This document is published in:
Robles, G.; Martinez-Tarifa, J.M.; Rojas-Moreno, M.V.; Albarracin, R.; Ardila-Rey, J.; , "Antenna
selection and frequency response study for UHF detection of partial discharges,"
Instrumentation and Measurement Technology Conference (I2MTC), 2012 IEEE International ,
vol., no., pp.1496-1499, 13-16 May 2012. DOI: 10.1109/I2MTC.2012.6229440
© 2012 IEEE. Personal use of this material is permitted. Permission from
IEEE must be obtained for all other uses, in any current or future media,
including reprinting/republishing this material for advertising or
promotional purposes, creating new collective works, for resale or
redistribution to servers or lists, or reuse of any copyrighted component of
this work in other works.
This research has been supported by the Madrid Region
al Government
and Universidad Carlos III de Madrid under Contract No. CCG10-
UC3M/DPI-4627.
Tests have been carried out in the High Voltage Research
and Tests Laboratory at Universidad Carlos III de Madrid (LINEALT).
Antenna selection and frequency response study for
UHF detection of partial discharges
Robles, G.; Martínez-Tarifa, J.M.; Rojas-Moreno, M.V.; Albarracín, R.; Ardila-Rey, J.
Department of Electrical Engineering
Universidad Carlos III de Madrid
Leganés, Spain
grobles@ing.uc3m.es
Abstract—Partial Discharge (PD) detection is a widely extended
technique for electrical insulation diagnosis. Classical PD
detection by means of phase resolved patterns require electrical
connections to the power equipment and is sensitive to many
noise sources. Ultra High Frequency (UHF) detection techniques
are being recently proposed to overcome these problems, and to
detect partial discharges on-line. In this paper, four antennas will
be tested in order to compare their response to this physical
phenomenon.
Keywords- partial discharges; UHF detection; antennas
response.
I. INTRODUCTION
Partial Discharge (PD) is a clear ageing agent in electrical
insulation of power systems. Power cables, transformers and
generators withstand PD even at rated voltages due to mixed
ageing agents arising from thermal, mechanical, electrical and
environmental stresses [1]. On the other hand, these
microscopic ionizations lead to small signal current pulses that
can be detected in electrical equipment. Thus, PD is also a
symptom of high voltage electrical apparatus ageing.
Normalized PDs measurements are made using resistive-
capacitive dividers [2]. These classical methods use Phase-
Resolved PD (PRPD) patterns to identify certain PD sources
(corona, internal and surface PD). However, the measurement
procedure is always done in industrial environments, where
low Signal-to-Noise Ratio (SNR) signals are obtained or
several microscopic sites are usually discharging
simultaneously. This is the reason to add new measurement
techniques where PD pulse waveform analysis is a fruitful
methodology for noise rejection and PD source separation [3].
In this case, Very High Frequency (VHF) detectors, such as
High Frequency Current Transformers (HFCT), are usually
selected.
However, these methods have drawbacks: power equipment
disconnection is sometimes necessary to adjust the
measurement and detection of discharges and PD geometric
location is not possible with these systems. In order to solve
this, acoustic and UHF PD detection techniques are recently
being applied to these systems, since these are non-contact
measurements that can help in on-line PD monitoring and
location [4]-[7]. Unfortunately, acoustic detection is restricted
to oil-paper insulation systems and inner PD sites can be hardly
detected because acoustic waves only propagate through oil.
Despite the fact that UHF detection seems to have clear
advantages over other techniques, it is not fully understood the
relationship between conventional HF or VHF signals and
UHF signals [8]. Moreover, UHF signals depend on the
selected antenna bandwidth and gain. Hence, the selection of
the proper antenna for PD detection and location in high
voltage assets is currently a clear technical challenge.
In this paper, a study about UHF signals from controlled
insulating test objects withstanding PD is proposed. Four
different antennas: zig-zag, monopoles with two lengths and
log-periodic, will be used to measure PD pulses and their
responses will be compared in the frequency domain.
Discussion will be made taking into account technical and
economic characteristics for each device.
II. MEASURING SETUP
The measuring setup consists of two different parts: partial
discharge generation and sensor deployment around the test
object.
