Measurement 42 (2009) Contents lists available at ScienceDirect. Measurement. journal homepage:

Transcription

1 Measurement 42 (2009) Contents lists available at ScienceDirect Measurement journal homepage: A new instrument for the measurement of peak value of non-sinusoidal asymmetric voltage over wide range of frequency Biswendu Chatterjee *, Debangshu Dey, Sivaji Chakravorti Electrical Engineering Department, High Tension Laboratory, Jadavpur University, 188, Raja S.C. Mullick Road, Kolkata , India article info abstract Article history: Received 28 July 2007 Received in revised form 3 March 2008 Accepted 4 April 2008 Available online 12 April 2008 Keywords: Peak voltmeter Voltage measurement Peak value High voltage This paper presents a novel approach towards measurement of peak value of high alternating voltage. Performance shows that the proposed method is suitable for laboratory applications as well as for industrial usage. The proposed scheme is low-cost and also portable and small enough to be mounted anywhere. It can be adjusted to adapt to wide range of values of high voltage arm of capacitive potential divider present at the working place for peak value measurement. The peak voltmeter (PVM) circuit is tested with the help of different waveshapes at different frequencies and the results show that it can measure the peak values for a wide spectrum (from 50 Hz to 1 khz) of waveforms efficiently. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction In high voltage applications breakdown of insulation is dependant on the peak value of the applied voltage and not on the effective value. So measurement of the peak value in high voltage systems is of immense importance. Various techniques are in use for the measurement of high voltage ranging from simple potential divider to sophisticated sensors like optical transducers. However, all these methods have specific areas of application as well as own advantages and limitations. Voltage transformers and current transformers have been used for decades to step down the magnitude to a safe level for instrumentation. But these transformers suffer from errors and have very limited frequency response [1]. Moreover these instruments are not peak reading. Peak value is obtained by multiplying the measured effective value with peak factor provided that the waveshape is known. This also introduces error in the case of non-sinusoidal waves. * Corresponding author. address: biswenduc@gmail.com (B. Chatterjee). A better choice is the use of an oscilloscope because apart from measuring the peak voltage, they also provide valuable information on waveshape [2]. But oscilloscopes are not only costly and bulky, but often also unsatisfactorily accurate. Besides that, the high voltage to be measured is scaled down by a potential divider to a level within the operating range of the oscilloscope. Thus the peak magnitude of peak obtained from oscilloscope recording is to be multiplied by a factor to calculate the original peak value. This calculation needs application specific skill for correct interpretation. The use of vacuum tube voltmeters for peak value measurements became widespread due to its direct reading capability for a fixed potential divider ratio [3]. But the vacuum tube circuitry is also bulky, needs auxiliary high voltage supply and its portability is poor. Another important fact is that, the use of vacuum tube voltmeters is at present not only obscure but it also may be rather hard to reach. One of the common methods used for peak reading is the modified Chubb Fortescue peak measuring circuit [4]. According to the basic principle the peak value V p is proportional to the arithmetic mean I m of the rectified current flowing through a high voltage capacitor of capacitance C /$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi: /j.measurement

2 72 B. Chatterjee et al. / Measurement 42 (2009) Fig. 1. A schematic of the PVM. Fig. 2. Photograph of the developed peak voltmeter. Fig. 3. Photograph of the high voltage measuring capacitor C 1. such that I m =2fCV p where f is the fundamental frequency of the wave. The limitation of such measurement is that the basic principle assumes that there are no secondary voltage peaks anywhere and that both positive and negative half-cycles are identical [5]. Although these methods are widely used, they have some notable limitations. As for example these methods are not capable of measuring the peak values of both the half- cycles. Moreover, all these schemes has a disadvantage that if the capacitance of the high voltage arm of the capacitive potential divider changes, the peak reading has to be adjusted according to the change in multiplication factor. In some cases, the use of bleeder resistance across the peak holding capacitor makes the circuits frequencydependent, i.e. for different frequency values the bleeder resistance should be different. This paper presents a novel approach towards measurement of peak value of high alternating voltage. Performance shows that the method is suitable for laboratory applications as well as for industrial usage. The proposed scheme has the following advantages, viz. (i) the development cost is low, (ii) is based upon analog signal processing, (iii) is portable and small enough to be mounted anywhere, (iv) can be adjusted to adapt to wide range of values of high voltage arm of capacitive potential divider present at the working place, (v) the two half-cycles can

