sensors

Article
Low-Cost Online Partial Discharge Monitoring System for
Power Transformers
Wojciech Sikorski * and Artur Wielewski

Institute of Electric Power Engineering, Poznan University of Technology, 60-965 Poznan, Poland
* Correspondence: [Link]@[Link]; Tel.: +48-616-652-035

Abstract: The article presents in detail the construction of a low-cost, portable online PD monitoring
system based on the acoustic emission (AE) technique. A highly sensitive piezoelectric transducer was
used as the PD detector, whose frequency response characteristics were optimized to the frequency
of AE waves generated by discharges in oil–paper insulation. The popular and inexpensive Teensy
3.2 development board featuring a 32-bit MK20DX256 microcontroller with the ARM Cortex-M4 core
was used to count the AE pulses. The advantage of the system is its small dimensions and weight,
easy and quick installation on the transformer tank, storage of measurement data on a memory card,
battery power supply, and immediate readiness for operation without the need to configure. This
system may contribute to promoting the idea of short-term (several days or weeks) PD monitoring,
especially in developing countries where, with the dynamically growing demand for electricity, the
need for inexpensive transformer diagnostics systems is also increasing. Another area of application
is medium-power transformers (up to 100 MVA), where temporary PD monitoring using complex
measurement systems requiring additional infrastructure (e.g., control cabinet, cable ducts for power
supply, and data transmission) and qualified staff is economically unjustified.

Keywords: partial discharge (PD); online monitoring; low-cost system; power transformer; transformer
diagnostics; acoustic emission (AE); piezoelectric transducer; microcontroller; arduino; teensyduino

1. Introduction
Citation: Sikorski, W.; Wielewski, A.
The results of the international survey on substation transformer failures published
Low-Cost Online Partial Discharge
by CIGRE showed that, in most cases, they were caused by partial discharges [1,2]. Apart
Monitoring System for Power
from the relatively rarely committed serious manufacturing errors (e.g., non-degassed or
Transformers. Sensors 2023, 23, 3405.
[Link]
under-dried insulation system, the use of conductive structural elements such as screws
or nuts with sharp edges, metal particles left after assembly works, incorrectly designed
Academic Editors: Silvia Sangiovanni elements of the insulation system, etc.), partial discharges are initiated in these areas of
and Marco Laracca the insulating system that are highly moist and degraded as a result of aging processes
Received: 1 February 2023 (pyrolysis, hydrolysis, and oxidation reactions) [3–5]. Partial discharges can also be gener-
Revised: 17 March 2023 ated in the vicinity of deformed windings due to overvoltage or mechanical shock during
Accepted: 21 March 2023 transformer transportation. For this reason, the phenomenon of partial discharges is not
Published: 23 March 2023 only the cause of failure but can be, and in fact is, more and more commonly treated as a
reliable indicator of the condition of the transformer insulation system [6,7].
For PD detection in laboratory conditions, the conventional electrical method IEC
60270 [8] is used, while in the place of installation of the power transformer, the acoustic
Copyright: © 2023 by the authors.
emission (AE) or ultra-high frequency (UHF) method is preferable due to the resistance
Licensee MDPI, Basel, Switzerland.
to external electromagnetic interferences [9,10]. These unconventional techniques of PD
This article is an open access article
detection have been adapted to both periodic diagnostic tests and tests carried out in
distributed under the terms and
the online monitoring mode [11–13]. The main advantage of monitoring over periodic
conditions of the Creative Commons
diagnostics is the possibility of immediate detection of PD ignition or an increase in their
Attribution (CC BY) license (https://
[Link]/licenses/by/
intensity and, thus, allows operating services to prepare quickly and implement procedures
4.0/).
minimizing the risk of failure. Based on the data collected by the monitoring system, one can

Sensors 2023, 23, 3405. [Link] [Link]

Sensors 2023, 23, 3405 2 of 21

dynamically change the requirements for periodic diagnostic tests (e.g., their acceleration
or postponement) and decide if and when it is necessary to plan downtime and perform
maintenance. Some online PD monitoring systems usually allow data exchange with a
substation SCADA system. This enables the implementation of advanced inference rules
based on statistical or machine learning methods, which allow detection of the relationships
between the PD activity and such monitored parameters as voltage, load, oil temperature,
tap position of the OLTP, or the amount of hydrogen dissolved in the oil [14–16].
In addition to the undoubted benefits of using online partial discharge monitoring
systems, a serious obstacle to their dissemination is still a small number of specialized
manufacturers, which translates into low supply and the high price of the device. In addi-
tion, the price is increased by the high costs of research and development and production.
For this reason, only the largest transformers of key importance for the power system are
currently equipped with online partial discharge monitoring systems. Therefore, in recent
years, there has been an intensification of research work related to the minimization of the
costs of PD monitoring. Two main areas of activity can be distinguished in the conducted
research. The first one is related to the development of simple and low-cost partial dis-
charge sensors. Castro et al. discussed [17] the possibility of using piezoelectric membranes
(buzzers) to detect PD in power transformers based on the acoustic emission method. Piezo-
electric membranes are readily available and often cheaper than conventional AE sensors,
making their use particularly attractive in applications where several sensors are used.
Laboratory tests have shown that a low-cost piezoelectric membrane provides acceptable
partial discharge detection sensitivity and can be an alternative to expensive AE sensors.
Besharatifard et al. [18] investigated the possibility of replacing piezoelectric technology
with microfiber composites (MFCs). Compared to piezoelectric membranes, MFC sensors
are slightly more expensive but offer higher sensitivity up to 500 kHz and a longer pulse
duration, which can help separate partial discharges from on-site noise. For several years,
also in the case of electromagnetic PD detectors operating in the radio frequency bands
(HF, VHF, and UHF), research has been carried out to simplify their design and reduce pro-
duction costs [19–22]. An example of such designs are miniaturized UHF antennas, which
can be mass-produced using printed circuit board (PCB) technology [23]. According to the
literature on the subject, the most promising designs of UHF PD detectors, which can be
easily manufactured in PCB technology, are microstrip patch antennas [24–26], bio-inspired
antennas [27,28], meander-line antennas [29], logarithmic spiral antennas [30], Vivaldi
antennas [31], different types of Archimedean spiral antennas [32,33], meandered planar
inverted-F antennas, and fractal antennas such as Hilbert curve fractal antennas [34,35],
Peano fractal antennas [36], H-fractal antennas [37], or Minkowski fractal antennas [38]. In
addition to UHF antennas, the most commonly used low-cost electromagnetic PD detectors
are high-frequency current transformers (HFCT) and transient earth voltage (TEV) sensors.
Their advantage is not only simple construction but also high sensitivity [39,40].
The second area of research concerns the construction of partial discharge monitoring
systems based on generally available single-chip microcontrollers, simple USB data loggers,
or field programmable gate arrays (FPGA), which are an alternative to expensive, multi-
channel data acquisition cards. Saeed et al. [41] presented a supervisory system for PD
monitoring designed around the Microchip PIC24EP512GU810 microcontroller, which is
characterized by a high speed of 70 million instructions per second (MIPS). The device
also offers a satisfying analog-to-digital conversion performance of 1.1 million samples per
second with up to four simultaneous channels. This ensures that the fast PD pulses are
adequately sampled. Chakrabarty et al. [42] have developed an online partial discharge
counting system using a microcontroller and FPGA technology. The signal from the
PD detector is sampled at a frequency of 20 MHz by the analog-to-digital converter of
the PIC 16F877A microcontroller. The signal is then transmitted to the input of a Xilinx
ML405 evaluation board equipped with an FPGA chipset type Virtex-4 (XC4VFX20) that
has been programmed to detect and count PD pulses in real time. Chang et al. [43]
have developed a reconfigurable partial discharge monitoring system based on FPGA
Sensors 2023, 23, 3405 3 of 21

technology. The proposed system uses the AMR Cortex M4 microprocessor to control the
system and data analysis, a fast AD9226 analog-to-digital converter, and Xilinx XC6SLX16
field programmable gate arrays with SDRAM memory. The signal processing procedure
consists of two steps. In the first step, PD pulses are captured by various PD detectors,
quickly converted to digital data by AD9226, and then temporarily stored in SDRAM by
an FPGA chip. In the second step, the signals are taken from the SDRAM and transmitted
to the microprocessor by the FPGA for further data analysis. Yan et al. [44] presented the
PD monitoring and locating system for medium-voltage switchgears, which is based on
low-cost TEV detectors. In order to automatically locate the PD source and minimize the
number of expensive high-speed acquisition cards, the authors proposed a time-sharing
access mechanism, which was implemented by multiple high-frequency surface mounting
relays integrated into each TEV detector. Mohamed et al. [45] proposed the use of a
spectrum analyzer based on the SDR (software-defined radio) technology to build a cheap
and portable online PD monitoring system operating in the VHF/UHF frequency range. A
system that is capable of automatically collecting PD data consists of only two hardware
components, i.e., a PC/laptop and a portable software-defined radio receiver type Realtek
RTL2832U that connects to the computer via a USB interface.
This article presents in detail the construction of a low-cost, portable PD online mon-
itoring system based on the acoustic emission (AE) technique, which has a chance to fill
the gap in the area of inexpensive monitoring systems that can be effectively used in the
diagnosis of small and medium power transformers. Currently designed systems and
commercial systems available on the market are intended—mainly due to the high price
and complicated construction—to monitor large power transformers that play a strategic
role in the power system. To the best of the authors’ knowledge, this manuscript is the first
complete description of a portable online PD monitoring system that can be assembled
from commercially available electronics and development boards compatible with the
Arduino architecture. As a module for processing and analyzing AE signals, the Teensy
3.2 board was used. Despite its small size and low price (20USD), it ensures high effi-
ciency of detection and counting of AE pulses generated by partial discharges. Laboratory
tests have shown that the system is capable of lossless, real-time counting of up to over
83,000 AE pulses.
The developed device is the world’s first PD monitoring system equipped with a
piezoelectric acoustic emission sensor optimized for the detection of partial discharges in
oil–paper insulation. The frequency characteristics and resonant frequencies of the sensor
coincide with the frequencies of AE pulses generated by the most dangerous and destruc-
tive types of discharges for the transformer insulation system, i.e., inter-turn discharges,
creeping and surface discharges. As a result, the presented low-cost monitoring system
has a very high sensitivity of PD detection. What is also essential is that the system is in
line with the strategy of modern monitoring of power transformers formulated in the Cigre
TB 630 brochure published in 2015: Guide On Transformer Intelligent Condition Monitoring
(TICM) Systems, which assumes the replacement of traditional sensors and transducers
with Intelligent Electronic Devices (IED), which are equipped with a microprocessor and
are capable of processing raw PD pulses and automatically determining the parameters
describing them.
The developed system may contribute to the promotion of the idea of short-term
(several days or weeks) PD monitoring, especially in developing countries where, with
the dynamically growing demand for electricity, the demand for inexpensive transformer
diagnostics systems is also growing. Another area of application is medium power trans-
formers (up to 100 MVA), on which temporary PD monitoring using expensive, often
requiring additional infrastructure (e.g., control cabinet, cable ducts for power and data
transmission), and qualified staff would be economically unjustified.
This paper is organized as follows. The hardware and software layers of the system
are discussed in Section 2, while Section 3 presents the results of laboratory and field tests
that allowed us to assess the possibility of using the system for short-term PD monitoring
Sensors 2023, 23, 3405 4 of 21

