ACCURACY OF DISTRIBUTION CURRENT TRANSFORMERS UNDER NON-SINUSOIDEAL EXCITATION

K. Debnath School of Engineering University of Tasmania Hobart, Australia 7000 Abstract The accuracy of CTs used for metering purposes are required to be very accurate as they are used in tariff calculations. Decades ago, the current flowing in distribution circuits could be considered more or less to be of a purely sinusoidal form. In recent times, however, such is not the case. Many loads such as rectifiers, inverters and a host of other electronic devices draw non-sinusoidal currents. A non-sinusoidal current waveform can be considered to be composed of a pure sinusoid of the fundamental power frequency and its harmonics. Since these harmonics are injected into the system the ammeters and energy meters read the composite current, not just the fundamental component. The measurements will, therefore, reflect the actual currents only if the CTs transform the harmonics in the same proportion as the fundamental component. This paper presents the results of an investigation on the accuracy of some commercial current transformers. 1. INTRODUCTION and the secondary were measured. The following two types of instruments were used for the current measurements: 1. HP digital oscilloscope (Model 54602A) 2. Fluke 41 Power Harmonic Analyzer. Both of the above meters can provide true rms current values.
]

Current transformers used in measuring currents in high voltage and high current circuits need to have high accuracy since they may be used for tariff calculations. A sinusoidal primary current is expected to produce a similar secondary current wave shape with a high transformation accuracy. However, the current may contain harmonic distortion due to rectifying and other types of loads such as fluorescent lamps [1]. The question therefore arises whether the CTs remain equally accurate in the presence of the harmonics. It is the purpose of this paper to present the results of an investigation performed on commercial current transformers to establish their behaviour in the presence of harmonics. Although both protection and metering type of CTs were used in the investigation, the results of only the latter type are presented in this paper. 2. TEST METHODS AND EQUIPMENT In addition to some others, the following two CTs were made available to us by a power utility for investigation: Delle Delle Type TAT 1 Type TAT 2 15VA 50 Hz 150/5A 30VA 50 Hz 300/5A

2.1 Harmonic sources Two different types of sources were used. The first one is a single-phase half-controlled bridge rectifier fed from the mains power supply. The current drawn by the rectifying circuit was passed through the CT primary winding as shown in Figure 1. The harmonic contents of the current can be varied by varying the firing angle of the thyristors. The advantage of this type of source is that it is realistic and representative of the kind of non-sinusoidal currents experienced in the real world [2]. The disadvantage is that the
A Secondary current Half-controlled rectifier A CT Ac mains Primary current Resistive load

Tests performed on both CTs yielded similar results. Therefore, results of only the first one are presented in this paper. The primary winding of the CT was excited by a harmonic source and the currents in both the primary

Figure 1 Test circuit with controlled rectifier as harmonic source

magnitudes of the individual harmonics can not be controlled at will for test purposes. The second type of source used is a programmable waveform generator [PWG]. The equipment consists of computer software and hardware including a digital to analog conversion circuit. Since the output of this generator is very low, a power amplifier was used to deliver the required test current. This equipment is capable of generating any arbitrary waveform by specifying the order as well as the magnitude of the harmonics, in addition to the fundamental. An additional advantage of this generator is that it could generate a "pure" sinusoid in contrast to the somewhat distorted mains power supply. The test circuit with this kind of harmonic source is shown in Figure 2.

Rp Ip I p/ N

Lp Re Ie

Ls

Rs Ib

Cs Le 1:N Id eal
Figure 3 Frequency dependent equivalent circuit of a current Transformer. current through the total secondary circuit impedance including the burden. Since the secondary current may vary over a wide range i.e. from zero to rated current. The secondary current Ib can be expressed as I e I cs N For a practical CT, Ie is not zero. Also, at high frequencies, the current Ics can be significant [5]. Therefore the secondary current Ib is different from the nominal Ip/N. 2.3 Ratio correction factor Ib = Ip

Ics

Secondary current A

Primary current A PWG CT Resistive Load

Figure 2 Test circuit with programmable waveform generator (PWG) as harmonic source. 2.2 Equivalent circuit An equivalent circuit of the test CT was determined by performing the open-circuit and short-circuit tests. Since a frequency dependent model is considered, test voltages of varying frequencies were used. Figure 3 shows the kind of CT equivalent circuit considered [3] in this investigation. Some authors use a slightly different equivalent circuit [4]. The parameters of the equivalent circuit are as follows: Rp = Primary winding resistance Lp = Primary leakage inductance Re = Resistive part of the magnetizing impedance Ze Le. = Inductive part of the magnetizing impedance Ze Cs = Stray capacitance Rs = Secondary winding resistance Ls = Secondary leakage inductance Rb = Resistive burden (not shown) through which the current Ib flows. The exciting current Ie is dependent on Ze and the secondary voltage required to drive the secondary

