150
Accurate CT Calibration for the Model 1133A
15.7
Selection of the Current Calibration Points
The proper choice of calibration points (test currents) satisfies two criteria: first, the points chosen
should allow complete characterization of the CT; and second, the points should allow for accurate
interpolation. Since CT’s are magnetic devices, they have error curves, which are smooth and do
not have jumps or breakpoints. Usually, either a linear (1, 2, 3, 4, 5) or logarithmic (0.1, 0.2, 0.5,
1.0, 2.0) progression of the test current values will provide an accurate characterization of the CT
performance.
Many interpolation algorithms do not easily deal with non-linear input progressions, often
yielding bizarre results. The algorithm chosen for the 1133A will operate well with either logarithmic
or linear progression of test points. No algorithm can be expected to yield acceptable performance
if the test points are not spaced out in some logical manner. For example, the test sequence (0.1,
0.2, 0.3, 10, 15, 20) might seem reasonable for a device which had relatively large errors at the ends
of its range, and small, consistent errors in the midrange. However, the large hole in the middle
is certain to confound any general-purpose interpolation algorithm with more complex-than-linear
interpolation. What will confuse the algorithm is the dramatic change in the slope of the x-axis
points at 0.3 and 10. Using points with a relatively predictable and consistent pattern, such as
logarithmic or linear progression, will eliminate this potential problem.
To determine which is the best choice for a particular type of CT, you could characterize the
CT initially using enough points to thoroughly describe its performance. Then, the points can
be plotted on graph paper (or using a spreadsheet program) with log and linear axes for the
current values. Whichever curve appears smoother and more representative of the CT to the eye is
probably the best choice for the calibration point sequence. In general, logarithmic sequences (1,
2, 5, and 10) emphasize the lower current values and linear (1, 2, 3, 4, 5) sequences emphasize the
higher current values. Where the errors are changing rapidly at the low end of the current range
(usually due to changes in permeability of the core as a function of current level), a logarithmic
progression will usually work best. When the errors are changing rapidly at the high end of the
current range (usually due to incipient saturation of the core with increasing burden voltage), then
a linear progression might be a better choice (although a reduction in burden will almost certainly
improve overall performance).
15.8
Conclusion
Taking advantage of the full accuracy of the 1133A Power Sentinel requires calibration of the user’s
CT’s. This paper presents a method to perform these calibrations.
Calibrating existing CT’s requires three extra pieces of equipment: a reference CT, an excitation
current source and a current comparator. Two newly developed products by Arbiter Systems should
provide users with these tools for on site calibrations. The Model 935 Current Source and Model
936 Reference CT are now available for this purpose. The popular Arbiter Systems Model 931
Power System Analyzer can accept three-phase primary currents up to 2000 amperes (higher under
some conditions), with a transfer accuracy of 5 ppm and an overall traceable accuracy of 0.01%, or
better.
