International Conference on Information Technology and Management Innovation (ICITMI 2015)

Investigation of High Frequency Current Transformers Used for Partial

Discharge Detection

S.H.HE1 , Q.H.ZHAN1, S.B.HUANG1 & J.B.WANG1&Y.M.ZHANG2 & Y.P.XU2

1Guangdong Power Grid Corporation Foshan Power Supply Bureau, Foshan, China
2
Shanghai Jiao Tong University, Shanghai, China

KEYWORD: partial discharge; online monitoring; onsite detection; HFCT; frequency characteristics
ABSTRACT: In the contribution, the equivalent circuit model of high frequency current transform-
ers (HFCT), used for partial discharge online monitoring or onsite detecting, is established. Next, the
major parameters of HFCT are derived and the influences on amplitude-frequency and phase-
frequency characteristics are discussed. In the end, an optimized design for HFCT is proposed.

INTRODUCTION
Partial discharge monitoring inspects electrical equipment insulation state. Thus, on-line monitoring
sensor shall meet the following requirements: easy installation, low noise, safety. The above ad-
vantages make HFCTs become one of the effective measures for on-line monitoring of the trans-
formers, generators, power cables and other equipment partial discharges. The relationships between
transfer function and various parameters are derived. On this basis, Optimization HFCT design pre-
sents.

MODEL ANALYSIS
HFCTs have similar structure with Rogowski coil (Fig.1).Partial discharge signal is small current sig-
nals. Thus, ultrachrystallite material having a high magnetic permeability is selected as HFCT core
(inner diameter D1, an outer diameter D2, the height h), in order to improve sensitivity. In one direc-
tion by the winding core N turn coils and integrally connected to a resistor, usually close to the sam-
pling circuit through a long coaxial cable. When the measurement happens, the current is measured
by placing HFCT center. Figure 2 is a schematic diagram of a magnetic circuit analysis HFCT under
common conditions. In the figure, i1 is measured currents; N1 is the turns (usually 1 ring) of to be
measured current through the center of the current HFCT. U1 is the voltage to be measured by the
HFCT coupled to the circuit. The secondary winding is HFCT output coil, i2 is the output current
coupled by the HFCT. The secondary coil of HFCT (integral characteristic impedance of the coaxial
cable resistance should be) is a coaxial cable, which is a few to tens of meters long connection to the
integrator resistor. It corresponds to CL and RL (integrating resistor) connected in parallel, as shown
in FIG. 2.
Then, working mechanism HFCTs magnetic field will form a view for analysis. Magneto motive
force (MMF) F1, F2 are generated by the primary and secondary coil :

(1)

(2)
uuv
and uuu
v
the primary coil and the secondary coils generate the magnetic fields H1 H2 . .
At this time, the remaining MMF system.
(3)
Where in, im is converted current, which generates FM, to the primary side. From Eq. (1), (2) and
(3) it can be deduced

© 2015. The authors - Published by Atlantis Press 247

 (4)
From Eq. (4), we see that the smaller im will make ratio of the primary current to the second-
ary more linear. Therefore, we reduce can to improve accuracy.
According to the principle of electromagnetic induction, HFCT U2 voltage induced in the sec-
ondary side is

(5)
Acquire the relationship of u 2 and im from Eq.(3) and (5):

(6)
is the core reluctance of the HFCT . Perform Fourier transform on Eq. (6):

(7)
From Ohm's Law, obtain the relationship between U 2 and I 2 :

(8)
Substitute Eq. (7) and (8) into Eq. (9), current I1 the transfer function to the voltage U 2 is:

(9)
N22
Where LL = . In Eq. (9), Section is very similar to a parenthesis RLC parallel resonance cir-
N
cuit with resistance R1 l, capacitors and inductors C1 . L1 . Therefore, we can analyze HFCT per-
formance using a parallel RLC resonant circuit theory.
Slightly modify the Eq. (9), we can get:

(10)
1 N CL RL
The center frequency wC = g ; Q = RL = g N CL is the quality factor, which
N 2 CL LL N 2
are two important parameters affecting HFCT performance.
According to Eq. (10) the amplitude- and phase-frequency characteristics of H ( jw ) can be
calculated:
1）A HFCT transfer function is a first-order band-pass filter, the center angular frequency is
wC . It meets:

(11)

(12)

248
 2）There is close relationship between the lower cutoff frequency w L and the higher cut−off
frequency w H of the transfer function and Q:

(13)