The transformation accuracy of a CT can be defined in terms of its ratio correction factor (RCF). The turns ratio N multiplied by the RCF equals the ratio of primary current to secondary current. An ideal CT will have an RCF of unity. RCF = Ip N Ib = I b + I e + I cs I + I cs =1+ e Ib Ib

Using this RCF factor, CT accuracy may also be expressed as an error percentage as follows: % error = 100 I e + I cs 1 = 100 1 I b + I e + I cs RCF

It is noted that the percent error approaches zero as the RCF approaches unity. RCF is a function not only of Ie and Ics but also of the Ib i.e. the burden supplied by the CT. With a particular secondary current, Ics increases with frequency and so does the percent error because the second term of the RCF increases. A direct measurement of the currents in the primary and the secondary windings of the CT was performed and the ratio correction factor determined. This was done at various frequencies and is shown in a later section.

2.4 Open circuit test With the "primary winding" kept open, a voltage of 1 V rms was applied at the secondary. The magnitude and phase angle of the current drawn at the secondary were measured. This was repeated with a wide range of frequencies (50 Hz to 400 kHz). Plotted on a log-log paper, the open circuit impedance shows a linear rise upto about 125 kHz and then falls off as shown in Figure 4. Obviously, the impedance between 50 Hz and 125 kHz is due to Ze. The straight line can be expressed as log (Zoc) = 0.372 log (f) + 0.615 from which Z oc = 4.123 f
0.372 Frequency (Hz)

for f < 125 kHz

Figure 5 Phase angle of open circuit impedance of 15VA 150/5A Delle current transformer 2.5 Short circuit test For the purpose of this test, the "primary" was handwound with sufficient number of turns so that rated primary current was obtained. Short circuit test was then performed for a wide range of frequencies. Voltage and current magnitudes as well as the phase difference were noted. The short circuit impedance curve is shown in Figure 6.

Again, the equation of the straight line for the Zoc which can be considered due to Zcs, between 150 kHz and 400 kHz is given by: log (Zoc) = -0.999 log(f) + 7.669 Z oc = Z cs = 46647112.32 f C s = 3.412 nF
0.999

Frequency (Hz)

Frequency (Hz)

Figure 6 Short circuit impedances (referred to secondary side) of 15VA 150/5A Delle current transformer Upto a certain maximum frequency, Z p + Z s << Z e and therefore the short circuit impedance can be considered to equal to Z p + Z s .

Figure 4 Open circuit impedance of 15VA 150/5A Delle current transformer The phase angle of this impedance remains almost constant at 450 for a frequency upto 4 kHz as shown in Figure 5.

The secondary winding resistance can be measured directly by using a suitable measuring device (a bridge circuit or a digital meter). The leakage inductance of the secondary winding can be estimated by the following methods: 1. A maximum value for Ls can be estimated by noticing any lack of resonance in the short circuit impedance curve. The RCF of the CT can be measured as a function of frequency and the value of Ls that gives the best fit for the RCF calculated from the CT equivalent circuit and the measured RCF.

3. MEASURED DATA Using the test circuit shown in Figure 1, the primary current was adjusted to 51 amps (about 33% of rated current) corresponding to a firing angle of 900. The expected secondary current is 1.7 amps. The actual measured current was 1.73 amps. With the same primary current but at a firing angle of 1260, the secondary current was measured to be 1.72 amps. The same procedure was repeated with 80% of rated current (120 amps). The results are shown in Table 1. Table 1 Measured currents of the CT excited by rectifier current. Firing Angle () 90 126 90 126 Primary Current (A) 51 51 120 120 Secondary Current (A) 1.73 1.72 4.05 4.04

2.

The slope of the short circuit impedances gives an approximation of the primary leakage inductance (Lp) of the hand-wound primary. It may be noticed in Figure 6 that the portion between 200 Hz and 30 kHz is straight which can be expressed as Log (Zsc) = 0.936 log(f) -2.96

( Lp + Ls ) = 1.097 10 3 f 0.936
where L p is the primary leakage inductance referred to the secondary side and Ls is the secondary leakage inductance. This total inductance as calculated from the above equation is 1.94 mH. As there is no sign of any resonance up to 30 kHz, Ls may be calculated as 1 Ls = 8.249 mH 2 (2 30,000) 3.412 10 9 Using MATLAB, the value of Ls that gives the best fit for the RCF is found to be 0.56 mH. The resistance of the secondary winding Rs was found to be 0.1 ohm. With all these values the equivalent circuit of the CT was constructed and is shown in Figure 7. Rp Lp 2.91 f 0.37 3 . 4 1 nF 0.47 / f 0.63 0 . 5 6 mH 0.1 ohm