in power transformers. General conclusions from the conducted research are included in
Section 4.

2. Materials and Methods

2.1. General Description
In designing the device, the following assumptions were made: low price and avail-
ability of all system components, easy assembly, high sensitivity of PD detection, resistance
to external electromagnetic and acoustic disturbances, registration of the PD data in real
time, collecting PD data and storing it on a portable, inexpensive storage medium, battery
or mains operation, small dimensions and weight, and immediate readiness to work.
The online PD monitoring system is based on the acoustic emission method, which,
along with electromagnetic HF/VHF/UHF methods, belongs to the group of unconven-
tional partial discharge detection techniques. The main advantages of the AE method
include relatively high sensitivity, which primarily depends on the position of the sensor
in relation to the PD source, galvanic separation of the tested object from the measure-
ment system, resistance to external electromagnetic interference, the ability to locate PD
using auscultation or TDOA (time difference of arrival) technique, and installation of the
measurement system does not require switching off the tested object. As shown in the
schematic diagram of the system (Figure 1), a contact piezoelectric transducer was used
to detect acoustic waves from PD. The electrical signals at the output of the piezoelectric
transducer usually have very small amplitude, from a few to tens of millivolts, therefore,
the monitoring system is equipped with a preamplifier circuit consisting of an instrumenta-
tion amplifier with adjustable gain (default gain is 40 dB), and a voltage follower. Another
element of the system is an active bandpass filter, whose task is to eliminate low-frequency
components of the acoustic background (e.g., oil pump and cooling fan noises, vibrations
caused by the magnetostrictive action of the transformer core, etc.) and high-frequency
electromagnetic interferences. The amplified and filtered signal is then processed by peak
detector and voltage comparator circuits. The use of both of these systems made it possible
to replace the expensive signal acquisition card with a simple microcontroller equipped
with an analog-to-digital converter. The operation of the monitoring system is controlled by
a program written in C/C++, which performs such functions as detection and counting
Sensors 2023, 23, x FOR PEER REVIEW 5 ofPD
21
pulses, saving measurement data on a memory card, and presenting current measurement
data on the LCD.

Figure
Figure 1.
1. Schematic
Schematic diagram
diagram of
of aa low-cost
low-cost online
online partial
partial discharge
discharge monitoring system.

2.2. Hardware
2.2. Hardware Layer
Layer
2.2.1. Partial Discharge Detector
2.2.1. Partial Discharge Detector
A contact piezoelectric transducer was used as a partial discharge detector and the
A contact piezoelectric transducer was used as a partial discharge detector and the
various design and production stages, described in detail in Reference [46]. The frequency
various design and production stages, described in detail in Reference [46]. The frequency
response of the transducer was determined on the basis of laboratory tests aimed at iden-
tifying the acoustic frequencies emitted by partial discharges in oil–paper insulation. The
test results showed that the most destructive forms of PD for the transformer insulation
system (inter-turn, surface, and creeping discharges) generate AE signals, the energy of
Sensors 2023, 23, 3405 5 of 21

response of the transducer was determined on the basis of laboratory tests aimed at
identifying the acoustic frequencies emitted by partial discharges in oil–paper insulation.
The test results showed that the most destructive forms of PD for the transformer insulation
system (inter-turn, surface, and creeping discharges) generate AE signals, the energy of
which is transferred in three bands: 20–45 kHz, 50–70 kHz, and 85–115 kHz. The dominant
frequencies in these bands are 40 kHz, 68 kHz, and 90 kHz, respectively. Since Barkhausen
noise from the transformer core can reach ultrasonic frequencies [47], the transducer has
been optimized to work in the other two higher frequency bands characteristic for partial
discharges, i.e., 50–70 kHz and 85–115 kHz. The optimal material and geometric properties
of the main transducer structures were selected using the Krimholtz–Leedom–Matthaei
(KLM) model. Based on the simulation results, two piezoelectric disks made of PZT-5A
(Navy Type II) ceramics with a diameter of 10.5 mm and a height of 18.8 mm and 25 mm,
respectively, were used to build the transducer. Piezoelectric elements with opposite
polarization directions were bonded to the matching layer using an electrically conductive
composite adhesive based on epoxy resin and silver. The matching layer is made of a
round plate with a diameter of 25 mm and a height of 1 mm, made of high-density alumina
(Figure 2a). The acoustic impedance of the matching layer is 37.9 MRayl, which ensures
an efficient transfer of acoustic wave energy from the mineral oil through the steel tank
of the power transformer to the piezoelectric elements of the transducer. Everything is
closed in a housing made of stainless steel. The transducer was placed centrally on the
front wall of the monitoring system housing, which was additionally equipped with four
height-adjustable magnetic holders (Figure 2b).
The selected geometric and material parameters, as well as a fully differential design,
allowed to obtain the desired properties of the transducer, i.e., a two-resonant (68 kHz
and 90 kHz) and wide (30–100 kHz) frequency response curve and high peak sensitivity
(−61.1 dB ref. V/µbar) (Figure 2c). The test results presented in reference [46] showed that
this transducer is characterized by high detection sensitivity of partial discharges generated
in paper-oil insulation. Compared to commonly used commercial AE sensors, the average
PD pulse amplitude recorded by the new transducer was a minimum of 5.2 dB higher and a
maximum of 19.8 dB higher. Figure 2d shows the AE waveforms recorded a needle partial
discharge in oil with an apparent charge of 82 pC and a surface discharge on a pressboard
sample in oil with an apparent charge of 387 pC.

2.2.2. Amplifier and Bandpass Filter

Figure 3 shows circuit diagrams and photographs of the amplifier and filter designed
for the PD monitoring system. The amplifier circuit is based on the AD8421BRZ instrumen-
tation amplifier from Analog Devices (Analog Devices, Norwood, MA, USA). The gain G is
regulated by the appropriate selection of the resistance RG value. In this case, a 100-ohm
resistor was used, giving a gain of 40 dB (G = 100). The main advantages of this circuit are
√
low current consumption (<2.3 mA), low noise level, which does not exceed 3.2 nV/ Hz,
ultra-low polarization current (<500 pA), wide bandwidth (2 MHz with gain G = 100) and
low price (around $10).
The active bandpass filter (20–500 kHz) is designed based on the Sallen–Key archi-
tecture, where the low-pass and high-pass sections have a fourth-order Butterworth filter
structure with unity gain. For this purpose, the Analog Devices ADA4898 voltage feedback
√
operational amplifier was used, which is characterized by ultralow noise (0.9 nV/ Hz),
ultralow distortion (−93 dBc at 500 kHz), and low supply current (8 mA). The filter is made
on a separate PCB and is plugged in using dedicated board-to-board connectors between
the instrumentation amplifier and the voltage follower. This solution allows the filter to be
easily disconnected or replaced with a filter with a different passband characteristic.
The designed low-cost PD monitoring system provides the possibility of transmitting
the amplified and filtered AE signal via a coaxial cable directly to an external measuring
device (oscilloscope, signal acquisition card) or to a substation SCADA system. Coaxial
cables have a relatively large capacity (50–100 pF/m) and, therefore, cannot be directly
 Sensors 2023, 23, 3405 6 of 21

Sensors 2023, 23, x FOR PEER REVIEW 6 of 21

connected to the output of the amplifier. This problem was solved by the use of a voltage
follower based on the AD810 op-amp, which can drive capacitive loads exceeding 1000 pF,
without parasitic oscillations.

(a) (b)

(c) (d)
Figure
Figure2. [Link]
Piezoelectrictransducer
transducerfor for detection
detection of AE signals
of AE signals generated
generatedbybypartial
partial discharges: (a)
discharges:
schematic diagram of the transducer construction; (b) photograph of the monitoring
(a) schematic diagram of the transducer construction; (b) photograph of the monitoring system system housing
with piezoelectric
housing transducer
with piezoelectric and magnetic
transducer holders;
and magnetic (c) frequency
holders; response
(c) frequency responsecharacteristics
characteristics of the
piezoelectric transducer;
of the piezoelectric (d) AE (d)
transducer; waveforms recorded
AE waveforms for a for
recorded PDafrom a needle
PD from in oil
a needle inwith an apparent
oil with an
charge of 82
apparent pC and
charge for
of 82 pCa and
surface
for a discharge on a pressboard
surface discharge sample
on a pressboard in oil
sample with
in oil anan
with apparent
apparentcharge
of charge
387 [Link] 387 pC.

