Design and Application of a Metamaterial Superstrate on a Bio-Inspired Antenna for Partial Discharge Detection Through Dielectric Windows

sensors

Article Design and Application of a Metamaterial Superstrate on a Bio-Inspired Antenna for Partial Discharge Detection through Dielectric Windows

George Victor Rocha Xavier 1,* , Alexandre Jean René Serres 2 , Edson Guedes da Costa 2 , Adriano Costa de Oliveira 1 , Luiz Augusto Medeiros Martins Nobrega 2 and Vladimir Cesarino de Souza 3

1 Post-Graduate Program of Electrical Engineering (PPgEE), Federal University of Campina Grande (UFCG), AprigioVeloso 882, Campina Grande 58429-900, Brazil; [email protected] 2 Department of Electrical Engineering (DEE), Federal University of Campina Grande (UFCG), AprigioVeloso 882, Campina Grande 58429-900, Brazil; [email protected] (A.J.R.S.); [email protected] (E.G.d.C); [email protected] (L.A.M.M.N.) 3 Companhia Hidrelétrica do São Francisco (Chesf), Grisbert de Oliveira Gonzaga Street, Campina Grande 58418-105, Brazil; [email protected] * Correspondence: [email protected]; Tel.: +55-079-99999-9317

Received: 18 July 2019; Accepted: 23 September 2019; Published: 30 September 2019

Abstract: The adaptation of dielectric windows as metamaterial superstrate over a bio-inspired Printed Monopole Antenna (PMA) was evaluated in order to improve the detection sensitivity of Ultra High Frequency (UHF) sensors designed for Partial Discharge (PD) measurement. For this purpose, rectangular and circular Split Ring Resonators (SRR) structures were designed and evaluated aiming to achieve a metamaterial superstrate that improves the characteristics of the bio-inspired PMA as the gain, bandwidth, and radiation pattern. Measurements of the PMA with metamaterial superstrate were carried out in an anechoic chamber and compared to the simulations performed. The results show that the metamaterial superstrate insertion did not impact the original operating bandwidth, covering most of the characteristic frequency range of PD activity. Moreover, this insertion resulted in a mean gain enhancement of 0.7 dBi regarding the reference PMA, resulting in an antenna with better sensitivity for PD detection (mean gain of 3.61 dBi). The PMA-metamaterial set PD detection sensitivity was evaluated through laboratory tests with a point-to-plane PD generator setup and in ﬁeld with measurements from a 230 kV current transformer. The developed PMA-metamaterial set was able to detect, successfully, the activity of PD for both tests, being classiﬁed as an optimized sensor for PD detection through dielectric windows.

Keywords: bio-inspired design; metamaterial; gain enhancement; dielectric windows; partial discharges; printed monopole antenna; UHF sensors; monitoring

1. Introduction High voltage equipment insulating systems are subjected to stressful conditions, such as intense electric fields, chemical reactions, mechanical efforts, temperature variations, and several environmental phenomena. Due to the occurrence of these stresses, low magnitude electrical discharges that partially short-circuit the insulating material may occur. This phenomenon is defined as partial discharge (PD) [1]. The continuous action of the PDs can induce a significant degradation of the insulating material, which can eventually lead to a full dielectric breakdown and, consequently, equipment failure.

Sensors 2019, 19, 4255; doi:10.3390/s19194255 www.mdpi.com/journal/sensors Sensors 2019, 19, 4255 2 of 27

