energies

Article
On-Line Partial Discharge Monitoring System for
Power Transformers Based on the Simultaneous
Detection of High Frequency, Ultra-High Frequency,
and Acoustic Emission Signals
Wojciech Sikorski 1, * , Krzysztof Walczak 1 , Wieslaw Gil 2 and Cyprian Szymczak 1
1 Institute of Electric Power Engineering, Poznan University of Technology, 60-965 Poznan, Poland;
[email protected] (K.W.); [email protected] (C.S.)
2 Mikronika, 60-001 Poznan, Poland; [email protected]
* Correspondence: [email protected]




Received: 5 June 2020; Accepted: 22 June 2020; Published: 24 June 2020


Abstract: The article presents a novel on-line partial discharge (PD) monitoring system for power
transformers, whose functioning is based on the simultaneous use of three unconventional methods
of PD detection: high-frequency (HF), ultra-high frequency (UHF), and acoustic emission (AE). It is
the first monitoring system equipped in an active dielectric window (ADW), which is a combined
ultrasonic and electromagnetic PD sensor. The article discusses in detail the process of designing
and building individual modules of hardware and software layers of the system, wherein the
most attention was paid to the PD sensors, i.e., meandered planar inverted-F antenna (MPIFA),
high-frequency current transformer (HFCT), and active dielectric window with ultrasonic transducer,
which were optimized for detection of PDs occurring in oil-paper insulation. The prototype of
the hybrid monitoring system was first checked on a 330 MVA large power transformer during
the induced voltage test with partial discharge measurement (IVPD). Next, it was installed on a
31.5 MVA substation power transformer and integrated according to the standard IEC 61850 with
SCADA (Supervisory Control and Data Acquisition) system registering voltage, active power, and oil
temperature of the monitored unit. The obtained results showed high sensitivity of the manufactured
PD sensors as well as the advantages of the simultaneous use of three techniques of PD detection and
the possibility of discharge parameter correlation with other power transformer parameters.

Keywords: power transformer; partial discharge (PD); on-line PD monitoring system; active dielectric
window; acoustic emission; UHF antenna; PIFA; HFCT sensor

1. Introduction
Power transformers are very important devices of public and industry use, requiring high capital
investment. Their reliable operation ensures continuity of electrical energy supply and stability of
the power energy system. A serious global problem is a high share of power transformers, for which
the designed lifetime (usually estimated at 30–35 years) expired [1–3]. However, considering the
high cost of purchase, transportation, and insurance as well as very long delivery time (even over
a year), power transformers are not, at the moment of expiring their lifetime, immediately replaced
with new ones. The majority of distribution companies adopt a management strategy based on the
use of the old units as long as it is possible, which unfortunately increases the risk of a catastrophic
failure [4]. The report of the CIGRE working group A2.37 indicates that dielectric (partial discharge,
tracking, arcing, flashover) and electrical failures (open circuit, short circuit, poor joint/contact, ground
deterioration, floating potential) were nearly half (49.3%) the cause of the serious damages of power

Energies 2020, 13, 3271; doi:10.3390/en13123271 www.mdpi.com/journal/energies

Energies 2020, 13, 3271 2 of 37

transformers in Europe [1]. The dynamics of the defect development caused by the activity of partial
discharges depends on a variety of factors, including the type of the discharge itself and its energy,
location of the defect, degree of aging and moisture content of the insulation system, varying load
affecting the oil temperature changes as well as moisture migration process. In some cases, the defect
may develop over many years of power transformer operation not leading it to damage. In other
cases, the defect may develop extremely rapidly, mostly as a consequence of a random event, such as
overvoltage or a sudden increase in oil temperature caused by overloading of a unit or malfunction of
the cooling system [1,5–7]. Very often the dynamics of the defect development grow in its final stage,
shortly before the damage itself [8]. For that reason, currently, the monitoring online systems have
gained popularity, since they have the advantage over periodic tests allowing for instant detection of
sudden growth of partial discharges activity [9–13]. Additionally, in combination with the SCADA
(supervisory control and data acquisition) system, which is installed on the majority of substations,
they allow to find the correlation between PD intensity and power transformer parameters, e.g., voltage
or oil temperature [14,15]. It increases the credibility of insulation system condition assessment and
enables the introduction of current limitations of power transformer exploitation, assuming that its
condition is not critical, and it does not require an immediate shutdown.
At present, the acoustic emission method (AE), electromagnetic methods (high-frequency (HF),
very-high frequency (VHF), and ultra-high frequency (UHF)), and dissolved gas analysis (DGA)
are some of the most important PD detection methods adjusted to operating in online monitoring
mode [16–25]. The last technique is indirect and because of its limitations, such as long time intervals
between successive measurements, the sensitivity depending on the dimensions of transformer tank and
distance of the gas sensor from the PD source as well as complex and non-standardized methods of DGA
results interpretation, it is not capable to detect very quickly developing defects. The acoustic emission
method and electromagnetic methods have both strengths and weaknesses, therefore none of them
may be recognized as more efficient or universal. For this reason, in the context of power transformers
diagnostics, the possibility of elaborating a hybrid system combining at least two measurement
methods is suggested more and more often. The main advantage of such a system application is
increase in PD detection reliability. This is because the individual methods are based on different
physical phenomena (detection of electromagnetic and acoustic waves) and different coupling methods
(inductive, capacitive, or acoustic), therefore each of them is sensitive to another type of disturbance.
For instance, in the AE method, an ultrasonic transducer may register core magnetostriction noise
(Barkhausen effect), pumped oil and fans noise, environmental noises (thunderstorms, rain, hail,
etc.), tap changer operation, and other random signals [26]. In the high-frequency method, sensor
HFCT is installed outside the transformer tank, therefore, it may register external electromagnetic
interferences (EMI) occurring on the substation, such as corona discharge from HV transmission
lines and signals accompanying switching operations. In turn, in the UHF method, the antenna may
receive a radio signal emitted by FM radio stations, aerial navigation systems, civilian and military
aviation, transmitters of civil services, DVB-T digital television, and GSM operators [27]. Therefore,
when a hybrid monitoring system, e.g., registers numerous UHF pulses and none of AE signals, it
may be assumed that the source of the signal is external EMI. In turn, when the system registers
pulses simultaneously by all methods and the fluctuations of their intensity are similar, then with high
probability one may assume that they are generated by partial discharges. As it was mentioned before,
the credibility of the diagnosis may be increased by finding a correlation between PD parameters and
power transformer parameters registered by the SCADA system. Those parameters that may affect the
partial discharge occurrence are voltage and oil temperature levels.
Other researchers also point to the benefits of combining two PD detection methods. Cavallini et
al. [28] believe that a combination of AE and UHF sensors can be successfully used as a tool to highlight
the presence of PD phenomena and can be used conveniently for monitoring and trending. Similar
conclusions were formulated by Witos and Gacek [29], who indicated that simultaneous registration of
PDs by two diagnostic techniques gives the possibility to elaborate new PD descriptors, which increase
Energies 2020, 13, x FOR PEER REVIEW 3 of 38
Energies 2020, 13, 3271 3 of 37

registration of PDs by two diagnostic techniques gives the possibility to elaborate new PD
descriptors,
the efficiencywhich increase
of defects the efficiency
recognition. of defects recognition.
Fuangsoongnern Fuangsoongnern
et al. [30] proved et al. [30]
that a combination of proved
the AE
that HF
and a combination
methods allows of eliminating
the AE and theHF methods
diagnostic allows
errors eliminating theindicate
and simultaneously diagnostic
if theerrors
problem andis
simultaneously indicate if the problem is caused by partial discharge, mechanical
caused by partial discharge, mechanical faults, arcing, or loose parts inside the transformer. Combining faults, arcing, or
loose
PD parts inside
detection methodsthe transformer.
allows obtaining Combining PD detection
of additional monitoring methods
systemallows obtaining of
functionalities. For additional
instance,
monitoring system
simultaneous functionalities.
registration of UHF For andinstance,
AE improves simultaneous
the accuracy registration
of PD of UHF and
sources AE improves
localization with
the accuracy of PD sources
trilateration technique [31–33]. localization with trilateration technique [31‒33].
The presented
The presented in in the
thearticle
articlenovel
novelon-line
on-linepartial
partialdischarge
discharge (PD)
(PD)monitoring
monitoring system
system is based
is basedon
simultaneous UHF, HF, and AE signal registration. It is a world first
on simultaneous UHF, HF, and AE signal registration. It is a world first system equipped in active system equipped in active
dielectric windows,
dielectric windows,which whichare are a new
a new concept
concept of combined
of combined acousticacoustic emission
emission and electromagnetic
and electromagnetic partial
partial discharge
discharge detectordetector
for power fortransformers.
power transformers. Each elaborated
Each elaborated prototype PD prototype
detector,PD detector,
i.e., meandered i.e.,
meandered planar inverted-F antenna, active dielectric window with
planar inverted-F antenna, active dielectric window with built-in ultrasonic transducer as well asbuilt-in ultrasonic transducer
as well as high-frequency
high-frequency current transformer
current transformer (HFCT) were (HFCT) wereoptimized
specially specially optimized to theofdetection
to the detection the partial of
the partial occurring
discharges discharges in occurring in oil-paper
oil-paper insulation. insulation.
Thanks to the Thanks to the implementation
implementation of new PD sensor of concepts,
new PD
sensor concepts, a significant improvement in sensitivity was achieved
a significant improvement in sensitivity was achieved for individual partial discharge detection for individual partial
discharge The
methods. detection
articlemethods.
shows that Thethearticle shows that
combination the combination
of different methods—in of different
this case,methods—in
electromagnetic this
case, electromagnetic and acoustic—increases the reliability
and acoustic—increases the reliability of partial discharge detection. of partial discharge detection.
The paper
The paperisisstructured
structured as as follows:
follows: thethe assumptions
assumptions and structure
and structure of the of the hybrid
hybrid monitoring monitoring
system
system
as as the
well as wellprocess
as the process
of designingof designing
and PD and PD detectors
detectors testing are testing are presented
presented in Sectionin Section 2. The
2. The results
results of the prototype system evaluation conducted on a new 330
of the prototype system evaluation conducted on a new 330 MVA large power transformer during MVA large power transformer
during
the the induced
induced voltage voltage
test, andtest, andexploited
on the on the exploited
31.5 MVA 31.5 MVAtransformer
power power transformer are discussed
are discussed in detail in in
detail in3.Section
Section General 3. General
conclusions conclusions andplans
and further further planstolinked
linked to the development
the development of the produced
of the produced system
system are included
are included in Section 4. in Section 4.

2. Design and Implementation of On-Line PD Monitoring System

2.1. General Structure of On-Line PD Monitoring System

In Figure 1, the schematic diagram of the general structure of thethe developed
developed on-line partial
discharges monitoring system is shown, whose unique feature is the possibility
possibility of simultaneous
simultaneous
registration of PD pulses with the use of three different diagnostic techniques,
techniques, i.e.,
i.e., high-frequency,
high-frequency,
ultra-high frequency,
frequency, and
and acoustic
acousticemission.
emission.

Figure
Figure 1. Schematic diagram
1. Schematic diagram ofof the
the general
general structure
structure of
of the
the developed
developed on-line
on-line partial
partial discharges
discharges
monitoring system for power transformers.
monitoring system for power transformers.

High-frequency current transformer (labeled in the system as sensor ISO-002-HF) and active
High-frequency current transformer (labeled in the system as sensor ISO-002-HF) and active
dielectric windows (ADW) with built-in acoustic emission sensor and mounted UHF antenna (the
dielectric windows (ADW) with built-in acoustic emission sensor and mounted UHF antenna (the set
set was labeled as sensor ISO-002-ADW/UHF) were employed as PD detectors. Each PD detector
was labeled as sensor ISO-002-ADW/UHF) were employed as PD detectors. Each PD detector has an
has an additional housing, in which there is a signal conditioning module and analog-to-digital
additional housing, in which there is a signal conditioning module and analog-to-digital (A/D) signal
(A/D) signal processing module. A module for fiber-optic signal transmission was included as well.
processing module. A module for fiber-optic signal transmission was included as well. This type of
Energies 2020, 13, 3271 4 of 37
Energies 2020, 13, x FOR PEER REVIEW 4 of 38

transmission allows to minimize

This type of transmission allows to the influence
minimize of external
the influence EMI occurring
of external on substations.
EMI occurring The
on substations.
transmitted
The transmitted digital data data
digital is received by the
is received bymulti-channel
the multi-channeldata concentrator labeled
data concentrator as MPD-001.
labeled It is
as MPD-001.
aItspecialized modular device, prepared for operation in a substation environment,
is a specialized modular device, prepared for operation in a substation environment, equipped with equipped with
multi-port
multi-port network
network communication
communication cards. cards. ItItperforms
performsthetheinitial
initialintegration
integrationof ofdata
dataframes
framesreceived
received
from
from many PD PD sensors
sensorsinto
intoa aform
form that
that facilitates
facilitates their
their transmission
transmission to server
to the the server and performing
and performing more
more advanced
advanced calculations
calculations (e.g., (e.g., phased-resolved
phased-resolved partial
partial discharge
discharge patterns
patterns ortrend
or data data trend analysis).
analysis).
The process of
The process ofdesigning
designingand and producing
producing PDPD sensors
sensors and and
mainmain elements
elements of the of the monitoring
monitoring system,
system, i.e., module for conditioning and digital processing of PD pulses,
i.e., module for conditioning and digital processing of PD pulses, data concentrator, and modules data concentrator, andof
modules of server
server software aresoftware
discussed areindiscussed in in
detail later detail later in the article.
the article.

2.2. Partial
2.2. PartialDischarge
DischargeDetectors
Detectors

2.2.1. Active
2.2.1. ActiveDielectric
DielectricWindow
Window
The active
The active dielectric
dielectric window
window(ADW),
(ADW),installed
installedininthe revision
the revision hole of the
hole power
of the powertransformer,
transformer,was
elaborated
was for the
elaborated fordetection of AE and
the detection of AEUHF andpulses
UHFgenerated by partialby
pulses generated discharges. The previously
partial discharges. The
conducted research demonstrated that ADW assures over five-times
previously conducted research demonstrated that ADW assures over five-times higher AE higher AE pulses detection
pulses
sensitivitysensitivity
detection comparedcompared
to commonly used contact
to commonly usedtransducer type PACtype
contact transducer R15PAC[33]. R15
It results
[33]. Itfrom the
results
fact that ADW is placed inside the power transformer tank and registers acoustic
from the fact that ADW is placed inside the power transformer tank and registers acoustic signal signal propagating
directly in oil,
propagating whereas
directly a conventional
in oil, AE sensor is
whereas a conventional AEmounted
sensor ison the outer
mounted on wall of thewall
the outer tank. For
of the
that reason,
tank. For thatitreason,
only registers
it only aregisters
small part of thepart
a small acoustic
of thewave
acousticenergy
wavegenerated by partial by
energy generated discharge.
partial
The remaining energy is lost in the result of the processes of the acoustic wave reflection
discharge. The remaining energy is lost in the result of the processes of the acoustic wave reflection and absorption
by the
and steel wallbyofthe
absorption thesteel
transformer
wall of thetank.
transformer tank.
The active dielectric window
The active dielectric window consists of consists of aa specially
specially formed
formed bowl-shaped
bowl-shaped ceramic
ceramic element,
element, in in
which all ultrasonic transducer components were built-in, i.e., piezoelectric crystal
which all ultrasonic transducer components were built-in, i.e., piezoelectric crystal with electrodes with electrodes and
electric
and wireswires
electric as well
as aswellmatching and backing
as matching layer (Figure
and backing 2). The2).
layer (Figure bowl-shape of the ADW
The bowl-shape of theenables
ADW
mounting-in of the UHF antenna and detection of radio signals generated
enables mounting-in of the UHF antenna and detection of radio signals generated by PDs. by PDs.

Technicaldrawing
Figure2.2.Technical
Figure drawingof
ofthe
theactive
activedielectric
dielectricwindow.
window.

The dielectric
The dielectricwindow
window waswas
produced from high-density
produced alumina ceramics,
from high-density alumina which is characterized
ceramics, which is
characterized by high thermal, mechanical, and chemical resistance (it does not oxidizenot
by high thermal, mechanical, and chemical resistance (it does not oxidize and does andreact
doeswith
not
mineral oils and esters). The main parameters of the used ceramic material are listed in Table
react with mineral oils and esters). The main parameters of the used ceramic material are listed in1.
Table 1.
Energies 2020,
Energies 2020,13,
13,3271
x FOR PEER REVIEW 55 of 38
of 37

Table 1. Properties of the high-strength aluminous porcelain used to construct the dielectric
window.
Table 1. Properties of the high-strength aluminous porcelain used to construct the dielectric window.

Parameter Parameter Symbol Symbol Value

Value Unit Unit
Apparent1 (bulk) density 1 ρb 2500 kg/m3
Apparent (bulk) density ρb 2500 kg/m3
Flexural strength 1 σ 140 MPa
Flexural strength 1 σ 140 MPa
Fracture
1 toughness 1 K K 2.0 MPa·m1/2
Fracture toughness 2.0 MPa·m1/2
Young’s modulus Young's
1 modulus 1 E E 100
100 GPa GPa
Abrasion resistance
Abrasion resistance (volume loss) (volume loss)
WV WV 172
172 mm3 mm3
Dielectric
Dielectric strength 1 strength 1 VB VB 20
20 kV/mm kV/mm
Resistivity 1 Resistivity 1 ρ ρ 10
1013
13 Ω·cm Ω·cm
Thermal shock
Thermal shock resistance 1 resistance T1 c Tc 150
150 K K
11 Minimum values.
Minimum values.

Regarding the
Regarding the specific
specific environment
environment of of the
the ultrasonic
ultrasonic transducer
transducer operation,
operation, whichwhich is is mineral
mineral oil,
oil,
the matching layer is a very important element. Its proper manufacturing
the matching layer is a very important element. Its proper manufacturing assures high transducer assures high transducer
sensitivity. So
sensitivity. So that
that the
theacoustic
acousticenergy
energytransmission
transmissionfrom frommineral
mineraloiloiltoto piezoelectric
piezoelectric crystal
crystalcould be
could
carried
be carriedoutout
lossless
lossless(without
(without reflections),
reflections),thetheacoustic
acoustic impedance
impedance of the
of thematching
matchinglayer should
layer shouldbe the
be
geometric mean of the mineral oil impedance (Z = 1.28 MRayl at
the geometric mean of the mineral oil impedance (Zoil = 1.28 MRayl at 25 C) and used piezoelectric
oil 25 °C)
◦ and used piezoelectric
ceramics type
ceramics type PZT-5A
PZT-5A(Z (ZPZT
PZT =
= 31
31 MRayl).
MRayl). The Thematching
matching layer layer was
was made
made of of epoxy/alumina composite
epoxy/alumina composite
with 0.3825 volume fraction of alumina, which allowed for obtaining
with 0.3825 volume fraction of alumina, which allowed for obtaining the required value of acoustic the required value of acoustic
impedance i.e., 6.3 MRayl. For the preparation of the composite, the
impedance i.e., 6.3 MRayl. For the preparation of the composite, the epoxy resin of low viscosityepoxy resin of low viscosity (200
mPa·s
(200 at 25at°C)
mPa·s 25 and high high
◦ C) and purity alumina
purity aluminapowderpowder with with
a grain thickness
a grain of 1 µm
thickness of 1 was
µm wasused.used.
The
backing layer was made of composite material as well. In this case,
The backing layer was made of composite material as well. In this case, epoxy/tungsten composite epoxy/tungsten composite with
0.08 volume
with 0.08 volumefraction of tungsten
fraction of acoustic
of tungsten impedance
of acoustic 4.56 MRayl
impedance 4.56 was
MRayl used.wasThe optimal
used. Thevalues
optimal of
acoustic
values of impedance and geometric
acoustic impedance dimensions
and geometric of all components
dimensions of the of
of all components ultrasonic transducer
the ultrasonic (PZT
transducer
crystal, matching, and backing layer), which affect the shape of the frequency
(PZT crystal, matching, and backing layer), which affect the shape of the frequency response curve were response curve were
designed based
designed based on on the
the simulations
simulations performed
performed with with the the model
model of of Krimholtz,
Krimholtz, Leedom,
Leedom, and and Matthaei
Matthaei
(KLM) [34]. In turn, proper proportions of epoxy, alumina, and tungsten
(KLM) [34]. In turn, proper proportions of epoxy, alumina, and tungsten for the elaborated composite for the elaborated composite
materials were
materials were selected
selected using
using thethe Devaney
Devaney and and Levine
Levine model, model, which
which is is based
based on on aa self-consistent
self-consistent
formulation of
formulation of multiple-scattering
multiple-scattering theory theory [35].
[35]. In
In the final effect,
the final effect, the
the transducer
transducer with with relatively
relatively flat
flat
frequency response in the range between 120 and 270 kHz and with first
frequency response in the range between 120 and 270 kHz and with first resonance frequency totaling resonance frequency totaling
134 kHz
134 kHz waswas elaborated.
elaborated. The The transducer
transducer frequency
frequency parameters
parameters are are aa compromise
compromise between between highhigh PD
PD
detection sensitivity
detection sensitivity and and the
the resistance
resistanceto tocore
coremagnetostriction
magnetostrictionnoise. noise.The The remaining
remaining parameters
parameters of
the elaborated ultrasonic transducer are listed in Table 2, whereas detailed
of the elaborated ultrasonic transducer are listed in Table 2, whereas detailed information on the information on the design,
modeling,
design, and production
modeling, processes
and production were discussed
processes were discussedin [33].in Figure 3 presents
[33]. Figure a photograph
3 presents of the
a photograph
front and back sides of the active dielectric window with visible components
of the front and back sides of the active dielectric window with visible components of the built-in of the built-in ultrasonic
transducer.
ultrasonic transducer.

Figure 3. Photographs
Figure 3. Photographs of
of the
the active
active dielectric
dielectric window
window produced.
produced.
Energies 2020, 13, 3271 6 of 37

Table 2. Parameters of piezoelectric transducer components.