2.2.3.
2.2.2. Peak Detector
Amplifier and VoltageFilter
and Bandpass Comparator
In order to reduce the required sampling frequency of the AE pulses, a passive peak
Figure 3 shows circuit diagrams and photographs of the amplifier and filter designed
detector (sometimes called an envelope detector) and a voltage comparator were used.
forThe
thepeak
PD monitoring system.
detector is made up ofThe amplifier
just circuit is based
three components: on the AD8421BRZ
fast-switching diode D typeinstru-
mentation
1N4148, capacitor C with capacitance 22 pF, and resistor R with resistance 1 MΩ MA,
amplifier from Analog Devices (Analog Devices, Norwood, (FigureUSA).
4a). The
gain G is regulated
Together, by the
these elements appropriate
form a half-waveselection of the
rectifier that resistance
charges RG value.
the capacitor C to In
thethis
peakcase, a
100-ohm
voltage resistor was used,
of the incoming AEgiving
burst. aAsgain
the of 40 dB (Gof= the
amplitude 100). Thesignal
input main increases,
advantages theof this
capacitor voltage is increased by a rectifier diode D. When the amplitude of
circuit are low current consumption (<2.3 mA), low noise level, which does not exceed 3.2 the input signal
decreases,
nV/√Hz, the capacitor
ultra-low is discharged
polarization through
current (<500apA),parallel
wide bleeder [Link]
resistor(2
bandwidth The discharge
with gain G =
rate of the capacitor depends
100) and low price (around $10). on the value of the time constant τ = RC. During the first time
constant, the capacitor discharges 63%, and after 5τ, it is almost completely discharged.
The active bandpass filter (20–500 kHz) is designed based on the Sallen–Key archi-
tecture, where the low-pass and high-pass sections have a fourth-order Butterworth filter
structure with unity gain. For this purpose, the Analog Devices ADA4898 voltage feed-
back operational amplifier was used, which is characterized by ultralow noise (0.9
nV/√Hz), ultralow distortion (−93 dBc at 500 kHz), and low supply current (8 mA). The
filter is made on a separate PCB and is plugged in using dedicated board-to-board con-
Sensors 2023, 23, x FOR PEER REVIEW 7 of 21
Sensors 2023, 23, 3405 7 of 21

Sensors 2023, 23, x FOR PEER REVIEW

(a) 8 of 21

discharge rate of the capacitor depends on the value of the time constant τ = RC. During
the first time constant, the capacitor discharges 63%, and after 5τ, it is almost completely
discharged.
A voltage comparator is a special type of operational amplifier with an unbalanced
input and high gain that compares the voltage value of the signal applied to the non-
inverting input with the reference voltage applied to the inverting input. The voltage
value at the comparator’s output results from the difference in voltages between its two
inputs. If the voltage at the non-inverting input is higher than at the inverting input, then
the output voltage is close to the positive pole of the supply. If the voltage at the non-
inverting input is lower than at the inverting input, then the output voltage is close to the
negative pole of the power supply. Thus, the comparator can be considered an elemen-
tary, one-bit analog-to-digital converter. The project uses a single-channel LM311N volt-
age comparator (Figure 4b), which is characterized by a very low current consumption
(up to 100 nA at an ambient temperature of 25 °C) and a fast response time (~200 ns). The
open collector (OC) comparator output is compatible with CMOS, TTL, and RTL-DTL in-
tegrated circuits and can switch voltages up to 50 V and currents up to 50 mA. The
LM311N is designed to operate (b) with a wide range of supply voltages, including ±15 V
power
Figure supplies for operational amplifiers and, asofin
thethis case,circuit
5 V for logic(a)circuits. The
Figure 3.
3. Electronic
Electroniccircuit
circuitdiagram
diagramand
andphotograph
photograph of printed
the printed board:
circuit board: amplifier, (b)
(a) amplifier,
voltage value
bandpass [Link] to the inverting input is set with the A20k rotary potentiometer.
(b) bandpass filter.

The designed low-cost PD monitoring system provides the possibility of transmitting

the amplified and filtered AE signal via a coaxial cable directly to an external measuring
device (oscilloscope, signal acquisition card) or to a substation SCADA system. Coaxial
cables have a relatively large capacity (50–100 pF/m) and, therefore, cannot be directly
connected to the output of the amplifier. This problem was solved by the use of a voltage
follower based on the AD810 op-amp, which can drive capacitive loads exceeding 1000
pF, without parasitic oscillations.

2.2.3. Peak Detector and Voltage Comparator

In order to reduce the required sampling frequency of the AE pulses, a passive peak
detector (sometimes called an envelope detector) and a voltage comparator were used.
The(a)peak detector is made up of just three components: (b) fast-switching diode D type
1N4148,
Figure capacitor C with capacitance 22 pF, and resistor R with resistance 1 MΩ (Figure
[Link]
Electroniccircuit
circuitdiagram:
diagram:(a)(a)Peak
Peakdetector;
detector;(b)
(b)Voltage
Voltagecomparator.
comparator.
4a). Together, these elements form a half-wave rectifier that charges the capacitor C to the
peakThe
Avoltage
voltageof comparator
the incoming
correctness isAE
of the peak burst. type
detector
a special As
andthe amplitude
voltage
of ofamplifier
the input
comparator
operational signalan
operation
with increases,
was tested the
unbalanced in
acapacitor
input andvoltage
measurement is increased
highset-up
gain shown
that inby
compares a rectifier
Figure Asdiode
[Link] [Link]
the value
AE When
of the the amplitude
source,
signalthe oftothe
Olympus
applied input
V101B
the non-
signal decreases,
ultrasonic
inverting probewith
input thethe
was capacitor
used, which
referenceis voltage
discharged
was through
acoustically
applied the ainverting
tocoupled parallel bleeder resistor
to a piezoelectric
input. The [Link]
The
transducer
voltage
with a USG gel. The ultrasonic probe was excited by rectangular pulses with a voltage of
10 to 500 mV and a duration of 1 μs to 3 μs, which were generated using a Keysight
DSOX2024A oscilloscope with a built-in waveform generator. The frequency with which
the pulses were generated varied in the range from 1 Hz to 1 kHz. The outputs of the
voltage follower, peak detector, and voltage comparator were connected with coaxial ca-
Sensors 2023, 23, 3405 8 of 21

at the comparator’s output results from the difference in voltages between its two inputs.
If the voltage at the non-inverting input is higher than at the inverting input, then the
output voltage is close to the positive pole of the supply. If the voltage at the non-inverting
input is lower than at the inverting input, then the output voltage is close to the negative
pole of the power supply. Thus, the comparator can be considered an elementary, one-bit
analog-to-digital converter. The project uses a single-channel LM311N voltage comparator
(Figure 4b), which is characterized by a very low current consumption (up to 100 nA at
an ambient temperature of 25 ◦ C) and a fast response time (~200 ns). The open collector
(OC) comparator output is compatible with CMOS, TTL, and RTL-DTL integrated circuits
and can switch voltages up to 50 V and currents up to 50 mA. The LM311N is designed
to operate with a wide range of supply voltages, including ±15 V power supplies for
operational amplifiers and, as in this case, 5 V for logic circuits. The voltage value given to
the inverting input is set with the A20k rotary potentiometer.
The correctness of the peak detector and voltage comparator operation was tested in
a measurement set-up shown in Figure 5a. As the AE signal source, the Olympus V101B
ultrasonic probe was used, which was acoustically coupled to a piezoelectric transducer
with a USG gel. The ultrasonic probe was excited by rectangular pulses with a voltage
of 10 to 500 mV and a duration of 1 µs to 3 µs, which were generated using a Keysight
DSOX2024A oscilloscope with a built-in waveform generator. The frequency with which
the pulses were generated varied in the range from 1 Hz to 1 kHz. The outputs of the
voltage follower, peak detector, and voltage comparator were connected with coaxial cables
to channels 1, 2, and 3 of the oscilloscope, respectively. Examples of registered waveforms
are shown in Figure 5b. In this case, the ultrasonic probe was excited by rectangular pulses
generated with a frequency of 1 kHz, with a duration of 3 µs, and an amplitude of 50 mV.

2.2.4. Single Board Microcontroller

The popular and inexpensive ($20) Teensy 3.2 development board featuring a 32-bit
MK20DX256 microcontroller with the ARM Cortex-M4 core was used to count the PD
pulses. The Teensy 3.2 module, despite its small dimensions (35 × 18 mm), has 34 digital
input/output lines tolerant of 5V voltage, of which 12 can be used as PWM outputs, as well
as one analog output. Between them, there are 21 high-resolution analog inputs (13 usable
bits), of which 16 lines are shared with the digital ones and one with the analog output line
(they cannot be used simultaneously). In addition, the board is equipped with 7 timers,
3 UART serial ports, SPI, I2C, I2S, CAN Bus, RTC module, 16 DMA channels, and touch
sensor inputs.
The output signal from the voltage comparator is fed directly to Pin 13 of the micro-
controller. If the amplitude of the signal at the input of the comparator is higher than
the set trigger level, then its output returns a logic “high” state (~3.3 V). Otherwise, a
logic “low” state (~0 V) is sent to Pin 13. The microcontroller continuously counts the PD
pulses and records their current number every minute on the microSD memory card (Open
Smart Technologies Limited, Hong Kong, China) and displays it on the 8 × 2 character
monochrome LCD display type WC0802C (Hubbell Wiegmann, Freeburg, IL, USA). The
schematic diagram of connecting the LCD driver and the microSD card adapter to the
Teensy 3.2 board is shown in Figure 6.