3 B. Chatterjee et al. / Measurement 42 (2009) Table 1 Response of the PVM for different peak values of sinusoidal voltage and C 1 ESV readings (kv rms ) Calculated peak (kv rms p 2) (kv peak ) Peak reading of the developed meter (kv peak) C 1 = 26.1 pf C 1 = 62.5 pf C 1 = pf C 1 = pf Pol + Pol Pol + Pol Pol + Pol Pol + Pol Fig. 4. Response of the PVM for a sinusoidal input. (a) The input voltage waveform and (b) The voltage across peak holding capacitor before and after discharge. be individually observed for peak value measurement and (vi) the peak can be held and discharged for accurate results. Performance of the peak voltmeter (PVM) circuit is tested with the help of different waveshapes and the results show that it can measure the peak values for a wide spectrum of waveforms efficiently. 2. Developed scheme A schematic circuit diagram is given in Fig. 1 elaborating the principle of operation of the developed circuit. The high alternating voltage (measurand) is impressed at the high voltage capacitor C 1. Another capacitor C 2,in series with C 1, forms the capacitive potential divider. The scaled down voltage is given to OP AMP (A 1 ) that acts as a buffer with very high input impedance for increasing the linearity of the divider. C 1 is provided in-site and C 2 is located inside the meter enclosure. OP AMP (A 2 ) acts as unity gain inverter. The pole of switch SW 1 can be Fig. 5. Response of the PVM for asymmetric sinusoidal input. (a) The input waveform; (b) Output waveform with polarity switch set to (+) position and (c) Output waveform with polarity switch set to ( ) position. connected to the inverted as well as the non-inverted input to investigate the peak for both positive and negative halfcycles separately. As it is evident that for different values of C 1 the multiplying factor of the capacitive potential divider changes, necessary modifications in the circuit have been introduced. The principle of the adjustment of the circuit for different values of C 1 is explained below. Say, the adjustment is to be made for a range of C 1 from C 1min to C 1max. Again, for C 1min, let the voltage at the point A is V in. Then, V in ¼ V p C 1min C 1min þc 2 (where C 1 and C 2 are in pf and V p is the actual peak value of measurand) = V p k. This k < 1 is a factor, which is dependent on the lowest possible value of C 1 under consideration and the value of C 2.

4 74 B. Chatterjee et al. / Measurement 42 (2009) So, V in ¼ V 0 in k 1 þ C 2 C 1 Therefore, the regular expression of V in can be given as V in ¼ V 0 in k þ V 0 in k 1 C 1 ð1þ Fig. 6. Effect of discharging the peak holding capacitor at the negative half-cycle of input when the polarity switch is at (+) position. (a) The input voltage waveform and (b) The voltage across peak holding capacitor before and after discharge. where k 1 = k C 2. Therefore, a fixed fraction of the voltage at point A is to be added with a variable fraction that depends on the value of capacitor C 1. It thus will give the proper scaling so that the output shows the same peak for a given input even if C 1 is changed, provided that the appropriate value of C 1 is set on the front panel. This value of C 1 is to be set in pf using the knobs provided on the SET-C 1 section of the front panel as shown in Fig. 2. This panel is actually a calibrated potential divider. Another potential divider formed by resistances R 5 and R 6, feeds a fraction of the scaled down input voltage at point A to OP AMP A 3, which then acts as the adder that is required as explained in Eq. (1). The ratio panel of this work is designed in such a way that the adjustment can be done for the values of C 1 ranging from 20 to 500 pf. However, the design is not limited to this range of C 1. It can be suitably extended, if required. This eliminates the need for any separate calculation and the peak value can be obtained directly from the meter display. A 4 acts as the half wave rectifier very near to ideal diode characteristics. Fig. 7. Effect of discharging the peak holding capacitor at the positive half-cycle of input when the polarity switch is at (+) position. (a) The input voltage waveform and (b) The voltage across peak holding capacitor before and after discharge. Now, if C 1 > C 1min, then V in is increased to V 0 in ¼ V p C 1 C 1 þ C 2 Fig. 8. Response of the PVM for asymmetric, non-sinusoidal waveform having local maxima. (a) The input waveform; (b) Output waveform with polarity switch set to (+) position and (c) Output waveform with Polarity Switch set to ( ) position.