Component Parameter Symbol Value Unit

Thickness d 5.05 mm
Density ρ 2079 kg/m3
Longitudinal velocity VL 3029 m/s
Matching layer 1 Shear velocity VS 1640 m/s
Bulk modulus K 11.62 GPa
Shear modulus G 5.59 GPa
Acoustic impedance ZM 6.3 MRayl
Thickness d 6.2 mm
Piezoelectric ceramics Radius r 6.35 mm
(PZT-5A disc) Area A 126.61 mm2
Acoustic impedance Z0 31.0 MRayl
Thickness d 8.8 mm
Density ρ 2546 kg/m3
Longitudinal velocity VL 1790 m/s
Backing layer 2 Shear velocity VS 845 m/s
Acoustic impedance ZB 4.56 MRayl
Bulk modulus K 5.74 GPa
Shear modulus G 1.82 GPa
1 epoxy/alumina composite with 0.3825 volume fraction of alumina; 2 epoxy/tungsten composite with 0.08 volume
fraction of tungsten.

2.2.2. Ultra-High Frequency Antenna

In the power transformers, the registration of the radio signals in the VHF/UHF band generated
by the partial discharges is carried out with the use of antennas mounted in special dielectric
windows [36,37] or antennas put in the tank interior through oil valves [38,39]. There are also
new solutions known, in which, antennas are placed outside the tank and detect the leakage of
electromagnetic waves through the gap (insulation gasket between the upper or lower ladle cover
and the main tank) [40] or bushing taps [41]. Antennas placed in the dielectric windows usually
allow to obtain the highest sensitivity in PD detection. The obstacle in the global popularization of
this solution is the fact that the new power transformers are still very rarely equipped with dielectric
windows. The factory installation of the interior UHF antennas takes place exclusively on demand of
the purchaser. Additionally, the exchange of the revision hatches in the transformer being exploited to
mount dielectric windows and antennas is an expensive, time-consuming, and complicated operation,
both regarding technical and logistic issues. Unfortunately, the remaining solutions, i.e., antennas
placed in the tank interior through oil valves and external antennas, regarding numerous limitations
are not a valuable alternative [39]. The first ones are simple monopole antennas of weak amplification
and a very narrow transmission band. Additionally, the shielding effect of the oil drain valve and
unfavorable placement inside the tank (usually it is a sidewall of the tank, near the bottom), have
a negative impact of their efficiency, because they may hamper the registration of the radio signals
generated by PDs [42]. In turn, the external antennas are vulnerable to the influence of external
electromagnetic disturbances, mainly corona from HV transmission lines. A significant problem is also
the fact that the signal registered by the antenna may have very low energy because of the EM wave
attenuation on the way from the PD source to the antenna [40]. To minimize the problem, it is necessary
to apply advanced algorithms of signals denoising, filtration, and recognition. It is worth mentioning
that only a small part of the useful signal gets out through gaps (if they are present in the given tank
construction at all) and bushing insulators—outside the transformer tank. Having considered all
advantages and disadvantages of the available methods of UHF signals registration, finally, it was
decided that for the needs of the monitoring system the antenna adjusted to installation in the dielectric
window would be elaborated. High EM signals detection sensitivity, which directly translates into
high effectiveness of the PD monitoring system operation, determined its choice.
Energies 2020, 13, x FOR PEER REVIEW 7 of 38
Energies 2020, 13, 3271 7 of 37

In the work of [21] the complex review of antennas applied for PD detection in the VHF/UHF
bandIn was
thepresented. Among
work of [21] the commercial
the complex solutions,applied
review of antennas circularforplate antenna (also
PD detection in thecalled UHF band
VHF/UHF disk
sensor) prevails, since it has a simple construction and because it may be easily adjusted to the
was presented. Among the commercial solutions, circular plate antenna (also called UHF disk sensor)
dimensions of the dielectric window, which usually is round-shaped. A disadvantage of this
prevails, since it has a simple construction and because it may be easily adjusted to the dimensions
construction is a narrow bandwidth and the possibility to regulate the resonant frequency practically
of the dielectric window, which usually is round-shaped. A disadvantage of this construction is a
only by changing the diameter d of the circular plate (d = λ/4 where λ is the wavelength). Therefore,
narrow bandwidth and the possibility to regulate the resonant frequency practically only by changing
in many research centers, the works on the elaboration of the antenna not only of compact dimensions
the diameter d of the circular plate (d = λ/4 where λ is the wavelength). Therefore, in many research
but also of high gain and wide bandwidth, which covers the frequency band of the EM signals
centers, the works on the elaboration of the antenna not only of compact dimensions but also of high
generated by PD in oil-paper insulation system are carried out. Among the most common wideband
gain and wide bandwidth, which covers the frequency band of the EM signals generated by PD in
antenna constructions proposed by researchers, dedicated to PD detection are meander-line antenna
oil-paper insulation system are carried out. Among the most common wideband antenna constructions
[43], logarithmic spiral antenna [44], different types of Archimedean spiral antennas [45,46], Vivaldi
proposed by researchers, dedicated to PD detection are meander-line antenna [43], logarithmic spiral
antenna [47], microstrip patch antennas [48‒50], bio-inspired antennas [51,52], and fractal antennas
antenna [44], different types of Archimedean spiral antennas [45,46], Vivaldi antenna [47], microstrip
such as Hilbert curve fractal antenna [53,54], H-fractal antenna [55], Peano fractal antenna [56], or
patch antennas [48–50], bio-inspired antennas [51,52], and fractal antennas such as Hilbert curve fractal
Minkowski fractal antenna [57].
antenna [53,54], H-fractal antenna [55], Peano fractal antenna [56], or Minkowski fractal antenna [57].
From the viewpoint of efficient PD detection in a power transformer, the key antenna parameter
From the viewpoint of efficient PD detection in a power transformer, the key antenna parameter is
is properly adjusted bandwidth. The research results presented in [58] showed that PDs in oil-
properly adjusted bandwidth. The research results presented in [58] showed that PDs in oil-immersed
immersed power transformers generate EM pulses of frequencies from 150 to 700 MHz, wherein most
power transformers generate EM pulses of frequencies from 150 to 700 MHz, wherein most of their
of their energy was transmitted in a band of lower frequencies (up to 400 MHz). In turn, in [59] the
energy was transmitted in a band of lower frequencies (up to 400 MHz). In turn, in [59] the authors
authors registered EM signals simultaneously with the use of three ultra-wideband antennas, which
registered EM signals simultaneously with the use of three ultra-wideband antennas, which jointly
jointly covered a wide measurement band from 20 MHz up to 18 GHz. The research results showed that
covered a wide measurement band from 20 MHz up to 18 GHz. The research results showed that
the main PD types in paper-oil insulation (PD in oil, PD in gas bubbles in oil, inter-turn discharges,
the main PD types in paper-oil insulation (PD in oil, PD in gas bubbles in oil, inter-turn discharges,
surface discharges and creeping sparks on pressboard) generate EM pulses, for which the majority
surface discharges and creeping sparks on pressboard) generate EM pulses, for which the majority of
of the energy is transmitted in the band from 250 to 390 MHz. For that reason, it was decided that the
the energy is transmitted in the band from 250 to 390 MHz. For that reason, it was decided that the
designed antenna would be optimized to operate in this band.
designed antenna would be optimized to operate in this band.
In search of the most efficient and scalable antenna design, the authors tested different
In search of the most efficient and scalable antenna design, the authors tested different constructions
constructions of wideband antennas, among others logarithmic spiral antenna, circular-cross
of wideband antennas, among others logarithmic spiral antenna, circular-cross antenna, H-fractal
antenna, H-fractal antenna, Hilbert curve fractal antenna, and broadband planar inverted-F antenna
antenna, Hilbert curve fractal antenna, and broadband planar inverted-F antenna (PIFA) [53,55].
(PIFA) [53,55]. The results of tests and simulations showed that the highest performance and
The results of tests and simulations showed that the highest performance and possibility of relatively
possibility of relatively easy bandwidth regulation is offered by the modified constructions of a
easy bandwidth regulation is offered by the modified constructions of a conventional PIFA. It is
conventional PIFA. It is currently the most popular antenna in the world applied in mobile devices
currently the most popular antenna in the world applied in mobile devices [60–62]. This is due to its
[60‒62]. This is due to its numerous advantages such as compact size compared to monopole antenna,
numerous advantages such as compact size compared to monopole antenna, high gain in both vertical
high gain in both vertical and horizontal states of polarization, close to omnidirectional radiation
and horizontal states of polarization, close to omnidirectional radiation pattern, and low manufacturing
pattern, and low manufacturing cost. Figure 4 presents a schematic diagram of the conventional
cost. Figure 4 presents a schematic diagram of the conventional PIFA, which consists of a ground
PIFA, which consists of a ground plane, a rectangular radiating element of length equal to the quarter-
plane, a rectangular radiating element of length equal to the quarter-wave, feeding pin and shorting
wave, feeding pin and shorting plate (or pin) to connect the ground plane with the top patch.
plate (or pin) to connect the ground plane with the top patch.

Figure 4. Schematic diagram of the conventional planar inverted-F antenna (PIFA).

Figure 4. Schematic diagram of the conventional planar inverted-F antenna (PIFA).
Resonant frequency fr of conventional PIFA may be approximately estimated with the use of
Resonant frequency fr of conventional PIFA may be approximately estimated with the use of
Equation (1), in which c is the speed of light in vacuum, εeff is the effective permittivity of the dielectric
Equation (1), in which c is the speed of light in vacuum, εeff is the effective permittivity of the dielectric
substrate, L1 and L2 are the width and length of the rectangular radiating element, H is the thickness of
substrate, L1 and L2 are the width and length of the rectangular radiating element, H is the thickness
Energies 2020, 13, x FOR PEER REVIEW 8 of 38
Energies 2020, 13, 3271 8 of 37
of the substrate, and W is the width of shorting plate, assuming that L1 > W > 0 [63,64]. In order to
more accurately estimate the resonant frequency of PIFA, some authors propose more complex
the substrate, and W is the width of shorting plate, assuming that L1 > W > 0 [63,64]. In order to more
formulas [65].
accurately estimate the resonant frequency of PIFA, some authors propose more complex formulas [65].
c
f= c (1)
fr = r 4√ εeff (L1 +L2 +H-W) (1)
4 εeff (L 1 +L2 +H − W )

The
The major
major limitation
limitation of of standard
standard PIFA
PIFA isis the
the narrow
narrow bandwidth.
bandwidth. Simple
Simple methods
methods for for widening
widening
the
the bandwidth of PIFA are based on reducing the size of the ground plane and the use of a substrate
bandwidth of PIFA are based on reducing the size of the ground plane and the use of a substrate
with the low
with the low value
value of of dielectric
dielectric constant
constant [64].
[64]. Another
Another technique
technique is is the
the employment
employment of of specially
specially
designed
designed slits in front patch [66–70], ground plane [71], or in both these structures [72–74]. A
slits in front patch [66‒70], ground plane [71], or in both these structures [72‒74]. A more
more
advanced method of bandwidth widening is based on the use of stacked patches,
advanced method of bandwidth widening is based on the use of stacked patches, often also equipped often also equipped
in
in slits
slits [75‒79].
[75–79]. Other
Other tuning
tuning techniques
techniques areare based
based on on varying
varying thethe position
position of of feeding
feeding pin, changing
pin, changing
the
the capacitance value of the shorting plate, or changing the widths of feed and shorting plates[80,81].
capacitance value of the shorting plate, or changing the widths of feed and shorting plates [80,81].
By
By applying
applying aa properly
properly selected
selected chip
chip resistor
resistor or
or chip
chip capacitor
capacitor load
load in
in place
place of
of the
the shorting
shorting plate
plate one
one
may
may reduce
reduce the
the resonant frequency of
resonant frequency of the
the antenna
antenna [82].[82].
For
For the
the needs
needs of of the monitoring system,
the monitoring system, it it was
was decided
decided to to apply
apply PIFA construction with
PIFA construction with the
the
meandered
meandered radiating
radiating structure loaded with
structure loaded with chip
chip resistor
resistor [83].
[83]. Meandering
Meandering the the radiating
radiating patch
patch allows
allows
to
to reduce
reduce thethe dimensions
dimensions of of the conventional PIFA
the conventional PIFA andand obtain
obtain large
large gain
gain [84–88].
[84‒88]. In turn, applying
In turn, applying
the chip resistor causes not only lowering of the resonant frequency but also
the chip resistor causes not only lowering of the resonant frequency but also widens the bandwidth widens the bandwidth
(even
(even several
several times
times compared
compared to to standard
standard PIFA)
PIFA) [83].
[83]. AA schematic
schematic diagram
diagram of of aa meandered
meandered PIFA PIFA isis
presented in Figure
presented in Figure 5. 5.

Figure 5. Schematic diagram of the meandered PIFA.

Figure 5. Schematic diagram of the meandered PIFA.
In the initial stage of designing, the necessity to adjust the antenna to the dielectric window
In
dimensionsthe initial
was takenstageinto
of designing, the necessity
account (Figure to adjust
2) of internal the antenna
diameter equal toto 15
thecmdielectric
and thewindow
need to
dimensions was taken
equip the antenna in ainto accounthousing.
protective (Figure 2)Since
of internal
the gain diameter
of PIFAequal to 15as
increases cmtheand the need
ground to
plane
equip the antenna in a protective housing. Since the gain of PIFA increases as
size increases [89], it was assumed that the largest possible diameter equaled 14 cm. The remaining the ground plane size
increases
dimensions [89], it was
of the assumed
prototype that thePIFA
meandered largest possible
(Table 3) were diameter
optimized equaled
through 14 iterative
cm. Thesimulations
remaining
dimensions
performed in ofCST
the prototype
Microwave meandered
Studio 2016 PIFA (Table Systèmes
(Dassault 3) were optimized through iterative
SE, Vélizy-Villacoublay, simulations
France).
performed in CST Microwave Studio 2016 (Dassault Systèmes SE, Vélizy-Villacoublay,
The results of the simulation proved that the application of chip-resistor in place of the shorting France).
The results
pin allows of thethe
to increase simulation proved
bandwidth that6),
(Figure the application
whereas of chip-resistor
by changing in place
the distance of thethe
between shorting
patch
pin
andallows
ground to plane
increase onethe bandwidth
may regulate(Figure 6), whereas
the resonant by changing
frequency the distance
of the antenna between
(Figure the patch
7). Satisfying
and ground
results plane onefor
were obtained may regulate the
chip-resistor withresonant
resistancefrequency
R = 5.4 of the antenna
Ω and distance (Figure 7). Satisfying
H = 5 mm, for which
results were obtained for chip-resistor with resistance R = 5.4 Ω and
the simulated voltage standing wave ratio was VSWR = 1.038 (S11 = -36.34 dB), resonantdistance H = 5 mm, for frequency
which the
simulated
fr = 347 MHz,voltage
andstanding
the −6 dB wave ratio waswas
bandwidth VSWR = 1.038The
73 MHz. (S11use
= -36.34 dB), resonant
of chip-resistor of frequency fr = 347
higher resistance
MHz, and the −6 dB bandwidth was 73 MHz. The use of chip-resistor of higher
allows to widen the bandwidth even more (e.g., for R = 7.2 Ω, the −6 dB bandwidth grew to 90 MHz). resistance allows to
widen the bandwidth
Unfortunately, it is at theeven more
expense of (e.g.,
VSWRfor R = 7.2
growth andΩ, the −6
a slight dB bandwidth
change of resonant grew to 90fr ,MHz).
frequency which
Unfortunately, it is at the expense of VSWR growth and a slight change of resonant
is shifted to the lower frequencies. In turn, change in distance H between the patch and ground plane frequency fr,
 Energies 2020, 13, x FOR PEER REVIEW 9 of 38

Energies 2020, 13, x FOR PEER REVIEW 9 of 38

which is shifted to the lower frequencies. In turn, change in distance H between the patch and ground
Energies 2020, 13, 3271 9 of 37
planeisinfluences
which shifted to thethe lower
resonant frequency
frequencies. Into a greater
turn, changeextent than the
in distance bandwidth.
H between The results
the patch of the
and ground
simulation
plane proved
influences the that for Hfrequency
resonant = 3 mm the to aresonant frequency
greater extent thanwas 319 MHz, and
the bandwidth. Thefor H = 6ofmm
results the it
increased
simulation
influences theto 357
proved MHz, wherein
that frequency
resonant −6
for H = 3 to dB
mm bandwidth
the resonant
a greater was, respectively,
frequency
extent than equal
was 319 MHz,
the bandwidth. to 47 and 56
and for
The results MHz.
H =simulation
of the 6 mm it
increased to 357
proved that H = 3 wherein
for MHz, mm the −6 dB bandwidth
resonant frequency was, respectively,
was fortoH47=and
equal
319 MHz, and 56 MHz.
6 mm it increased to
Table 3. Dimensions of the prototype meandered PIFA.
357 MHz, wherein −6 dB bandwidth was, respectively, equal to 47 and 56 MHz.
Table 3. Dimensions of the prototype meandered PIFA. Value
Parameter
Table 3. Dimensions Indication
of the prototype meandered PIFA. (mm)
Value
Parameter
Width of the patch Indication
L1 62.5
Parameter Indication (mm)
Value (mm)
Length of the
Width of the patch patch L1 L2 100
62.5
Width of the Length
patch of the slit L1 S1 50 62.5
Length
Length of the patch of the patch L L2 100 100
Widthofofthetheslit
slit 2
Length
Length of the slit S1 S1S2 502 50
Distance between
Width patch and ground plane S2H 25 2
Width of the slit of the slit S2
Diameter
Distancepatch of
between the circular ground plane D 514 5
Distance between and patch
ground and ground plane H
plane H
Distance
Diameter ofDiameter between
of the
the circular feeding
circular
ground point and resistor
ground plane
plane D D d 26.3 14
14
Distance between
Distancefeeding
betweenpoint point and resistor d
and resistor
feeding d 26.3 26.3

Figure 6. Simulated VSWR for meandered PIFA with various loading chip resistors and constant H =
Figure
5.0 6. Simulated VSWR for meandered PIFA with various loading chip resistors and constant
mm.
Figure 6. Simulated VSWR for meandered PIFA with various loading chip resistors and constant H =
H = 5.0 mm.
5.0 mm.

Figure 7. Simulated VSWR for prototype meandered PIFA with various distances H between patch
Figure 7. Simulated VSWR for prototype meandered PIFA with various distances H between patch
and ground plane and with the same loading chip resistor with resistance R = 5.4 Ω.
Figureground
and plane and
7. Simulated with
VSWR forthe same loading
prototype chip resistor
meandered with various
PIFA with resistance R = 5.4 Ω.
distances H between patch
and ground
Based plane
on the and with the
simulation same loading
results, chip resistor
the prototype with resistance
construction R = 5.4 Ω. PIFA of dimensions
of meandered
Based on the simulation results, the prototype construction of meandered PIFA of dimensions
presented in Table 3 was manufactured. The antenna was equipped in cylindrical protective housing
presented
Based in Table 3 was manufactured. The antenna was equipped in cylindrical protective housing
made fromon the simulation results,(PTFE)
polytetrafluoroethylene the prototype construction
and matched of meandered
to the cavity PIFA
of the active of dimensions
dielectric window.
made from
presented polytetrafluoroethylene
in Table 3 was manufactured. (PTFE)
The and matched
antenna to the cavity cylindrical
was equipped of the activeprotective
dielectrichousing
window.
The antenna with the active dielectric window, inspection hole cover,inand the module for conditioning
The
made antenna with the active
from polytetrafluoroethylene dielectric window, inspection hole cover, and the module for
and digital processing of PD pulses(PTFE) and matched
constitutes to the
a complete PDcavity
sensoroflabeled
the active dielectric window.
as ISO-002-ADW/UHF
The antenna
(Figure 8). with the active dielectric window, inspection hole cover, and the module for
Energies 2020, 13, x FOR PEER REVIEW 10 of 38
Energies 2020, 13, x FOR PEER REVIEW 10 of 38
conditioning and digital processing of PD pulses constitutes a complete PD sensor labeled as ISO-
conditioning
002-ADW/UHF and(Figure
digital8).
processing of PD pulses constitutes a complete PD sensor labeled as ISO-
Energies 2020, 13, 3271 10 of 37
002-ADW/UHF (Figure 8).

Figure 8. Components of
Figure of the
the ISO-002-ADW/UHF
ISO-002-ADW/UHF sensor: sensor:11==meandered PIFA,2 2= =
meanderedPIFA, N-type
N-type connector,
connector,3
Figure 8. Components
3= =cylindrical
cylindrical of the ISO-002-ADW/UHF
PTFE-housing,
PTFE-housing, 4 4= =active
activedielectricsensor:
dielectricwindow,1 = meandered
window, 55==enclosurePIFA,
enclosure for2partial
for = N-type
partial connector,
discharge (PD)3
=pulses
cylindrical
pulses PTFE-housing,
conditioning module 4and
module and = active dielectric window,
analog-to-digital
analog-to-digital processing5 =module,
processing enclosure
module,6 =6for partial
= fiber
fiber optic discharge
cable
optic (PD)
gland,
cable 7=
gland,
pulses
7cable conditioning
= cable gland
gland module
for power
for power and
supply,
supply, analog-to-digital
= shielding
8 = 8shielding tubetubeprocessing
for signal
for signal module,
cables,
cables, 6 = fiber optic
9 = inspection
9 = inspection cable
holehole gland,
cover.
cover. 7 =
cable gland for power supply, 8 = shielding tube for signal cables, 9 = inspection hole cover.
Having
Havinginstalled
installedthe antenna
the antenna in thein active dielectric
the active window,
dielectric the VSWR
window, thecharacteristic was measured
VSWR characteristic was
using Having
a vector
measured installed
using network the antenna
analyzer
a vector network in the
typeanalyzer active
KC901S+type dielectric
(Deepace,
KC901S+ window,
Guangdong, the VSWR
(Deepace,China). characteristic
China). was
The measurements
Guangdong, The
measured
showed using
that a
compared vector
to network
the simulationanalyzer results,typethe KC901S+
VSWR (Deepace,
curve is
measurements showed that compared to the simulation results, the VSWR curve is slightly shifted Guangdong,
slightly shifted China).
towards The
lower
measurements
frequencies
towards lower showed
(Figure 9). that
frequencies compared
It is(Figure
probably9). Itto
an isthe simulation
effect
probably of the results,
antenna
an effect theantenna
VSWR curve
surroundings
of the is slightly
influence,
surroundings shifted
i.e., metal
influence,
towards
transformerlower frequencies
tank and inspection
i.e., metal transformer (Figure
tank and hole 9). It is probably
cover, hole
inspection which an effect
werewhich
cover, of the
not taken antenna
wereinto surroundings
notaccount in the
taken into influence,
simulations.
account in the
i.e.,
The metal
simulations.transformer
measured Theresonant tank resonant
frequency
measured and inspection
of the antenna
frequency holeof
wascover,
equal
the which
to 329
antenna were
MHz
was not
(VSWR
equal taken
to = 1.13;
329 MHz 11 = −24.3
into Saccount
(VSWR in the
dB),
= 1.13;
simulations.
S11 =−6
and dB dB),
−24.3 The
bandwidth measured
and −6 equal resonant
to 73 MHz
dB bandwidth frequency
(from
equal to 280of
73 MHz the antenna
to 353(from
MHz). was equal
280 to 353 MHz).to 329 MHz (VSWR = 1.13;
S11 = −24.3 dB), and −6 dB bandwidth equal to 73 MHz (from 280 to 353 MHz).