2.2.5. Power Supply

The monitoring system is powered by a package of four series-connected lithium-
ion cells, type US18650VTC5A (Murata Manufacturing Co., Nagaokakyo, Japan), with a
nominal voltage of 3.6 V and a capacity of 2600 mAh, which cooperate with a dedicated
module of the BMS 4S battery management system (Figure 7a). The module has the function
of charging cells with the option of a balancer, discharging, and a function protecting the
cells against excessive discharge. The cells can be charged with a continuous current of
up to 10 A. Since all digital modules (microcontroller, LCD driver, memory card adapter)
require a voltage of 5 V, behind the battery management system, there is a step-down pulse
Sensors 2023, 23, 3405 9 of 21

converter module whose output voltage value is set with a potentiometer (from 1.0 V to
17 V). In turn, the instrumentation amplifier, voltage follower, and operational amplifiers
used to build an active bandpass filter require a symmetrical voltage of 15 V. For this reason,
Sensors 2023, 23, x FOR PEER the
REVIEW
power module is additionally equipped with a voltage converter type ICL7662CPA+
(Figure 7b). This converter is a monolithic charge pump voltage inverter that converts a
positive voltage in the range of +4.5 V to +20 V to a corresponding negative voltage of
−4.5 V to −20 V.

(a)

(b)
Figure Figure 5. Testing
5. Testing of individual
of individual electronicelectronic
modules ofmodules of the PD system;
the PD monitoring monitoring system; di-
(a) schematic (a) schema
agram ofgram of the measurement
the measurement system: TRA—wideband
system: TRA—wideband piezoelectricpiezoelectric transducer
transducer Olympus V101BOlympus
for V10
generating
generating AE pulsesAE(transmitter),
pulses (transmitter), REC—piezoelectric
REC—piezoelectric transducer oftransducer of thesystem
the monitoring monitoring
(re- syste
ceiver),ceiver), DUT—system
DUT—system under test,under test, OSC—oscilloscope
OSC—oscilloscope KeysightGEN—signal
Keysight DSOX2024A, DSOX2024A, GEN—signal g
generator,
tor, AES—signal
AES—signal recorded
recorded at the output of atthe
theamplifier,
output of the amplifier,
DET—signal DET—signal
recorded recorded
at the output at the outpu
of the peak
peak
detector; detector; recorded
COM—signal COM—signal recorded
at the output at voltage
of the the output of the voltage
comparator; comparator;
(b) waveforms recorded(b)atwaveform
orded
the output at the
of the outputpeak
amplifier, of the amplifier,
detector, peak detector,
and voltage andrespectively.
comparator, voltage comparator, respectively.

2.2.4. Single Board Microcontroller

The popular and inexpensive ($20) Teensy 3.2 development board featuring a
MK20DX256 microcontroller with the ARM Cortex-M4 core was used to count t
pulses. The Teensy 3.2 module, despite its small dimensions (35 × 18 mm), has 34
input/output lines tolerant of 5V voltage, of which 12 can be used as PWM outp
 set trigger level, then its output returns a logic “high” state (~3.3 V). Otherwise, a logic
“low” state (~0 V) is sent to Pin 13. The microcontroller continuously counts the PD pulses
and records their current number every minute on the microSD memory card (Open
Smart Technologies Limited, Hong Kong, China) and displays it on the 8 × 2 character
monochrome LCD display type WC0802C (Hubbell Wiegmann, Freeburg, IL, USA). The
Sensors 2023, 23, 3405 schematic diagram of connecting the LCD driver and the microSD card adapter 10 to ofthe
21
Teensy 3.2 board is shown in Figure 6.

Sensors 2023, 23, x FOR PEER REVIEW 11 of 21

Figure 6. The
The schematic
schematic diagram
diagram of
of connecting
connecting the
the microSD
microSD card adapter and the 14-pin LCD driver
to the Teensy 3.2 board.
to the Teensy 3.2 board.

2.2.5. Power Supply

The monitoring system is powered by a package of four series-connected lithium-ion
cells, type US18650VTC5A (Murata Manufacturing Co., Nagaokakyo, Japan), with a nom-
inal voltage of 3.6 V and a capacity of 2600 mAh, which cooperate with a dedicated mod-
ule of the BMS 4S battery management system (Figure 7a). The module has the function
of charging cells with the option of a balancer, discharging, and a function protecting the
cells against excessive discharge. The cells can be charged with a continuous current of up
to 10 A. Since all digital modules (microcontroller, LCD driver, memory card adapter)
require a voltage of 5 V, behind the battery management system, there is a step-down
pulse converter module whose output voltage value is set with a potentiometer (from 1.0
(a) V to 17 V). In turn, the instrumentation amplifier, (b) voltage follower, and operational am-
plifiers used to build an active bandpass filter require a symmetrical voltage of 15 V. For
[Link]
Figure Components ofof
thethe power
power supply
supply system:
system: (a) battery
(a)equipped
battery management
management systemsystem
BMSBMS 4S with
4S with the
this reason, the power module is additionally with a voltage converter type
the package of four series-connected lithium-ion cells US18650VTC5A; (b) electronic circuit diagram
package of
ICL7662CPA+ four series-connected lithium-ion cells US18650VTC5A; (b) electronic
(Figure 7b). This converter is a monolithic charge pump voltage inverter circuit diagram of
of the voltage converter ICL7662CPA+.
the voltage
that converts converter ICL7662CPA+.
a positive voltage in the range of +4.5 V to +20 V to a corresponding negative
voltage of −4.5
[Link] V to
Assemblyofofthe −20 V.
theSystem
System
2.2.6.
Allelectronic
All electronic modules
modules of of
thethe monitoring
monitoring system
system together
together withwith
the the power
power supply
supply systemsys-
are placed in a waterproof (IP67) aluminum housing with dimensions of 120 × 120 × 75 mm.×
tem are placed in a waterproof (IP67) aluminum housing with dimensions of 120 × 120
75 the
On [Link]
On the front
panel ofpanel of the system,
the system, there isthere
an LCD is anscreen
LCD screen
and a andswitcha switch for resetting
for resetting the
the pulse
pulse counter.
counter. The front
The front panelpanel elements
elements are protected
are protected by a transparent
by a transparent inspection
inspection windowwin-
dow of
made made of polycarbonate.
polycarbonate. In addition,
In addition, the housing
the housing is equipped
is equipped with a with a DC power
DC power connectorcon-
nector through
through which the which the charger
charger is connected,
is connected, a microaSD micro
cardSD card
slot, slot, a grounding
a grounding screw,
screw, and a
and connector
BNC a BNC connector for optional
for optional wired AE wired AEtransmission
signal signal transmission
(Figure (Figure
8). 8).

2.3. Firmware
Each Teensy 3.2 module has a bootloader uploaded, thanks to which it can be pro-
grammed using the built-in USB connector (no external programmer required). Programs
can be written in any environment supporting the C language or—as in the case of the
discussed monitoring system—in the Arduino IDE programming environment with the
Teensyduino extension installed. The computer program performs three main functions,
i.e., counting PD pulses, displaying the number of pulses on the LCD display, and storing
data (time, date, and number of pulses) on the memory card. The complete source code
with comments can be found in the Appendix A.
 tem are placed in a waterproof (IP67) aluminum housing with dimensions of 120 × 120 ×
75 mm. On the front panel of the system, there is an LCD screen and a switch for resetting
the pulse counter. The front panel elements are protected by a transparent inspection win-
dow made of polycarbonate. In addition, the housing is equipped with a DC power con-
Sensors 2023, 23, 3405 nector through which the charger is connected, a micro SD card slot, a grounding screw,
11 of 21
and a BNC connector for optional wired AE signal transmission (Figure 8).

Sensors 2023, 23, x FOR PEER REVIEW 12 of 21

can be written in any environment supporting the C language or—as in the case of the
discussed monitoring system—in the Arduino IDE programming environment with the
Figure 8.
Figure Photoofofthe
8. Photo the system
system assembly:
assembly: 1—aluminum electronic enclosure, 2—power switch,
Teensyduino extension installed. The 1—aluminum electronic
computer program enclosure,
performs 2—power
three switch,
main functions, 3—
3—polycarbonate
polycarbonate inspection
inspection window,
window, 4—LCD4—LCD display,
display, 5—reset
5—reset switch,
switch, 6—handles,
6—handles, 7—BNC 7—BNC con-
connector
i.e., counting PD pulses, displaying the number of pulses on the LCD display, and storing
nector (bandpass
(bandpass filterfor
filter output) output) forAE
optional optional AE signal transmission,
signal transmission, 8—DC power8—DC power
connector, connector,
9—piezoelec-
data (time, date, and number of pulses) on the memory card. The complete source code
9—piezoelectric
tric transducer, 10—magnetic
transducer, 10—magnetic holders,SD
holders, 11—micro 11—micro
card slot,SD card slot, 12—memory
12—memory card
card adapter, adapter,
13—LCD
with comments can be found in the Appendix A.
driver board, 14—Teensy 3.2 development board, 15—amplifier board, 16—bandpass filter board,
13—LCD driver board, 14—Teensy 3.2 development board, 15—amplifier board, 16—bandpass filter
17—peak detectordetector
board, 17—peak and voltage comparator
and voltage board, 18—grounding
comparator screw. screw.
board, 18—grounding
3. System Testing
3. System
2.3. The test
FirmwareTesting
of the partial discharge monitoring system was carried out in three stages.
In theEach
firsttest
The stage,
Teensy the
of the efficiency
3.2partial
module of athe
discharge
has AE pulseuploaded,
monitoring
bootloader counting function
system was towas
carried
thanks checked
out
which canusing
init three a
bestages.
pro-
signal
grammedgenerator
In the firstusing
stage, and
thethea piezoelectric
efficiency
built-in USB of transmitter.
the AE pulse
connector The second
counting
(no external stage of the
function was
programmer test
checked
required). was carried
using a
Programs
out in controlled
signal generatorlaboratory conditions,
and a piezoelectric during which
transmitter. partial discharges
The second stage of thegenerated in the
test was carried
out in controlled
transformer laboratory
tank model wereconditions,
[Link] which
The third partial
stage discharges
of the test wasgenerated
carried out in the
in
transformer
field conditions,tankand
model wereobject
the test monitored. TheMVA
was a 10 thirdpower
stage of the test was carried out in field
transformer.
conditions, and the test object was a 10 MVA power transformer.
3.1. Testing the Efficiency of AE Pulse Counting
3.1. Testing the Efficiency of AE Pulse Counting
To assess the efficiency of counting AE pulses, a measurement system was used con-
sistingToof assess the efficiency
an oscilloscope of counting
Keysight AE pulses,
with a built-in a measurement
signal generator, system
a reference wascoun-
pulse used
consisting of an oscilloscope Keysight with a built-in signal generator,
ter, and a wideband piezoelectric transducer Olympus V101B, which was excited with a reference pulse
counter, andpulses
rectangular a wideband
with a piezoelectric
duration of 1transducer Olympus V101B,
us and an amplitude of 100 which was excited
mV (Figure 9). with
rectangular pulses with a duration of 1 us and an amplitude of 100 mV (Figure 9).