The maximum errors here are 1.8% (1.73 amps instead of 1.7) and 1.3% (4.05 amps instead of 4). It may be noted from the measured data that the error tends to decrease as the current approaches the rated value. The above measurements were repeated with the second harmonic source i.e. the programmable waveform generator. Primary current consisting of the fundamental and any one of the odd harmonics between 3 to 11 were injected at one time. The magnitude of the harmonic was 20% of the fundamental. The secondary current was measured to be 1.71 or 1.72 amps for a primary current of 51 amps as shown in Table 2. The secondary current does not change when all of this frequency components are added so long as their combined magnitude remains within 20%. The theoretical transformation ratio of 30:1 is not available even for the pure 50 Hz input (for a primary current of 120 amps, the secondary current is 3.99 amps instead of 4). However, it is noted with interest that exactly 4 amps is measured in Table 2 Measured secondary currents of the CT excited by a programmable waveform generator. Harmonic order 1 1,3 1,5 1,7 1,11 1,3,5,7,11 Secondary Current with Ip = 51 A (A) 1.71 1.71 1.72 1.72 1.71 1.71 Secondary Current with Ip = 120 A (A) 3.99 3.99 4.01 4.02 4.02 4.00

1:N I d e al T r a n s f o r m e r Figure 7 Frequency dependent equivalent circuit for the CT under investigation

the secondary when the primary consisted of many frequencies. The ratio correction factor curves for the current transformer with various burden resistances are shown in Figure 8. It may be noted from this figure that for a burden of 1 ohm or less, the ratio correction factor is very close to unity for frequencies upto 2 kHz. As the burden increases, the transformation error increases for the same frequency range ( 2 kHz). Beyond this frequency range of 2 kHz (40th harmonic) the transformation accuracy deteriorates regardless of the burden value. f (Hz) 50 100 150 200 250 300 350 400 450 500 550 600 650 700 800 900 1k 1.2k 1.4k 2.0k

Table 3 Phase angle error Burden, Ohms 0.2 1.0 3.0 5.0 Phase angle error, deg 0.58 0.20 0.67 1.64 0.77 0.79 1.31 1.32 1.20 0.54 1.17 1.42 0.86 1.35 1.32 1.29 1.34 1.15 1.28 1.68 1.51 1.31 1.36 1.77 1.19 1.54 1.17 2.05 0.59 1.24 1.40 1.82 1.35 1.06 1.20 1.63 2.10 1.23 1.24 1.93 1.67 1.62 1.18 1.80 1.72 1.55 1.58 1.73 1.21 1.37 1.85 1.79 1.15 1.43 1.47 1.96 1.66 1.69 1.92 1.88 1.61 1.63 1.62 1.48 1.53 1.66 1.37 1.95 1.30 1.75 1.44 2.19 1.45 1.58 1.59 2.04 1.33 1.47 1.57 1.89

10.0 2.34 2.03 2.11 1.85 1.89 1.93 1.78 2.04 2.38 2.23 2.09 2.37 2.43 2.28 2.19 2.07 2.27 2.18 2.25 2.17

Frequency (Hz)

2 degrees has been found for a burden of 10 ohms at a frequency of 2 kHz. Considerable time has been devoted in developing a frequency dependent equivalent circuit of the CT. Such a circuit can be utilized for the computation of actual current transformation ratio at any frequency. 5. REFERENCES [1] P. G. Kendall, "Harmonics in Power System", The Electricity Council, Power System Engineering Series, 1981. [2] J. Arrillaga, D.A. Bradley and P.S. Bodger. "Power System harmonics", 1985, Wiley, New York. [3] D.A.Douglass, "Current Transformer Accuracy with Asymmetric and High Frequency Fault Currents", IEEE Transactions on Power Apparatus and Systems, Vol. PAS-100 No. 3, March 1981. [4] V.J. Gosbell and G.J. Sanders, "Frequency Response of distribution CTs", Proceedings of AUPEC96, pp. 77-82. [5] Wright, "Current Transformers: Their transient and Steady State Performance", 1968, Chapman and Hall.

Figure 8 Ratio correction factor curves for the current transformer with different burden. The phase error of the current transformer has been determined for 50 Hz and its harmonics up to 2 kHz for burdens of 0.2 ohm, 1 ohm, 3 ohms, 5 ohms and 10 ohms respectively. Phase angle errors can exceed 20 even at 50 Hz with high burden say 10 ohms. However, this error remains well within the bound of 20 for upto the 40th harmonic if the burden is 3 ohms or less. Even with a burden of 5 ohms, the phase error is very close to 20 as shown in Table 3. 4. SUMMARY & CONCLUSIONS Measured data on CT accuracy has been reported in this paper. Two different types of harmonic sources were used with varying degrees of representation of practical situations. It has been seen that the ratio correction factor is very close to unity for low burden (1 ohm or less) for a frequency range of upto about 2 kHz i.e. 40th harmonic. There is a sharp deterioration of the RCF beyond the 40th harmonic even with very low burden. It has also been observed that the phase angle error increases, in general, with frequency for a particular burden. A maximum phase error of slightly above