Figure 9.
Figure 9. Simulated and measured VSWR of the prototype meandered PIFA.
PIFA.
Figure 9. Simulated and measured VSWR of the prototype meandered PIFA.
In order
order to check the performance of meandered PIFA, PIFA, comparative
comparative research
research waswas conducted,
conducted, in
which
which In aorder to checkUHF
commercial the performance
disk sensor
disk of meandered
sensor with
with aadiameter
diameter PIFA,
ofof1515comparative
cmand
cm research
andmeasured
measured was conducted,
resonant
resonant frequency
frequency infr
which
f=r = 475
475 a MHz
MHzcommercial
wasused
was usedUHF
asasa adisk sensor
reference
reference with a diameter
antenna.
antenna. Bothsensors
Both of 15were
sensors cm and
were measured
installed
installed anresonant
in an
in oil-filled frequency
transformer fr
=tank
475 MHz was
tank ofofdimensions used
dimensions 1200as
1200 a reference
× 800
× 800× 750 antenna.
mm,mm,
× 750 Both
equipped sensors were
in twoinceramic
equipped installed in
dielectric
two ceramic an oil-filled
windowswindows
dielectric transformer
and an electrode
and an
tank
system of dimensions
for generating 1200 ×
surface 800 × 750
dischargesmm, on equipped
the in
pressboard two ceramic
sample. dielectric
The
electrode system for generating surface discharges on the pressboard sample. The distance between distance windows
between and an
the PD
electrode
the PDand
source system
source for
and the
the central generating
central
point of thesurface
point discharges
of theinantenna,
antenna, on
the case of the
inboth pressboard
the case
sensors, sample.
of both sensors,
was the The
same wasdistance between
the same
and equal and
to about
the
equal
45 PDto
cm. source
Theabout and the The
45 were
pulses cm. central point
pulses
registered of athe
were
with antenna,with
registered
four-channel inoscilloscope
the case of type
a four-channel both Tektronix
sensors, was
oscilloscope MDO the3104
type same anda
Tektronix
with
equal
MDO to
sampling 3104about
with45a cm.
frequency of 5The
sampling pulses
GS/s. wereof
frequency
The level ofregistered
5 GS/s.
the withlevel
The
apparent a four-channel
charge ofofthe oscilloscope
theapparent
generated partialoftype
charge the Tektronix
generated
discharges was
MDO 3104 with
partial discharges
monitored a sampling
wasof
by the use frequency
monitored by the
the standard of 5 GS/s. The level
use of the standard
measurement of the apparent
measurement
set compliant with the charge
setIEC of
compliant the
with
60270 standard generated
the [90]
IEC
partial discharges
60270 standard
(Figure 10). was
[90] monitored
(Figure 10). by the use of the standard measurement set compliant with the IEC
60270 standard [90] (Figure 10).
of the surface discharges of the average apparent charge equal to qavg = 895 pC, the growth of the
amplitude of the registered UHF pulses compared to the disk antenna by 7.80 ± 0.9 times on average
was found. In turn, in case of high energy creeping sparks with an average apparent charge equal to
qavg = 5990 pC, lower growth of the amplitude, equal to 1.75 ± 0.7 times on average was observed. The
comparison of3271
Energies 2020, 13, the exemplary time waveforms of the PD pulses registered by the prototype antenna
11 of 37
and commercialFOR
Energies 2020, 13, x PEER
disk REVIEW
antenna is demonstrated in Figure 11. 11 of 38

The research results confirmed the high sensitivity of the prototype antenna. For the initial form
of the surface discharges of the average apparent charge equal to qavg = 895 pC, the growth of the
amplitude of the registered UHF pulses compared to the disk antenna by 7.80 ± 0.9 times on average
was found. In turn, in case of high energy creeping sparks with an average apparent charge equal to
qavg = 5990 pC, lower growth of the amplitude, equal to 1.75 ± 0.7 times on average was observed. The
comparison of the exemplary time waveforms of the PD pulses registered by the prototype antenna
and commercial disk antenna is demonstrated in Figure 11.

Schematic diagram
Figure 10. Schematic diagram of of the
the measurement
measurement setup
setup in
in which
which PD detection sensitivity
sensitivity of the
prototype UHF antenna was tested: U = high-voltage supply, Z = short-circuit
prototype UHF antenna was tested: U = high-voltage supply, Z = short-circuit current current limitinglimiting
resistor,
OSC = four-channel
resistor, oscilloscope,
OSC = four-channel TT = oil-filled
oscilloscope, transformer
TT = oil-filled transformer = electrode
tank, PDtank, system for
PD = electrode surface
system for
partial
surfacedischarge generation,
partial discharge PA = prototype
generation, meandered
PA = prototype meandered = commercial
PIFA, CA PIFA, UHF disk antenna
CA = commercial UHF diskas
antenna K = coupling
reference,asCreference, CK =capacitor,
couplingCD = coupling
capacitor, CD device (measuring
= coupling CC = connecting
impedance),impedance),
device (measuring CC =
cable, M = conventional
connecting partial discharge
cable, M = conventional partialdetector,
discharge = computer.
PCdetector, PC = computer.

The research results confirmed the high sensitivity of the prototype antenna. For the initial form
of theFigure
surface
10. discharges of the average
Schematic diagram apparent charge
of the measurement equal
setup in whichto PD = 895 pC,
qavgdetection the growth
sensitivity of the
of the
amplitude of the
prototype registered
UHF antenna UHF pulses U
was tested: compared to the disk
= high-voltage antenna
supply, by 7.80 ± 0.9
Z = short-circuit timeslimiting
current on average
was found.
resistor, OSC = four-channel oscilloscope, TT = oil-filled transformer tank, PD = electrode systemequal
In turn, in case of high energy creeping sparks with an average apparent charge for to
qavg =surface
5990 pC, lower growth of the amplitude, equal to 1.75 ± 0.7 times on average
partial discharge generation, PA = prototype meandered PIFA, CA = commercial UHF disk was observed.
The comparison of the exemplary
antenna as reference, time waveforms
CK = coupling of =the
capacitor, CD PD pulses
coupling registered
device by the
(measuring prototypeCC
impedance), antenna
=
and commercial disk antenna is demonstrated in Figure 11.
connecting cable, M = conventional partial discharge detector, PC = computer.

(b)
(a)

Figure 11. The comparison of the exemplary UHF pulses registered by the commercial UHF disk
antenna and prototype meandered PIFA for (a) surface PD of an apparent charge of q = 625 pC, (b)
creeping spark of apparent charge of q = 4670 pC.

2.2.3. High-Frequency Current Transformer

The high-frequency current transformer is an inductive PD detector, (b) which is usually installed
on the wire grounding the (a) investigated high voltage device. That wire is simultaneously a primary
winding of the current transformer. A typical sensor HFCT consists of toroidal ferrite core, on which
Figure 11.
Figure 11. The comparison of the exemplary UHF pulses registered by the commercial commercial UHFUHF disk
disk
from several to over a dozen turns (secondary winding) are wound-up. This winding is terminated
antenna and prototype
prototype meandered
meanderedPIFA PIFAfor for(a)
(a)surface
surfacePD
PDofofanan
apparent
apparent charge
chargeof of = 625
q =q625 pC,pC, (b)
with creeping
a low-inductance sense resistor. A current pulse of high frequency flowing through the
spark
(b) creeping of apparent
spark charge
of apparent charge = 4670
of qof= q4670 pC.pC.
grounding wire (primary winding) causes the occurrence of a time-varying magnetic field, which
passes
2.2.3. through the secondary
High-Frequency
2.2.3. High-Frequency Current winding and induces a voltage across the sense resistor. Figure 12
Current Transformer
Transformer
presents a model and equivalent circuit of the HFCT sensor, where i is the current flowing through
The
The high-frequency
high-frequency current transformer is an inductive PD
PD detector, which is is usually installed
the primary winding, R iscurrent transformer
the sense resistor, VisR an an inductive
induced voltage detector,
in the which
secondary usually
winding, installed
Lm is
on
on the wire grounding the investigated high voltage device. That wire is simultaneously a primary
the wire grounding the investigated high voltage device. That wire is simultaneously a primary
winding
winding of of the
the current
current transformer.
transformer. A A typical
typical sensor
sensor HFCT
HFCT consists
consists of
of toroidal
toroidal ferrite
ferrite core,
core, on
on which
which
from several to over a dozen turns (secondary winding) are wound-up. This
from several to over a dozen turns (secondary winding) are wound-up. This winding is terminated winding is terminated
with
with aalow-inductance
low-inductance sense resistor.
sense A current
resistor. pulse of
A current highof
pulse frequency flowing through
high frequency flowingthe grounding
through the
wire (primary winding) causes the occurrence of a time-varying magnetic field,
grounding wire (primary winding) causes the occurrence of a time-varying magnetic field, which which passes through
the secondary
passes through winding and induces
the secondary a voltage
winding and across
induces theasense resistor.
voltage acrossFigure 12 presents
the sense a model
resistor. Figureand 12
presents a model and equivalent circuit of the HFCT sensor, where i is the current flowing through
the primary winding, R is the sense resistor, VR an induced voltage in the secondary winding, Lm is
Energies 2020, 13, 3271 12 of 37

Energies 2020, 13, x FOR PEER REVIEW 12 of 38

equivalent circuit of the HFCT sensor, where i is the current flowing through the primary winding, R is
the sense
the magnetizing
resistor,inductance of the
VR an induced currentintransformer,
voltage the secondaryRc winding,
is the coreL parallel equivalent resistance, Ll
m is the magnetizing inductance
and
of Rscurrent
the are the leakage inductance
transformer, andcore
Rc is the series resistance
parallel of the secondary
equivalent resistance, winding,
Ll and Rsrespectively. The
are the leakage
remaining parasitic components included in the model are stray capacitance C and inductance
inductance and series resistance of the secondary winding, respectively. The remaining parasitic L R of

the sense resistor

components [91].in the model are stray capacitance C and inductance L of the sense resistor [91].
included R

Figure
Figure 12. High-frequency current
12. High-frequency current transformer
transformer model
model with
with parasitic
parasitic components.
components.

The HFCT sensors are applied in the diagnostics of high-voltage power cables, power generators,
The HFCT sensors are applied in the diagnostics of high-voltage power cables, power
metal-enclosed switchgears, and gas-insulated systems (GIS) [92–94]. In the case of power transformers,
generators, metal-enclosed switchgears, and gas-insulated systems (GIS) [92‒94]. In the case of power
the HFCT sensor may be installed on a grounding wire of the tank, on a grounded measuring tap
transformers, the HFCT sensor may be installed on a grounding wire of the tank, on a grounded
of the bushing insulator, or a grounding of the neutral point of the transformer winding. The last
measuring tap of the bushing insulator, or a grounding of the neutral point of the transformer
option assures the highest detection sensitivity of the PDs occurring in the power transformer main
winding. The last option assures the highest detection sensitivity of the PDs occurring in the power
insulation system.
transformer main insulation system.
The toroidal HFCT sensor core is most commonly made from soft ferrite. This material is
The toroidal HFCT sensor core is most commonly made from soft ferrite. This material is
ferromagnetic and consists of iron oxides mixed with oxides of manganese and zinc (MnZnFe2 O4 ),
ferromagnetic and consists of iron oxides mixed with oxides of manganese and zinc (MnZnFe2O4), or
or nickel and zinc (NiZnFe2 O4 ). Manganese-zinc soft ferrites have high relative permeability
nickel and zinc (NiZnFe2O4). Manganese-zinc soft ferrites have high relative permeability (µr ≈ 350–20,000
(µr ≈ 350–20,000 at 10 kHz), medium saturation flux density (Bs ≈ 320−545 mT at 1200 A/m),
at 10 kHz), medium saturation flux density (Bs ≈ 320−545 mT at 1200 A/m), operating frequency range
operating frequency range from a few kilohertz to 4 MHz, and low resistivity ρ (from 10 Ωm at 10 kHz
from a few kilohertz to 4 MHz, and low resistivity ρ (from 10 Ωm at 10 kHz to 0.01 Ωm at 100 MHz).
to 0.01 Ωm at 100 MHz). In turn, nickel-zinc soft ferrites are characterized by low relative permeability
In turn, nickel-zinc soft ferrites are characterized by low relative permeability (µr ≈ 15−2000 at 10 kHz),
(µr ≈ 15−2000 at 10 kHz), slightly lower saturation flux density (Bs ≈ 220−380 mT at 1200 A/m), much
slightly lower saturation flux density (Bs ≈ 220−380 mT at 1200 A/m), much higher maximum
higher maximum operating frequency (up to about 200 MHz), and much higher resistivity ρ (from
operating frequency (up to about 200 MHz), and much higher resistivity ρ (from 105 Ωm at 10 kHz to
10 3 Ωm at 10 kHz to 10 Ωm at 100 MHz) [95].
5 3
10 Ωm at 100 MHz) [95].
The clamp-type HFCT sensor devoted to PD monitoring in power transformers should have high
The clamp-type HFCT sensor devoted to PD monitoring in power transformers should have
sensitivity, wide bandwidth, closer to the constant line frequency response characteristics, and be
high sensitivity, wide bandwidth, closer to the constant line frequency response characteristics, and
equipped in enclosure shielding from external electromagnetic interferences. The sensitivity S of the
be equipped in enclosure shielding from external electromagnetic interferences. The sensitivity S of
HFCT sensor depends on resistance value R and the number of windings N, as shown in Equation (2).
the HFCT sensor depends on resistance value R and the number of windings N, as shown in Equation
(2). R
S= (2)
N
R
S= (2)
The lower −3 dB cutoff frequency of HFCT sensor N is given by

The lower −3 dB cutoff frequency of HFCT Rc ||(Rsensor

+ Rs ) is given
R by
fL = ≈ (3)
2πLm 2πLm
Rc ||(R+Rs ) R
where Lm is the magnetizing inductance. fL = The HFCT sensors≈ (3)
2πLm 2πLmare mostly loaded with non-inductive
resistor R = 50 Ω, in order to assure a good impedance matching with the transmission line (usually it
is a standard
where 50-ohm
Lm is the coaxial cable)
magnetizing and with
inductance. ThetheHFCT analog inputare
sensors of mostly
signal acquisition
loaded withunit. Therefore,
non-inductive
to achieve
resistor R =a 50
lowΩ,value of the
in order to cutoff
assurefrequency fL , the magnetizing
a good impedance matching withinductance should be line
the transmission possibly the
(usually
highest. The magnetizing inductance L on the secondary side is given by
it is a standard 50-ohm coaxial cable) and with the analog input of signal acquisition unit. Therefore,
m
to achieve a low value of the cutoff frequency fL, the magnetizing inductance should be possibly the
µr µ0 N2 Ac
highest. The magnetizing inductance Lm on the
Lm = secondary side is given by (4)
lc
µr µ0 N2 Ac
Lm = (4)
lc
Energies 2020, 13, 3271 13 of 37

where µ0 is the vacuum permeability (magnetic constant), Ac is the core cross-sectional area, and lc
is the toroidal mean length. Equation (4) shows that the high value of the magnetizing inductance
Lm may be obtained the easiest using a large number of turns N. Unfortunately, with increasing the
number of turns, the sensor sensitivity is reduced. The other way is a choice of a core made of a
material with high relative permeability µr (e.g., MnZn soft ferrites) and possibly the smallest toroidal
mean length lc [95,96].
In turn, the higher –3 dB cutoff frequency of the HFCT sensor mainly depends on stray capacitance
C and the leakage inductance Ll (Equation (5)). To obtain a high value of fH , both parameters should
be reduced.
1
fH = r (5)
2
 L 2
l
2π (RC) + Rc

The stray capacitance C consists of several capacitive couplings, which are present in high-frequency
current transformer, i.e., turn-to-turn capacitance, winding-to-core capacitance, winding-to-shielded
enclosure capacitance, and primary-to-secondary winding capacitance. The stray capacitance can be
minimized by reducing the number of turns N, increasing the insulation layer of core and secondary
winding (using a material with low dielectric constant εr ), increasing the distance between turns, and
using a Faraday electrostatic shield [95].
The reduction of the number of turns N is also favorable from the viewpoint of leakage inductance
Ll minimizing. Additionally, the secondary winding should be made wide and have a small thickness.
As one may notice, the factors which have the greatest influence on the sensitivity and width of
the HFCT sensor band, are the number of windings N and the type of material, which the toroidal
core is made from. The cores made from MnZn and NiZn soft ferrites were selected for the tests.
The physical parameters of the materials are listed in Table 4.

Table 4. Specifications of soft ferrite materials used in the study.

Ferrite Type Parameter Symbol Value Unit Conditions

Initial permeability µi 900 ± 20% − 25 ◦ C;≤ 10 kHz; 0.25 mT
Amplitude permeability µa ≈ 1700 − 100 ◦ C; 25 kHz; 200 mT
Magnetic flux density B ≈ 410 mT 25 ◦ C; 10 kHz; 1200 A/m
Specific power loss Pv ≈ 130 kW/m3 100 ◦ C; 1 MHz; 30 mT
MnZn
Loss factor tan n/a − −
Resistivity ρ ≈ 10 Ωm DC; 25 ◦ C
Curie temperature Tc ≥ 220 ◦C −
Density D ≈ 4700 kg/m3 −
Initial permeability µi 125 ± 20% − 25 ◦ C; ≤ 10 kHz; 0.25 mT
Amplitude permeability µa n/a − −
Magnetic flux density B ≈ 380 mT 25 ◦ C; 10 kHz; 3000 A/m
Specific power loss Pv n/a − −
NiZn
Loss factor tan ≤ 80 × 10-6 − 25 ◦ C; 3 MHz; 0.25 mT
Resistivity ρ ≈ 105 Ωm DC; 25 ◦ C
Curie temperature Tc ≥ 350 ◦C −
Density D ≈ 4500 kg/m3 −

The dimensions of the applied ferrite cores were selected so that, after equipping them in shielding
enclosure, they could be easily installed on the wire grounding the transformer tank or the neutral
point of the star windings. Since the width of the grounding wire/rail of power transformer usually
does not exceed 40 mm, it was decided to choose toroidal cores with an internal diameter equal to
65 mm.
The external diameter MnZn ferrite core was 107 mm, and its height was 18 mm. For NiZn
ferrite core they were, respectively, 103 and 16 mm. To reduce the skin effect, for manufacturing of the
secondary winding, the high-frequency Litz-wire with a diameter of 0.48 mm and external insulation
with a thickness of 0.02 mm made from natural silk of dielectric constant εr = 2.6 was used.
 Energies 2020, 13, x FOR PEER REVIEW 14 of 38

Energies 2020, 13, 3271 14 of 37

secondary winding, the high-frequency Litz-wire with a diameter of 0.48 mm and external insulation
with a thickness of 0.02 mm made from natural silk of dielectric constant εr = 2.6 was used.
Litz-wire consists
Litz-wire consists ofof 920
920 strands
strands insulated
insulated electrically
electrically from
fromeacheachother,
other,wherein
whereinaadiameter
diameterofofa
asingle
singlewire
wireis is0.016
0.016mm mm (AWG54).
(AWG54). TheThecores were
cores splitsplit
were in half using
in half a diamond
using cutting
a diamond disc.disc.
cutting The
secondary winding with a counter-wound compensation turn was applied
The secondary winding with a counter-wound compensation turn was applied to one of the core to one of the core halves.
To reduce
halves. the stray
To reduce capacitance
the stray (or more
capacitance precisely,
(or more the
precisely, theturn-to-turn
turn-to-turncapacitances),
capacitances), the distance
the distance
betweenindividual
between individualturns turnswas
wasmaximized.
maximized. TheThe enclosure
enclosure ofof HFCT
HFCT sensor
sensorwaswas made
madefrom
fromaluminum
aluminum
withaathickness
with thicknessof of22mmmmtotoattenuate
attenuateradiofrequency
radiofrequencyelectromagnetic
electromagneticinterferences
interferences and
and improved
improved
performance in
performance in ananelectrically
electrically noisy
noisy substation
substation environment.
environment. It It was
was equipped
equipped with
with aa 11 mm
mm slit
slit for
for
better magnetic coupling between primary and secondary winding [97]. The
better magnetic coupling between primary and secondary winding [97]. The sensor HFCT housing sensor HFCT housing
and the
and the housing
housing of of the
the conditioning
conditioning and
and A/D
A/D processing
processing module
module werewere integrated
integrated and
and jointly
jointly are
are aa
completesensor
complete sensorISO-002-HF
ISO-002-HFwith withthe
thefunction
functionofoffiber
fiberoptic
opticsignal
signaltransmission
transmission(Figure
(Figure13).
13).

