7th International Symposium on H.V. Engineering - ISH '91 Dresden, Germany

DR. STRAUSS

Abstract 

Three digital recorders with sampling rates respectively rated resolutions of 100 MHz - 10 Bit, 100 MHz - 8 Bit and 25 MHz - 8 Bit were calibrated at full and chopped lightning impulses using a calibration pulse generator with a peak voltage up to 1000 V. The peak and time errors of the recorders were evaluated in compliance with the new IEC standard.

1.  Introduction

The new IEC-Standard /1/ "DIGITAL RECORDERS FOR MEASUREMENTS IN HIGH-VOLTAGE IMPULSE TESTS, Part I - Requirements for Digital Recorders", specifies the measuring characteristics and calibrations required to meet the measuring accuracies and procedures specified in IEC Pub. 60-1: 1989. This international standard defines the terms specifically related to the digital recorders used for measurement during high voltage and high current impulse tests, specifies the necessary requirements for such digital recorders to ensure their compliance with the requirements for high voltage and high current impulse tests and establishes the tests and procedures which are necessary in order to fulfil these requirements.
The paper summarizes the results of performance tests by pulse calibration and step calibration on three impulse voltage measuring systems including digital recorders designed for high voltage and high current impulse tests with sampling rates respectively rated resolutions of 100 MHz - 10 Bit (TRAS 100-10), 100 MHz - 8 Bit (TRAS 100-8) and 25 MHz - 8 Bit (TRAS 25-8). The measuring systems are available with several input amplifiers with calibrated deflection factors from 0.05 V to 10 V, 2 V to 400 V and 5 V to 1000 V each with 24 steps in 0.8 sequence to get optimal adaption to the measuring signal amplitude. The calibration pulses applied had a peak value of 927 V for full  lightning impulses and 800 V for chopped impulses to produce a deflection of 80% to 92.7% of the full scale deflection of 1000 V.

2.  Pulse Calibration Procedure

The calibration pulse generator used can generate full lightning impulses 0.84/60 usec with peak values from 400 V to 1000 V, front chopped lightning impulses 0.84/60 us and linearly rising voltage impulses with peak values from 200 V to 800 V with a time-to-chopping of 500 ns.
The input voltage range of the tested digital recorders was 1000 V, so the applied calibration pulses fulfil
the requirements to produce a deflection of between 40% and 100% of the full scale deflection.
The digital recorders tested are part of a complete impulse voltage measurement system with an integrated personal computer AT 386 class that controls all digital recorder settings, performs automatically the measuring procedure, evaluates and displays the measuring data on color screen and storages them onto hard disk. Details can be found in /2/.
The evaluated pulse parameters are separately stored into a protocol file to generate a test report or allow additional sta

tistic analysis of the test results.
The measuring data stored or displayed are not subject for any interpolation, filtering or curve fitting algorithmen to increase measuring accuracy.
The record length of the 100 MHz recorders TRAS 100-10 and TRAS 100-8  at maximal sampling rate is 640 us corresponding to 64K samples, the record length  of the 25 MHz recorder is 1.28 ms corresponding to 32K samples, therefore full and chopped impulses can be recorded with the maximal sampling rate of the individual digital recorder.
The algorithm for the determination of the pulse parameters are as follows:
- determination of the base magnitude as an average of 10 samples in the flat part of the record
- determination of the top magnitude as the highest sample recorded
- determination of 30% point with help of a dynamic smooting algorithm to eliminate initial oscillations
- determination of 50% and 90% points without interpolation between samples
- determination of instant of chopping

Figure 1:  full impulse voltage 0.93/55 us measured with TRAS 100-10. The actual time parameters of the calibration generator at full lightning impulses had a small systematic deviation in respect to the required parameters of 0.84/60 us according to IEC standard, but could not be corrected for this test. The influence of this deviation to the test results regarding front time error, time-to-half-value  error  and peak voltage error is insignificant.
At last, twenty pulses of each shape were recorded and the digital measuring data were processed by the implemented software with regard to the peak value Us, the front time T1, the time-to-half-value   T2 at full impulses as also the equivalent front time T1 and the time-to-chopping Tc at front chopped impulses.

Figure 2: linearly rising voltage chopped at 500 ns, recorded with TRAS 100-10. The corresponding record with TRAS 100-8 is nearly identical. The marked points on the curve shows the real digital measuring points of the digital recorder in the distance of the sample intervall of 10 ns.

Figure 5:  front-chopped lightning impulse with time-to-chopping of 500 ns, recorded with TRAS 25-8.
The front time was calculated to T1 = 1,67 . Tx  where Tx is the time interval between the two samples greater than the 30% and 90% points. The applied circles on the curve at 30% and 90% respectively the straight line from calculated virtual origin at 0.0 us through 30% and 90% points have deviatins from the curve because they are not interpolated in respect to the sample intervall.

Figure 3:  front chopped lightning impulse with time-to-chopping of 500 ns, recorded with TRAS 100-10. The corresponding record with TRAS 100-8 is nearly identical.

Figure 6:  Step calibration with d.c. voltage of 800 V.
A d.c. voltage of 800 V +-0.1% was applied to the input of the digital recorders calibrated and then short circuit by an appropriate switching device. The resulting transition to zero level was recorded as shown for TRAS 100-10.

Figure 4:  full impulse voltage recorded with TRAS 25-8, the marked points on the curve shows the real digital measuring points of the digital recorder in the distance of the sample intervall of 40 ns.