Therefore, the continuous monitoring of PD activity in high voltage equipment is crucial to prevent the developmentSensors 2019, 19, ofx dielectric problems [2]. 2 of 27 The most traditional method used for PD detection is the one established by the IEC 60270 standardThe [most3], which traditional allows themethod measurement used for ofPD current detection pulses is the emitted one established during the occurrenceby the IEC of 60270 PDs. However,standard [3], this which method allows is considered the measurement highly invasive, of curre sincent pulses it requires emitted an during electrical the connection occurrence with of PDs. the monitoredHowever, this equipment method through is considered a coupling highly capacitor. invasive, since it requires an electrical connection with the monitoredIn order to equipment overcome through the practical a coupling limitations capacitor. of the IEC 60270 standard method for online monitoring,In order researchers to overcome have the been practical studying limitations several alternative of the IEC methodologies 60270 standard for PDmethod detection for online [4–8]. Amongmonitoring, these researchers methods, the have radiometric been studying ones are several highlighted alternative due to theirmethodologies non-invasive for nature, PD detection since there [4– is8]. no Among need of these an electrical methods, connection the radiometric between theones sensor are highlighted and the monitored due to equipment. their non-invasive The radiometric nature, orsince Ultra there High is Frequencyno need of (UHF)an electrical method connection consists of between the detection the sensor of electromagnetic and the monitored waves equipment. emitted by PDThe current radiometric pulses, or whichUltra High propagate Frequency through (UHF) the insulationmethod consists system of of the high detection voltage of equipment electromagnetic in the UHFwaves frequency emitted by range PD (300current MHz pulses, to 3 GHz). which The propagate UHF operation through frequency the insulation range system isolates of thehigh detection voltage systemequipment from in most the electromagneticUHF frequency interferencerange (300 inMHz substations, to 3 GHz). such The as UHF power operation electronics frequency switching range and coronaisolates discharges, the detection since system the signals from generatedmost electromagnetic by these phenomena interference have in components substations, with such significant as power energyelectronics for frequencies switching and between corona 200 discharges, and 300 MHz sinc [9e]. the Therefore, signals thegenerated UHF method by these is characterizedphenomena have as a non-invasivecomponents andwith interference-resistant significant energy for technique, frequenc makingies between it attractive 200 and for 300 the MHz continuous [9]. Therefore, monitoring the ofUHF high method voltage is equipment. characterized as a non-invasive and interference-resistant technique, making it attractiveIn order for tothe improve continuous the monitoring detection sensitivity of high voltage of UHF equipment. sensors, researchers studied the spectral featuresIn order of the to PD improve radiation, the concluding detection sensitivity that, generally, of UHF most sensors, of the radiatedresearchers energy studied is concentrated the spectral betweenfeatures of 300 the MHz PD radiation, and 1.5 GHz concluding [9,10]. One that, of genera the pioneeringlly, most of papers the radiated regarding energy the is application concentrated of UHFbetween sensors 300 inMHz PD and detection 1.5 GHz is presented[9,10]. One in of [11 the], which pioneering reports papers the detection regarding and the location application of PDs of inUHF a 420 sensors kV Gas in PD Insulated detection Substation is presented (GIS), in [11], attesting which the reports efficiency the detection of the UHF and method. location Overof PDs the in years,a 420 kV other Gas researches Insulated haveSubstation proved (GIS), the e ffiattestingciency andthe efficiency reliability of of the the UHF UHF method. method Over in GISs the [12 years,,13], andother its researches application have was proved extended the efficiency to other high and voltage reliability equipment, of the UHF mainly method power in GISs transformers [12,13], and and its highapplication voltage was cable extended connections to other [10,14 high,15]. voltage equipment, mainly power transformers and high voltageFor cable power connections transformers [10,14,15]. and GIS, the allocation of UHF sensors is usually made through dielectric windows,For power as presented transformers in Figure and1. GIS, Generally, the allocati a dielectricon of windowUHF sensors can beis definedusually asmade an opening through covereddielectric by windows, dielectric as material presented on the in equipmentFigure 1. Ge tank,nerally, providing a dielectric a propagation window waycan be between defined the as PD an radiatedopening wavescovered and by the dielectric installed UHFmaterial sensor on [16th].e equipment tank, providing a propagation way between the PD radiated waves and the installed UHF sensor [16].

Cable

Sensor

Dielectric Window

Dielectric Layer Tank FigureFigure 1.1. Sensor allocation throughthrough aa dielectricdielectric windowwindow onon aa powerpower transformertransformer [[14,17].14,17].