Energies 2020, 13, x FOR PEER REVIEW 15 of 38

visible for sensors with MnZn soft ferrite core, whose magnetizing inductance is high regarding the
high value of permeability µ at low frequencies. For comparison, the value fL for current transformer
with Figure
MnZn 13. ferrite core and 10 turns is lower than 100 kHz, in turn, for a current transformer with
The components of sensor ISO-002-HF: 1 = MnZn ferrite core with secondary winding,
NiZn2Figure
ferrite core with the same
= insulation, 3components
13. The = N-type ofnumber ofhigh-frequency
turns, fL 1equals
sensor4ISO-002-HF:
connector, = = MnZn 450
current kHz.
ferrite Forwith
core
transformera lower number
secondary
(HFCT) of
winding,
sensor turns,
housing,2 = the
5insulation,
differences
= slitare 3 = N-type
in even more
metallic connector,
clear.
housing, For 4 = high-frequency
instance,
6 = adjustable fLlatch, 7 = current
for HFCT sensortransformer
waterproof with MnZn
enclosure (HFCT)
ferrite sensor
core and
for electronics, housing,
8N 3 5is=also
= =cable
very lowslit in
andmetallic
equals housing,
250 6
kHz. = adjustable
The current latch,
gland for power supply, 9 = fiber optic cable gland. 7 = waterproof
transformer of theenclosure
same for electronics,
construction, 8
but = cable
with gland
the ferrite
for power supply, 9 = fiber optic cable gland.
core NiZn, has fL equal to even 3 MHz. It should be emphasized that for the HFCT sensors devoted
to PDThe investigation
detection, the more on important
the influence of the number
parameter than theoffL turns N and
is higher material
−3 dB cutoff of the usedfHcore
frequency , since on
The investigation
the frequency
the discharges response
in oil-paperon theinsulation
of the influence emit
prototype of thehigh-frequency
HFCT number
sensorsof wasturns N and mainly
performed
waves material
using the inofthe
the used
rangecore
measurement on
setup
between the
frequency
presented
approx. 2 andresponse
in Figure
20 MHz of [98].
14. the
Theprototype
measurement
In HFCT
this case, is sensors
it setup
more was performed
consisted
favorable of the using
oscilloscope
to apply a lower theTektronix
measurement
number MDO
of turns.setup
3104 It
presented
with the
allows in
tosampling Figure
obtain high 14.
frequency The measurement
of 5 GS/s,
sensitivity 1 GHz
S of the setup
HFCT consisted
bandwidth,
sensor and of
and high the oscilloscope
built-in signal
value Tektronix
of generator.
the parameter MDO
The sinusoidal
fH, which3104
with the
signal
improves wasthe sampling
generated
performancefrequency
in theof the of
frequency 5 range
sensor GS/s,
in the 1 range
fromGHz100ofbandwidth,
kHz andwith
to 30frequencies
higher MHz built-in
a(>step signal generator.
frequency
1 MHz). One of 100
may The
kHz.
notice
insinusoidal
It allowed
Figure 14, signal
to cover
that thewas generated
the positive
full rangeeffect in the frequency
of high-frequency
resulting from (HF) range from
band, i.e.,
reduction 100
of 3–30 kHz
MHz. The
the number to 30 MHz
of current with
turns isflow more a step
in for
the
frequency
primary
sensors with ofthe
winding100core
kHz.made
was It allowed
forced to
by using
from NiZncover
a 50 the
Ω
soft full range
since of
noninductive
ferrite high-frequency
resistor
permeability of (HF)
RL inµ series.
the The band,
same
applied i.e., 3‒30 MHz.
resistor
material was
is a
The current
connected in flow in
parallel the
with primary
the winding
secondary was
winding forced
of the
practically constant value in the wide range between 1 MHz and approx. 50 MHz. by
HFCT using a 50
sensor. Ω noninductive resistor R L in

series. The same resistor was connected in parallel with the secondary winding of the HFCT sensor.
The measurement of the frequency response was automatically realized by the application
implemented in LabVIEW environment and launched on a computer connected to the oscilloscope.
The individual points of the frequency response curve FRHFCT(f) were calculated as

VR (𝑓)
𝐹𝑅 (𝑓)=20∙log , (6)
VGEN (𝑓)

where VR(f) is the RMS value of voltage measured on the output of the HFCT sensor at frequency f,
VGEN(f) is the RMS value of voltage measured on the output of signal generator at frequency f.
Figure 15 presents the frequency response curves of the prototype HFCT sensors. As one may
notice, simultaneously with the increase in the number of turns, the sensor sensitivity decreases. It
results from the fact that the values of that parameter depend directly proportional to the load
Figure14.
Figure Schematicdiagram
14.Schematic diagram ofthethe measurement setup for measuring the frequency response of
resistance R (which in this caseofwas measurement
constant andsetup forto
equal measuring the inversely
50 Ω) and frequency response of the
proportional to the
the HFCT
HFCTofsensor;sensor; OSC = oscilloscope type Tektronix MDO3104, GEN = built-in arbitrary function
number turns OSC = oscilloscope
N. Increasing type Tektronix
the number MDO3104,
of turns raises theGEN = built-in arbitrary
magnetizing inductancefunction
Lm, which
generator, PC ==computer,
generator, PC HFCT==high-frequency
computer,HFCT high-frequencycurrent
current transformerunder
undertest.
test.
causes the reduction of lower −3 dB cutoff frequency fL oftransformer
the HFCT sensor. This effect is even more
Energies 2020,
Energies 2020, 13,
13, 3271
x FOR PEER REVIEW 15
15 of
of 38
37

visible for sensors with MnZn soft ferrite core, whose magnetizing inductance is high regarding the
high The
valuemeasurement
of permeabilityof µ
the
at frequency response
low frequencies. For was automatically
comparison, realized
the value by the transformer
fL for current application
implemented in LabVIEW environment and launched on a computer connected to the
with MnZn ferrite core and 10 turns is lower than 100 kHz, in turn, for a current transformer withoscilloscope.
The
NiZnindividual points
ferrite core with of
thethe frequency
same numberresponse
of turns,curve FRHFCT
fL equals 450(f) were
kHz. Forcalculated as
a lower number of turns, the
differences are even more clear. For instance, fL for HFCT sensor with MnZn ferrite core and N = 3 is also
VR ( f )
very low and equals 250 kHz. The current FRHFCT (transformer
f )= 20· log of the same , construction, but with the ferrite (6)
core NiZn, has fL equal to even 3 MHz. It should be emphasized VGEN ( f ) that for the HFCT sensors devoted
to PD detection,
where the
VR (f ) is the RMSmore important
value parameter
of voltage measured than
onthe
thefLoutput
is higher −3 dB
of the cutoff
HFCT frequency
sensor fH, sincef,
at frequency
the discharges in oil-paper insulation emit high-frequency waves mainly
VGEN (f ) is the RMS value of voltage measured on the output of signal generator at frequency f. in the range between
approx. 2 and
Figure 20 MHz the
15 presents [98].frequency
In this case, it is more
response favorable
curves to apply aHFCT
of the prototype lowersensors.
numberAs of one
turns.may It
allows to
notice, obtain high sensitivity
simultaneously S of the in
with the increase HFCT sensor and
the number high value
of turns, of thesensitivity
the sensor parameter decreases.
fH, which
It results from the fact that the values of that parameter depend directly proportional to thenotice
improves the performance of the sensor in the range of higher frequencies (> 1 MHz). One may load
in Figure 14, that the positive effect resulting from reduction of the number
resistance R (which in this case was constant and equal to 50 Ω) and inversely proportional to the of turns is more for
number of turns N. Increasing the number of turns raises the magnetizing inductance Lm , which causesa
sensors with the core made from NiZn soft ferrite since permeability µ of the applied material is
practically
the reductionconstant
of lowervalue in cutoff
−3 dB the wide range between
frequency 1 MHzsensor.
f of the HFCT and approx. 50 MHz.
This effect is even more visible for
L
sensors with MnZn soft ferrite core, whose magnetizing inductance is high regarding the high value of
permeability µ at low frequencies. For comparison, the value fL for current transformer with MnZn
ferrite core and 10 turns is lower than 100 kHz, in turn, for a current transformer with NiZn ferrite
core with the same number of turns, fL equals 450 kHz. For a lower number of turns, the differences
are even more clear. For instance, fL for HFCT sensor with MnZn ferrite core and N = 3 is also very
low and equals 250 kHz. The current transformer of the same construction, but with the ferrite core
NiZn, has fL equal to even 3 MHz. It should be emphasized that for the HFCT sensors devoted to PD
detection, the more important parameter than the fL is higher −3 dB cutoff frequency fH , since the
discharges in oil-paper insulation emit high-frequency waves mainly in the range between approx.
2 and 20 MHz [98]. In this case, it is more favorable to apply a lower number of turns. It allows to
obtain high sensitivity S of the HFCT sensor and high value of the parameter fH , which improves the
performance of the sensor in the range of higher frequencies (> 1 MHz). One may notice in Figure 14,
Figure
that the 14. Schematic
positive diagram of
effect resulting the measurement
from reduction ofsetup for measuring
the number the is
of turns frequency response
more for sensorsofwith
the the
core made from NiZn soft ferrite since permeability µ of the applied material is a practically constant
HFCT sensor; OSC = oscilloscope type Tektronix MDO3104, GEN = built-in arbitrary function
valuegenerator, PC =range
in the wide computer, HFCT
between = high-frequency
1 MHz and approx.current transformer under test.
50 MHz.

(a) (b)

Figure 15.
Figure 15. Frequency
Frequencyresponse
responseofofthe
theprototype
prototypeHFCT
HFCTsensors
sensorsdepending
depending onon
the number
the numberof of
turns N
turns
and the core material used: (a) MnZn soft ferrite, (b) NiZn soft ferrite, compared with frequency
N and the core material used: (a) MnZn soft ferrite, (b) NiZn soft ferrite, compared with frequency
response of
response of commercial
commercial HFCT
HFCT sensors
sensors (RFCT-4
(RFCT-4 and
and HFCT
HFCT 140/100).
140/100).

Based on the analysis of the measured frequency response curves, for further research, the two
prototype HFCT
prototype HFCT sensors
sensors were
were selected.
selected. The first of the chosen sensors was manufactured from ferrite
core MnZn with the three-turn secondary winding. The main advantages of that construction are a
Energies 2020, 13, 3271 16 of 37

Energies 2020, 13, x FOR PEER REVIEW 16 of 38

Energies 2020, 13, x FOR PEER REVIEW 16 of 38
core MnZn with the three-turn secondary winding. The main advantages of that construction are a
very
veryhigh
highsensitivity
sensitivity(higher
(higherthan
thanpopular
popularcommercial
commercialsensors),
sensors),wide
widebandwidth
bandwidth(250 (250kHz–25.5
kHz–25.5MHz),
MHz),
very high sensitivity (higher than popular commercial sensors), wide bandwidth (250 kHz–25.5 MHz),
and
and relatively
relatively flat
flatfrequency
frequency response
response curve.
curve. The
The second
second from
from the
the selected
selected constructions
constructions was
was built
builtfrom
from
and relatively flat frequency response curve. The second from the selected constructions was built
ferrite
ferrite core NiZn
core core
NiZnNiZnwith four
withwith turns.
four four
turns. It is characterized
It is characterized by slightly
by slightly lower
lower sensitivity
sensitivity (≤ 2.67
(≤ 2.67 dB) than
dB)dB)
than the
the
from ferrite turns. It is characterized by slightly lower sensitivity (≤2.67 than
transformer
transformer with MnZn
withwithMnZn core, but
core,core, its
but but frequency
its frequency response
response curve
curve is almost
is almost ideally
ideally flat in a
flat inflatwide
a wide band
band
the transformer MnZn its frequency response curve is almost ideally in a wide
between
between 1.8
1.8 and
and 30
30 MHz.
MHz. In
In the
the first
first stage
stage of the
of of
the laboratory tests, measurements of the impulse
band between 1.8 and 30 MHz. In the first stage thelaboratory
laboratorytests,tests,measurements
measurements of the impulse impulse
response
responseof ofthe
theselected prototype
selectedprototype
prototypeHFCT HFCT sensors
HFCTsensors were
sensorswere conducted.
wereconducted.
conducted.For For this
Forthis purpose,
purpose,aaastandard
thispurpose, standard LDC-
standardLDC-5
LDC-
response of the selected
55 PD
PD calibrator
calibrator (dDoble
(dDoble Engineering
Engineering Company,
Company, Marlborough,
Marlborough, USA)
USA) was
was used,
used, which
which allowed
allowed to
to
PD calibrator (dDoble Engineering Company, Marlborough, USA) was used, which allowed to inject
inject pulses
inject pulses with an apparent charge of 500 pC and duration of 50 ns into the primary circuit. The
pulses with anwith an apparent
apparent charge ofcharge
500 pCofand 500duration
pC and of duration of the
50 ns into 50 ns into the
primary primary
circuit. circuit. The
The waveforms
waveforms
waveforms of
of the
the calibration
calibration pulses induced
induced in the
the secondary circuit of the
the HFCT sensor were
of the calibration pulses inducedpulses
in the secondary incircuit secondary
of the HFCT circuit
sensorofwere HFCT
registeredsensor
usingwere
an
registered
registered using
using anan oscilloscope
oscilloscope (Figure
(Figure 16).
16).
oscilloscope (Figure 16).

(a) (b)
(a) (b)
Figure 16. Examination of HFCT sensors: (a) Schematic diagram of the measurement setup for
Figure 16.
Figure 16. Examination
Examination of
of HFCT
HFCT sensors:
sensors: (a) Schematic diagram
(a) Schematic diagram ofof the
the measurement
measurement setup
setup for
for
measuring the impulse response of the HFCT sensor using standard PD calibrator: OSC = oscilloscope
measuring the impulse response of the HFCT sensor using standard PD calibrator: OSC = oscilloscope
measuring the impulse response of the HFCT sensor using standard PD calibrator: OSC = oscilloscope
Tektronix MDO3104, HFCT = high-frequency current transformer under test, CAL = partial discharge
TektronixMDO3104,
Tektronix MDO3104,HFCTHFCT==high-frequency
high-frequencycurrent
currenttransformer
transformerunder
undertest, CAL==partial
test,CAL partialdischarge
discharge
calibrator type Doble LDC-5; (b) time waveform of PD calibration pulse with an apparent charge of
calibrator type
calibrator type Doble
Doble LDC-5;
LDC-5; (b)
(b)time
timewaveform
waveform ofof PD
PDcalibration
calibration pulse
pulse with
with an
anapparent
apparentcharge
chargeofof
500 pC.
500pC.
500 pC.

Figure
Figure17
Figure 17shows
17 showsthe
shows therecorded
the recordedwaveforms
recorded waveformsof
waveforms ofthe
of theimpulse
the impulseresponse
impulse responseof
response ofthe
of thetested
the testedHFCT
tested HFCTsensors.
HFCT sensors.
sensors.
The
The highest
highest detection
detection sensitivity
sensitivity of
of the
the pulse
pulse of
of apparent
apparent charge
charge 500
500 pC
pC was
was shown
shown
The highest detection sensitivity of the pulse of apparent charge 500 pC was shown by the prototype by
by the
the prototype
prototype
current
current transformer
current transformer made
transformer madefrom
made from MnZn
from MnZnferrite
MnZn ferrite core
ferrite corewith
core with three
withthree turns.
turns. At
threeturns. At the
At the output
the output of
output of this
of thiscurrent
this current
current
transformer,
transformer, the
transformer,the pulses
thepulses of amplitude
pulsesofofamplitude A max = 146 mV were registered, which means that its sensitivity
amplitudeAAmaxmax =
= 146 mV were registered, which which means
means that
that its
its sensitivity
sensitivity
was
wasby
was byabout
by about70%
about 70% higher
70%higher than
higherthan the
thanthe prototype
theprototype sensor
prototypesensor made
sensormade from
madefrom NiZn
fromNiZn ferrite
NiZnferrite core
ferritecore (A
core(A(Amax = 86 mV) and
max== 86
86 mV)
mV) and
and
max
commercial
commercial HFCT
commercialHFCT sensor
HFCTsensor type
sensortype RFCT-4
typeRFCT-4
RFCT-4(A(A max = 88 mV).
(Amax = 88 mV).
= 88
max

Figure 17. Impulse response of the prototype (MnZn ferrite core with three turns, NiZn ferrite core with
Figure 17. Impulse response of the prototype (MnZn ferrite core with three turns, NiZn ferrite core
Figure
four 17. Impulse
turns) responseHFCT
and commercial of thesensors
prototype (MnZn
to partial ferrite core
discharge with three
calibration turns,with
impulse NiZnanferrite core
apparent
with four turns) and commercial HFCT sensors to partial discharge calibration impulse with an
with four
charge turns)
of 500 pC. and commercial HFCT sensors to partial discharge calibration impulse with an
apparent charge of 500 pC.
apparent charge of 500 pC.

Time
Time parameters
parameters of
of the
the pulses
pulses generated
generated by
by standard
standard PDPD calibrator
calibrator are
are optimized
optimized for
for the
the
conventional
conventional method IEC 60270, in which case the registration of PD pulses is performed in aa
method IEC 60270, in which case the registration of PD pulses is performed in
Energies 2020, 13, 3271 17 of 37

Time parameters of the pulses generated by standard PD calibrator are optimized for the
Energies 2020, 13, x FOR PEER REVIEW 17 of 38
conventional method IEC 60270, in which case the registration of PD pulses is performed in a relatively
low frequency
relatively low band (100–500
frequency bandkHz). For that
(100‒500 kHz). reason, the reason,
For that tests with thethe usewith
tests of thethe
PDuse calibrator
of the PDdo
not allow unambiguous statement that the prototype HFCT sensor
calibrator do not allow unambiguous statement that the prototype HFCT sensor with core MnZn is with core MnZn is an optimal
construction of the detector
an optimal construction of of
thehigh-frequency PD pulses. Nevertheless,
detector of high-frequency PD pulses. measurements
Nevertheless, with the use of
measurements
calibrator
with the use showed that dueshowed
of calibrator to the lowest
that duelower −3 dB
to the cutoff
lowest frequency
lower L = 250
−3 dB(fcutoff kHz) among
frequency (fL =the
250tested
kHz)
HFCT sensors, it may be successfully employed both in the standardized
among the tested HFCT sensors, it may be successfully employed both in the standardized frequency frequency range according to
IEC
range60270 as welltoasIEC
according at higher
60270 as frequencies
well as at up to 25.5
higher MHz.
frequencies up to 25.5 MHz.
In the second stage of the laboratory research,
In the second stage of the laboratory research, the detection the detection sensitivity
sensitivity of of
PDPD pulses
pulses generated
generated in
oil-paper insulation was tested, which were simultaneously registered
in oil-paper insulation was tested, which were simultaneously registered by four HFCT sensors (two by four HFCT sensors (two
prototype
prototype and and twotwocommercial).
commercial).The The research
research waswas conducted
conducted with with the of
the use use of a transformer
a transformer tank
tank filled
filled
with mineral oil. The electrode systems devoted to generating PD in oil and surface PD on the
with mineral oil. The electrode systems devoted to generating PD in oil and surface PD on the
pressboard
pressboard sample
sample were were put
put inside
inside the
the tank.
tank. Detailed
Detailed information
information concerning
concerning the the geometry
geometry of of the
the
applied
applied electrode
electrode systems
systems waswas presented
presented in in [24]. The HFCT
[24]. The HFCT sensors
sensors were
were installed
installed on grounding wire
on grounding wire
going out of the tank through the ceramic bushing insulator. The PD
going out of the tank through the ceramic bushing insulator. The PD pulses were registered with pulses were registered with
the
the
use use
of aoffour-channel
a four-channel oscilloscope
oscilloscope Tektronix
Tektronix MDO3104
MDO3104 withthe
with thesampling
samplingfrequency
frequencyofof 55 GS/s.
GS/s.
Additionally,
Additionally, the the standard
standard IECIEC 60270
60270 measurement
measurement setup setup waswas applied
applied to to control
control the
the level
level ofof the
the PD
PD
apparent
apparent charge (Figure 18). Discharges in oil of apparent charge in the range between about 200 and
charge (Figure 18). Discharges in oil of apparent charge in the range between about 200 and
850
850 pC
pC were
weregenerated
generatedatat2929kV. kV.InIn
turn,
turn,thethesurface discharges
surface dischargeshadhada significantly
a significantlyhigher intensity
higher and
intensity
apparent charge (q = 450–5100 pC). This type of discharge was generated
and apparent charge (q = 450–5100 pC). This type of discharge was generated by a voltage of 15.4 kV.by a voltage of 15.4 kV.

measurement setup for generating partial discharges in oil-paper

Figure 18. Schematic diagram of the measurement
insulation and assessing the sensitivity of high-frequency PD pulses detection by prototype and
commercial HFCT
HFCT sensors:
sensors:U U = high-voltage
= high-voltage supply,
supply, Z = short-circuit
Z = short-circuit current
current limiting
limiting resistor,resistor,
OSC =
OSC = four-channel oscilloscope, HFCT = high-frequency current transformers under
four-channel oscilloscope, HFCT = high-frequency current transformers under test, TT test, TT == oil-filled
oil-filled
transformer tank, PD
transformer tank, PD==electrode
electrodesystem
systemforfor partial
partial discharge
discharge generation,
generation, = coupling
C =Ccoupling capacitor,
capacitor, CD
CD = coupling
= coupling device
device (measuring
(measuring impedance),
impedance), CC = connecting
CC = connecting cable, Mcable, M = conventional
= conventional partial
partial discharge
discharge detector,
detector, PC PC = computer.
= computer.

Figure
Figure 1919 presents
presents typical
typical partial
partial discharge
discharge time
time waveforms
waveforms and and power
power spectral
spectral density
density (PSD)
(PSD)
characteristics registered with the use of HFCT sensors. The results of frequency
characteristics registered with the use of HFCT sensors. The results of frequency analysis analysis showed that
showed
PDs
that inPDsoilin
generated pulsespulses
oil generated of highoffrequency in the band
high frequency in thebetween 12.4 and
band between 18.3
12.4 MHz,
and 18.3whereas surface
MHz, whereas
PD generated pulses in a narrow band between 10.8 and 11.6 MHz. The
surface PD generated pulses in a narrow band between 10.8 and 11.6 MHz. The PD detectionPD detection sensitivity of
both prototypes was higher than commercial sensors. In the case of discharges in
sensitivity of both prototypes was higher than commercial sensors. In the case of discharges in oil, oil, the highest
sensitivity
the highestwas observed
sensitivity wasforobserved
the prototype
for theHFCT sensor
prototype with sensor
HFCT four turns
withandfourNiZn
turnssoft
andferrite
NiZncore.
soft
Its
ferrite core. Its sensitivity was 36% higher on average than the sensors from MnZn soft ferriteturns.
sensitivity was 36% higher on average than the sensors from MnZn soft ferrite with three with
On
threetheturns.
other On
hand,
thethe construction
other hand, thebased on the MnZn
construction basedcore showed
on the MnZn higher sensitivity
core showed in thesensitivity
higher detection
of surface discharges on the pressboard sample. The average growth in sensitivity
in the detection of surface discharges on the pressboard sample. The average growth in sensitivity compared to the
prototype with NiZn core was 46%. The detection possibility of surface discharges
compared to the prototype with NiZn core was 46%. The detection possibility of surface discharges is very important
from
is very the viewpoint
important of the
from the performance andperformance
viewpoint of the effectivenessandof the monitoring
effectiveness ofsystem operation.
the monitoring It is
system
operation. It is one of the most energetic and destructive types of partial discharges. Therefore, for
the designed monitoring system, finally, the HFCT sensor with MnZn core was selected.
Energies 2020, 13, 3271 18 of 37

one of the most energetic and destructive types of partial discharges. Therefore, for the designed
monitoring system,
Energies 2020, 13, finally,
x FOR PEER the HFCT sensor with MnZn core was selected.
REVIEW 18 of 38

(a) (b)

Figure 19.
Figure 19. Comparison of typical partial discharge time waveforms and power spectral density (PSD)
characteristics registered for (a) PD
characteristics PD in
in oil,
oil, and
and (b)
(b) surface
surface PD
PD with
with the
the use
use of
of selected
selected prototype
prototype and
and
commercial (RFCT-4,
commercial (RFCT-4,HFCT
HFCT140/100)
140/100)HFCT
HFCTsensors.
sensors.