A. Partial Discharge generation
A constant and predictable partial discharges activity is
necessary to ensure the repeatability of results. To achieve this,
a controlled experiment is carried out in the laboratory with a
test object consisting of a vessel filled with transformer oil and
two electrodes separated by 5 sheets of transformer insulating
paper, see Fig. 1.
Figure 1. Test object
1
1
2
3
4
5
6
1
2
3
45
6
Monopole 10 cm
Monopole 5 cm
Zig-zag
Log-periodic
Coupling capacitor
Test object
Figure 2. Measuring setup with four antennas. The test object and the
coupling capacitor are also visible
An electrode is connected to a high voltage source and the
other to ground. According to Standard IEC 60270, a coupling
capacitor is connected in parallel to the test object to provide a
path to ground for high frequency current pulses created by
partial discharges in the test object, see Fig. 2. Pulses are
measured in VHF with a HFCT with a bandwidth up to 40
MHz connected to a commercial PDs detector (PD-Check from
TechImp Systems S.r.l.) capable to identify PRPD patterns.
This test setup in VHF is used to confirm that the detected
UHF pulses are a consequence of PDs activity.
It has been found, that using new and dry transformer paper
sheets, the partial discharges activity (internal PDs in
microscopic air voids between papers) starts around 2 kV and
is stable during at least three hours which is enough to acquire
proper signals in VHF with the HFCT and in UHF with the
antennas. Hence, the high voltage source is slowly set slightly
above the inception voltage and the measuring campaign starts.
Pulses were acquired at 3600 V.
B. Antennas deployment
Four antennas are used in the experiments with different
frequency ranges: UHALP 91088A log-periodic from 250 to
2400 MHz, two monopoles, 5 and 10 cm long, which ensures
a wide frequency range [7], and a zig-zag antenna, see Fig. 2
and Fig. 3. The antennas are deployed around the test object
a
nd their outputs connected to an oscilloscope. The distances
between the test object and the antennas are not critical
parameters at this stage because the experiment is focused on
studying the frequency response of the signals and not the
pulses in the time domain; nevertheless, distances were of a
similar range, as can be seen in Fig. 2. Since PD source
location is not the focus of this paper, the lengths of the
coaxial cables are random too.
UHF acquisitions were made in a Tektronix DPO7254 8-
bit, 40 GS/s, 4 channel oscilloscope, where the response of
each antenna to PD pulses was registered. PDs activity is a
stochastic phenomenon that depends on several factors such as
applied voltage level, insulation ageing status, environmental
conditions, etc. Despite the fact that, during the experiments in
laboratory, most factors were controlled to assure uniformity in
Figure 3. Zig-zag, 10 cm monopole and 5 cm antennas
the measurements, series of 500 pulses were recorded and
processed to guarantee that the results were statistically
reliable.
III. SIGNAL ACQUISITION AND PROCESSING
Prior to starting the acquisition of partial discharge pulses,
the background noise is characterized to be compared with the
FFT in the presence of pulses. This is done by measuring the
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5
x 10
9
10
-5
10
-4
10
-3
10
-2
Frequency [Hz]
[V]
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5
x 10
9
10
-5
10
-4
10
-3
10
-2
Frequency [Hz]
[V]
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5
x 10
9
10
-5
10
-4
10
-3
10
-2
Frequency [Hz]
[V]
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5
x 10
9
10
-5
10
-4
10
-3
10
-2
Frequency [Hz]
[V]
Figure 4. FFT voltage amplitude. Top: 5 cm monopole antenna; Second one:
10 cm monopole antenna. Third one: zig-zag antenna.; Bottom: log-periodic
antenna
2
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 10
-7
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
Time [s]
Amplitude [V]
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 10
-7
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
Time [s]
Amplitude [V]
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 10
-7
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
Time [s]
Amplitude [V]
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 10
-7
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
Time [s]
Amplitude [V]
Figure 5. Voltage amplitud versus time for PD at 3600 V. Top: 5 cm
monopole antenna; Second one: 10 cm monopole antenna. Third one: zig-zag
antenna.; Bottom: log-periodic antenna
amplitude in absence of applied high voltages and shown in
Fig. 4, where the FM radio, Digital Audio Broadcasting
(DAB), TV broadcast, GSM and WiFi signals are clearly
visible. The vertical axis for the four antennas is set to a
logarithmic scale in volts. The noise floor is around 6·10
-5
V,
5.8·10
-5
V, 6·10
-5
V and 7·10
-5
V for the 5 cm monopole, 10
cm monopole, zig-zag and log-periodic antennas, respectively.