TheThe UHFUHF sensorssensors appliedapplied inin dielectricdielectric windowswindows forfor PDPD detection can assumeassume several structures, suchsuch asas looploop electrode,electrode, diskdisk electrode,electrode, conicalconical coupler,coupler, andand antennasantennas (monopole,(monopole, dipole,dipole, microstrip,microstrip, andand others)others) [[18,19].18,19]. Among these sensors,sensors, printedprinted monopolemonopole antennasantennas (PMA)(PMA) areare highlightedhighlighted forfor havinghaving desirabledesirable characteristics characteristics for for practical practical applications, applications, such assuch low as cost, low attractive cost, attractive radiation radiation patterns, wideband,patterns, wideband, and ease ofand installation ease of installation/construction/construction [20]. However, [20]. However, PMA with PMA traditional with traditional radiating elementradiating (patch) element shapes (patch) designed shapes to designed operate into UHFoperate range in wouldUHF range assume would relatively assume large relatively dimensions, large limitingdimensions, their limiting use for practicaltheir use applicationsfor practical inapplications dielectric windows. in dielectric windows. The operating frequency of PMA is directly related to the perimeter of the patch. The greater the perimeter, the lower the operating frequency. Therefore, techniques used for PMA miniaturization mainly seek optimized geometries that maximize the patch perimeter/area ratio. These optimized geometries can be addressed in the shapes of living beings developed in order to provide greater efficiency regarding the ability to survive. Hence, aiming to achieve a better radiating efficiency, bio-inspired antennas use the shapes of plants or animals as basis for their design [21–26].

Sensors 2019, 19, 4255 3 of 27

The operating frequency of PMA is directly related to the perimeter of the patch. The greater the perimeter, the lower the operating frequency. Therefore, techniques used for PMA miniaturization mainly seek optimized geometries that maximize the patch perimeter/area ratio. These optimized geometries can be addressed in the shapes of living beings developed in order to provide greater efficiency regarding the ability to survive. Hence, aiming to achieve a better radiating efficiency, bio-inspired antennas use the shapes of plants or animals as basis for their design [21–26]. In addition, in order to detect the inception of PD (low intensity pulses), the application of PMAs with high mean gain is required [27], making the evaluation of gain enhancement technique assimilation on PMAs designed for PD detection interesting. Among the gain improvement techniques for PMA, the use of metamaterials stands out in the literature due to its characteristic of double negativity (DNG) for magnetic permeability and for electrical permittivity values [28–31]. The DNG feature results in the decreasing of material reflection and refractive indexes, acting as a converging lens that concentrates the electromagnetic waves in a given propagation medium, resulting in gain enhancement [32,33]. The application of metamaterials can be implemented in different ways, such as changes in the antenna patch [34], modifications in the ground plane [35], and applications as dielectric superstrates [36]. As can be seen from Figure1, dielectric windows have a dielectric layer that naturally works as a superstrate. Thus, it is possible to take advantage of the dielectric layer to improve the gain of the antennas coupled to it by adding metamaterials in its structure. Therefore, the application of metamaterials in dielectric superstrates presents a greater practical feasibility for the detection of PD, since dielectric windows can be easily adapted for this purpose. In addition, the use of dielectric superstrates with metamaterials presents better gain enhancement results for broadband applications, as PD detection, in comparison to the application of modifications in the ground plane or antenna patch [37–39]. Therefore, the main objective of this paper is to design a metamaterial for gain enhancement that can be used as superstrate over a PMA developed for PD detection through dielectric windows. The PMA used as reference is a bio-inspired structure based on the Jatropha mollissima (Pohl) Baill leaf that was developed and presented in a previous work [26] and meets the size, bandwidth, and gain values requirements for PD detection through dielectric windows in high voltage equipment. Hence, this work aims to improve the PD sensitivity of detection of the structure presented in [26] through the application of a metamaterial superstrate representing a dielectric window, allowing a more accurate detection of PD inception (low intensity pulses). The designed metamaterial is composed by the combination of Split Ring Resonators (SRR) and Capacitive Load Strips (CLS). The bio-inspired PMA with metamaterial superstrate was subjected to bandwidth and gain measurement tests (in an anechoic chamber) and sensitivity tests of PD detection by means of comparative results with the IEC 60270 standard method connected to a PD generator laboratory setup. Finally, in order to evaluate the PD sensitivity of detection in practice, the bio-inspired PMA with metamaterial superstrate was tested in a substation to detect the PD activity on a 230 kV current transformer.