2.3. Module for Conditioning and Analog-to-Digital Signal Processing of PD Pulses

2.3. Module for Conditioning and Analog-to-Digital Signal Processing of PD Pulses
The schematic diagram of electronic modules for conditioning and analog-to-digital processing of
The schematic diagram of electronic modules for conditioning and analog-to-digital processing
PD pulses is presented in Figure 20. Conditioning of the UHF signals is realized in an electronic circuit
of PD pulses is presented in Figure 20. Conditioning of the UHF signals is realized in an electronic
consisting of two low noise RF amplifiers basing on a chipset SPF5043Z of maximal amplification of
circuit consisting of two low noise RF amplifiers basing on a chipset SPF5043Z of maximal
G = 18 dB at 900 MHz, tunable bandpass filter, and a system of peak detector based on multistage
amplification of G = 18 dB at 900 MHz, tunable bandpass filter, and a system of peak detector based
demodulating logarithmic amplifier type AD8313. Application of a peak detector circuit (also called
on multistage demodulating logarithmic amplifier type AD8313. Application of a peak detector
frequency converter or envelope detector) is a common practice in the case of the registration of UHF
circuit (also called frequency converter or envelope detector) is a common practice in the case of the
pulses because it allows to significantly reduce the sampling frequency [99,100]. Due to that, it is
registration of UHF pulses because it allows to significantly reduce the sampling frequency [99,100].
possible to process the PD pulses and determine their parameters in real-time. Unfortunately, this
Due to that, it is possible to process the PD pulses and determine their parameters in real-time.
is done at the expense of losing information on time-domain parameters (e.g., rise and fall time) of
Unfortunately, this is done at the expense of losing information on time-domain parameters (e.g., rise
individual pulses. The AE signals conditioning module consists of a low-noise preamplifier (with 40 dB
and fall time) of individual pulses. The AE signals conditioning module consists of a low-noise
gain) based on high-speed instrumentation amplifier AD8421BRZ, active bandpass filter (20–500 kHz)
preamplifier (with 40 dB gain) based on high-speed instrumentation amplifier AD8421BRZ, active
based on Sallen–Key architecture and voltage follower (based on AD813ARZ op-amp) to buffer the
bandpass filter (20‒500 kHz) based on Sallen–Key architecture and voltage follower (based on
output. Since the PD pulses registered with the use of sensitive HFCT sensor may reach the amplitude
AD813ARZ op-amp) to buffer the output. Since the PD pulses registered with the use of sensitive
of even a few volts, it was decided to give up on the dedicated conditioning system and to process the
HFCT sensor may reach the amplitude of even a few volts, it was decided to give up on the dedicated
conditioning system and to process the raw HF signals. At the input of the analog-to-digital signal
processing module, there is A/D converter type AD9266BCPZ-40 with 40 MS/s sampling rate and 16-
bit resolution. To each PD sensor, a separate A/D converter was allocated.
Energies 2020, 13, x FOR PEER REVIEW 19 of 38

Energies 2020, 13, 3271 19 of 37

raw HF signals. At the input of the analog-to-digital signal processing module, there is A/D converter
type AD9266BCPZ-40 with 40 MS/s sampling rate and 16-bit resolution. To each PD sensor, a separate
Energies
A/D 2020, 13, xwas
converter FORallocated.
PEER REVIEW 19 of 38

Figure 20. Schematic diagram of PD signal conditioning modules and A/D signal processing module
and 3D drawing showing the mechanics of these modules: A = signal conditioning boards, B = analog-
to-digital signal processing board, C = internal power supply module.

Figure 21 shows exemplary time waveforms of conditioned AE and UHF pulses and raw HF
pulses registered during laboratory tests of the signal conditioning modules and A/D signal
processing module. During the tests, in the oil-filled transformer tank, the surface discharges on a
20. Schematic
Figure sample
pressboard diagram of PD
were generated. signal conditioning modules and A/D signal processing
Figure 20. Schematic diagram of Their apparent
PD signal charge was
conditioning in
modules the range
and A/Dbetween 400 and
signal processing 5300
module pC. As
one canmodule and 3D drawing showing the mechanics of these modules: A = signal conditioning boards,
andsee in Figure
3D drawing 21a, the
showing theapplication
mechanics ofof themodules:
these peak detector circuit
A = signal at a relatively
conditioning boards, Blow sampling
= analog-
B = analog-to-digital signal processing board, C = internal power supply module.
frequency equal
to-digital to 40
signal MS/s allows
processing to Ceffectively
board, detect supply
= internal power the UHF pulses. A comparative analysis of the
module.
resultsFigure 21 shows exemplary time waveforms of conditionedmodule
obtained with the HF method demonstrated that the UHF AE anddetected all PDand
UHF pulses pulses
rawand HF
it Figure 21
confirmed the shows
high exemplaryoftime
sensitivity the waveformssensor.
ISO-002-HF of conditioned
Despite AEresignation
the and UHF pulses
from and
the HF raw HF
signal
pulses registered during laboratory tests of the signal conditioning modules and A/D signal processing
pulses
conditioningregistered
module,during
pulses laboratory tests of the signal conditioning modules and A/D signal
module. During the tests, in registered
the oil-filledwith the high-frequency
transformer tank, the current
surface transformer
discharges on had on average
a pressboard
processing
1.8 times module.
higher During than
amplitude the tests,
the in the oil-filled
strongly amplified transformer
UHF tank,The
pulses. theplacement
surface discharges
of theone on a
HFCT
sample were generated. Their apparent charge was in the range between 400 and 5300 pC. As can
pressboard
sensor on sample
the werewire
ground generated.
of the Their apparent
electrode charge
system, was inthe
nearby the range
PD between
source, 400 and had
certainly 5300apC. As
high
see in Figure 21a, the application of the peak detector circuit at a relatively low sampling frequency
one can
influence see in
onMS/s Figure
obtaining 21a, the application of the peak detector circuit at a relatively low sampling
equal to 40 allowssuch a good result.
to effectively detect the UHF pulses. A comparative analysis of the results
frequency
In the equal
case tothe
of 40 MS/s
AE allows to
method, effectively
the pulses ofthatdetect the UHF pulses.
comparable A amplitude
comparative analysis of the
obtained with the HF method demonstrated the UHF to the UHF
module detected all PDwere registered.
pulses and it
results
In obtained
Figure 21b with the HF method demonstrated that the UHF module detected all PD pulses and
confirmed thethe
high visible time differences
sensitivity (T = 272sensor.
of the ISO-002-HF µs) in UHF andthe
Despite AEresignation
signal arrival
from result
the HF from the
signal
it confirmed
different the high
velocity of EMsensitivity
wave of the ISO-002-HF
propagation (approx. sensor.
2/3 Despite
speed of the in
light resignation
vacuum) fromacoustic
and the HF signal
wave
conditioning module, pulses registered with the high-frequency current transformer had on average
conditioning
in module,
m/s). pulses registered with the high-frequency current transformer had oncm average
1.8oil (1400‒1500
times higher amplitude The than
piezoelectric transducer
the strongly amplified built
UHF in pulses.
the ADW Thewas placed about
placement of the 39.5
HFCT sensor from
1.8
the times
PDgroundhigher
source, wire amplitude
which than the strongly amplified UHF pulses. The placement of the HFCT
on the of means that system,
the electrode acoustic wavethe
nearby velocity in thecertainly
PD source, transformer
had atank
highfilled with oil
influence on
sensor
was 1452onm/s.
the ground wire of the electrode system, nearby the PD source, certainly had a high
obtaining such a good result.
influence on obtaining such a good result.
In the case of the AE method, the pulses of comparable to the UHF amplitude were registered.
In Figure 21b the visible time differences (T = 272 µs) in UHF and AE signal arrival result from the
different velocity of EM wave propagation (approx. 2/3 speed of light in vacuum) and acoustic wave
in oil (1400‒1500 m/s). The piezoelectric transducer built in the ADW was placed about 39.5 cm from
the PD source, which means that the acoustic wave velocity in the transformer tank filled with oil
was 1452 m/s.

(a) (b)

Figure
Figure 21.
21. Exemplary
Exemplarytime
timewaveforms
waveforms registered
registered during
during laboratory
laboratory tests
tests of
of the
the PD
PD signal
signal conditioning
conditioning
modules: (a) conditioned
modules: (a) conditionedUHF
UHFPD PDpulses
pulses
andand
rawraw
HFHF
PD PD pulses,
pulses, (b) conditioned
(b) conditioned singlesingle
UHF UHF PD
PD pulse
pulse and conditioned acoustic emission (AE)
and conditioned acoustic emission (AE) PD pulses.PD pulses.

In the case of the AE method, the pulses of comparable to the UHF amplitude were registered.
In Figure 21b the visible time differences (T = 272 µs) in UHF and AE signal arrival result from the
(a)
different velocity of EM wave propagation (approx. 2/3 speed of light(b)
in vacuum) and acoustic wave in
Figure 21. Exemplary time waveforms registered during laboratory tests of the PD signal conditioning
modules: (a) conditioned UHF PD pulses and raw HF PD pulses, (b) conditioned single UHF PD
pulse and conditioned acoustic emission (AE) PD pulses.
Energies 2020, 13, 3271 20 of 37

oil (1400–1500 m/s). The piezoelectric transducer built in the ADW was placed about 39.5 cm from the
PD source, which means that the acoustic wave velocity in the transformer tank filled with oil was
1452 m/s.
Digital signals from the A/D converter then go to the FPGA (Field-Programmable Gate Array) type
Intel 10M50DAF256I7G with 256 MB cache memory type DDR2. In that logic circuit, two operating
modes were implemented: basic (monitoring mode) and extended (diagnostic mode).
In the monitoring mode, for HF and UHF method, separate data frames of duration time 55.5 µs
were analyzed. Such data frame length was selected for two reasons. First, it is the time corresponding
to the angle 1◦ of voltage period (20 ms per 360◦ which is approximately 55.5 µs per 1◦ ), which enables
determination of phase-resolved partial discharge patterns (PRPD). Second, the duration time of a
single UHF PD pulse usually amounts to less than 2 µs. It allows to minimize the risk that the UHF PD
pulse would not be detected by the monitoring system. In the case of the AE method, the data frames
are aggregated in one-second records. It is caused by the fact that the AE pulses of the high-energy
partial discharges (e.g., creeping sparks or interturn discharges), which are registered with resonant
piezoelectric transducer may have a time duration of even over a dozen milliseconds [24].
The algorithm implemented in the FPGA for each data record (55 µs for the HF and UHF method,
and 1 s for the AE method) determines the following parameters:

• the number of pulses detected (n),

• the maximal amplitude of the pulses (Amax ),
• the maximal energy of the pulses (Emax ),
• the average amplitude of the pulses (Amean ),
• the average energy of the pulses (Emean ).

In the diagnostic mode, the FPGA does not do the calculations. Data frames are sent via a
fiber-optic Ethernet cable to the multi-channel data concentrator, and afterward to the monitoring
system server. The 1-Gigabit Ethernet physical layer (PHY) was made using the Marvell Alaska
88E1512 chip. That system supports Synchronous Ethernet (SyncE) and Precise timing control protocol
(PTP) time stamping, which is based on IEEE1588 [101] and IEEE802.1AS [102] standards. The 88E1512
system on the first side is connected using the RGMII interface to the FPGA chip and on the second it
is connected using SERDES interface to the small form-factor pluggable (SFP) optical transceiver.

2.4. Multi-Channel Data Concentrator Module

Measurement data is sent to the multi-channel data concentrator (labeled as MPD-001 module)
using protocol elaborated based on the so-called Process Bus (PB) according to the guidelines of the
international substation automation standard IEC 61850-9-2 [103]. As it was mentioned before, the
fiber-optic cables allow to effectively reduce the influence of external electromagnetic interferences
occurring on the substation. The schematic diagram and photograph of the MPD-001 module are
demonstrated in Figure 22.
The data into the MPD-001 module are entered through an integrated, specialized MSO-001
network card containing up to eight fiber-optic interfaces (ISO). The MSO-001 network card transmits
the measurement data via the PCX/1 bus to the PJC-900 module based on the E660T ATOM processor.
Individual data streams are decoded by the FPGA type CYCLONE V implemented in the PJC-900
module. The FPGA also supports the following external network transmission channels:

• the basic channel of the Ethernet network in standard FX1000 used for data transmission, warning
signals, download the parameterization from the server,
• configuration channel in TX100 standard dedicated to programming, local parameterization,
and service,
• synchronizing channel in 1PPS or IRIGB standard.
 Measurement data is sent to the multi-channel data concentrator (labeled as MPD-001 module)
using protocol elaborated based on the so-called Process Bus (PB) according to the guidelines of the
international substation automation standard IEC 61850-9-2 [103]. As it was mentioned before, the
fiber-optic cables allow to effectively reduce the influence of external electromagnetic interferences
occurring on the substation. The schematic diagram and photograph of the MPD-001 module are
Energies 2020, 13, 3271 21 of 37
demonstrated in Figure 22.

Figure Schematic
22.Schematic
Figure 22. diagram
diagram andand photograph
photograph of developed
of developed multi-channel
multi-channel data concentrator
data concentrator module
module
MPD-001.MPD-001.

The PCX/2 bus is used to transfer data streams from FPGA to the embedded PC (EPC) computer
module. As with the signal processing module, MPD-001 can work in the monitoring or diagnostic
mode. In the monitoring mode, based on the data sent from the analog-to-digital signal processing
module, the embedded PC calculates for three different time intervals (one minute, fifteen minutes, and
one hour) parameters such as the number of PD pulses, maximum and average pulse amplitude, as
well as maximum and average pulse energy. In diagnostic mode, the computer prepares data for PRPD
patterns. For this purpose, it aggregates 1-s data frames with a duration of 55.5 µs. The data prepared
in this way is sent to the server. If necessary, data from the MPD-001 module can be transferred to
other local or remote SCADA systems using the DNP-3.0 or IEC 61850 protocol.
The module MPD-001 is powered with the use of the 220 V AC/DC power supply type MZA-502,
in which the voltage level is supervised by the PJC-900 module.
After manufacturing the complete prototype of MPD-001 module, the EMC properties of the
housing ports and power ports were tested in accordance with IEC 60255-26 standard [104].

2.5. Modules of Server Software

In the firmware of the monitoring system server, the following software modules were
implemented: virtual data concentrator (VDC), analytic module, the module of graphic data
presentation, and reporting module. In the monitoring mode, the VDC module receives from
the MPD-001 concentrator notifications of the PD pulse occurrence. This causes that information about
the event with calculated partial discharge parameters (number of pulses, their energy, and amplitude)
is saved to the database on the server. In turn, in the diagnostic mode, VDC enables direct access to the
transmitted data frames, including the preview in real-time the PD waveforms of the registered in
particular channels. This functionality also enables gathering the samples to perform further, more
advanced analysis, including the elaboration of PRPD patterns.
Based on the data gathered in the database parameters, the analytical module regularly checks
if the threshold values of the PD parameters have not been exceeded. Additionally, using the least
square method, a linear trend estimation is conducted for the data from the last week and the last
month. All information is shared with the user through the module of the graphic data presentation,
which displays in the application window, i.e.,

• the number of detected PD pulses and the average value of their energies and amplitudes,
• the result of self-diagnosis of measurement channels,
• warning or alarm state after exceeding the threshold values for the sum of registered pulses, their
average energy, and amplitude for various time intervals: 15 min, 60 min, 24 h, and 30 days,
• the warning or alarm state when the values of PD parameters collected during last week and last
month show a rising trend.
Energies 2020, 13, x FOR PEER REVIEW 22 of 38
Energies 2020, 13, 3271 22 of 37
images are created based on the 20 ms aggregation of data frames from HFCT and UHF sensors and
voltage sine wave registered by the current/voltage transformer (CVT) on the substation.
The functions of the reporting module enable elaborating summary reports or details for the PD
The system also provides the possibility to wirelessly measure the high voltage in a power
parameters registered in the database and events description.
transformer, which is realized with the use of a special capacity probe. The structure of the probe and
As it was mentioned before, the system in the diagnostic mode functions as a multichannel partial
assumptions of the measuring method is discussed in detail in [105]. The signal from the probe is
discharges detector allowing to observe the PD waveforms or PRPD pattern images. PRPD images are
registered with an additional, A/D converter of sampling frequency equal to 4 MS/s dedicated
created based on the 20 ms aggregation of data frames from HFCT and UHF sensors and voltage sine
exclusively for this task. This converter is placed in the circuit of every analog-to-digital signal
wave registered by the current/voltage transformer (CVT) on the substation.
processing module.
The system also provides the possibility to wirelessly measure the high voltage in a power
The angular resolution of the created PRPD patterns is 1°, whereas the amplitude resolution
transformer, which is realized with the use of a special capacity probe. The structure of the probe
equals 16 bits for the output voltage range equal to 0–3.6 V. The software function responsible for
and assumptions of the measuring method is discussed in detail in [105]. The signal from the probe
preparing PRPD patterns was tested in the laboratory conditions. For this purpose, the comparative
is registered with an additional, A/D converter of sampling frequency equal to 4 MS/s dedicated
research, during which PRPD patterns were simultaneously registered with a standard partial
exclusively for this task. This converter is placed in the circuit of every analog-to-digital signal
discharges detector type Doble PD-Smart and monitoring system under test. A schematic diagram of
processing module.
the measurement setup was presented in Figure 23.
The angular resolution of the created PRPD patterns is 1◦ , whereas the amplitude resolution equals
The PRPD images obtained during the research were the effect of about 60–90 min registrations
16 bits for the output voltage range equal to 0–3.6 V. The software function responsible for preparing
of the partial discharges pulses. In this case, instead of pressboard samples, glass samples with a
PRPD patterns was tested in the laboratory conditions. For this purpose, the comparative research,
thickness of 6 mm were used. This change was caused by the fact that the pressboard samples quickly
during which PRPD patterns were simultaneously registered with a standard partial discharges detector
degraded under the influence of the action of high-intensity surface discharges with the apparent
type Doble PD-Smart and monitoring system under test. A schematic diagram of the measurement
charge of a few nC. This, in turn, led to PD extinction or breakdown.
setup was presented in Figure 23.

Figure 23. Schematic

Figure 23. Schematic diagram
diagram ofof the
the measurement
measurement setup
setup in
in which
which the
the monitoring
monitoring system
system software
software
module for performing PRPD patterns was tested: MS =
module for performing PRPD patterns was tested: MS = partial discharge monitoring system,system,
partial discharge monitoring MPD =
MPD = multichannel
multichannel signal concentrator,
signal concentrator, S == server,
S = server, PC PC U
computer, ==computer, = high-voltage
U supply,
high-voltage supply,
Z = short-circuit
Z = short-circuit current limiting resistor, TT = oil-filled transformer tank; PD = electrode system for
current limiting resistor, TT = oil-filled transformer tank; PD = electrode system for partial discharge
partial discharge generation, C = coupling capacitor, CD = coupling device (measuring impedance),
generation, C = coupling capacitor, CD = coupling device (measuring impedance), CC = connecting
CC = connecting cable, M = conventional partial discharge measuring instrument.
cable, M = conventional partial discharge measuring instrument.

The PRPD images obtained during the research were the effect of about 60–90 min registrations
In Figure 24 the exemplary results of the 80-min registration of the surface discharges of average
of the partial discharges pulses. In this case, instead of pressboard samples, glass samples with a
apparent charge qmean = 283 pC and maximal qmax = 5702 pC are presented. A typical PRPD distribution
thickness of 6 mm were used. This change was caused by the fact that the pressboard samples quickly
for surface discharges was obtained. The most PD pulses were registered in the first (0–90°) and third
degraded under the influence of the action of high-intensity surface discharges with the apparent
quarter (180–270°) of a sine wave of test voltage.
charge of a few nC. This, in turn, led to PD extinction or breakdown.
In Figure 24 the exemplary results of the 80-min registration of the surface discharges of average
apparent charge qmean = 283 pC and maximal qmax = 5702 pC are presented. A typical PRPD distribution
for surface discharges was obtained. The most PD pulses were registered in the first (0–90◦ ) and third
quarter (180–270◦ ) of a sine wave of test voltage.
Energies 2020, 13,
Energies 2020, 13, 3271
x FOR PEER REVIEW 23 of
23 of 37
38
Energies 2020, 13, x FOR PEER REVIEW 23 of 38

(a)
(a)

(b)
(b)
Figure 24. Measurement results obtained with conventional electric method: (a) instantaneous value
of apparent charge q, (b) PRPD pattern.
Figure 24.
24. Measurement results obtained with
with conventional
conventional electric
electric method:
method: (a) instantaneous value
apparent
of apparent charge
charge
The results q, (b)
q, (b) PRPD
obtained PRPD pattern.
pattern.
by the prototype monitoring system (Figure 25) confirmed the properly
implemented functions of PRPD pattern creation. In this case, also most of the pulses were registered
The
in theresults
first andobtained by the
third quarters prototype
prototype
of the monitoring
voltage sine systemcompared
wave. However, (Figure 25) confirmed
to the IEC 60270the properly
method
implemented
implemented functions
functions of
of PRPD
PRPD pattern
pattern creation.
creation. In this case, also most of the
(where the number of pulses did not depend clearly on the voltage polarization), both pulses were registered
in theelectromagnetic
first and thirdmethods
quartersregistered
of the voltage sine wave.
more pulses However,
of positive compared to the IEC 60270 method
polarization.
(where
(where the thenumber
number
Currently, of pulses
there did conducted
ofis work
pulses notdid
depend
notonclearly
depend on clearly
equipment the
of voltage
on polarization),
voltageinboth
the module
the analytical the electromagnetic
polarization),
function of both
automatic
methods recognition
registered
electromagnetic more of
methods PD type
pulses based
moreonpulses
of positive
registered the analysis
polarization. of the polarization.
of positive PRPD patterns.
Currently, there is work conducted on equipment of the analytical module in the function of
automatic recognition of PD type based on the analysis of the PRPD patterns.