Schematic
[Link]
Figure diagram
diagram of of
thethe measurement
measurement system
system for testing
for testing the efficiency
the efficiency of counting
of counting AE
AE pulses by the monitoring system: TRA—wideband piezoelectric transducer
pulses by the monitoring system: TRA—wideband piezoelectric transducer Olympus V101B Olympus V101B
for
for generating
generating AE pulses
AE pulses (transmitter),
(transmitter), REC—piezoelectric
REC—piezoelectric transducer
transducer of theofmonitoring
the monitoring
systemsystem
(re-
ceiver), DUT—monitoring
(receiver), DUT—monitoring system (device
system under
(device test),
under OSC—oscilloscope
test), OSC—oscilloscopeKeysight
KeysightDSOX2024A,
DSOX2024A,
GEN—signal
GEN—signalgenerator,
generator,CNT—universal
CNT—universalcounter
counterAgilent/HP
Agilent/HP53132A.
53132A.

The measurement results listed in Table 1 show that the system is characterized by
high efficiency and is capable of lossless counting of up to 83,400 AE pulses per minute.
Above this value, the counting accuracy decreases.

Table 1. The results of the AE pulse counting performance test by the monitoring system.
Sensors 2023, 23, 3405 12 of 21

The measurement results listed in Table 1 show that the system is characterized by
high efficiency and is capable of lossless counting of up to 83,400 AE pulses per minute.
Above this value, the counting accuracy decreases.

Table 1. The results of the AE pulse counting performance test by the monitoring system.

Pulse Generation The Number of Pulses Average Amount Maximum Amount

Frequency (Hz) Generated per Minute of Lost Pulses (%) of Lost Pulses (%)
≤1380 ≤83,400 0 0
1400 84,000 0.60 0.79
1500 90,000 2.99 4.08
1600 96,000 4.74 6.23
1700 102,000 7.91 8.28

3.2. Online Monitoring of Partial Discharges in Laboratory Conditions

The tests were carried out in a shielded high-voltage laboratory using a model of
a power transformer tank with dimensions of 1200 × 800 × 730 mm, which was filled
with mineral oil. A system of electrodes for generating surface discharges on a round
sample of pressboard with a diameter of 100 mm and a thickness of 3 mm was mounted to
the internal connecting terminal of the bushing. The tip of a point electrode was located
400 mm from the wall of the ladle on which the tested system was installed. The PDtracker
Portable system (Poznan University of Technology, Poznan, Poland) was used as a reference
system, which is equipped with eight analog inputs with parallel sampling up to 20 MS/s.
As partial discharge detectors, the system accepts both piezoelectric acoustic emission
sensors (default configuration) and high-frequency current transformers. The AE sensor
of the tested system and the A6890 sensor of the reference PDtracker system were placed
exactly at the height of the PD source. The distance between the sensors was 200 mm,
while the distance from the PD source to each sensor was the same and was about 400 mm
(Figure 10). One of the basic functions performed by the PDtracker Portable system is also
the counting of PD pulses. For this purpose, the system continuously records the signal,
which is divided into time frames with a fixed width set by the user (in this case, 2 ms).
If, in the analyzed time frame, the amplitude of the AE waveform exceeds the threshold
value, then the system treats this event as the occurrence of a PD pulse and counts it. The
threshold value is usually twice the average amplitude of the acoustic background noise,
which in laboratory conditions does not exceed 10 mV.
During the tests, in addition to the reference PDtracker Portable system discussed
above, a conventional PD meter (PD-Smart, Doble Engineering Company, Marlborough,
MA, USA) and a measuring circuit compliant with the IEC60270 standard were also used
(Figure 11). This made it possible to control the value of the apparent charge of the partial
discharges and their intensity during the experiment.
A voltage in the range from 0 to 17 kV was applied to the electrode system, which
allowed to generate partial discharge pulses of various intensity and energy. The inception
voltage of surface discharges was Ui = 9.2 kV, and their apparent charge ranged from 45 pC
to over 5 nC. Exemplary test results are shown in Figure 12.
The analysis of the obtained test results shows that the discussed system is capable of
online monitoring of partial discharges occurring in the oil–paper insulation system of the
transformer. Due to the different implementations of the pulse counting procedure by the
tested and reference acoustic emission systems, correlation analysis was performed instead
of quantitative analysis. The Pearson correlation coefficient determined for the number
of AE pulses recorded by both systems was 0.775 (strong correlation), while Spearman’s
correlation coefficient was 0.91 (very strong correlation). In turn, the comparison with
the number of PD pulses recorded by a conventional meter was more favorable for the
PDtracker Portable reference system, as in this case, the Pearson correlation coefficient was
0.797, while for the low-cost monitoring system it was 0.639, which means a moderate
 placed exactly at the height of the PD source. The distance between the sensors was
mm, while the distance from the PD source to each sensor was the same and was a
400 mm (Figure 10). One of the basic functions performed by the PDtracker Portable
tem is also the counting of PD pulses. For this purpose, the system continuously rec
Sensors 2023, 23, 3405 the signal, which is divided into time frames with a fixed width set by 13 ofthe
21 user (in
case, 2 ms). If, in the analyzed time frame, the amplitude of the AE waveform exceed
threshold value, then the system treats this event as the occurrence of a PD pulse
counts it. The threshold value is usually twice the average amplitude of the acoustic b
correlation. Spearman’s correlation coefficient was 0.891 and 0.793, respectively, which can
ground noise, which in laboratory conditions does not exceed 10 mV.
be interpreted as a strong correlation.

(a) (b)
Figure 10. TestFigure
of the10. Test of thesystem
monitoring monitoring system inconditions:
in laboratory laboratory conditions:
(a) electrode(a)system
electrode
for system
gen- for gen
erating surface discharges; (b) arrangement of the tested and reference monitoring system on the on the t
ing surface discharges; (b) arrangement of the tested and reference monitoring system
former
transformer tank tank
model: model:
TT—oil TT—oil
filled filled transformer
transformer tank with tank with system
electrode electrode
forsystem for generating
generating partial partia
Sensors 2023, 23, x FOR PEER REVIEW charges; TB—transformer bushing; PD—electrode system for generating surface 14 of partial
21 disch
discharges; TB—transformer bushing; PD—electrode system for generating surface partial discharges
PT—piezoelectric transducer A6890; DUT—system under test.
PT—piezoelectric transducer A6890; DUT—system under test.

During the tests, in addition to the reference PDtracker Portable system discu
above, a conventional PD meter (PD-Smart, Doble Engineering Company, Marlboro
MA, USA) and a measuring circuit compliant with the IEC60270 standard were also
(Figure 11). This made it possible to control the value of the apparent charge of the pa
discharges and their intensity during the experiment.
A voltage in the range from 0 to 17 kV was applied to the electrode system, w
allowed to generate partial discharge pulses of various intensity and energy. The in
tion voltage of surface discharges was Ui = 9.2 kV, and their apparent charge ranged
45 pC to over 5 nC. Exemplary test results are shown in Figure 12.

Figure
Figure 11.
11. The
Themeasuring
measuringset-up
set-upused
usedduring
duringtesting ofof
testing the monitoring
the monitoringsystem in in
system laboratory condi-
laboratory con-
tions: U—high-voltage
ditions: U—high-voltagesupply; Z—short-circuit
supply; current
Z—short-circuit limiting
current resistor;
limiting MS—reference
resistor; MS—referenceonline PD
online
monitoring
PD monitoringsystem PDtracker;
system PT—piezoelectric
PDtracker; transducer
PT—piezoelectric A6890;
transducer DUT—system
A6890; underunder
DUT—system test; TT—
test;
oil filled transformer tank with electrode system for generating partial discharges; CK—coupling
TT—oil filled transformer tank with electrode system for generating partial discharges; CK—coupling
capacitor; CD—measuring impedance; CC—connecting cable; M—conventional partial discharge
capacitor; CD—measuring impedance; CC—connecting cable; M—conventional partial discharge
measuring device Doble PD-Smart; PC—computer.
measuring device Doble PD-Smart; PC—computer.

3.3. Online Monitoring of Partial Discharges in a 10 MVA Power Transformer

The field test of the system took place during partial discharge monitoring on a power
transformer with a voltage of 115,000 ± 10% kV and a power of 10 MVA, which was
manufactured in 1992. The reference measuring device was the eight-channel PDtracker
Portable monitoring system, whose six piezoelectric transducers were placed near the
phases of the low (LV1, LV2, LV3) and high (HV1, HV2, HV3) voltage sides. The last
two transducers were installed on the side walls of the transformer tank (under the oil
conservator and opposite the on-load tap-changer). The tested, low-cost PD monitoring
system was always mounted in close proximity (approx. 15–20 cm) to the piezoelectric
transducer of the reference system (Figure 13).
 Figure 11. The measuring set-up used during testing of the monitoring system in laboratory condi-
tions: U—high-voltage supply; Z—short-circuit current limiting resistor; MS—reference online PD
monitoring system PDtracker; PT—piezoelectric transducer A6890; DUT—system under test; TT—
oil filled transformer tank with electrode system for generating partial discharges; CK—coupling
Sensors 2023, 23, 3405 14 of 21
capacitor; CD—measuring impedance; CC—connecting cable; M—conventional partial discharge
measuring device Doble PD-Smart; PC—computer.