2. Printed Monopole Antenna (PMA) PMA have a simple structure and great ease of construction and installation. In addition, features such as large bandwidth make this type of antenna suitable for PD application [26]. The general structure of a PMA is presented in Figure2, in which W0 and L0 are the antenna length and width, respectively, L and W are the length and width of the patch, respectively, Lg and Wg are the length and width of the truncated ground plane, respectively, Wf is the width of the transmission line, g is the distance between the patch and the ground plane, and h is the substrate thickness. Sensors 2019, 19, x 3 of 27

In addition, in order to detect the inception of PD (low intensity pulses), the application of PMAs with high mean gain is required [27], making the evaluation of gain enhancement technique assimilation on PMAs designed for PD detection interesting. Among the gain improvement techniques for PMA, the use of metamaterials stands out in the literature due to its characteristic of double negativity (DNG) for magnetic permeability and for electrical permittivity values [28–31]. The DNG feature results in the decreasing of material reflection and refractive indexes, acting as a converging lens that concentrates the electromagnetic waves in a given propagation medium, resulting in gain enhancement [32,33]. The application of metamaterials can be implemented in different ways, such as changes in the antenna patch [34], modifications in the ground plane [35], and applications as dielectric superstrates [36]. As can be seen from Figure 1, dielectric windows have a dielectric layer that naturally works as a superstrate. Thus, it is possible to take advantage of the dielectric layer to improve the gain of the antennas coupled to it by adding metamaterials in its structure. Therefore, the application of metamaterials in dielectric superstrates presents a greater practical feasibility for the detection of PD, since dielectric windows can be easily adapted for this purpose. In addition, the use of dielectric superstrates with metamaterials presents better gain enhancement results for broadband applications, as PD detection, in comparison to the application of modifications in the ground plane or antenna patch [37–39]. Therefore, the main objective of this paper is to design a metamaterial for gain enhancement that can be used as superstrate over a PMA developed for PD detection through dielectric windows. The PMA used as reference is a bio-inspired structure based on the Jatropha mollissima (Pohl) Baill leaf that was developed and presented in a previous work [26] and meets the size, bandwidth, and gain values requirements for PD detection through dielectric windows in high voltage equipment. Hence, this work aims to improve the PD sensitivity of detection of the structure presented in [26] through the application of a metamaterial superstrate representing a dielectric window, allowing a more accurate detection of PD inception (low intensity pulses). The designed metamaterial is composed by the combination of Split Ring Resonators (SRR) and Capacitive Load Strips (CLS). The bio-inspired PMA with metamaterial superstrate was subjected to bandwidth and gain measurement tests (in an anechoic chamber) and sensitivity tests of PD detection by means of comparative results with the IEC 60270 standard method connected to a PD generator laboratory setup. Finally, in order to evaluate the PD sensitivity of detection in practice, the bio-inspired PMA with metamaterial superstrate was tested in a substation to detect the PD activity on a 230 kV current transformer.

2. Printed Monopole Antenna (PMA) PMA have a simple structure and great ease of construction and installation. In addition, features such as large bandwidth make this type of antenna suitable for PD application [26]. The general structure of a PMA is presented in Figure 2, in which W0 and L0 are the antenna length and width, respectively, L and W are the length and width of the patch, respectively, Lg and Wg are the Sensors 2019, 19, 4255 4 of 27 length and width of the truncated ground plane, respectively, Wf is the width of the transmission line, g is the distance between the patch and the ground plane, and h is the substrate thickness. h W0

W L L0 Substrate g Patch L g

Wf Ground Plane

Wg FigureFigure 2. 2. PrintedPrinted Monopole Monopole Antenna Antenna (PMA) (PMA) general general structure. structure.