Using the FM band to compare the response of the antennas to
external radiation, it can be observed that the peaks are located
in 2·10
-4
V, 4·10
-4
V, 6·10
-4
V and 5·10
-3
V for the 5 cm
monopole, 10 cm monopole, zig-zag and log-periodic antennas,
respectively. This means that, considering wide band
behaviour, the log-periodic antenna has better sensitivity than
the others. The horizontal axis is the frequency in Hz with a
scale of 250 MHz/div, from 0 to 2.5 GHz. This plot is done by
averaging the FFT of 500 time signals acquired with the
oscilloscope.
Once this noise is characterized, the voltage is raised up to
the inception voltage and the pulses are synchronized with the
trigger set to channel one in the oscilloscope where the 10 cm
monopole is connected. An example of the time domain signals
acquired is shown in Fig. 5 for pulses at 3600 V. The sampling
frequency is 10 GS/s, the acquisition time is 200 ns and the
voltage peaks for the three signals are below 60 mV. Notice
TABLE I. CUMULATIVE POWER BY BANDS
Antenna type
Cumulative Power by Bands
(V
2
)
0 V 3.6 kV Ratio
5 cm monopole
0 – 600 MHz
6.53·10
-7
1.98·10
-6
3.0
5 cm monopole
1300 – 1900 MHz
2.51·10
-7
3.661·10
-6
14.6
10 cm monopole
0 – 600 MHz
3.62·10
-7
1.034·10
-5
28.6
10 cm Monopole
1300 – 1900 MHz
3.20·10
-7
1.82·10
-6
5.7
Zig-zag
0 – 600 MHz
1.30·10
-6
2.96·10
-5
22.8
Zig-zag
1300 – 1900 MHz
2.57·10
-7
1.90·10
-6
7.4
Log-periodic
0 – 600 MHz
4.22·10
-5
3.13·10
-4
7.4
Log-periodic
1300 – 1900 MHz
3.86·10
-7
1.87·10
-6
4.8
that the signal for the log-periodic antenna is detected before
the signals of the monopoles and zig-zag antennas. This is due
to the fact that, even when the monopoles and zig-zag are
closer to the test object, they have longer coaxial cables, (3 m
long for log-periodic antenna and 5 m long for the others),
connected to the oscilloscope.
The same FFT analysis is done for signals with pulses from
partial discharges and the result is shown in Fig. 6. It is a
remarkable fact that the amplitudes of the spectra in the low
frequency range of the FFT, up to 600 MHz, are noticeably
larger. However, the importance of these results lays on the
FFT content that appears in the range from 1300 to 1900 MHz
and that is directly related to the UHF emission of the partial
discharge pulse, compare Fig. 6 and Fig. 4. These peaks are
captured by the four antennas being the 5 cm monopole
antenna the best one to visualize them. Moreover, it is also
remarkable the performance of the 10 cm monopole
considering its simplicity and inexpensive manufacture
compared to the log-periodic antenna.
Table I. summarizes the behavior of the antennas showing
the cumulative power by bands, in order to make a comparison
between the frequency response of the antennas, in two ranges
of frequency: from 0 to 600 MHz and from 1300 to 1900 MHz.
The column on the right represents the ratio between the power
content in squared volts in those ranges with PD and without
PD. It can be clearly seen that the spectral power of the PD
pulses detected with the antennas is notably larger than in the
case of absence of PD, specially, in the lower frequency band
of study, from 0 to 600 MHz. The best frequency response in
this band is for the 10 cm monopole, and the zig-zag antennas,
reaching a ratio of 28.6, and 22.8 respectively. In the higher
frequency band, from 1300 to 1900 MHz, the cumulative
power increase due to PDs activity is not so high but is more
relevant because there is energy only when PD occurs
otherwise, this band is flat. In this band, the best behavior is
found for the 5 cm monopole with a remarkable ratio of 14.6.