(14)
In Fig. 3, the amplitude and HFCT transfer function, which can be seen that the phase - fre-
quency curve: (1) when the angular frequency of wC , the transfer function reaches its maximum am-
N1
plitude g RL with a 20 dB/dec attenuation slope; (2) The smaller Q is, transfer function is in a
N2
wider pass-band, better phase - frequency characteristics. This is a great validation Eq. (11) to (14).
In short, a HFCT transfer function is a first-order band pass filter. The two main parameters that
affect its performance is the center frequency wC and quality factor Q. We can widen the HFCT
R
pass band by reducing Q to improve performance. We know Q = L g N CL , so we can reduce Q by:
N2
R
(1) reducing N, (2)reducing C L , (3)reducing L 。
N2

Fig.1 the relationship between the amplitude-frequency and phase-frequency characteristics and
Q of H ( jw )
1) Reducing N
Toroidal cores without air gap reluctance dimensions in Figure 1 is

(15)
Where m0 the permeability of air is, m is the relative permeability of the core material. As can be
seen, if there is no air gap reluctance annular core N ' is inversely proportional to its relative per-
meability μ. The μ values of magnetic materials are nonlinear. They change because of changes in
magnetic flux density. It is very possible that this core is made so that the third harmonic HFCTs into
the test results even penetrate the core. This is unacceptable HFCTs and must take measures to curb
this phenomenon.
Typically, the gap is cut off to improve the linearity of the core. Having an air gap in size as
shown in Fig. 1 toroidal core, whose length is δ Magnetic field analysis is shown in Fig.4. Accord-
ing to Maxwell's equations, you can understand, the magnetic flux density B(r) of the radius r is the
same gap in the air. The relationship between B(r) and MMF F meets:

249
 (16)
core reluctance N is:

(17)
p D2
By comparison, we find that the Eq. (17) in the general form of Eq. (15). When d ? ,
( m - 1)
Eq. (17) can be:

(18)
The s is a cross-sectional area of the core. At this point, the permeability μ, nonlinear may intro-
duce been eliminated from the equation. (18).

Fig.2 Core magnetic field analysis

From (17), the core reluctance N increases with the air gap width δ . The smaller δ and N, the
p D2
better HFCT is performance. However, considering linearity, the gap length d ? has been
( m - 1)
selected to suppress non-linearity.

Fig.3 magnetic saturation curve

N reduce the problems caused by the increase in δ can be obtained by increasing the core cross-
sectional area s to compensate. Increasing s is also to prevent magnetic saturation. Figure 5 is a
graph of magnetic saturation of various ferromagnetic material, can be seen, there is saturation mag-
netic induction BS of variety of ferromagnetic materials. Magnetic induction hardly increases after
the core reaches the BS . While the partial discharge signal is typically low, current signal does not
cause HFCT saturation, there is all some of the leakage capacitance of the electrical equipment,

250
when it works at its rated voltage, a number of power-frequency current to be introduced to the ob-
ject being measured. It can cause very easy to be saturated, if not handled properly.
For example, the measured maximum power frequency component of the current is , the
magnetic induction in the core is B, following relationships are met:

(19)
A general magnetic material satisfies B < BS / 2 , when working, whereby the core cross-
sectional area constraint can be obtained:

(20)
2) Reducing C L
C L is HFCT coil drain capacitance. It mainly consists of two parts - the coil and the output ter-
minal, and its coil wire lead leakage capacitance between the leakage capacitance.
C L impendence becomes small when the signal frequency is high, thus affecting HFCT response
sensitivity. Through theoretical analysis, we can know the parasitic capacitance of the coil is propor-
tional to the length of the coil. Thus, in the production of coils, minimizing the length of the wire be-
tween the output terminal, and the core may be designed to have a square cross-section, so as to
minimize the coil retaining a constant cross-sectional area of the length.
RL
3) Reducing
N2
Partial discharge monitoring of high voltage equipment insulation condition is detected. Gener-
ally, HFCT attached to the device under test. In power cables, for example, the wavelength by the
partial discharge occurs in the body of the discharge signal is usually about 30 meters. It is the same
length as the magnitude of the coaxial cable. Thus coaxial cable distribution parameters and the fol-
lowing circuit must be considered. So integrating resistor RL must match the coaxial cable (usually
50Ω) of the characteristic impedance. Only in this way, the "coaxial cable + integral resistance"
structure can be equivalent to a lumped resistance RL

Fig.4 HFCT Frequency Characteristic Diagram

And it can increase N2 to reduce Q. But we also need to note: the HFCT transfer function in
N
the amplitude of the center frequency of 1 g RL . N 2 increasing will reduce the sensitivity of the re-
N2
action, which is on-line monitoring system for partial discharge is very important. A lot of interfer-
ence present in the operating environment of electrical equipment and partial discharge signal is very
subtle. If insufficient sensitivity, signal to noise ratio of the system (SNR), will be significantly re-
duced. Therefore, the discharge signal or even is buried in a complex electromagnetic environment.

