DEVELOPMENT AND EVALUATION OF A HIGH RESOLUTION MEASURING SYSTEM FOR A SWITCHING IMPULSE MEASUREMENT
Hong Tang Anders Bergman
Swedish National Testing and Research Institute
Abstract
This paper describes development and evaluation of a high accuracy high voltage measuring system for switching impulse. The noise level is less than 0.05%, the estimated uncertainties for the peak value measurement is better than M.1%; for the time parameter measurement is better than +1%.
Introduction
High voltage switching impulses with front time and half value time of 250/25000 ps are mainly used in high voltage industries and .research laboratories for testing high voltage equipment and investigating discharge phenomena. IEC 60060-2 [l] requires for a reference measuring system that the uncertainty for peak value measurement should be less than +1% and for time parameter measurement less than +5%. In practise, a high voltage impulse measurement is always related to interference problems, tempe
rature rise, non-linearity of components, parasite parameters in large measuringcircuit and curve form evaluations. Therefore, even for a switching impulse (normally in the frequency range of DC to 5 kHz), it also exhibits difficulties for engineer and electrical metrology to improve the accuracy of a measuring system. In SP (Swedish National Testing and Research Institute), an 200kV high accuracy reference measuring system for switching impulse measurements was developed. Some details of the measuring system concerning designing and evaluation procedures will be discussed in this paper.
DescriDtion of measuring system
Fig. 1 shows a diagram of the measuring system. Basically a RC type circuit was used as a voltage divider. For obtaining a constant ratio for both high and low frequency, the measuring system should fulfil condition as formula (1):
R1C1= R2C LOW (1)
Capacitance in the low voltage arm C Low consist of low voltage capacitor C2and other capacitance from the measuring system. The resistors R1 and R2 are used for damping resonance in the circuit. They will not be heavily loaded as only transient input voltage cause currents through, thereby requirements for the resistors are comparatively low.R, is a matching resistor for the measuring cable t
o avoid high frequency oscillations.Low frequency behaviour of a divider mainly depends upon the capacitors in this circuits. A compressed gas capacitor C1 with cylindrical configuration is used as a high voltage capacitor. The concept used for compressed gas capacitors, ensures that it exhibits a low tun6 value, low temperature coefficient and negligible proximity effect. This provides a very defined high voltage capacitor with excellent frequency performance. The low voltage capacitor C2is composed by the combination of ceramic capacitors type NPO (also designated COG) with temperature coefficient less than 30ppm/"C. A low inductance in the low voltage arm is of vital importance for obtaining a good response for a RC voltage divider. Therefore, the circuit board has been designed with utmost care to obtain low inductance. Further care has been exercised to ensure that all components and their connections exhibit a low inductive coupling.
Fig. 2 shows a step response of the measuring system for different time scale. From the transfer chara
cteristics, the settling time is calculated to be less than 3.7 ps (less than 10 p required by IEC 60060-2), and response time430ns. A frequency evaluation bandwidth for the divider cables can reach to 50kHz.
A computer controlled HP 3458A is used for waveform capture. The DCV (DC voltage measurement) mode is used for digitising the waveform. The input impedance of Hp multimeter is 10 MQ; the frequency bandwidth 30kHz;and the vertical resolution 16 bit. A program developed under Labview platform is used to control the multimeter and to analyse the captured waveform.。Error caused bv transfer characteristics of divider
characterise
The step response test is a traditional way for evaluation of an impulse measuring system. However, none of the parameters related to step response method can be directly used for uncertainty evaluation for a measuring system. For analysing errors of a measuring system characterised by step r
esponse, FIT and IFFT transformations are used to computation the difference between the input and output signal. Paper [3]presents a detailed description on this method. Table 1 shows front time Tp changes between input and output curve forms for the step response shown in Fig. 2. It shows that the maximum front time changes Sr,is less than 0.32%. The peak value changes 6V is less than0.005%.
Three parameters are used to characterise a switching impulse, peak value, front time and half value time.Here only the front time determination is discussed. In IEC 60060-1 the time to peak is defined as "the time interval between the actual origin and the instance when the voltage reached its peak value". Both these points are difficult to determine precisely even though a switching impulse has less noise and distortion compared with a lightning impulse. Furthermore, for an ideal switching impulse, the time span for an interval where instantaneous value exceeds 99% of the peak value can be as long as over one hundred microseconds. It could bring a large uncertainty for
determination of such kind of measured curve form if the noise level is only 1% of the peak value. How
ever the time parameters can be determined on a fitted double exponential curve form. Especially in switching impulse calibration one can normally expect excellent double exponential curve form since only a simple capacitive loading exists. In our analysis program, an approximation function derived in [4] is used to obtainthe front time Tp of a curve form.
where T, i s time difference between the time rise to 30% and 90% of the peak value; T2 actual origin time
to the half peak value time. The maximum relative error of the front time from formula (2) and the double exponential curve form is less than 0.02%. (In IEC recommendation method, the error is less than 1.5%)
Error budpe t
The estimated total uncertainty (k=2)for the measuring system is better than 0.1% for the peak value measurement and 1% for the time parameter measurement under interference level of 0.05%. The unc
ertainty contributions (k=2) for the measuring system can be summarised as:
*This error caused by the divider can be considered as an offset and should be linearly added to the expanded uncertainty [3].
It can be observed that the uncertainties are mainly caused by the evaluation of the waveform instead of uncertainties arising from measuring instrumentation. Interference level of measured curve form will however play an important role for influencing the measuring accuracy and necessitates evaluation of stochastic variations between several impulses.
Conclusions
A developed measuring system using compressed gas capacitor combined with specially designed low voltagearm shows a good characteristics for measuring switching impulses. The use of a high resolution multimeter with a new evaluation program has made it possible to exploit the full potential for high accuracy for this measuring system.
References
（1）IEC 60060-2, High voltage test techniques, part 2,measuring system, 1994-1 1
（2）E.Kuffe1, W.S.Zaeng1: High Voltage Bngineering,Fundamentals, Pergamon Press, Oxford, 1988
（3）Hong Tang, Anders Bergman: "Uncertainty Calculation for an Impulse Voltage Divider Characterised by Step Response", 1 1" International Symposium on High Voltage Engineering, London,UK, August 1999
（4）Wakimoto Takayuki, Sat0 Shuji and Harada Tatsuya: "Time to Peak Value Determination for the Measured Switching Impulse Wave", 10" International Symposium on High Voltage Engineering, Montrkal, Canada, August 1997

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。