Sensors 2019, 19, x 4 of 27 According to [21], the electric current density in a PMA is greater at the edges of the patch. Therefore,According an increase to [21], in the the electric patch perimeter current density would alsoin a increasePMA is thegreater wavelength at the edges and, consequently,of the patch. decreaseTherefore, the an lower increase operating in the patch frequency. perimeter Hence, would the loweralso increase frequency the ofwavelength the PMA operating and, consequently, band can bedecrease approximated the lower by operating the patch frequency. perimeter, Hence, as presented the lower in Equation frequency (1) of [21 the]. PMA operating band can be approximated by the patch perimeter, as presented300 in Equation (1) [21]. f (GHz) = , (1) p √300εre f 𝑓(𝐺𝐻𝑧) = , (1) 𝑝𝜀 where p is the patch perimeter and εre f is the relative permittivity of the dielectric, approximated aswhere [21]: p is the patch perimeter and εre f is the relative permittivity of the dielectric, approximated as [21]: 300 f (GHz) = 300 . (2) 𝑓(𝐺𝐻𝑧) = p √εre f . (2) 𝑝𝜀 3. Metamaterials: Concept and Design 3. Metamaterials:According to [Concept28], from and Maxwell Design equations, it is possible to classify the materials in four diﬀerent classesAccording according to to their[28], valuesfrom ofMaxwell magnetic equations, permeability it is (µ possible) and electrical to classify permittivity the materials (ε), as presented in four indifferent Figure 3classes. according to their values of magnetic permeability (µ) and electrical permittivity (ε), as presented in Figure 3. µ

(0,µε < > 0) (0,µε > > 0)

(0,µ<ε < 0)(0,µε > < 0)

Figure 3. Material classiﬁcationclassification based on electrical permittivitypermittivity andand magneticmagnetic permeabilitypermeability values.values.

TheThe firstfirst quadrantquadrant ofof FigureFigure3 (3 (εε >> 0,0, µµ >> 0)0) representsrepresents thethe materialsmaterials inin whichwhich thethe interactioninteraction betweenbetween thethe electric electric and and magnetic magnetic vectors vectors (E and(E and H, respectively)H, respectively) results results in the in propagation the propagation of a vector of a wavevector K wave with K direction with direction defined defined by the right by the hand right rule, hand being rule, denominated being denominated as RHM (Rightas RHM Handed (Right Materials).Handed Materials). The RHM The represents RHM represents most of the most materials of the commonlymaterials commonly found innature found andin nature presents and a straightpresents wave a straight propagation wave propagation mode [40]. mode [40]. TheThe secondsecond quadrantquadrant ((εε << 0, µµ >> 0) describesdescribes thethe electricelectric plasmasplasmas inin whichwhich evanescentevanescent waveswaves propagatepropagate withwith higher higher e efficiencyfficiency [41 [41].]. The The materials materials that that are are represented represented by by this this quadrant quadrant are are known known as ENGas ENG (Epsilon (Epsilon Negative). Negative). Similarly, Similarly, the fourth the quadrantfourth quadrant (ε > 0, µ (<ε 0)> also0, µ presents < 0) also greater presents propagation greater propagation of evanescent waves, but for magnetic media. Therefore, the materials in this quadrant are called MNG (Mu Negative). Finally, the third quadrant (ε < 0, µ < 0) represents the metamaterials, or Double Negative (DNG) materials. Due to its double negative characteristic, the refractive index, calculated from Maxwell’s equations, also assumes negative values. The negative refractive index condition reflects in an interaction between E and H that results in the propagation of a K wave vector governed by the left hand rule. Thus, metamaterials can also be referred as LHM (Left Handed Materials) and have a reverse propagation mode regarding RHM. Besides the presentation of the characteristics described for ENG and MNG materials, the LHM have a negative refractive index that enables these materials to be able to concentrate incident electromagnetic waves in the medium in which they are applied. Hence, LHM can be defined as converging lenses that increase the focus of electromagnetic waves, both incident and evanescent waves, in a given dielectric medium. The use of LHM as electromagnetic wave lenses has a useful practical application in microwave engineering, since a higher concentration of electromagnetic waves on the surface of an antenna, for example, can result in a considerable increase in its gain [42–44].