Sensors 2023, 23, x FOR PEER REVIEW 15 of 21

The analysis of the obtained test results shows that the discussed system is capable
of online monitoring of partial discharges occurring in the oil–paper insulation system of
the transformer. Due to the different implementations of the pulse counting procedure by
the tested and reference acoustic emission systems, correlation analysis was performed
instead of quantitative analysis. The Pearson correlation coefficient determined for the
number of AE pulses recorded by both systems was 0.775 (strong correlation), while
Spearman’s correlation coefficient was 0.91 (very strong correlation). In turn, the compar-
ison with the number of PD pulses recorded by a conventional meter was more favorable
for the PDtracker Portable reference system, as in this case, the Pearson correlation coeffi-
cient was 0.797, while for the low-cost monitoring system it was 0.639, which means a
moderate correlation. Spearman’s correlation coefficient was 0.891 and 0.793, respectively,
which can be interpreted as a strong correlation.

3.3. Online Monitoring of Partial Discharges in a 10 MVA Power Transformer

The field test of the system took place during partial discharge monitoring on a
power transformer with a voltage of 115,000 ± 10% kV and a power of 10 MVA, which
was manufactured in 1992. The reference measuring device was the eight-channel
PDtracker Portable monitoring system, whose six piezoelectric transducers were placed
near the phases of the low (LV1, LV2, LV3) and high (HV1, HV2, HV3) voltage sides. The
last two
Figure
Figure [Link]
12. Testvoltage
Test were
voltagevalue
value installed
(a),
(a), PDPD on thecharge
apparent
apparent side walls
charge
(b),(b),ofnumber
number theoftransformer
of pulses
PD PD pulsestank (under
recorded
recorded bythe
by a con-a
oil conservator
conventional
ventional partial and
partial opposite
discharge
discharge meter thePD-Smart,
meter on-load
PD-Smart,tap-changer).
(c)(c) number
number The
ofofAE tested,
AEpulses low-cost
pulsesrecorded PD
recordedby
bythemonitoring
the reference
reference
system was
monitoring
monitoring always
system
system mounted
PDtracker
PDtracker in close
Portable,
Portable, (d) proximity
(d) number
number of (approx.
of AE
AE pulses 15–20 cm)bytothe
pulses registered
registered the piezoelectric
tested, low-cost
PD
PD monitoring
transducer
monitoringof system (e).
the reference
system (e). system (Figure 13).

Figure 13.
Figure 13. Monitoring
Monitoringofofpartial
partialdischarges in the
discharges HV3
in the phase
HV3 of a of
phase 10 MVA power
a 10 MVA transformer:
power MS—
transformer:
reference monitoring
MS—reference system
monitoring PDtracker
system Portable;
PDtracker PT—piezoelectric
Portable; PT—piezoelectrictransducer
transducerof
of the reference
the reference
monitoring system;
monitoring system;DUT—tested
DUT—testedlow-cost
low-costPD
PDmonitoring
monitoringsystem.
system.

The measurement data analysis showed that in the tested transformer, both monitor-
ing systems recorded a large number of AE pulses only near the HV3 phase (Figure 14).
The period of increased intensity of the acoustic emission phenomenon lasted about 20
min and began at the moment when the OLTC (on-load tap changer) position changed.
After another tap change, both systems recorded only single AE events. The Pearson cor-
 Sensors 2023, 23, 3405 15 of 21

The measurement data analysis showed that in the tested transformer, both monitoring
systems recorded a large number of AE pulses only near the HV3 phase (Figure 14). The
period of increased intensity of the acoustic emission phenomenon lasted about 20 min
and began at the moment when the OLTC (on-load tap changer) position changed. After
another tap change, both systems recorded only single AE events. The Pearson correlation
coefficient calculated for the number of AE pulses detected by both systems was 0.980 (very
Sensors 2023, 23, x FOR PEER REVIEW
strong correlation), and the Spearman correlation coefficient was 0.869 (strong correlation).
16 of 21
This showed that the performance of a low-cost PD monitoring system could be comparable
to much more expensive and complex commercial systems.

Figure 14.14.
Figure TheThenumber of of
number AEAEpulses recorded
pulses recorded during a 2-h
during partial
a 2-h discharge
partial monitoring
discharge monitoringnear thethe
near
HV3 phase of a 10 MVA transformer by the reference PDtracker Portable system (a), and the tested
HV3 phase of a 10 MVA transformer by the reference PDtracker Portable system (a), and the tested
low-cost PD monitoring system (b).
low-cost PD monitoring system (b).

4. 4.
Conclusions
Conclusions
TheThearticle
articlepresents
presentsin in detail thethe design
designofofa alow-cost,
low-cost, portable
portable online
online partial
partial dis-
discharge
charge monitoring
monitoring system system
based based on a non-invasive
on a non-invasive methodmethod of acoustic
of acoustic emission, emission,
which meetswhich the
guidelines
meets of the IEC
the guidelines ofTSthe62478
IEC TSstandard
62478 in terms ofingeneral
standard terms of requirements for the AE for
general requirements mea-
surement
the system and
AE measurement for theand
system measured
for the PD quantities.
measured The reason The
PD quantities. for developing this device
reason for develop-
was
ing to device
this eliminatewasthetomain obstacle
eliminate thetomain
the widespread
obstacle to usethe of PD online monitoring
widespread systems,
use of PD online
which is their
monitoring highwhich
systems, price and complicated
is their high priceoperation, which requires
and complicated operation, a lot
whichof experience.
requires
In the
a lot case of the discussed
of experience. In the casesystem, the total cost
of the discussed of all the
system, components
total cost used
of alltocomponents
build it does
used to build it does not exceed 300USD, which is a small fraction of the price of a Thanks
not exceed 300USD, which is a small fraction of the price of a commercial system. com-
to this,system.
mercial the system
Thanks canto contribute to the dissemination
this, the system can contributeof tothe
theidea of short-term
dissemination PDidea
of the moni-
oftoring, especially
short-term in developing
PD monitoring, countries,
especially where with
in developing the dynamically
countries, where with growing demand
the dynam-
for electricity,
ically the demand
growing demand for inexpensive,
for electricity, easy-to-manufacture
the demand for inexpensive, transformer diagnostic
easy-to-manufacture
systems is diagnostic
transformer also growing. This is
systems system can also be
also growing. This used during
system canlaboratory
also be used measurements
during la-
as a supplement or alternative to a stationary partial discharge
boratory measurements as a supplement or alternative to a stationary partial discharge detector. Another area
of application for the device may be medium power transformers,
detector. Another area of application for the device may be medium power transformers, for which online PD
monitoring using complex measurement systems requiring
for which online PD monitoring using complex measurement systems requiring addi- additional infrastructure and
qualified
tional personneland
infrastructure is usually
qualifiedeconomically
personnel isunjustified.
usually economically unjustified.
In addition to the low production
In addition to the low production price, price, the
the advantage
advantage ofofthe
thesystem
systemis:is:
• • a high
a high detection
detection sensitivity
sensitivity ofof acoustic
acoustic signals
signals from
from partial
partial discharges,
discharges, resulting
resulting from
from
the
the useofoftrue
use truedifferential
differential AE
AE sensor
sensor with
with optimized
optimizedfrequency
frequencyresponse
responsecharacteristics;
character-
• istics;
the possibility of equipping the system with inexpensive modules dedicated to the
Arduino/Teensyduino platform that increase its functionality, such as a Bluetooth
• the possibility of equipping the system with inexpensive modules dedicated to the
module for wireless transmission of measurement data, or additional sensors to monitor
Arduino/Teensyduino platform that increase its functionality, such as a Bluetooth
other parameters of the transformer’s operation (e.g., an accelerometer for measuring
module for wireless transmission of measurement data, or additional sensors to mon-
transformer tank vibrations);
itor other parameters of the transformer’s operation (e.g., an accelerometer for meas-
• easy installation on the transformer tank (thanks to the low weight and magnetic holders);
uring transformer tank vibrations);
• no need to configure the device;
• easy installation on the transformer tank (thanks to the low weight and magnetic
holders);
• no need to configure the device;
• high resistance to external electromagnetic interference due to the shielding of elec-
tronic modules and piezoelectric transducer elements.
Sensors 2023, 23, 3405 16 of 21

• high resistance to external electromagnetic interference due to the shielding of elec-

tronic modules and piezoelectric transducer elements.
The system also has some limitations, the most important of which is the battery
power supply and the relatively short, several-day working time. Another disadvantage
is related to the PD detection method used. In the case of the acoustic emission method,
the sensitivity of partial discharge detection strongly depends on the distance between the
AE sensor and the PD source. Therefore, before assembling the system, especially on large
power transformers, it may be necessary to locate the place on the tank where the AE pulses
are recorded in advance. A standard auscultatory technique (SAT) can be used for this
purpose [48]. The possibility of real-time execution of only one task (counting PD pulses)
can also be considered a disadvantage of the system. This is due to the limited computing
power of the Teensy 3.2 microcontroller, which ranges from a few to several dozen MIPS.
However, this problem can be relatively easily solved by equipping the system with a
second microcontroller or using—unfortunately at the cost of many times more power
consumption—a microcontroller with a much more efficient processor, such as Teensy 4.1
with an ARM Cortex-M7 processor clocked at 600 MHz.