The 10 cm monopole and the zig-zag antennas only reach
ratios of 5.7 and 7.4, respectively. The differences between 10
cm and 5 cm monopoles arise from their different sensitivity to
3
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5
x 10
9
10
-5
10
-4
10
-3
10
-2
Frequency [Hz]
[V]
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5
x 10
9
10
-5
10
-4
10
-3
10
-2
Frequency [Hz]
[V]
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5
x 10
9
10
-5
10
-4
10
-3
10
-2
Frequency [Hz]
[V]
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5
x 10
9
10
-4
10
-3
10
-2
10
-1
Frequency [Hz]
[V]
Figure 6. FFT voltage amplitude during the presence of PD at 3600 V. Top:
5 cm monopole antenna; Second one: 10 cm monopole antenna. Third one:
zig-zag antenna.; Bottom: log-periodic antenna
electromagnetic radiation for different wavelengths, so the 5
cm monopole must have a better response at higher
frequencies.
The poor performance of the log-periodic antenna can be
seen in Fig. 5 where the direct electromagnetic wave of the PD
had to be highlighted putting it into a box. These types of
antennas capture PD and signals without diagnosis interest in a
broadband of frequencies, and PD, generally with low emitting
powers, are hidden by the rest of signals.
For all these reasons, if measurements are taken in noisy
electromagnetic environments with high Radio Frequency (RF)
spectral amplitudes, the 5 cm monopole would be the best
option because it would detect PDs in the 1300-1900 MHz
band with good sensitivity.
This study opens a research trend to characterize the
behavior of monopole antennas with different lengths and
shapes. The frequency response for these inexpensive
monopoles shows a broadband behavior appropriate for UHF
emissions from insulation systems. Their significant
components above 1.3 GHz are an interesting characteristic
for PDs location, where the geometric sensitivity is
determined by the bandwidth of the antennas. Thus, several
inexpensive monopoles can be used for PDs sites location in
power systems.
IV. C
ONCLUSION
The 10 cm monopole and zig-zag antennas show an overall
better performance in the two frequency bands of study after
comparing the behaviour of the four antennas. However, the 5
cm monopole antenna is the best detecting PDs emitting in
UHF from 1300 to 1900 MHz band. Monopoles and zig-zag
antennas are strong candidates for further studies because
these antennas are inexpensive, simpler, smaller, easier to
manufacture and can be tuned to a frequency band of interest
by changing their lengths. The paper also gives practical
indications about PDs detection with different antennas.
R
EFERENCES
[1] G. Stone, E.A. Boutler, I. Culbert, H. Dhirani, “Electrical insulation for
rotating machines: design, evaluation, aging, testing and repair”. New
Jersey; IEEE Press Series on Power Engineering, Wiley Interscience;
2004;p. 295-307.
[2] IEC Document No 60270, “High voltage test techniques—Partial
discharge measurements”, 2000.
[3] A. Cavallini, G.C. Montanari, A. Contin, F. Pulletti, “A new approach to
the diagnosis of solid insulation systems based on PD signal inference,”
Electrical Insulation Magazine, IEEE , vol.19, pp.23-30, March-April
2003.
[4] J. Ramírez-Niño, A. Pascacio, “Acoustic measuring of partial discharges
in power transformers,” Measurement Science and Technology, 2009,
vol. 20, 115108, November 2009.
[5] S. Markalous, S. Tenbohlen, K. Feser, “Detection and location of partial
discharges in power transformers using acoustic and electromagnetic
signals,” Dielectrics and Electrical Insulation, IEEE Transactions on ,
vol.15, pp.1576-1583, December 2008.
[6] P.J. Moore, I. Portugues, I.A. Glover, “A nonintrusive partial discharge
measurement system based on RF technology,” Power Engineering
Society General Meeting, 2003, IEEE , vol.2, pp. 4 vol. 2666, 13-17 July
2003.
[7] J. López-Roldán, T. Tang, M. Gaskin, “Optimisation of a sensor for
onsite detection of partial discharges in power transformers by the UHF
method,” Dielectrics and Electrical Insulation, IEEE Transactions on ,
vol.15, pp.1634-1639, December 2008.
[8] A.J. Reid, M.D. Judd, R.A. Fouracre, B.G. Stewart, D.M. Hepburn,
“Simultaneous measurement of partial discharges using IEC60270 and
radio-frequency techniques,” Dielectrics and Electrical Insulation, IEEE
Transactions on , vol.18, pp.444-455, April 2011.
4