Sensors 2019, 19, 4255 5 of 27 of evanescent waves, but for magnetic media. Therefore, the materials in this quadrant are called MNG (Mu Negative). Finally, the third quadrant (ε < 0, µ < 0) represents the metamaterials, or Double Negative (DNG) materials. Due to its double negative characteristic, the refractive index, calculated from Maxwell’s equations, also assumes negative values. The negative refractive index condition reflects in an interaction between E and H that results in the propagation of a K wave vector governed by the left hand rule. Thus, metamaterials can also be referred as LHM (Left Handed Materials) and have a reverse propagation mode regarding RHM. Besides the presentation of the characteristics described for ENG and MNG materials, the LHM have a negative refractive index that enables these materials to be able to concentrate incident electromagnetic waves in the medium in which they are applied. Hence, LHM can be defined as converging lenses that increase the focus of electromagnetic waves, both incident and evanescent waves, in a given dielectric medium. The use of LHM as electromagnetic wave lenses has a useful practical application in microwave engineering, since a higher concentration of electromagnetic waves on the surface of an antenna, Sensorsfor example, 2019, 19, canx result in a considerable increase in its gain [42–44]. 5 of 27 However, materials such as ENG, MNG, and LHM are not found in nature, requiring the manufactureHowever, of materials resonant structuressuch as ENG, that emulate MNG, suchand LHM materials. are not found in nature, requiring the manufactureThe behavior of resonant of an ENG-likestructures material that emulate can be such emulated materials. from Capacitive Load Strips (CLS) that, for certainThe behavior frequency of ranges,an ENG-like may have material higher can electron be emulated concentrations from Capacitive and an increase Load Strips in their (CLS) effective that, formass, certain resulting frequency in a negative ranges, electrical may have permittivity higher electron [29]. The concentrations basic structure and of aan CLS increase is presented in their in effectiveFigure4. mass, resulting in a negative electrical permittivity [29]. The basic structure of a CLS is presented in Figure 4.

Figure 4. ExampleExample of of Capacitive Capacitive Load Load Strips Strips (CLS) (CLS) basic basic structure.

For MNG-type materials, the emulation of their characteristics can be obtained from the use of a For MNG-type materials, the emulation of their characteristics can be obtained from the use of a Split Ring Resonator (SRR). For speciﬁc frequency ranges, the electromagnetic interactions between Split Ring Resonator (SRR). For specific frequency ranges, the electromagnetic interactions between the characteristic inductance of the rings and the capacitance present between their spacing results in the characteristic inductance of the rings and the capacitance present between their spacing results in high energy densities that characterize the material as MNG [30]. high energy densities that characterize the material as MNG [30]. Therefore, from the joint application of CLS and SRR, it is possible to obtain a metamaterial that can Therefore, from the joint application of CLS and SRR, it is possible to obtain a metamaterial that be used as superstrate of antennas, resulting in the gain enhancement of these structures [36,39,45–47]. can be used as superstrate of antennas, resulting in the gain enhancement of these structures A more detailed physic-mathematical explanation regarding the functioning of SRR and CLS [36,39,45–47]. structures,A more as detailed well as thephysic-mathematical interaction between explanation them, can beregarding found in the [29 –functioning31,41,48,49]. of SRR and CLS structures,The metamaterial as well as the operating interaction frequency between can them, be deﬁnedcan be found from thein [29–31,41,48,49]. resonance frequency of the SRR geometryThe metamaterial adopted. Although operating the frequency SRR can be can designed be defined with from several the resonance shapes [50 frequency–52], the circular of the SRR and geometryrectangular adopted. shapes representAlthough thethe mostSRR can adopted be designed geometries with in several practice shapes due to [50–52], their lower the circular complexity and rectangularand ease of construction.shapes represent Therefore, the most in adopted this paper, geomet only circularries in practice and rectangular due to their SRRs lower were complexity evaluated andfor the ease design of construction. of the proposed Therefore, metamaterial. in this paper, only circular and rectangular SRRs were evaluated for theThe design general of the structures proposed of circularmetamaterial. and rectangular SRR are presented in Figure5, respectively. The general structures of circular and rectangular SRR are presented in Figure 5, respectively. l1

r 2 Scirc Srect Wm

(a) (b)

Figure 5. Examples of Split Ring Resonator (SRR) geometries: (a) circular; (b) rectangular.