Author Contributions: Conceptualization, W.S.; methodology, W.S.; software, A.W.; investigation,

W.S. and A.W.; data curation, W.S.; writing—original draft preparation, W.S.; writing—review and
editing, W.S. and A.W.; visualization, W.S.; supervision, W.S.; project administration, W.S.; funding
acquisition, W.S. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by the Polish National Centre for Research and Development,
within the Applied Research Programme, grant No. PBS3/A4/12/2015.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

Appendix A. Source Code for the Firmware of the Online PD Monitoring System
//declare libraries
#include <TimeLib.h> //RTC library
#include <FreqCount.h> //reading pulses library
#include <LiquidCrystal.h> //LCD library
#include <SPI.h> //Serial Peripheral Interface library
#include <SD.h> //SD card
//declare LCD pinout
const int rs = 18,, en = 17,
, d4 = 23,, d5 = 22,
, d6 = 21,
, d7 = 20;
;
LiquidCrystal lcd((rs, en,
, d4,, d5,
, d6,
, d7)
);
//declare count variable
unsigned long count = 0;;
//declare count in previous iteration variable
unsigned long prior_count=0;;
//auxiliary variable
bool x = 0;;
File myFile;;
//define pin connected to "RESET" button
const byte resetPin = 15;;
//void setup is executing only once, after start
() {
void setup()
//PULLDOWN the NC "RESET" button
(resetPin, INPUT_PULLDOWN);
pinMode( );
//set SCK in SPI connection
Sensors 2023, 23, 3405 17 of 21

[Link]((14);
);
();
[Link]();
setSyncProvider( (getTeensy3Time);
);
//begin frequency counting
[Link]( (100);
);
//define LCD size
(8,, 2);
[Link]( );

[Link]( (115200););
//try to connect to external RTC and sync
()!= timeSet)
if ( timeStatus()!= ) {
[Link]( ("Unable to sync with the RTC");
);
} else {
[Link]( ("RTC has set the system time");
);
}
//setup message
[Link]( ("Initializing SD card..."); );
[Link]( (0,, 0);
);
[Link]( ("SD:"););
//connect to SD, if failed print a message
if (!
()
)) {
[Link]( ("Initialization failed!"););
[Link]( (4,,0);
);
[Link]( ("FAIL"););
}
else
else{{
[Link]( ("Initialization done."););
[Link]( (4,0););
[Link]( ("OK");
);
}
//starting delay after initialization to make a communicate readable
delay((4000);
);
}
//void loop is running repeatedly
void loop()() {
if ([Link]()) ()) {
count = count + [Link](); ();
//if the number of counted pulses changed print a new one on the LCD
if ( count != prior_count)) {
//next line can be uncommented if the console connection is enabled
//[Link](count);

//print the number of counted pulses

(0,, 1);
[Link]( );
(count);
[Link]( );

//overwrite the number of pulses

prior_count = count;
}
}
//print the time on the LCD
(0,, 0);
[Link]( );
printDigitsLcd((hour());
());
(":");
[Link]( );
Sensors 2023, 23, 3405 18 of 21

printDigitsLcd( (minute());
());
[Link]((":");
);
printDigitsLcd( (second());
());
//save number of pulses to file in 1 minute intervals (seconds==0)
//bool x is to avoid saving more than once in one second
()
if ( second() &&
()==0&& ){
&&x==0){
digitalClockDisplay();();
;
x=1;
//declare file name and open it
myFile = [Link](("[Link]", , FILE_WRITE);
);
//save time and number of pulses to the file
if ( myFile)) {
[Link]( ("Writing to [Link]... "); );
( ());
[Link]( day());
[Link]( (".");
);
[Link]( (month());
());
[Link]( (".");
);
[Link]( (year());
());
[Link]( (" ");
);
[Link]( (hour());
());
printDigitsFile( (minute());
());
printDigitsFile( (second());
());
[Link]( (",");
);
[Link]( (count);
);
();
[Link]();
}
();
[Link]();
}
//if the second turn to 1, overwrite x value as 0
()
if ( second()
()==1){){
;
x=0;
}
//execute function after the button "RESET' is pushed
if((digitalRead(
if (resetPin))==0))
{
();
zero();
}
}
//save data in a text file; "PULSES:" and number of pulses counted
void digitalClockDisplay() () {
[Link]( (day());
());
[Link]( (".");
);
( ());
[Link]( month());
[Link]( (".");
);
[Link]( (year());
());
[Link]( (" ");
);
[Link]( (hour());
());
printDigits( (minute());
());
printDigits( (second());
());
[Link]( (" ");
);
[Link]( ("PULSES: "););
[Link]( (count);
);
();
[Link]();
}
Sensors 2023, 23, 3405 19 of 21

//get time from RTC

()
time_t getTeensy3Time()
{
();
return [Link]();
}
#define TIME_HEADER “T”
() {
unsigned long processSyncMessage()
unsigned long pctime = 0L;
const unsigned long DEFAULT_TIME = 1357041600; // Jan 1 2013

if(([Link](
if (TIME_HEADER))
)) {
();
pctime = [Link]();
return pctime;
if
if(( pctime < DEFAULT_TIME)) { //check if the time if valid
; //return 0 if the time is invalid
pctime = 0L;
}
}
return pctime; ;
}
//add 0 when digits<10 (Serial output)
void printDigits( (int digits){
){
[Link]( (":");
);
if((digits < 10)
if )
[Link](('0');
);
[Link]( (digits);
);
}
//add 0 when digits<10 (LCD output)
void printDigitsLcd((int digits){
){
if (
if( digits < )
10)
[Link](('0');
);
[Link]((digits););
}
//add 0 when digits<10 (file output)
void printDigitsFile((int digits){
){
[Link]( (":");
);
if((digits < 10))
[Link](('0');
);
[Link]( (digits);
);
}
//function for resetting the number of pulses
() {
void zero()
count=0;;
[Link]( (0, 1);
);
[Link]((" ");
);
delay((500);
);
}

References
1. Tenbohlen, S.; Jagers, J.; Vahidi, F.; Bastos, G.; Desai, B.; Diggin, B.; Fuhr, J.; Gebauer, J.; Krüger, M.; Lapworth, J.; et al. Transformer
Reliability Survey; Technical Brochure 642 CIGRE: Paris, France, 2015.
2. Martin, D.; Marks, J.; Saha, T. Survey of Australian power transformer failures and retirements. IEEE Electr. Insul. Mag. 2017, 33,
16–22. [CrossRef]
3. Thiviyanathan, V.A.; Ker, P.J.; Leong, Y.S.; Abdullah, F.; Ismail, A.; Zaini Jamaludin, M. Power transformer insulation system: A
review on the reactions, fault detection, challenges and future prospects. Alex. Eng. J. 2022, 61, 7697–7713. [CrossRef]
Sensors 2023, 23, 3405 20 of 21