(a) (b)

FigureFigure 25. PRPD

25. PRPD patterns
patterns obtained
obtained inina amonitoring
monitoring system
system for
forpartial
partialdischarges
dischargespulses registered
pulses in in a
registered
a measurement channel: (a) of high frequency, (b) of ultra-high frequency.
measurement channel: (a) of high frequency, (b) of ultra-high frequency.
(a) (b)
3. Tests of the Developed PD Monitoring System on Power Transformers
Currently, there is work conducted on equipment of the analytical module in the function of
Figure 25. PRPD patterns obtained in a monitoring system for partial discharges pulses registered in
automatic
3.1. recognition
Test of PDSystem
of the Monitoring type based
During on the analysis
Induced Voltage of the
Test PRPD
of 330 MVApatterns.
Power Transformer
a measurement channel: (a) of high frequency, (b) of ultra-high frequency.
After
3. Tests of the successfully
Developed finished laboratory
PD Monitoring tests, iton
System was decided
Power to check the effectiveness of the
Transformers
3. Tests of the
system Developed
operation PDpower
on a real Monitoring System
transformer. Firston Power
such Transformers
measurement was carried out during the
induced
3.1. Test of thevoltage test with
Monitoring partial
System discharge
During measurement
Induced Voltage Test(IVPD) on aMV
of 330 power transformer
A Power of 330 MVA.
Transformer
3.1. Test
The of the of
tank Monitoring Systemtransformer,
the investigated During Induced Voltage
already at theTest of 330 MVA
production Power
stage, wasTransformer
equipped in two
After successfully
ceramic finished
dielectric windows laboratory
placed tests,ofitthe
in side-walls wastank, decided
i.e., neartothe
check the effectiveness
L1 phase (Figure 26) andof the
systemAfter successfully
operation on a finished
real power laboratory
transformer. tests, it was
First such decided
measurement to checkwas the effectiveness
carried out during of the
the
the L3 phase. This unfortunately prevented the active dielectric window with the built-in acoustic
system
induced operation
voltage
emission sensor
on(ISO-002-AWD/UHF)
test a real
with power
partial transformer.
discharge First such
measurement
from installation and
measurement
(IVPD) on aThe
testing. powerwastransformer
induced
carried outofduring
voltage test 330
wasMVA.
the
induced
The tank voltage
of the
carried test with
out investigated partial discharge
transformer,
strictly according measurement
already
to the scheme (IVPD)
at the production
described on a power transformer
stage, standard
in the IEC 60076-3 was equipped of
in two
[106] and 330 MVA.
ceramic
taking
The tank
dielectric of the investigated
into account
windows the strict in
placed transformer,
internal safety
side-walls thealready
of regulations at near
in
tank, i.e., the high
the production
the L1 phasestage,
voltage was
laboratory.
(Figure equipped
26)Due
and to L3in
thethese two
phase.
ceramic
This dielectric windows
unfortunately preventedplaced in side-walls
the active dielectric of the tank,
window i.e.,
with thenear the L1
built-in phase emission
acoustic (Figure 26) and
sensor
the L3 phase. This unfortunately prevented the active dielectric window
(ISO-002-AWD/UHF) from installation and testing. The induced voltage test was carried out strictly with the built-in acoustic
emission sensor
according (ISO-002-AWD/UHF)
to the scheme described in thefrom installation
IEC 60076-3 standardand testing.
[106] and The induced
taking voltage the
into account teststrict
was
carried out strictly according to the scheme described in the IEC 60076-3 standard
internal safety regulations in the high voltage laboratory. Due to these restrictions, the installation of the [106] and taking
into account the strict internal safety regulations in the high voltage laboratory. Due to these
Energies 2020, 13, x FOR PEER REVIEW 24 of 38
Energies 2020, 13, 3271 24 of 37

restrictions, the installation of the high-frequency current transformer (ISO-002-HF sensor) on the
grounding of the
Energies 2020,
high-frequency 13, xneutral
FOR PEER
current point
REVIEW
transformer was(ISO-002-HF
not allowed. Therefore,
sensor) on the the registration
grounding of theofneutral
the PD pulses
24
point 38 was
of was not
performed using two UHF antennas, whereas the employees of the HV test station
allowed. Therefore, the registration of the PD pulses was performed using two UHF antennas, whereas conducted the
restrictions,ofthe
measurement installation of the high-frequency current
60270transformer (ISO-002-HF sensor) on the PD
the employees of apparent
the HV test charge
station(according
conductedtothe
IEC standard)
measurement using
of apparent multichannel digital
charge (according
grounding of the neutral point was not allowed. Therefore, the registration of the PD pulses was
to IEC
detector type DDX
60270 standard) using9121b (Haefely, Basel,
multichannel digitalSwitzerland).
PD detector type DDX 9121b (Haefely, Basel, Switzerland).
performed using two UHF antennas, whereas the employees of the HV test station conducted the
measurement of apparent charge (according to IEC 60270 standard) using multichannel digital PD
detector type DDX 9121b (Haefely, Basel, Switzerland).

Figure 26.
Figure The installation
26. The installation ofof the
the UHF
UHF antenna
antenna (ISO-002-UHF
(ISO-002-UHF sensor)
sensor) in
in the
the dielectric
dielectric window
window of of the
the
330 MVA
330 MVA power
Figurepower transformer.
26. Thetransformer.
installation of the UHF antenna (ISO-002-UHF sensor) in the dielectric window of the
330 MVA power transformer.
The time sequence for the application of test voltage for induced voltage test with partial discharge
The time sequence for the application of test voltage for induced voltage test with partial
measurement (IVPD) is shown for in Figure 27. Inof thetest
first stage, theinduced
voltage voltage
was increased to partial
the level of
dischargeThe time
measurement
√ sequence(IVPD) the application
is shown in Figure voltage
27. In thefor
first stage, the test was
voltage with increased to
U )/ 3, measurement
(0.4 ×discharge which for a transformer
(IVPD) is shown with inrated
Figurevoltage
27. In theoffirst
a winding
stage, the = 410 was
Uvoltage kV corresponded
increased to to
the levelr of (0.4 × Ur)/√3, which for a transformer with rated voltage rof a winding Ur = 410 kV
the value Um =
the level of 94.7
(0.4 kV.
× UrAt)/√3,
that which
voltage for alevel,
transformer with ratednoise
the background (q =of3.5
voltage a pC)
winding
was U r = 410 kVThen,
registered.
corresponded to the value Um = 94.7 kV. At that voltage level,√the background noise (q = 3.5 pC) was
corresponded
the voltage to the value
was increased forU1m min
= 94.7tokV. theAtlevel
that ofvoltage
(1.2 ×level, = 284 kV, atnoise
Ur )/the3 background which(q =partial
3.5 pC)discharge
was
registered. Then,
registered. thethe
Then, voltage
voltage waswas increased
increased for
for 11 min
minto tothe
thelevel
levelof of (1.2
(1.2 × U×r)/√3
Ur)/√3
= = 284
284 kV, kV,
at at which
which
pulses of the apparent charge from a several dozen to 150 pC were registered occasionally. According
partial discharge pulses of the apparent charge from a several dozen
partial discharge pulses of the apparent charge from a several dozen to 150 pC were registered to 150 pC were registered
to the scheme of the IVPD test, the voltage was increased to the 1-h PD measurement voltage level
occasionally. According
occasionally. According to tothethescheme
schemeof of the
the IVPD test,the
IVPD test, thevoltage
voltage waswas increased
increased to theto 1-h
thePD 1-h PD
(Um = 374 kV). This voltage level was held for 5 min. During that time stable PD pulses of apparent
measurement
measurement voltage level
voltage level(U(Um m= =374374kV).
kV). This voltagelevel
This voltage levelwaswas held
held forfor 5 min.
5 min. During
During that time
that time
charge q = PD600–700 pC were registered. Then the voltage was raisedThen to thetheenhancement voltage level
stablestable
PD pulses pulses of apparent
of apparent chargeqq==600‒700
charge 600‒700 pC pC were
wereregistered.
registered. Then voltage
the was was
voltage raised to
raised to
(Um = the473.4 kV)
enhancement and held
voltage there
level (Ufor =1 min.
473.4 kV) Atandthis test
held stage,
there for 1the
min. apparent
At this charge
test stage, grew
the to
apparent approx.
the enhancement voltage level (Um = 473.4 kV) and held there for 1 min. At this test stage, the apparent
m
1500–1800
charge pC. to approx. 1500‒1800 pC.
charge grewgrew
to approx. 1500‒1800 pC.

Figure
Figure 27. 27.
TimeTime sequencefor
sequence forthe
theapplication
application of
of test
testvoltage
voltageforfor
induced voltage
induced test test
voltage withwith
partial
partial
discharge
discharge measurement (IVPD).
Figure 27.measurement
Time sequence(IVPD).
for the application of test voltage for induced voltage test with partial
discharge measurement (IVPD).
For theFor IVPD
the IVPD test,
test, thethe monitoringsystem
monitoring system was
wasconfigured
configured toto
thetheoperation
operationin a in
diagnostic mode,mode,
a diagnostic
whichwhich enabled
enabled the the observation
observation andregistration
and registration of
of UHF
UHF time
time waveforms
waveforms in real-time.
in real-time.In Figure
In 28 the28 the
Figure
For the IVPD
exemplary timetest, the monitoring
waveforms of duration system
20 ms was
wereconfigured to thewere
presented, which operation in afor
registered diagnostic
three main mode,
exemplary
which timethewaveforms of duration 20 ms were presented, which were registeredInfor three28 main
IVPD test voltage levels, i.e., 284, 374, and 473 kV. PD pulses were detected by the UHF probe the
enabled observation and registration of UHF time waveforms in real-time. Figure
IVPD test
exemplary voltage
time levels, i.e., 284, 374, and 473 kV. PD pulses were detected by the UHF probe installed
installed nearwaveforms
the phase L3. of duration
The probe20installed
ms were onpresented,
the opposite which
side were
of theregistered
tank, near for three
phase L1, main
near the phase
IVPDindependently L3.
test voltage from The probe
levels,
the i.e.,
installed
level284,
of the374,
on the opposite
and 473registered
test voltage,
side
kV. PD only of the
pulses tank, near
were detected
background
phase L1,
by the
noise (about
independently
20‒30UHFmV).probe
fromThis
the near
installed level
meanstheofthat
the testPD
the
phase voltage,
L3. source registered
The probewas probably onlyinon
installed background
phase L3 or noise
the opposite (about
its neighborhood.
side of the 20–30 For
tank, mV).
the This
near means
partial
phase L1,
that the PD
discharges source
of low was probably
intensity and in
low phase
apparentL3 or its
charge neighborhood.
(max. 150 pC), the For
UHF
independently from the level of the test voltage, registered only background noise (about 20‒30 mV). the partial
probe discharges
registered pulses, of low
intensity
This however,
means andthatlowthe
their apparent
amplitude
PD sourcecharge
did notwas(max.
exceed 150 pC),
45 mV
probably the UHF
in(approx.
phase L3probe
1.5‒2or itsregistered
times higher than pulses,
neighborhood. however,
the background
For their
the partial
noise). did
amplitude In turn, by Um = 45
not exceed 374mV kV,(approx.
for partial discharges
1.5–2 of apparent
times higher than charge equal to 600‒700
the background noise).pC,In the
turn, by
discharges of low intensity and low apparent charge (max. 150 pC), the UHF probe registered pulses,
Um = 374 kV,
however, foramplitude
their partial discharges
did notofexceed
apparent
45 charge equal to1.5‒2
mV (approx. 600–700 pC,higher
times the UHF pulses
than of definitely
the background
higher In
noise). amplitude
turn, by (120–135
Um = 374mV) were
kV, for registered.
partial For the
discharges enhancement
of apparent voltage
charge equallevel (473.4 kV),
to 600‒700 PD
pC, the
Energies 2020, 13, 3271 25 of 37
Energies 2020, 13, x FOR PEER REVIEW 25 of 38

UHF
pulses pulses of charge
of apparent definitely higher amplitude
1500–1800 (120‒135 mV)
pC corresponded were
to the UHF registered.
pulses ofFor the enhancement
maximal amplitude 175
voltage
mV. The level (473.4
obtained kV), PD pulses
measurement of apparent
results charge
during the 1500‒1800
IVPD test of pC corresponded
power to the
transformer UHF
330 MVApulses
showed
of maximal amplitude 175 mV. The obtained measurement results during the IVPD test
the correct design and implementation of the sensor ISO-002-UHF mechanics and high sensitivity of of power
transformer 330 MVA showed the correct design and implementation of the sensor ISO-002-UHF
the UHF antenna, which enabled detection of partial discharges with relatively low apparent charge
mechanics and high sensitivity of the UHF antenna, which enabled detection of partial discharges
(100–150 pC).
with relatively low apparent charge (100‒150 pC).

(a)

(b)

(c)

FigureFigure 28. Exemplary

28. Exemplary UHFPD
UHF PD pulses
pulses registered
registeredforfor
different voltage
different levelslevels
voltage duringduring
the IVPD
thetest:
IVPD(a) test:
(a) 284 kV, (b) 374 kV, (c) 473 kV. 284 kV, (b) 374 kV, (c) 473 kV.

3.2. of
3.2. Test Test
theofSystem
the System on 31.5
on 31.5 MV MVA PowerTransformer
A Power Transformer
The next
The next object,
object, on on which
which thePD
the PDmonitoring
monitoring system
systemwas wastested
testedwas a transformer
was withwith
a transformer a rated
a rated
power of 31.5 MVA and voltage of 115/33 kV. Before the prototype system installation, the
power of 31.5 MVA and voltage of 115/33 kV. Before the prototype system installation, the localization
localization of PD sources was carried out by the trilateration technique. For this purpose, eight
of PD sources was carried out by the trilateration technique. For this purpose, eight piezoelectric AE
piezoelectric AE sensors type PAC R15α and 8-channel AE system type DiSP (Physical Acoustic
sensors type PAC Princeton,
Corporation, R15α and NJ,8-channel AE system
USA) were type
used. The DiSP (Physical
coordinates of theAcoustic Corporation,
AE sensors mounted on Princeton,
the
NJ, USA) were used. The coordinates of the AE sensors mounted on
transformer tank of dimensions 4 × 1.3 × 2.6 m are listed in Table 5. the transformer tank of dimensions
4 × 1.3 × 2.6 m are listed in Table 5.
Table 5. Coordinates of AE sensors.
Table 5. Coordinates of AE sensors.
Sensor Index X (m) Y (m) Z (m)
Sensor Index 1 X 1.20
(m) 1.40Y (m)1.30 Z (m)
2 2.05 0.45 1.30
1 3 1.20
3.05 1.40 1.401.30 1.30
2 4 2.05
4.00 1.10 0.450.65 1.30
3 5 3.05
3.00 0.75 1.400.00 1.30
4 6 4.00
2.00 0.45 1.100.00 0.65
5 7 3.00
1.06 1.05 0.750.00 0.00
6 8 2.00
0.00 0.95 0.450.65 0.00
7 1.06 1.05 0.00
8 0.00 0.95 0.65
Energies 2020, 13, 3271 26 of 37
Energies 2020, 13, x FOR PEER REVIEW 26 of 38

The amplitude
The amplitude of of the
the registered
registered signals
signals waswas relatively
relatively low
low and
and equal
equal toto 30–45
30‒45 dB
dB ref
ref 100
100 µV
µV
(peakatataround
(peak around 66 66 dB).
dB). TheThe results
results of theoflocalization
the localization obtained
obtained with
with the the trilateration
trilateration techniquetechnique
showed
showed
that in thethat in the investigated
investigated power transformer
power transformer there are twothere are two
sources sources
of partial of partial The
discharges. discharges. The
first source
firstdetected
was source wason detected on the high
the high voltage sidevoltage
(115 kV), side (115 kV),
between between
phases phases
L1 and x = 1.9 m,
L1 and L2 (coordinates:
L2 (coordinates:
yx== 1.0
1.9 m, zy == 1.0
1.25m, z =In1.25
m). m).the
turn, In second
turn, the PDsecond
sourcePDwassource was on
detected detected
the lowonvoltage
the lowside
voltage side
(33 kV),
(33
at kV),
the at the
point point
of LV of LV connections
connections (coordinates: x = 1.3, y =x0.1
(coordinates: = 1.3,
m, yz = 0.1
1.5 m,
m).zThe
= 1.5graphic
m). The graphic
results results
of the PD
of the PD
source source localization
localization are presentedare in
presented
Figure 29. in Figure 29.

Figure 29. Graphic presentation of the localization results of two partial discharges sources by the
Figure 29. Graphic
trilateration presentation
technique of the localization
and the distribution results
of the eight of two(blue
AE sensors partial discharges sources by the
dots).
trilateration technique and the distribution of the eight AE sensors (blue dots).
After performing the measurements which showed the existence of partial discharge sources,
Afterpreceding
the works performing
thethe measurements
prototype which showed
system installation werethe existence
carried of partial
out. First, discharge
the power sources,
transformer
the works preceding the prototype system installation were carried out. First, the power transformer
was turned off and disconnected from the HV transmission lines. Then, the oil excess was drained to
was turned off and disconnected from the HV transmission lines. Then, the oil excess was drained to
the level enabling opening the revision hole in the upper tank cover. After opening the revision hole,
 Energies 2020, 13, 3271 27 of 37

Energies 2020, 13, x FOR PEER REVIEW 27 of 38

Energies 2020, 13, x FOR PEER REVIEW 27 of 38
the level enabling opening the revision hole in the upper tank cover. After opening the revision hole,
the
theactive dielectric windows and the remaining elements of the ISO-002-ADW/UHF sensors were
active dielectric windows and the remaining elements of the ISO-002-ADW/UHF sensors were
immediately
immediatelyinstalled, in order to minimize the time of contact
contactof
ofthe
theoil
oilwith
withhumid
humidair
air(Figure
(Figure30).
30).
installed, in order to minimize the time of contact of the oil with humid air (Figure 30).

Figure
Figure30.
30.The
Theinstallation
installationofof
ofthe
theactive
activedielectric
dielectricwindows with UHF antennas (ISO-002-ADW/UHF-
Figure 30. The installation the active dielectric windows
windows with
with UHF
UHF antennas
antennas(ISO-002-ADW/UHF-1
(ISO-002-ADW/UHF-
1and
and ISO-002-ADW/UHF-2)
ISO-002-ADW/UHF-2) in in the
the revision
revision holes
holes of the upper cover of the 31.5 MVA
MVA power
1 and ISO-002-ADW/UHF-2) in the revision holes of the upper cover of the 31.5
of the upper cover of the 31.5 MVA power
power
transformer tank.
transformer tank.
tank.
transformer

The investigatedpower
The power transformerwas was equippedinintwo two sensorsISO-002-ADW/UHF
ISO-002-ADW/UHF andone one
Theinvestigated
investigated powertransformer
transformer wasequipped
equipped in twosensors
sensors ISO-002-ADW/UHFand and one
sensor ISO-002-HF installed
sensor installed the grounding wire of the neutral point ofof
the HV winding (Figure 31).
sensorISO-002-HF
ISO-002-HF installedon on thegrounding
groundingwire
wireofofthe
theneutral
neutralpoint
point ofthe
theHVHVwinding
winding(Figure
(Figure
TheThe
31). optic fiber
optic andand
fiber power supply
power cables
supply werewere
cables protected by putting
protected by themthem
putting into into
the corrugated
the PCV PCV
corrugated pipes
31). The optic fiber and power supply cables were protected by putting them into the corrugated PCV
with awith
pipes diameter of 25 mm. A waterproof enclosure of the monitoring system with data concentrator
pipes with aa diameter
diameter ofof 2525 mm.
mm. AA waterproof
waterproof enclosure
enclosure ofof the
the monitoring
monitoring system
system with
with data
data
module MPD-001,
concentrator server, and power supply was placed opposite toplaced
the power transformer. The last
concentrator module MPD-001, server, and power supply was placed opposite to the power
module MPD-001, server, and power supply was opposite to the power
stage of the installation
transformer. works was configuring thewasdata transmission between the PD monitoring
transformer.The Thelast
laststage
stageofofthe
theinstallation
installationworks
works wasconfiguring
configuringthe thedata
datatransmission
transmissionbetween
between
system monitoring
the and SCADA substation system type SYNDIS ES (Mikronika, Poznan, Poland), which registered
thePDPD monitoringsystem systemand andSCADA
SCADAsubstation
substationsystem
systemtypetypeSYNDIS
SYNDISES ES(Mikronika,
(Mikronika,Poznan,
Poznan,
basic operationregistered
Poland), parameters, such as voltage, load, andsuch
oil temperature.load, and oil temperature.
Poland),which
which registeredbasic basicoperation
operationparameters,
parameters, suchasasvoltage,
voltage, load, and oil temperature.