5 B. Chatterjee et al. / Measurement 42 (2009) The function of the capacitor C hold is to hold the peak value of the voltage obtained from A 4. This capacitor is of very low leakage polystyrene capacitor and used with a value of 0.1 lf. Since this capacitor holds the highest peak and does not respond to a change in peak value lower than that held by C hold, a discharge section is provided in the front panel of the instrument to discharge this capacitor. To read the peak, C hold is to be discharged manually by using the MANUAL switch provided in the front panel to clear the voltage held by C hold, if any. The function of the AUTO switch is to discharge the capacitor for 1 ms at an interval of 10 s. It refreshes C hold automatically to read the peak within a time interval of 10 s. The interval of 10 s is chosen as a specific case in this work. This interval can be pre-adjusted during construction according to any practical requirement. It helps the user to keep track of the change in peak value for a waveform, which has time varying peak. The OFF switch is provided to stop discharging C hold so that it can hold the global peak until the user presses MANUAL. The circuit is tested from 50 Hz to 1 khz signal and is found to give satisfactory results over this wide frequency range. Fig. 3 shows the high voltage measuring capacitor C Results and discussion The peak values of high voltage sinusoidal AC source as measured by the developed PVM for different values of high voltage capacitor C 1 are tabulated in Table 1. The values of Fig. 9. Error curves for different values of peak voltage and high voltage capacitors. Fig. 10. Response of the PVM for different peak values of the input voltage.

6 76 B. Chatterjee et al. / Measurement 42 (2009) C 1 are set accordingly in the SET-C 1 section of the front panel and the readings are compared with a precision electrostatic voltmeter (ESV) that gives true RMS value. The ESV used is ESH-8X from sensitive research. The accuracy of the ESV is less than 0.5% according to the manufacturers specification. Table 1 shows the response of the PVM for different values of high voltage capacitor C 1 and peak voltage in both the half-cycles, where both the half-cycles were identical and sinusoidal. Fig. 4a shows a sinusoidal voltage waveform to the input of the circuit for the evaluation of its performance. Points A and E represents the two positive peaks, B and D represents the two zero crossings and at point C the discharge switch is pressed. Fig. 4b shows the voltage across the peak holding capacitor. It is observed from the figure that C hold holds the previous peak until it is discharged at C where the voltage drops to zero almost instantly. Since the polarity switch in the front panel is kept at (+) position, and the discharge occurs at the negative half-cycle, the voltage remains zero until the next zero crossing at D. From D, the voltage follows the input upto the peak at E and holds this new peak value. The performance of the PVM is also evaluated for asymmetric waveform having different peak values in the positive and negative half-cycles. With the help of polarity selection switch provided in the front panel, the PVM almost accurately shows the peak of the two half-cycles individually as shown in Fig. 5b and in Fig. 5c. Fig. 6a is another input waveform to test the circuit. It is a non-sinusoidal waveform having 27% 5th harmonic in addition to the 50 Hz fundamental sine wave. Points A and G represents the two global positive peaks, C represents the zero crossing, B is the time when the discharge switch is pressed and D is the local peak. Fig. 6b shows the voltage across the peak holding capacitor. It is observed from the figure that C hold holds the previous peak until it is discharged at B where the voltage drops to zero almost instantly. Since the polarity switch in the front panel is kept at (+) position, and the discharge occurs at the negative half-cycle, the voltage remains zero until the next zero crossing at C. From C, the voltage follows the input upto the local peak at D and holds this new peak value upto F while the input voltage drops from D to E and again rises to the magnitude of D at F. This observation clearly reveals the characteristics the instrument as a peak voltmeter. From F, since the voltage is again rising, the output follows the input upto the global peak at G and holds this peak value. Fig. 7a is another case with same input waveform as in Fig. 7a with polarity switch set to (+) position. Contrary to Fig. 7b, the discharge takes place at the positive half-cycle of the input at point B. It is observed from Fig. 7b that before discharge, the capacitor held the global peak upto the point B. Following discharge for 1 ms, the voltage rises to point C,increases upto point D, the local peak and remains there upto the point E neglecting any fall in the input voltage. From E, the voltage rises to the global peak F and remains there. As the function of the polarity switch is to pass the signal inverted or non-inverted as stated earlier, it is evident from these two figures that the discharging operation will give similar response in the negative cycle also, keeping the polarity switch at ( ) position. Fig. 8b and c shows the typical case of asymmetric, nonsinusoidal waveform along with the presence of local peak. Success for measurement of the peak values for each of the individual half-cycle shows that the PVM also performs efficiently for this type of waveforms. All the waveforms of Figs. 4 8 are obtained from data of a dual-trace digital storage oscilloscope. The error curves for different peak values with different high voltage capacitors are shown in Fig. 9. The variations are almost linear and the error decreases with the increase in the measured peak value. These curves thus show that the characteristic of the PVM is nearly a linear one. Further evidences of this are shown in Figs. 10 and 11. Fig. 10 shows the voltage linearity and Fig. 11 gives the frequency depen- Fig. 11. Response of the PVM for different input values of the peak and frequency.