251
 Although the N1 increase may make up for the increase N 2 defects to a certain extent, this
also increases the HFCT power frequency components. This will inevitably lead to the core in a
higher operating point, making it more possible to be more saturated (see Eq. (20)).

A B C

Portable XLPE
Cable Partial
35kV Cable
Discharge
Terminal
Detection System

Fig.5 Field Test Schematic

T Therefore, adjusting the number of turns designed to improve the performance of only the accu-
rate calculation (core saturation) weighed (sensitivity and the pass-band) and after take.

EXPERIMENT DATA
After the theoretical calculation, a HFCT (see Figure 6 is a frequency characteristic) of 35kV and be-
low XLPE cable line partial discharge detection is designed. It is tested to 35kV XLPE cable termi-
nal substation partial discharge conditions. UHF sensor is used to help verify the test results.

(A) Multi-Period Screenshot (B) PD pulse waveform

Fig.6 Test Results
Figure 7 shows a schematic view of the cable terminal substation test: place near the end of the
cable, and the three HFCTs to the body on the cable terminals A, B and C of the ultra-high frequency
sensor. 4 sensor output signals is connected to CH1 "portable cable partial discharge detection sys-
tem", respectively, after installation, the output signal of each sensor of methane observed on "detec-
tion system" screen.

Discharge Point
in PhaseA
Failure

Fig.7 discharge source

Figure 8 is the test results. From top to bottom is the ultra-high frequency signals and high-
frequency current signal cable phase A, B and C of the body. From (a), and can be seen in HFCT
UHF sensor and detected phase A significant discharge pulse has a significant cyclical characteristics.
From (b), phase A HFCT phase discharge measurement signal is approximately 5MHz; in stage B
and C simultaneously detect signals of the same frequency of the discharge, but small, and π ampli-
tude ratio A phase retardation.

252
 It is comprehensive analysis of this issue. It is believed that there are in some of the cable end of
the discharge and the main discharge power is in phase A. The cable holder was opened. Significant
fault discharge traces was found on the cable phase A (Fig. 9) body.

CONCLUSION
In this article, we have HFCTs modeling and analysis. The following conclusions:
1) Reduce the core reluctance N expand HFCT pass-band. Toroids reluctant, this core produc-
tion HFCTs is serious nonlinear. We can reduce HFCT nonlinearity by cutting an air gap of width
d to satisfy d ? p D2 / ( m - 1) .
2) A gapped core's reluctance
N is proportional to the gap width d and inversely proportional to the cross−sectional area
S. Small d can cause HFCT saturation. Then, we should increase S to suppress such saturation.
3) The HFCT pass band can also be improved by reducing the leakage capacitance C L of the
coil. A square cross-section is designed to minimize the length of the coil, reducing C L
4) Integral resistor must match the characteristic impedance of the coaxial cable, to prevent re-
flection HFCT output signal, which can interfere with the measurement.
Finally, the sensor is a testament to experimental verification, in partial discharge monitoring
good performance.

ACKNOWLEDGMENT
This work was supported by the Guangdong Power Grid Corporation (K-GD2013-047), The re-
search project of field acceptance and operation maintenance’s tests and detection methods of medi-
um voltage cable.

REFERENCES
[1] DAI Wei, TANG Xiu-song, High-frequency Pulse Current Method Applied in Partial Discharge
of Online Detection of Power Transformer [J]. Automation Application, 2014, (12).
[2] DU You-wei. The Development of Magnetic Materials[J]. physics, 2000, 29(6):323-332.
[3] CHEN Teng-biao, WU Tao, WEI Qian-hu, et al. Application of High Frequency Pulse Current
Method in Detecting and Positioning for High Pressure Cable Electrification Partial Discharge[J].
Guangdong Electric Power, 2014, (1):114-119.
[4] GU Jiang-guo, WANG Shi-ping, SUN Xiao-qian. Property and Development of Normal Dielec-
tric Insulating Material [J]. Chemical Engineer, 2002, (4):54-55.

253