Figure31.
Figure Photograph
31.Photograph of PD
of the the on-line
PD on-line monitoring
monitoring system
system installed
installed on 31.5on 31.5substation
MVA MVA substation
power
Figure 31. Photograph of the PD on-line monitoring system installed on 31.5 MVA substation power
power transformer.
transformer.
transformer.
After fillingthe
After the power transformer
transformer tank with oil and checking the tightness of of
revision holes, the
Afterfilling
filling thepower
power transformertank tankwith
withoil
oiland
andchecking
checkingthe thetightness
tightness ofrevision
revisionholes,
holes,
power
the transformer was turned on and the partial discharges monitoring system was started. Figure 32
thepower
powertransformer
transformerwas wasturned
turnedononandandthe thepartial
partialdischarges
dischargesmonitoring
monitoringsystem
systemwas
wasstarted.
started.
presents
Figure the characteristics of PD intensity (pulse rate per 15 s), active power, voltage, and top-oil
Figure3232presents
presentsthethecharacteristics
characteristicsofofPD
PDintensity
intensity(pulse
(pulserate
rateper
per15
15s),
s),active
activepower,
power,voltage,
voltage,and
and
temperature
top-oil obtained within 8-days-operation of the PD monitoring system.
top-oiltemperature
temperatureobtained
obtainedwithin
within8-days-operation
8-days-operationofofthe thePD
PDmonitoring
monitoringsystem.
system.
Energies 2020, 13, 3271 28 of 37
Energies 2020, 13, x FOR PEER REVIEW 28 of 38

.
Figure 32. Partial discharges intensity characteristics, voltage, load, and oil temperature
Figure 32. Partial
characteristics discharges
registered intensity characteristics,
during 8-days-operation voltage,
of the monitoring load,
system and
on the 31.5oil
MVAtemperature
substation
characteristics
power registered during 8-days-operation of the monitoring system on the 31.5 MVA
transformer.
substation power transformer.
The conducted data analysis proves that in the monitored power transformer:
The conducted data analysis proves that in the monitored power transformer:
• Partial discharges were registered every day, and their activity was the highest mainly for hours
• Partial discharges
from 8:45 were
a.m. to 5:45 registered
p.m. every
(these time day, and
intervals aretheir activity
indicated was the
in Figure 32highest
as grey mainly
bands). for
Thehours
sum
from
of PD8:45 a.m.from
pulses to 5:45
thisp.m.
time (these time
intervals intervals 68.9%
constitutes are indicated in Figurepulses.
of all registered 32 as grey bands). The
sum of PD pulses from this time intervals constitutes 68.9% of all registered pulses.
Energies 2020, 13, 3271 29 of 37
Energies 2020, 13, x FOR PEER REVIEW 29 of 38

•• The
Thepeak
peakintensity
intensityof
ofpartial
partialdischarges
dischargesoccurred
occurredbetween
between 1:00
1:00 p.m.
p.m. and
and 3:00
3:00 p.m.
p.m. In
In this
this short
short
period, almost one-quarter (24.4 %) of all PD pulses were detected.
period, almost one-quarter (24.4 %) of all PD pulses were detected.
•• The
ThePearson
Pearsoncorrelation
correlationcoefficient
coefficientbetween
betweenthe thehourly
hourlydistribution
distributionofofthe
the UHF
UHF and
and AE
AE pulses
pulses
for
for sensor ISO-002-ADW/UHF-1 was r = 0.87 with significance p = 0.001 (very strong positive
sensor ISO-002-ADW/UHF-1 was r = 0.87 with significance p = 0.001 (very strong positive
correlation), and for sensor ISO-002-ADW/UHF-2 it was r = 0.45 with significance p = 0.027
correlation), and for sensor ISO-002-ADW/UHF-2 it was r = 0.45 with significance p = 0.027
(moderate positive correlation). Both hourly distributions of the UHF and AE pulses are
(moderate positive correlation). Both hourly distributions of the UHF and AE pulses are presented
presented in
in Figure 33.
Figure 33.

(a) (b)

Figure
Figure 33.
33.Hourly
Hourlydistribution
distributionofofthe UHF
the and
UHF AEAE
and pulses forfor
pulses sensor: (a) ISO-002-ADW/UHF-1
sensor: andand
(a) ISO-002-ADW/UHF-1 (b)
ISO-002-ADW/UHF-2.
(b) ISO-002-ADW/UHF-2.

•• The
Thehourly
hourlydistribution
distributionofofthe
thePDPDpulses
pulsesregistered
registeredwith
withthethehigh-frequency
high-frequencycurrent current transformer
transformer
ISO-002-HF
ISO-002-HFwas was correlated themost
correlated the mostwith
withthethedistribution
distribution obtained
obtained forfor
the the
AE AE transducer
transducer (r = 0.68,
(r =
p < 0.001)
0.68, p < 0.001)
and and the UHF
the UHF antenna (r = (r
antenna 0.62; p = 0.001)
= 0.62; p = 0.001) of the
of the sensorsensor ISO-002-ADW/UHF-1.
ISO-002-ADW/UHF-1. In
In the
case of the sensor ISO-002-ADW/UHF-2 the correlation was lower at the level of r = 0.44 with
the case of the sensor ISO-002-ADW/UHF-2 the correlation was lower at the level of r = 0.44 with
significancepp==0.029
significance 0.029for
forthe
thedistribution
distributionobtained
obtained forfor the
the AE
AE transducer
transducer and and rr == 0.22 0.22 with
with
significance p = 0.031 for the distribution obtained for the UHF antenna.
significance p = 0.031 for the distribution obtained for the UHF antenna.
•• The
Thevoltage
voltagevalue
valuewas
waschanging
changingininaanarrow
narrowrange
rangefrom
from113.3
113.3toto117.6
117.6kV,
kV,wherein
whereinthe theaverage,
average,
variance, and standard deviation were, respectively, equal to M = 115.3 kV,
variance, and standard deviation were, respectively, equal to M = 115.3 kV, s = 0.47, and SD =SD
2 s 2 = 0.47, and =
0.69.
0.69. Due to the small voltage variation, it was decided not to investigate its
Due to the small voltage variation, it was decided not to investigate its influence on partial influence on partial
discharges
dischargesintensity.
intensity.
•• During
During the 8-days-operationofofthe
the 8-days-operation thesystem,
system,the thepower
power transformer
transformerwas was loaded
loaded with with relatively
relatively
low
low power from only 0.68 to 11.34 MVA, which constituted, respectively, 2.15% and 36.00%
power from only 0.68 to 11.34 MVA, which constituted, respectively, 2.15% and 36.00% of of its
its
rated power. The average value of the load was equal to M = 5.31 MVA,
rated power. The average value of the load was equal to M = 5.31 MVA, variance s = 6.16, and variance s22 = 6.16, and
standard
standarddeviation
deviationSD SD==2.48.
2.48.
• The oil temperature was changing in a wide band from 14.0 to 36.1 °C. It was observed that daily
• The oil temperature was changing in a wide band from 14.0 to 36.1 ◦ C. It was observed that daily
“hills” and local peaks of the oil temperature curve overlap the period of higher activity of partial
“hills” and local peaks of the oil temperature curve overlap the period of higher activity of partial
discharges. (from 8:45 a.m. to 5:45 p.m.) (see Figure 32). In turn, for daily “valleys” of the oil
discharges. (from 8:45 a.m. to 5:45 p.m.) (see Figure 32). In turn, for daily “valleys” of the oil
temperature curve, a clear reduction or complete disappearance of the partial discharges activity
temperature curve, a clear reduction or complete disappearance of the partial discharges activity
was observed. In order to test the hypothesis relating to the influence of the level of oil
was observed. In order to test the hypothesis relating to the influence of the level of oil temperature
temperature on partial discharges intensity in the monitored power transformer, it was decided
on partial discharges intensity in the monitored power transformer, it was decided to conduct a
to conduct a one-way analysis of variance (ANOVA) for independent samples. One of the main
one-way analysis of variance (ANOVA) for independent samples. One of the main assumptions
assumptions of the ANOVA is that the independent variable consists of two or more categorical,
of the ANOVA is that the independent variable consists of two or more categorical, independent,
independent, and equinumerous groups. The categorization of the daily oil temperature values
and equinumerous groups. The categorization of the daily oil temperature values (independent
(independent variable) was carried out using tertiles, in order to obtain three equinumerous
variable) was carried out using tertiles, in order to obtain three equinumerous groups. The “low”
groups. The “low” temperature category was ascribed to lower tertile, whereas “moderate” and
temperature category was ascribed to lower tertile, whereas “moderate” and “high” temperature
“high” temperature categories to middle and higher tertiles, respectively. Since the assumption
categories to middle and higher tertiles, respectively. Since the assumption of homogeneity of
of homogeneity of variance was violated (Levene's test for equality of variances was statistically
variance was violated (Levene’s test for equality of variances was statistically significant), therefore
significant), therefore the Brown–Forsythe and Welch F-ratios are reported. The results of both
the Brown–Forsythe and Welch F-ratios are reported. The results of both robust tests of equality
robust tests of equality of means are listed in Table 6. The results of the analysis showed that the
of means are listed in Table 6. The results of the analysis showed that the compared groups differ
compared groups differ from one another statistically, which means that the oil temperature level
from one another statistically, which means that the oil temperature level differentiates the number
differentiates the number of PD pulses. Figure 34 illustrates the value of the average number of
the AE, UHF, and HF pulses between particular oil temperature levels (low, moderate, and high),
Energies 2020, 13, 3271 30 of 37

of PD pulses. Figure 34 illustrates the value of the average number of the AE, UHF, and HF pulses
between
Energies 2020, 13,particular
x FOR PEERoil temperature levels (low, moderate, and high), which indicate a strong30causal
REVIEW of 38
relationship between these variables. Figures obtained for the sensors ISO-002-ADW/UHF1 and
which indicate a strong
ISO-002-ADW/UHF2 show causal relationship
that the higher the between these variables.
oil temperature level was, theFigures
higherobtained for the
was the number
sensors ISO-002-ADW/UHF1
of the registered AE and UHF and ISO-002-ADW/UHF2
pulses. The exception isshow that the highercurrent
the high-frequency the oil temperature
transformer
level was, thesensor,
ISO-002-HF higher forwaswhich
the number of theaverage
the highest registered AE and
number of UHF pulses.
HF pulses wasThe exceptionfor
registered is the
the
high-frequency current transformer
average oil temperature ISO-002-HF
level. To investigate whichsensor, for which
compared the are
groups highest averagesignificantly
statistically number of
HF pulsesfrom
different was registered
one another, for the
the Gamesa–Howell
average oil temperature post hoc level.
test,To investigate
which which compared
is recommended in the
groups
situation when the assumption of homogeneity of variance in the compared groups waspost
are statistically significantly different from one another, the Gamesa–Howell hoc
violated.
test,
Thewhich is recommended
post hoc in the situation
comparisons revealed whensignificant
statistically the assumption of homogeneity
differences (p < 0.0005) ofbetween
variance all
in
the compared groups was violated. The post hoc comparisons revealed
groups (oil temperature levels) except the groups “moderate” and “high” temperature for the statistically significant
differences
UHF pulses(pregistered
< 0.0005) between
with the all groups
sensor (oil temperature levels)
ISO-002-ADW/UHF-2 (Tableexcept
7). Thethehighest
groupsdifferences
“moderate” in
and “high” temperature
the number of PD pulses werefor the UHF pulses
observed between registered
groups “low”with and
the sensor ISO-002-ADW/UHF-2
“high” temperature. As it was
(Table 7). The
mentioned highest
before, differences
the voltage valueinvariability
the number wasof PDTherefore,
low. pulses werewhile observed
interpreting between groups
the obtained
“low”
resultsand “high”
it may temperature.
be concluded that inAsthe
it monitored
was mentionedpowerbefore, the voltage
transformer, value variability
the oil temperature growthwasis
low. Therefore, while interpreting the obtained results it may be concluded that
a significant factor inducing partial discharges. This hypothesis is in line with the research results, in the monitored
power
which transformer,
showed that the the oil temperature
temperature growth
growth is a significant
in combination factor
with inducingmoisture
an increased partial discharges.
content of
This hypothesis
the oil-paper is in linesystem
insulation with the mayresearch
induceresults, whichgrowth
the activity showed ofthat the discharges
partial temperature growth
[107–110].
in combination with an increased moisture content of the oil-paper insulation
Additionally, the presence of water molecules in oil as a result of the moisture migration process, system may induce
the activity growth of partial discharges [107‒110]. Additionally, the presence
as well as the effect of gas bubbles exuding from cellulose insulation, may significantly influence of water molecules
in oilintensification
the as a result of ofthethe
moisture migration [15,111–116].
PD phenomenon process, as well as the effect of gas bubbles exuding
from cellulose insulation, may significantly influence the intensification of the PD phenomenon
[15,111‒116].
Table 6. Results of robust tests of equality of means.

PD Sensor Label Table 6. Results ofTest

Pulses robust tests of Statistics
equality of1 means.
df1 df2 p
PD Sensor Label Pulses Test
Welch Statistics
289.93 1 df12 28,969.85 p < 0.0005
df2
UHF
Brown–Forsythe
Welch 371.11
289.93 2 2 < 0.0005
32,492.03< 0.0005
28,969.85
ISO-002-ADW/UHF-1 UHF
Brown–Forsythe
Welch 371.11
149.51 2 2 32,492.03
24,644.93< 0.0005
< 0.0005
ISO-002-ADW/UHF-1 AE
Welch
Brown–Forsythe 149.51
176.34 2 2 24,644.93 < 0.0005
20,351.35< 0.0005
AE
Brown–Forsythe
Welch 176.34
60.18 2 2 20,351.35
25,693.85< 0.0005
< 0.0005
UHF Welch 60.18 2 2 25,693.85
UHF Brown–Forsythe 39.05 35,227.18< 0.0005
< 0.0005
ISO-002-ADW/UHF-2
Brown–Forsythe 39.05 2 35,227.18 < 0.0005
ISO-002-ADW/UHF-2 Welch 1772.95 2 23,518.60 < 0.0005
AE Welch 1772.95 2 2 23,518.60 < 0.0005
AE Brown–Forsythe 977.32 32,563.54 < 0.0005
Brown–Forsythe 977.32 2 32,563.54 < 0.0005
Welch 1486.10 2 25,652.90 < 0.0005
ISO-002-HF HFCT Welch 1486.10 2 25,652.90 < 0.0005
ISO-002-HF HFCT Brown–Forsythe 914.83 2 26,939.20 < 0.0005
Brown–Forsythe 914.83 2 26,939.20 < 0.0005
1 Asymptotically F distributed.
1 Asymptotically F distributed.

Figure 34. Estimated marginal means of PD pulses.

Figure 34. Estimated marginal means of PD pulses.
Energies 2020, 13, 3271 31 of 37

Table 7. Results of multiple comparisons of mean PD pulses within groups using post hoc
Games–Howell test.

PD Sensor Label Pulses Temperature Groups Mean Difference p

Low vs. Moderate −18.60 * < 0.0005
UHF Low vs. High −79.08 * < 0.0005
Moderate vs. High −60.49 * < 0.0005
ISO-002-ADW/UHF-1
Low vs. Moderate −4.95 * < 0.0005
AE Low vs. High −22.10 * < 0.0005
Moderate vs. High −17.15 * < 0.0005
Low vs. Moderate −0.74 * < 0.0005
UHF Low vs. High −1.00 * < 0.0005
Moderate vs. High −0.26 0.154
ISO-002-ADW/UHF-2
Low vs. Moderate −386.96 * < 0.0005
AE Low vs. High −534.64 * < 0.0005
Moderate vs. High −147.68 * < 0.0005
Low vs. Moderate −39.46 * < 0.0005
ISO-002-HF HFCT Low vs. High −25.98 * < 0.0005
Moderate vs. High 13.48 * < 0.0005
* The mean difference is significant at the 0.05 level.

4. Conclusions
The article presents a prototype, hybrid on-line PD monitoring system. Its unique feature is a
combination of the functionality of three methods of PD detection (AE, HF, and UHF) and integrating
them into one complementary system. Another important function of the system is the possibility
of its integration using IEC 61850 protocol with remaining substation automation devices, but first
of all with SCADA system. Due to that, the monitoring system can analyze in real-time the PD
parameters registered using various detectors as well as the basic transformer parameters that affect
the phenomenon of partial discharges, i.e., voltage and oil-temperature. Tests conducted on 31.5 MVA
power transformer showed that in its case partial discharges ignited in particular times of day (from
approx. 8:45 a.m. to 5:45 p.m.), and their increased activity was observed during daily “hills” and
local peaks of the oil temperature curve. The results of the statistical analysis of the HF, UHF, and
AE signals presented in the paper allow to conclude with a high probability that in these periods, the
partial discharge phenomenon was the source of the majority of pulses registered by detectors.
Laboratory tests of partial discharge detectors developed for the monitoring system have confirmed
their high performance. The active dielectric window with built-in ultrasound transducer assures on
average 5.80 times higher sensitivity than the popular contact AE sensor type PAC R15D. Such high
sensitivity was obtained because the ultrasonic transducer is located inside the oil-filled transformer
tank and is equipped with a specially designed matching layer that facilitates the transfer of acoustic
energy from mineral oil to a piezoelectric crystal. In turn, the prototype HFCT sensor with MnZn
ferrite core showed—in comparison to the commonly used RFCT-4 sensor—on average 1.48 times
higher sensitivity of PD in oil detection and on average 1.50 times higher detection sensitivity of surface
PD. Additionally, very good results were obtained for the prototype construction of the meandered
planar inverted-F antenna. The antenna bandwidth (280–353 MHz) was optimized for the detection of
PDs typical for oil-paper insulation. Tests performed in a shielded laboratory and with the use of a
power transformer tank equipped in the dielectric windows showed that the prototype meandered
PIFA compared to standard disk antenna is on average 7.80 times more sensitive in the detection of
low-radiating PD (an initial form of surface PD with apparent charge below 1 nC) and on average 1.75
more sensitive in the detection of high energy creeping sparks with apparent charge about 6 nC.
Future research will focus primarily on the development of the system’s software layer, including
development and implementation of an algorithm for identifying the type of partial discharge pulses
based on PRPD patterns (for HF and UHF method) and implementation of the FFT (Fast Fourier
Energies 2020, 13, 3271 32 of 37

Transform) algorithm in the FPGA for the acoustic emission channel, which will create the ability to
identify the type of PDs based on selected frequency domain features and artificial neural network
(ANN) based classification. In the case of the hardware layer, research works are underway to adapt the
system to diagnose power equipment other than transformers. The first successful implementations of
the prototype monitoring system on 110 kV gas-insulated switchgear are described in [117].

Author Contributions: Conceptualization, W.S., K.W. and W.G.; methodology, W.S, K.W. and W.G.; software,
W.G., W.S. and K.W.; validation, W.S., W.G. and K.W.; formal analysis, W.S.; investigation, W.S, W.G., C.S.,
K.W.; resources, W.S and W.G; data curation, W.S.; writing—original draft preparation, W.S.; writing—review
and editing, W.S., K.W., W.G. and C.S.; visualization, W.S.; supervision, W.S.; project administration, W.S. and
W.G.; funding acquisition, W.S., W.G and K.W. 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.
Conflicts of Interest: The authors declare no conflict of interest.

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. Foros, J.; Istad, M. Health Index, Risk and Remaining Lifetime Estimation of Power Transformers. IEEE
Trans. Power Deliv. Early Access 2020, 7. [CrossRef]
3. Ortiz, F.; Fernandez, I.; Ortiz, A.; Renedo, C.J.; Delgado, F.; Fernandez, C. Health indexes for power
transformers: A case study. IEEE Electr. Insul. Mag. 2016, 32, 7–17. [CrossRef]
4. Krüger, M.; Koch, M.; Kraetge, A.; Rethmeier, K.; Energy, O.; Dierks, A.A. Diagnostic measurements on
power transformers. Energetyka 2009, 5, 299–305.
5. Hori, M.; Nishioka, M.; Ikeda, Y.; Noguchi, K.; Kajimura, K.; Motoyama, H.; Kawamura, T. Internal winding
failure due to resonance overvoltage in distribution transformer caused by winter lightning. IEEE Trans.
Power Deliv. 2006, 21, 1600–1606. [CrossRef]
6. Metwally, I.A. Failures, monitoring and new trends of power transformers. IEEE Potentials 2011, 30, 36–43.
[CrossRef]
7. Kassi, K.S.; Fofana, I.; Meghnefi, F.; Yeo, Z. Impact of local overheating on conventional and hybrid insulations
for power transformers. IEEE Trans. Dielectr. Electr. Insul. 2015, 22, 2543–2553. [CrossRef]
8. Paoletti, G.; Golubev, A. Partial discharge theory and applications to electrical systems. In Proceedings of the
Conference Record of 1999 Annual Pulp and Paper Industry Technical Conference (Cat. No. 99CH36338),
Seattle, WA, USA, 21–25 June 1999; pp. 124–138. [CrossRef]
9. Sikorski, W.; Walczak, K. PDtracker–system monitoringu wyładowań niezupełnych w transformatorze
energetycznym. Prz. Elektrotechniczny 2014, 10, 45–49.
10. Badicu, L.V.; Broniecki, U.; Koltunowicz, W.; Körber, S.; Krüger, M.; Voegel, E. Prevention of transformer
failure through continuous monitoring. In Proceedings of the XIXth International Symposium on High
Voltage Engineering (ISH), Pilsen, Czech Republic, 23–28 August 2015. Paper Number 274.
11. Gockenbach, E.; Borsi, H. Condition monitoring and diagnosis of power transformers. In Proceedings of the
2008 International Conference on Condition Monitoring and Diagnosis, Beijing, China, 21–24 April 2008;
pp. 894–897. [CrossRef]
12. Mondal, M.; Kumbhar, G.B. Detection, measurement, and classification of partial discharge in a power
transformer: Methods, trends, and future research. IETE Tech. Rev. 2018, 35, 483–493. [CrossRef]
13. IEEE Guide for Application for Monitoring Equipment to Liquid-Inmersed Transformers and Components; IEEE:
Piscataway, NJ, USA, 2012.
14. Zhao, W.B.; Zhang, G.J.; Sun, Y.; Yan, Z.; Xu, D.K. Online Multisensor Monitoring System For Insulation
Condition Of Oil-Immersed Power Transformer. In Proceedings of the IEEE International Symposium on
Electrical Insulation, Indianapolis, IN, USA, 19–22 September 2004; pp. 89–92. [CrossRef]
15. Sikorski, W.; Walczak, K.; Przybylek, P. Moisture migration in an oil-paper insulation system in relation to
online partial discharge monitoring of power transformers. Energies 2016, 9, 1082. [CrossRef]
Energies 2020, 13, 3271 33 of 37