3.  Test Results

Figures 7, 8 and 9 summarizes the results of the pulse and step calibration with the test impulses described before. The applied test voltages for calibration of the digital recorders were as follows:
A full impulse 0.93/55 us according to Fig. 1 and 4
B linearly rising voltage chopped at 500 ns, Fig. 2
C front-chopped impulse chopped at 500 ns, Fig. 3 and 5
D Step calibration according to Fig. 6
The following digital recorders where calibrated and marked as:
1 TRAS 25-8  25 MHz, 8 Bit
2 TRAS 100-8 100 MHz, 8 Bit
3 TRAS 100-10 100 MHz, 10 Bit

4. TRAS 100-10 100 MHz, 10 Bit, PTB calibration
   

approximately 20 MHz bandwidth (3). TRAS 25-8 with a bandwidth of 15 MHz cannot meet the accuracy requirements for a time to chopping below 1.2 us /3/ because of the smaller sample rate of 25 MHz.
The peak voltage errors for TRAS 100-10 found by PTB (4) with the 5 V waveform generator are in good accordance with the errors found with the 1000 V calibration generator (3).
The step calibration (D) shows for all recorders a negativ error of -0.5% to -0.8%, the peak voltage errors found before can be corrected by this value to improve accuracy.

Figure 8:  Front time errors dT1
Both 100 MHz recorders TRAS 100-8 and TRAS 100-10 meets the accuracy requirements supposing that the calibration generator causes an error perhaps of 0.5% which is included in the errors found. All errors are typically positiv and it must be assumed that the implemented algorithm for front time determination can be improved so that all calibrated recorders can meet the requirements.
The front time errors for TRAS 100-10 found by PTB (4) with the 5 V waveform generator are in good accordance with the errors found with the 1000 V calibration generator (3).

Figure 7:  Peak voltage error dU
For full impulses (A) all calibrated digital recorders meet the accuracy requirements.
For front chopped impulses (B,C) chopped at 500 ns the influence of  bandwidth and sample rate to the peak voltage error is significant, an mathematical model for easy determination can be found in /3/. Therefore TRAS 100-8 with a bandwidth > 40 MHz has a smaller error (2) than TRAS 100-10 with

Figure 9:  Time-to-half-value errors (A) and time-to-  chopping errors (B,C)  dT2.

For full impulses (A) all calibrated digital recorders meet the accuracy requirements regarding the time-to-half-value errors
For front chopped impulses (B,C) chopped at 500 ns both 100 MHz recorders TRAS 100-8 and TRAS 100-10 meet the requirements. TRAS 25-8 cannot meet the accuracy requirements for a time to chopping below 1 us because of the limited sample rate of 25 MHz.
The time-to-half-value  error  for TRAS 100-10 found by PTB (4) with the 5 V waveform generator is in good accordance with the error found with the 1000 V calibration generator (3). The time-to-chopping error for TRAS 100-10 found by PTB (4) is more negative than error (3). Because the internal timebase of all digital recorders calibrated is quarz-stable with  0.01 %, errors only can occur through the analog bandwidth limit, through the quantization error or through the implemented evaluating algorithm.
The relative high errors of front time T1 and time-to-chopping Tc of the digital recorder TRAS 25-8 are caused by the missing interpolation between samples of the implemented evaluating algorithm. The influence of this missing interpolation to the respective errors of the 100 MHz recorders is smaller but not negligible. Using the digital recorder TRAS 25-8 with the relativ low sampling rate of 25 MHz for measuring front chopped lightning impulse voltages the software must be extended to reduce this errors, for the 100 MHz recorders the errors can be decreased too. Note that only the evaluating accuracy is increased by interpolation between samples without processing of raw data record. Algorithm to increase measuring accuracy of a 25 MHz digital recorder are described in /3/.

4.  Conclusion

The new IEC-Standard is an applicable instruction for the tests and procedures to find the relevant errors of digital recorders used for measurements during high voltage and high current impulse tests. Only digital recorders with a calibrated input amplifier with stable impulse scale factor, low internal noise and quarz-stable timebase will meet the accuracy requirements.
The calibrated digital recorders TRAS 100-8 and TRAS 100-10 fulfil  all requirements for measurements during high voltage and high current tests. The digital recorder TRAS 100-8 is recommended for tests where only the impulse parameters are to be evaluated. For tests which require comparison of records e.g. for testing transformers the digital recorder TRAS 100-10 is recommended. The digital recorder TRAS 25-8 is recommended for tests with full or chopped impulse voltages with a time-to-chopping greater than 1.2 us, where only the impulse parameters are to be evaluated.

References

/1/ IEC ...: Digital recorders for measurements in high voltage impulse tests, Part 1 - Requirements for digital recorders, to be published in 1991 (based on document IEC 42(CO)43, 1990)
/2/ W. Strauss: 100 MHz Transienten-Mess-System für Blitzstoßspannung, Conference Proceedings of 4th Exhibition and Conference for Industrial Measurement 1990, Wiesbaden, published by Network GmbH, pp. 380-386.
/3/ W. Strauss: Automatisierung der Stoßspannungsprüfung mit Mikrocomputern in Echtzeitbetrieb unter Einsatz eines Transienten-Recorders. Dissertation TU Berlin 1983.

Address of the author:
Dr.-Ing. Werner Strauss
DR. STRAUSS SYSTEM-ELEKTRONIK GMBH
D-96163 Gundelsheim/Bamberg
Federal Republic of Germany