4. Kaliappan, G.; Rengaraj, M. Aging assessment of transformer solid insulation: A review. Mater. Today Proc. 2021, 47, 272–277.
[CrossRef]
5. Dombek, G.; Nadolny, Z. Influence of paper type and liquid insulation on heat transfer in transformers. IEEE Trans. Dielectr.
Electr. Insul. 2018, 25, 1863–1870. [CrossRef]
6. Liu, Y.; He, S.; Sun, L.; Xie, Y.; Zhao, S.; Qin, J.; Wan, X. Transformer winding Deformation Detection Analysis and Typical Cases.
In Proceedings of the 2016 IEEE International Conference on High-Voltage Engineering and Applications (ICHVE), Piscataway,
NJ, USA, 18–22 September 2016.
7. Hanbo, Z.; Yuquan, L.; Xiaohui, W.; Wei, W.; Zhenyu, W.; Yingbiao, S.; Lun, M.; Xiaogang, L. Low-voltage Impedance Diagnosis
Method of Transformer Winding Deformation and Case Study. Insul. Mater. 2016, 49, 66–70.
8. IEC 60270; High-Voltage Test Techniques—Partial Discharge Measurements. IEC: Geneva, Switzerland, 2015.
9. IEC-TS, PD. 62478; 2016 High Voltage Test Techniques—Measurement of Partial Discharges by Electromagnetic and Acoustic
Methods. BSI: London, UK, 2016.
10. C57.127-2018; IEEE Guide for the Detection, Location and Interpretation of Sources of Acoustic Emissions from Electrical
Discharges in Power Transformers and Power Reactors. Institute of Electrical and Electronics Engineers: Piscataway, NJ, USA,
11 March 2019.
11. Siegel, M.; Beltle, M.; Tenbohlen, S.; Coenen, S. Application of UHF sensors for PD measurement at power transformers. IEEE
Trans. Dielectr. Electr. Insul. 2017, 24, 331–339. [CrossRef]
12. Mirzaei, H.; Akbari, A.; Gockenbach, E.; Miralikhani, K. Advancing new techniques for UHF PD detection and localization in the
power transformers in the factory tests. IEEE Trans. Dielectr. Electr. Insul. 2015, 22, 448–455. [CrossRef]
13. Judd, M.; Yang, L.; Hunter, I. Partial discharge monitoring for power transformer using UHF sensors. Part 2: Field experience.
IEEE Electr. Insul. Mag. 2005, 21, 5–13. [CrossRef]
14. Soh, D.; Krishnan, S.B.; Abraham, J.; Xian, L.K.; Jet, T.K.; Yongyi, J.F. Partial Discharge Diagnostics: Data Cleaning and Feature
Extraction. Energies 2022, 15, 508. [CrossRef]
15. Wotzka, D.; Kozioł, M.; Boczar, T.; Kunicki, M.; Nagi, Ł. Latest Trends in the Improvement of Measuring Methods and Equipment
in the Area of NDT. Sensors 2021, 21, 7293. [CrossRef]
16. Lu, S.; Chai, H.; Sahoo, A.; Phung, B.T. Condition Monitoring Based on Partial Discharge Diagnostics Using Machine Learning
Methods: A Comprehensive State-of-the-Art Review. IEEE Trans. Dielectr. Electr. Insul. 2020, 27, 1861–1888. [CrossRef]
17. Castro, B.; Clerice, G.; Ramos, C.; Andreoli, A.; Baptista, F.; Campos, F.; Ulson, J. Partial Discharge Monitoring in Power
Transformers Using Low-Cost Piezoelectric Sensors. Sensors 2016, 16, 1266. [CrossRef] [PubMed]
18. Besharatifard, H.; Hasanzadeh, S.; Heydarian-Forushani, E.; Alhelou, H.H.; Siano, P. Detection and Analysis of Partial Discharges
in Oil-Immersed Power Transformers Using Low-Cost Acoustic Sensors. Appl. Sci. 2022, 12, 3010. [CrossRef]
19. Kunicki, M. Comparison of capacitive and inductive sensors designed for partial discharges measurements in electrical power
apparatus. E3S Web Conf. 2017, 19, 01035. [CrossRef]
20. Md Rahim, N.H.; Ahmad, M.H.; Nawawi, Z.; Sidik, M.A.B.; Jambak, M.I. Partial Discharge Detection using Developed Low-
cost High Frequency Current Transformer. In Proceedings of the 2019 International Conference on Electrical Engineering and
Computer Science (ICECOS), Batam, Indonesia, 2–3 October 2019; pp. 74–78.
21. Álvarez, F.; Garnacho, F.; Ortego, J.; Sánchez-Urán, M.Á. Application of HFCT and UHF Sensors in Online Partial Discharge
Measurements for Insulation Diagnosis of High Voltage Equipment. Sensors 2015, 15, 7360–7387. [CrossRef] [PubMed]
22. Kim, J.; Kim, K.-I. Partial Discharge Online Detection for Long-Term Operational Sustainability of On-Site Low Voltage Distribu-
tion Network Using CNN Transfer Learning. Sustainability 2021, 13, 4692. [CrossRef]
23. Sarkar, B.; Koley, C.; Roy, N.K.; Kumbhakar, P. Low cost RF sensor for partial discharge detection of high voltage apparatus. In
Proceedings of the 2013 IEEE 1st International Conference on Condition Assessment Techniques in Electrical Systems (CATCON),
Kolkata, India, 6–8 December 2013; pp. 259–264. [CrossRef]
24. Sarkar, B.; Mishra, D.K.; Koley, C.; Roy, N.K. Microstrip patch antenna based UHF Sensor for detection of partial discharge in
high voltage electrical equipments. In Proceedings of the 2014 Annual IEEE India Conference (INDICON), Pune, India, 11–13
December 2014; pp. 1–6.
25. Polak, F.; Sikorski, W.; Siodla, K. Location of partial discharges sources using sensor arrays. In Proceedings of the 2014 ICHVE
International Conference on High Voltage Engineering and Application, Poznan, Poland, 8–11 September 2014; pp. 1–4.
26. Ju, T.; Zhongrong, X.; Xiaoxing, Z.; Caixin, S. GIS partial discharge quantitative measurements using UHF microstrip antenna
sensors. In Proceedings of the 2007 Annual Report-Conference on Electrical Insulation and Dielectric Phenomena, Vancouver, BC,
Canada, 14–17 October 2007; pp. 116–119.
27. Nobrega, L.A.M.M.; Xavier, G.V.R.; Aquino, M.V.D.; Serres, A.J.R.; Albuquerque, C.C.R.; Costa, E.G. Design and development of
a bio-inspired UHF sensor for partial discharge detection in power transformers. Sensors 2019, 19, 653. [CrossRef]
28. Cruz, J.N.; Serres, A.J.R.; de Oliveira, A.C.; Xavier, G.V.R.; de Albuquerque, C.C.R.; da Costa, E.G.; Freire, R.C.S. Bio-inspired
printed monopole antenna applied to partial discharge detection. Sensors 2019, 19, 628. [CrossRef]
29. Li, M.; Guo, C.; Peng, Z. Design of meander antenna for UHF partial discharge detection of transformers. Sens. Transducers 2014,
171, 232–238.
30. Li, T.; Rong, M.; Zheng, C.; Wang, X. Development simulation and experiment study on UHF Partial Discharge Sensor in GIS.
IEEE Trans. Dielectr. Electr. Insul. 2012, 19, 1421–1430. [CrossRef]
Sensors 2023, 23, 3405 21 of 21

31. Robles, G.; Albarracín, R.; Vázquez-Roy, J.L.; Rajo-Iglesias, E.; Martínez-Tarifa, J.M.; Rojas-Moreno, M.V.; Ardila-Rey, J. On the
use of vivaldi Antennas in the detection of partial discharges. In Proceedings of the IEEE International Conference on Solid
Dielectrics (ICSD), Bologna, Italy, 30 June–4 July 2013; pp. 302–305.
32. Khosronejad, M.; Gentili, G.G. Design of an archimedean spiral UHF antenna for pulse monitoring application. In Proceedings of
the Loughborough Antennas and Propagation Conference (LAPC), Loughborough, UK, 2–3 November 2015; pp. 1–4.
33. Wu, Q.; Liu, G.; Xia, Z.; Lu, L. The study of archimedean spiral antenna for partial discharge measurement. In Proceedings of
the IVth International Conference on Intelligent Control and Information Processing (ICICIP), Beijing, China, 9–11 June 2013;
pp. 694–698.
34. Sikorski, W.; Szymczak, C.; Siodła, K.; Polak, F. Hilbert curve fractal antenna for detection and online monitoring of partial
discharges in power transformers. Eksploat. I Niezawodn. Maint. Reliab. 2018, 20, 343–351. [CrossRef]
35. Li, J.; Jiang, T.; Wang, C.; Cheng, C. Optimization of UHF Hilbert antenna for partial discharge detection of transformers. IEEE
Trans. Antennas Propag. 2012, 60, 2536–2540.
36. Li, J.; Li, X.; Du, L.; Cao, M.; Qian, G. An intelligent sensor for the ultra-high-frequency partial discharge online monitoring of
power transformers. Energies 2016, 9, 383. [CrossRef]
37. Szymczak, C.; Sikorski, W. Design and optimization of UHF antennas for partial discharge monitoring in power transformer. Prz.
Elektrotechniczny 2016, 92, 75–79.
38. Wang, F.; Bin, F.; Sun, Q.; Fan, J.; Liang, F.; Xiao, X. A novel UHF Minkowski fractal antenna for partial discharge detection.
Microw. Opt. Technol. Lett. 2017, 59, 1812–1819. [CrossRef]
39. Abdul Malek, N.A.; Ahmad, M.H.; Nawawi, Z.; Sidik, M.A.B.; Jambak, M.I. Low Cost Transient Earth Voltage Probe for Partial
Discharge Detection. In Proceedings of the 2019 International Conference on Electrical Engineering and Computer Science
(ICECOS), Batam, Indonesia, 2–3 October 2019; pp. 63–67. [CrossRef]
40. Zhang, C.; Dong, M.; Ren, M.; Huang, W.; Zhou, J.; Gao, X.; Albarracín, R. Partial Discharge Monitoring on Metal-Enclosed
Switchgear with Distributed Non-Contact Sensors. Sensors 2018, 18, 551. [CrossRef]
41. Saeed, B.; Upton, D.; Vieira, M.; Torres, F.; Mohamed, H.; Mistry, K.; Mather, P.; Lazaridis, P.; Khan, U.; Tachtatzis, C. A
supervisory system for partial discharge monitoring. In Proceedings of the 2018 2nd URSI Atlantic Radio Science Meeting
(AT-RASC), Melonaras, Spain, 28 May–1 June 2018.
42. Emilliano; Chakrabarty, C.K.; Abdul Ghani, A.B.; Ramasamy, A.K. Online partial discharge counting system using microcontroller
PIC 16F877A and FPGA technology. In Proceedings of the 2010 IEEE Student Conference on Research and Development
(SCOReD), Kuala Lumpur, Malaysia, 13–14 December 2010; pp. 314–319. [CrossRef]
43. Chang, H.C.; Jheng, Y.M.; Kuo, C.C.; Hsueh, Y.M.; Cheng, Y.J. Development of a reconfigurable FPGA-based partial discharge
detection device. In Proceedings of the 2017 International Conference on Applied System Innovation (ICASI), Sapporo, Japan,
13–17 May 2017; pp. 22–24. [CrossRef]
44. Yan, Y.; Ren, S.; Lu, Y.; Yang, S.; Zhao, K.; Li, H. Efficient and Low-Cost PD Monitoring and Locating System for MV Switchgears
Using TEV Detectors. IEEE Trans. Power Deliv. 2021, 36, 3266–3269. [CrossRef]
45. Mohamed, H.; Lazaridis, P.; Upton, D.; Khan, U.; Saeed, B.; Jaber, A.; Zhang, Y.; Mather, P.; Vieira, M.F.; Barlee, K. Partial
discharge detection using low cost RTL-SDR model for wideband spectrum sensing. In Proceedings of the IEEE 23rd International
Conference on Telecommunications (ICT), Thessaloniki, Greece, 16–18 May 2016; pp. 1–5.
46. Sikorski, W. Development of Acoustic Emission Sensor Optimized for Partial Discharge Monitoring in Power Transformers.
Sensors 2019, 19, 1865. [CrossRef] [PubMed]
47. IEEE Std C57.127-2007 (Revision of IEEE Std C57.127-2000); IEEE Guide for the Detection and Location of Acoustic Emissions
from Partial Discharges in Oil-Immersed Power Transformers and Reactors. Institute of Electrical and Electronics Engineers:
Piscataway, NJ, USA, 2007.
48. Sikorski, W.; Siodla, K.; Moranda, H.; Ziomek, W. Location of partial discharge sources in power transformers based on advanced
auscultatory technique. IEEE Trans. Dielectr. Electr. Insul. 2012, 19, 1948–1956. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.