7 B. Chatterjee et al. / Measurement 42 (2009) Table 2 Response of the PVM for different peak values of non-sinusoidal voltage having multiple frequency components and C 1 Screenshot of waveforms Actual peak (kv peak ) Peak reading of the developed meter (kv peak) C 1 = 26.1 pf C 1 = 62.5 pf C 1 = pf C 1 = pf +Peak Peak Pol + Pol Pol + Pol Pol + Pol Pol + Pol dence of the developed meter. For depicting the clarity of the voltage linearity curve, the results for high voltage capacitor C 1 = pf are presented as a representative case. Similar linear results were also obtained for other values of C 1, too. To judge the frequency dependence, sinusoidal waveforms with different peak values and frequencies are generated by a signal generator and the peak values of the respective voltages are measured. The dependence is investigated for a wide range of frequencies, i.e. from 50 Hz to 1 khz. The response curves of the PVM for different frequency values show acceptable results. For added interest, Table 2 shows the response of the PVM for different values of high voltage capacitor C 1 and peak voltage in both the half-cycles, where the waveform is non-sinusoidal having a wide range of frequency components. 4. Conclusions From the above results it is evident that the developed PVM is capable of measuring the peak values of different kinds of waveform. The results obtained from the experiments show that the PVM is capable of measuring the global peak independent of the waveshapes over wide range of frequency. The results also show that the percentage full-scale error in peak value measurement decreases with the increase in measured value. For low values of measurand voltage peak the error lies within 0.85% as obtained from comparison with electrostatic voltmeter. For higher values of peak voltage it lies within 0.5% which is practically well-acceptable for this type of instrument. The linearity of the instrument is also good for a wide range of frequency and input voltage. As the scale of the instrument is automatically adjusted for different values of high voltage capacitor using the front panel dial setting, it behaves like a direct reading instrument which is a significant feature of this developed instrument. References [1] W.Z. Fam, A novel transducer to replace current and voltage transformers in high-voltage measurements, IEEE Transactions on Instrumentation and Measurement 45 (4) (1996) [2] A.R. Ondrejka, Peak pulse voltage measurement (baseband pulse), Proceedings of the IEEE 55 (6) (1967) [3] C.J. Creveling, L. Mautner, An automatic-slideback peak voltmeter for measuring pulses, Proceedings of the I.R.E.-Waves and Electrons Section 35 (2) (1947) [4] R. Marx, R. Zirpel, Präzisions-Messeinrichtung zur Messung hoher Wechsel-und Gleichspannungen, PTB-Mitteilungen 2 (1990) [5] A. Bergman, R. Marx, K. Schon, E.-P. Suomalainen, J. Hallstrom, Intercomparison of AC peak voltage measurement, Proceedings of the Eleventh International Symposium on High Voltage Engineering 1 (1999) 9 12.