16. Sikorski, W.; Walczak, K.; Moranda, H.; Gil, W.; Andrzejewski, M. Partial discharge on-line monitoring system
based on acoustic emission method—Operational experiences. Prz. Elektrotechniczny 2012, 88, 117–121.
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]
18. Tenbohlen, S.; Beltle, M.; Siegel, M. PD monitoring of power transformers by UHF sensors. In Proceedings of
the International Symposium on Electrical Insulating Materials (ISEIM), Toyohashi, Japan, 12–15 September
2017; Volume 1, pp. 303–306. [CrossRef]
19. Judd, M.D.; Yang, L.; Hunter, I.B. Partial discharge monitoring of power transformers using UHF sensors.
Part I: Sensors and signal interpretation. IEEE Electr. Insul. Mag. 2005, 21, 5–14. [CrossRef]
20. Judd, M.D. Experience with UHF partial discharge detection and location in power transformers. In
Proceedings of the Electrical Insulation Conference (EIC), Annapolis, MD, USA, 5–8 June 2011; pp. 201–205.
[CrossRef]
21. Chai, H.; Phung, B.T.; Mitchell, S. Application of UHF sensors in power system equipment for partial
discharge detection: a review. Sensors 2019, 19, 1029. [CrossRef]
22. Braunlich, R.; Hassig, M.; Fuhr, J.; Aschwanden, T. Assessment of insulation condition of large power
transformers by on-site electrical diagnostic methods. In Proceedings of the IEEE International Symposium
on Electrical Insulation, Anaheim, CA, USA, 2–5 April 2000; pp. 368–372. [CrossRef]
23. International Electrotechnical Commission (IEC). High Voltage Test Techniques—Measurement of Partial
Discharges by Electromagnetic and Acoustic Methods; IEC TS 62478:2016; International Electrotechnical
Commission (IEC): Geneva, Switzerland, 2016.
24. Sikorski, W. Development of acoustic emission sensor optimized for partial discharge monitoring in power
transformers. Sensors 2019, 19, 1865. [CrossRef] [PubMed]
25. Mackenzie, E.A.; Crossey, J.; DePablo, A.; Ferguson, W. On-line monitoring and diagnostics for power
transformers. In Proceedings of the IEEE International Symposium on Electrical Insulation, San Diego, CA,
USA, 6–9 June 2010; pp. 1–5. [CrossRef]
26. IEEE Guide for the Detection, Location and Interpretation of Sources of Acoustic Emissions from Electrical
Discharges in Power Transformers and Power Reactors; IEEE Std C57.127-2018 (Revision of IEEE Std
C57.127-2007); IEEE 2019. Available online: https://ieeexplore.ieee.org/document/8664690 (accessed on
24 June 2020). [CrossRef]
27. Cavallini, A.; Ramos, C.G.A.; Montanari, G.C.; Saad, H.; Tozzi, M. UHF detection of PD in power transformers:
The influence of disturbances. In Proceedings of the Conference on Electrical Insulation and Dielectric
Phenomena (CEIDP), Cancun, Mexico, 16–19 October 2011; pp. 436–439. [CrossRef]
28. Cavallini, A.; Rubio Serrano, J.; Azcarraga Ramos, C.G.; Garcia Souto, J.A.; Montanari, G.C. Comparison of
ultrasonic, electrical and UHF characteristics of partial discharge emission in oil/paper insulation systems. In
Proceedings of the Conference on Electrical Insulation and Dielectric Phenomena (CEIDP), Cancun, Mexico,
16–19 October 2011; pp. 440–443. [CrossRef]
29. Witos, F.; Gacek, Z. Application of the joint electro-acoustic method for partial discharge investigations
within a power transformer. Eur. Phys. J. Spec. Top. 2008, 154, 1951–6401. [CrossRef]
30. Fuangsoongnern, U.; Plueksawan, W. A measurement technique to identify and locate partial discharge in
transformer with AE and HFCT. Int. J. Eng. Res. Appl. 2014, 4, 51–58.
31. Coenen, S.; Müller, A.; Beltle, M.; Kornhuber, S. UHF and acoustic partial discharge localisation in power
transformers. In Proceedings of the XVII International Symposium on High Voltage Engineering, Hannover,
Germany, 22–26 August 2011.
32. Markalous, S.M.; Tenbohlen, S.; Feser, K. Detection and location of partial discharges in power transformers
using acoustic and electromagnetic signals. IEEE Trans. Dielectr. Electr. Insul. 2008, 15, 1576–1583. [CrossRef]
33. Sikorski, W. Active Dielectric Window: A new concept of combined acoustic emission and electromagnetic
partial discharge detector for power transformers. Energies 2019, 12, 115. [CrossRef]
34. Krimholtz, R.; Leedom, D.A.; Matthaei, G.L. New equivalent circuits for elementary piezoelectric transducers.
Electron. Lett. 1970, 6, 398–399. [CrossRef]
35. Devaney, A.J.; Levine, H. Effective elastic parameters of random composites. Appl. Phys. Lett. 1980, 37,
377–379. [CrossRef]
Energies 2020, 13, 3271 34 of 37

36. Meijer, S.; Gulski, E.; Smit, J.J.; Reijnders, H.F. Sensitivity check for UHF PD detection on power transformers.
In Proceedings of the Conference Record of the 2004 IEEE International Symposium on Electrical Insulation,
Indianapolis, IN, USA, 19–22 September 2004; pp. 58–61. [CrossRef]
37. Judd, M.D.; Farish, O.; Pearson, J.S.; Hampton, B.F. Dielectric windows for UHF partial discharge detection.
IEEE Trans. Dielectr. Electr. Insul. 2001, 8, 953–958. [CrossRef]
38. 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]
39. Jahangir, H.; Akbari, A.; Werle, P.; Szczechowski, J. UHF PD measurements on power transformers-advantages
and limitations. IEEE Trans. Dielectr. Electr. Insul. 2017, 24, 3933–3940. [CrossRef]
40. Zhang, X.; Zhang, G.; Li, Y.; Zhang, J.; Huang, R. On the feasibility of gap detection of power transformer
partial discharge UHF signals: Gap propagation characteristics of electromagnetic waves. Energies 2017, 10,
1531. [CrossRef]
41. Wang, H.; Cheng, J.; Ding, D.; Liu, W.; Liang, H. A feasibility study on bushing tap UHF sensing method
for partial discharge in transformer. In Proceedings of the VIIth International Conference on Condition
Monitoring and Diagnosis (CMD), Perth, Australia, 23–26 September 2018.
42. Coenen, S.; Tenbohlen, S. Location of PD sources in power transformers by UHF and acoustic measurements.
IEEE Trans. Dielectr. Electr. Insul. 2012, 19, 1934–1940. [CrossRef]
43. Li, M.; Guo, C.; Peng, Z. Design of meander antenna for UHF partial discharge detection of transformers.
Sens. Transducers 2014, 171, 232–238.
44. 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]
45. 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.
46. 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. [CrossRef]
47. 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.
[CrossRef]
48. 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. [CrossRef]
49. Polak, F.; Sikorski, W.; Siodla, K. Location of partial discharges sources using sensor arrays. Proceedings of
2014 ICHVE International Conference on High Voltage Engineering and Application, Poznan, Poland, 8–11
September 2014; pp. 1–4. [CrossRef]
50. 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. [CrossRef]
51. 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]
52. 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]
53. Sikorski, W.; Szymczak, C.; Siodła, K.; Polak, F. Hilbert curve fractal antenna for detection and on-line
monitoring of partial discharges in power transformers. Eksploat. I Niezawodn. – Maint. Reliab. 2018, 20,
343–351. [CrossRef]
54. 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. [CrossRef]
55. Szymczak, C.; Sikorski, W. Design and optimization of UHF antennas for partial discharge monitoring in
power transformer. Prz. Elektrotechniczny 2016, 92, 75–79.
Energies 2020, 13, 3271 35 of 37

56. 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]
57. 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]
58. Jahangir, H.; Akbari, A.; Werle, P.; Akbari, M.; Szczechowski, J. UHF characteristics of different types of PD
sources in power transformers. In Proceedings of the Iranian Conference on Electrical Engineering (ICEE),
Tehran, Iran, 2–4 May 2017. [CrossRef]
59. Szymczak, C.; Sikorski, W.; Siodla, K.; Grabia, M. Construction UHF antenna PIFA type for partial discharges
monitoring in power transformer. Prz. Elektrotechniczny 2018, 94, 138–142.
60. Palantei, E.; Madika, R.S.; Baharuddin, M.; Syarif, S. Design of reconfigurable planar inverted-F antenna for
5g implementation. In Proceedings of the IEEE International Conference on Communication, Networks and
Satellite (Comnetsat), Makassar, Indonesia, 1–3 August 2019; pp. 41–46. [CrossRef]
61. El Halaoui, M.; Asselman, H.; Kaabal, A.; Ahyoud, S. Design and simulation of a planar inverted-F antenna
(PIFA) for Wi-Fi and LTE Applications. Int. J. Innov. Appl. Stud. 2014, 9, 1048.
62. Wong, H.; Luk, K.M.; Chan, C.H.; Xue, Q.; So, K.K.; Lai, H.W. Small antennas in wireless communications.
Proc. IEEE 2012, 100, 2109–2121. [CrossRef]
63. Khan, N. Design of planar Inverted-F antenna. Int. J. Adv. Technol. Eng. Sci. 2014, 2, 20–31.
64. Jean-Charles, Y.T.; Ungvichian, V.; Barbosa, J.A. Effects of substrate permittivity on planar inverted-F antenna
performances. J. Comput. 2009, 4, 610–614. [CrossRef]
65. Chattha, H.T.; Huang, Y.; Zhu, X.; Lu, Y. An empirical equation for predicting the resonant frequency of
planar inverted-F antennas. IEEE Antennas Wirel. Propag. Lett. 2009, 8, 856–860. [CrossRef]
66. Khan, M.M.; Abbasi, Q.H.; Ashique, R.H. Comprehensive design and propagation study of a compact dual
band antenna for healthcare applications. J. Sens. Actuator Netw. 2015, 4, 50–66. [CrossRef]
67. Luhaib, S.W.; Quboa, K.M.; Abaoy, B.M. Design and simulation dual-band PIFA antenna for GSM systems.
In Proceedings of the International Multi-Conference on Systems, Signals and Devices, Chemnitz, Germany,
20–23 March 2012; pp. 1–4. [CrossRef]
68. AbuTarboush, H.F.; Nilavalan, R.; Budimir, D.; Al-Raweshidy, H.S. Compact planar inverted-F antenna (PIFA)
for WiMAX application. In Proceedings of the 2009 IEEE Antennas and Propagation Society International
Symposium, North Charleston, SC, USA, 1–5 June 2009; pp. 1–4. [CrossRef]
69. Lawan, A.M.; Su, H.T.; Then, Y.L.; Zhang, J. Parallel dual-slot PIFA for 2.45 GHz rectenna applications. In
Proceedings of the 2018 IEEE Sensors Applications Symposium (SAS), Seoul, Korea, 12–14 March 2018;
pp. 1–4. [CrossRef]
70. Fakih, M.A.; Diallo, A.; Le Thuc, P.; Staraj, R.; Mourad, O.; Rachid, E.A. Optimization of efficient dual
band pifa system for MIMO half-duplex 4G/LTE and full-duplex 5G communications. IEEE Access 2019, 7,
128881–128895. [CrossRef]
71. Gomez-Villanueva, R.; Linares-y-Miranda, R.; Tirado-Mendez, J.A.; Jardon-Aguilar, H. Ultra-wideband
planar inverted-F antenna (PIFA) for mobile phone frequencies and ultra-wideband applications. Prog.
Electromagn. Res. 2013, 43, 109–120. [CrossRef]
72. Gómez-Villanueva, R.; Jardón-Aguilar, H.; Linares-y-Miranda, R. Broadband PIFA antenna for mobile
communications terminals. In Proceedings of the XIth International Conference on Electrical Engineering,
Computing Science and Automatic Control (CCE), Campeche, Mexico, 29 September–3 October 2014; pp. 1–6.
[CrossRef]
73. Anguera, J.; Sanz, I.; Mumbrú, J.; Puente, C. Multiband handset antenna with a parallel excitation of PIFA
and slot radiators. IEEE Trans. Antennas Propag. 2009, 58, 348–356. [CrossRef]
74. Naji, D.K.; Abdul-kareem, A. A dual-band U-slot PIFA antenna with ground slit for RFID applications. J.
Emerg. Trends Comput. Inf. Sci. 2013, 4, 213–219.
75. Mohammad, T.I.; Md. Amanath, U.; Touhidul, A.; Mandeep, J.S.; Mengu, C. Microwave imaging sensor
using low profile modified stacked type planar inverted-F Antenna. Sensors 2018, 18, 2949. [CrossRef]
76. Garbaruk, M.; Mieczynska, M. Design and characteristics of multilayer PIFA antenna for GPS and Bluetooth
systems. Prz. Elektrotechniczny 2018, 94, 91–94.
77. Liu, W.C.; Yeh, F.M.; Ghavami, M. Miniaturized implantable broadband antenna for biotelemetry
communication. Microw. Opt. Technol. Lett. 2008, 50, 2407–2409. [CrossRef]
Energies 2020, 13, 3271 36 of 37

78. Lee, C.M.; Yo, T.C.; Luo, C.H.; Tu, C.H.; Juang, Y.Z. Compact broadband stacked implantable antenna for
biotelemetry with medical devices. Electron. Lett. 2007, 43, 660–662. [CrossRef]
79. Nayeri, P.; Lee, K.F.; Elsherbeni, A.Z.; Yang, F. Dual-band circularly polarized antennas using stacked patches
with asymmetric U-slots. IEEE Antennas Wirel. Propag. Lett. 2011, 10, 492–495. [CrossRef]
80. Panayi, P.K.; Al-Nuaimi, M.; Ivrissimtzis, L.P. Tuning techniques for the planar inverted-F antenna. In
Proceedings of the IEE National Conference on Antennas and Propagation, York, UK, 1 April–March 1999.
[CrossRef]
81. Chattha, H.T.; Huang, Y.; Lu, Y. PIFA bandwidth enhancement by changing the widths of feed and shorting
plates. IEEE Antennas Wirel. Propag. Lett. 2009, 8, 637–640. [CrossRef]
82. Wong, K.L. Compact and Broadband Microstrip Antennas; John Wiley & Sons: Hoboken, NJ, USA, 2004.
83. Wong, K.L.; Yang, K.P. Modified planar inverted-F antenna. Electron. Lett. 1998, 34, 7–8. [CrossRef]
84. Verma, A.; Raghava, N.S. Compact slotted meandered PIFA antenna for wireless applications. J. Adv. Res.
Dyn. Control. Syst. 2017, 9, 28–34.
85. Karilainen, A.O.; Ikonen, P.M.; Simovski, C.R.; Tretyakov, S.A.; Lagarkov, A.N.; Maklakov, S.A.;
Starostenko, S.N. Experimental studies on antenna miniaturisation using magneto-dielectric and dielectric
materials. Iet Microw. Antennas Propag. 2011, 5, 495–502. [CrossRef]
86. Tang, I.T.; Lin, D.B.; Chen, W.L.; Horng, J.H.; Li, C.M. Compact five-band meandered PIFA by using
meandered slots structure. In Proceedings of the IEEE Antennas and Propagation Society International
Symposium, Honolulu, HI, USA, 9–15 June 2007; pp. 653–656. [CrossRef]
87. Chen, H.T.; Wong, K.L.; Chiou, T.W. PIFA with a meandered and folded patch for the dual-band mobile
phone application. IEEE Trans. Antennas Propag. 2003, 51, 2468–2471. [CrossRef]
88. Verma, A.; Saini, G. Design of meandered PIFA antenna for wireless application. Int. J. Adv. Commun.
Technol. 2014, 1, 27–29.
89. Azremi, A.A.H.; Saidatul, N.A.; Ahmad, R.B.; Soh, P.J. A parametric study of broadband planar inverted F
antenna (PIFA) for WLAN application. In Proceedings of the International Conference on Electronic Design,
Penang, Malaysia, 1–3 December 2008; pp. 1–6. [CrossRef]
90. International Electrotechnical Comission (IEC). High-Voltage Test Techniques—Partial Discharge Measurements;
IEC 60270:2000+AMD1:2015 CSV; IEC: Geneva, Switzerland, 2015.
91. Kondrath, N.; Kazimierczuk, M.K. Bandwidth of current transformers. IEEE Trans. Instrum. Meas. 2008, 58,
2008–2016. [CrossRef]
92. Álvarez, F.; Garnacho, F.; Ortego, J.; Sánchez-Urán, M.Á. Application of HFCT and UHF sensors in on-line
partial discharge measurements for insulation diagnosis of high voltage equipment. Sensors 2015, 15,
7360–7387. [CrossRef]
93. Albarracín-Sánchez, R.; Álvarez-Gómez, F.; Vera-Romero, C.A.; Rodríguez-Serna, J.M. Separation of partial
discharge sources measured in the high-frequency range with HFCT sensors using PRPD-teff patterns.
Sensors 2020, 20, 382. [CrossRef]
94. Rodrigo, M.A.; Castro Heredia, L.C.; Munoz, F.A. A Novel Approach for partial discharge measurements on
GIS using HFCT sensors. Sensors 2018, 18, 4482. [CrossRef]
95. Kazimierczuk, M.K. High-Frequency Magnetic Components; John Wiley & Sons: Hoboken, NJ, USA, 2009.
96. Zhu, J.; Yang, L.; Jia, J.; Zhang, Q. The design of Rogowski coil with wide band using for partial discharge
measurements. In Proceedings of the International Symposium on Electrical Insulating Materials (ISEIM),
Kitakyushu, Japan, 5–9 June 2005; Volume 2, pp. 518–521. [CrossRef]
97. Zachariades, C.; Shuttleworth, R.; Giussani, R.; MacKinlay, R. Optimization of a high-frequency current
transformer sensor for partial discharge detection using finite-element analysis. IEEE Sens. J. 2016, 16,
7526–7533. [CrossRef]
98. Kunicki, M. Comparison of capacitive and inductive sensors designed for partial discharges measurements
in electrical power apparatus. In Proceedings of the E3S Web of Conferences, EDP Sciences, Polanica-Zdrój,
Poland, 13–15 September 2017; Volume 19, p. 01035.
99. Vedral, J.; Kříž, M. Signal processing in partial discharge measurement. Metrol. Meas. Syst. 2010, 7, 55–64.
[CrossRef]
100. Baker, P.C.; Judd, M.D.; McArthur, S.D.J. A frequency-based RF partial discharge detector for low-power
wireless sensing. IEEE Trans. Dielectr. Electr. Insul. 2010, 17, 133–140. [CrossRef]
Energies 2020, 13, 3271 37 of 37

101. IEEE Instrument and Measurement Society. IEEE Standard for a Precision Clock Synchronization Protocol
for Networked Measurement and Control Systems. IEEE 1588–2008 Standard. Available online: https:
//ieeexplore.ieee.org/document/4579760 (accessed on 24 June 2020).
102. IEEE Standard for Local and Metropolitan Area Networks—Timing and Synchronization for Time-Sensitive
Applications. P802.1AS-Rev. Available online: https://ieeexplore.ieee.org/document/5741898 (accessed on
24 June 2020).
103. International Electrotechnical Comission (IEC). Communication Networks and System for Utility
Automation—Part 1: Introduction and Overview, 2nd ed.; IEC 61850-1; IEC: Geneva, Switzerland, 2012.
104. International Electrotechnical Comission (IEC). Measuring Relays and Protection Equipment—Part 26:
Electromagnetic Compatibility Requirement; IEC 60255-26:2013; IEC: Geneva, Switzerland, 2013.
105. Walczak, K. Zdalny pomiar napiecia na stanowisku probierczym w laboratorium wysokich napiec. Prz.
Elektrotechniczny 2008, 84, 207–209.
106. International Electrotechnical Comission (IEC). Power Transformers—Part 3: Insulation Levels, Dielectric Tests
and External Clearances in Air; IEC 60076-3:2013; IEC: Geneva, Switzerland, 2013.
107. Wang, H.; Li, C.; He, H. Influence of temperature to developing processes of surface discharges in oil-paper
insulation. In Proceedings of the IEEE International Symposium on Electrical Insulation (ISEI), San Diego,
CA, USA, 6–9 June 2010. [CrossRef]
108. Lia, R.; Hao, J.; Chen, G.; Ma, Z.; Yang, L. A comparative study of physicochemical, dielectric and thermal
properties of pressboard insulation impregnated with natural ester and mineral oil. IEEE Trans. Dielectr.
Electr. Insul. 2011, 18, 1626–1637. [CrossRef]
109. Buerschaper, B.; Kleboth-Lugova, O.; Leibfried, T. The electrical strength of transformer oil in a
transformerboard-oil system during moisture non-equilibrium. In Proceedings of the Conference on
Electrical Insulation and Dielectric Phenomena, Albuquerque, NM, USA, 19–22 October 2003; pp. 269–272.
[CrossRef]
110. Niasar, M.G. Partial Discharge Signatures of Defects in Insulation Systems Consisting of Oil and
Oil-impregnated Paper. Ph.D. Thesis, KTH Royal Institute of Technology, Stockholm, Sweden, 2012.
111. Borsi, H.; Schroder, U. Initiation and Formation of partial discharges in mineralbased insulating oil. IEEE
Trans. Dielectr. Electr. Insul. 1994, 1, 419–425. [CrossRef]
112. Dai, J.; Wang, Z.D.; Jarman, P. Moisture and aging effect on the creepage discharge characteristics at
the oil/transformer-board interface under divergent field. In Proceedings of the Conference on Electrical
Insulation Dielectric Phenomena, Quebec, QC, Canada, 26–29 October 2008. [CrossRef]
113. Partial Discharges in Transformer Insulation-CIGRE Task Force 15.01.04. Paris. 2000. Available online:
https://e-cigre.org/publication/676-partial-discharges-in-transformers (accessed on 24 June 2020).
114. Przybylek, P. Water saturation limit of insulating liquids and hygroscopicity of cellulose in aspect of moisture
determination in oil-paper insulation. IEEE Trans. Dielectr. Electr. Insul. 2016, 23, 1886–1893. [CrossRef]
115. Gielniak, J.; Graczkowski, A.; Moranda, H.; Przybylek, P.; Walczak, K.; Nadolny, Z.; Moscicka-Grzesiak, H.;
Feser, K.; Gubanski, S.M. Moisture in cellulose insulation of power transformers-statistics. IEEE Trans.
Dielectr. Electr. Insul. 2013, 20, 982–987. [CrossRef]
116. Przybylek, P.; Nadolny, Z.; Moscicka-Grzesiak, H. Bubble effect as a consequence of dielectric losses in
cellulose insulation. IEEE Trans. Dielectr. Electr. Insul. 2010, 17, 913–919. [CrossRef]
117. Gil, W.; Maslowski, W.; Wronek, P.; Jechorek, E.; Sikorski, W.; Walczak, K. Development and Implementation of
Partial Discharges On-Line Monitoring Module in GIS 110kV Switchgears; CIGRE Session 2020; CIGRE: Paris,
France, 2020; accepted for printing.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).