
Current under test (A)
Relative deviation (%)
0
-0,003
0,5
-0,020
1,0
-0,005
1,5
0,006
2,0
-0,007
2,5
-0,017
3,0
-0,011
3,5
-0,020
4,0
0,002
4,5
-0,006
5,0
0,006
Table 2 Relative rest deviation in range [0A, 5A]
5 Conclusions
The proposed error correction algorithm is tested on our
current sensor testing system with EA-PS 8080-120 and
Agilent 34401A. From the test results one can draw the
following conclusions:
For testing Hall Effect current sensors, the measure-
ment system needs at least an accuracy of ±0.05%.
High precision systems are more expensive. It results
in higher testing costs for sensors.
The error correction algorithm contains two parts:
error determination and error correction.
The measuring error of a testing system can be de-
termined with a reference sensor or resistor. The ref-
erence sensor should be measured with a high preci-
sion measurement system in advance. The reference
resistor should have a high accuracy and a low ther-
mal drift.
The absolute deviations of the inaccurate system are
saved as a deviation data matrix in PC system for er-
ror correction.
During the error correction, the deviation data matrix
is used to compensate the systematic errors of the
testing system.
Linear interpolation can be used for getting a devia-
tion value between two known neighbor values. In
this way the data space can be saved.
Testing system under correction can be improved to
an accuracy of about ±0.03%. It fulfills the criterion
for testing current sensors.
The error correction algorithm offers a low-cost solu-
tion for improving the measuring accuracy of existing
testing systems.
It is very suitable for different testing systems and
can be easily integrated in custom software.
This principle can also be applied to measurements of
other electrical and physical quantities.
6 References
[1] E. Ramsden: Hall Effect Sensors – Theory and Ap-
plication. Amsterdam, London, New York etc.: Else-
vier Inc., 2006
[2] J.-G. Liu, A. Sanli, Y. Wang, C. Liu: Error Compen-
sation of Closed Loop Hall Effect Current Sensors.
2012 IEEE International Workshop on Applied
Measurements for Power Systems (AMPS), Aachen,
Germany, September 26-28, 2012
[3] M. Cable: Calibration: A Technician's Guide. ISA –
Instrumentation, Systems, and Automation Society,
2005
Sensoren und Messsysteme ∙ 03. – 04.06.2014 in Nürnberg
ISBN 978-3-8007-3622-5
4
© VDE VERLAG GMBH · Berlin · Offenbach
[4] Data Sheet Agilent 34401A Multimeter:
http://cp.literature.agilent.com/litweb/pdf/5968-
0162EN.pdf
[5] Data Sheet EA-PS 8000 2U:
http://www.elektroautomatik.de/fileadmin/pdf/datash
eets/datasheet_ps8000-2u.pdf
[6] User Guide and Specifications NI USB-6008/6009:
http://www.ni.com/pdf/manuals/371303m.pdf
[7] J.-G. Liu: Verbesserung der Messgenauigkeitin
rechnergesteuerten Mess- und Prüfsystemen. Tech-
nisches Messen, tm 9/2002, pp.390-398
[8] J.-G. Liu, U. Frühauf: Self-Calibration Measuring
Methods and Application to Measurements of Elec-
trical Quantities. Measurement, vol. 26, no.2, August
1999, pp. 129-141
[9] H. Heuermann, B. Schiek: Error corrected imped-
ance measurements with a network analyzer. IEEE
Trans. Instrum. Meas. 44 (2) (1995) 295-299
[10] G. Adria, M. Savino, A. Trotta: Windows and inter-
polation algorithms to improve electrical measure-
ment accuracy. IEEE Trans. Instrum. Meas. 38 (4)
(1989) 856-863
[11] J.-G. Liu, U. Frühauf, A. Schönecker: Accuracy Im-
provement of Impedance Measurements by Using
the Self-Calibration. Measurement, Vol. 25, no. 3,
April 1999, pp. 213-225
[12] J.-G. Liu, A. Schönecker, U. Frühauf, U. Keitel, H.-
J. Gesemann: Application of discrete fourier trans-
form to electronic measurements. Proceedings of the
First International Conference on Information,
Communications & Signal Processing 3 (1997)
1257-1261
[13] M. Sedlacek, M. Titera: Increased accuracy Meas-
urement of parameters of non-synchronously sam-
pled periodic signals. IMEKO TC4 Symposium on
Development in Digital Measuring Instrumentation
and 3rd Workshop on ADC Modelling and Testing,
Naples, Italy, vol. I, 1998, pp. 169-173
[14] J.-G. Liu, A. Schönecker: Recursive self-correction
algorithm of discrete Fourier transform and applica-
tion to electrical measurements. 43rd International
Scientific Colloqium, Ilmenau, Germany, September
21-24, 1998, pp. 515-520
[15] E. Schrüfer: Elektrische Messtechnik – Messung
elektrischer und nichtelektrischer Größen, 6th Editi-
on. Carl Hanser Verlag Munich, 1995
Sensoren und Messsysteme ∙ 03. – 04.06.2014 in Nürnberg
ISBN 978-3-8007-3622-5
5
© VDE VERLAG GMBH · Berlin · Offenbach