For a circular SRR, the resonance frequency can be approximated by Equations (3) and (4) [53]. 𝑓 = , 𝑓 = , (3)

𝑅 =2𝜋𝑟 −𝑆 , 𝑅 =2𝜋𝑟 −𝑆, (4) where c represents the speed of the light, r1 and r2 represent the external and internal mean radius length, respectively, and Scirc is the existing gap in each ring with external and internal resonance frequencies defined by fC1 and fC2, respectively. Similarly, for the rectangular SRR [53]:

Sensors 2019, 19, x 5 of 27

However, materials such as ENG, MNG, and LHM are not found in nature, requiring the manufacture of resonant structures that emulate such materials. The behavior of an ENG-like material can be emulated from Capacitive Load Strips (CLS) that, for certain frequency ranges, may have higher electron concentrations and an increase in their effective mass, resulting in a negative electrical permittivity [29]. The basic structure of a CLS is presented in Figure 4.

Figure 4. Example of Capacitive Load Strips (CLS) basic structure.

For MNG-type materials, the emulation of their characteristics can be obtained from the use of a Split Ring Resonator (SRR). For specific frequency ranges, the electromagnetic interactions between the characteristic inductance of the rings and the capacitance present between their spacing results in high energy densities that characterize the material as MNG [30]. Therefore, from the joint application of CLS and SRR, it is possible to obtain a metamaterial that can be used as superstrate of antennas, resulting in the gain enhancement of these structures [36,39,45–47]. A more detailed physic-mathematical explanation regarding the functioning of SRR and CLS structures, as well as the interaction between them, can be found in [29–31,41,48,49]. The metamaterial operating frequency can be defined from the resonance frequency of the SRR geometry adopted. Although the SRR can be designed with several shapes [50–52], the circular and rectangular shapes represent the most adopted geometries in practice due to their lower complexity andSensors ease2019 of, 19 construction., 4255 Therefore, in this paper, only circular and rectangular SRRs were evaluated6 of 27 for the design of the proposed metamaterial.

r 2 Scirc Srect Wm

(a) (b)

FigureFigure 5. 5. ExamplesExamples of of Split Split Ring Ring Re Resonatorsonator (SRR) (SRR) geometries: geometries: ( (aa)) circular; circular; ( (bb)) rectangular. rectangular.

For a circular SRR, the resonance frequency can be approximated by Equations (3) and (4)(4) [[53].53]. c c 𝑓 = , 𝑓 = , (3) fC1 = , fC2 = , (3) 2R1 √εre f 2R2 √εre f 𝑅 =2𝜋𝑟 −𝑆 , 𝑅 =2𝜋𝑟 −𝑆, (4) R1 = 2πr1 Scirc , R2 = 2πr2 Scirc, (4) where c represents the speed of the light, r−1 and r2 represent −the external and internal mean radius length,where crespectively,represents the and speed Scirc is of the the existing light, r1 andgap rin2 representeach ring thewith external external and and internal internal mean resonance radius frequencieslength, respectively, defined by and fC1 Sandcirc isfC2, the respectively. existing gap Similarly, in each for ring the with rectangular external SRR and [53]: internal resonance frequencies deﬁned by fC1 and fC2, respectively. Similarly, for the rectangular SRR [53]: c c fR1 = , fR2 = , (5) 2L1 √εre f 2L2 √εre f

L = 4l Srect 4Wm , L = 4l Srect 4Wm, (6) 1 1 − − 2 2 − − where l1 and l2 represent the external and internal side length, respectively, and Wm is the width of each resonance ring with external and internal resonance frequencies deﬁned by fR1 and fR2, respectively. In order to verify if the designed SRR–CLS set represents a metamaterial (ε < 0, µ < 0), the ε and µ values can be estimated from the refraction index (η) and impedance (z) values, as presented in Equations (7) and (8), respectively [54]. η ε = , (7) z µ = ηz, (8)

where z and η can be defined in terms of coefficients of reflection (S11) and transmission (S21) of the designed set, as presented in Equations (9) and (10), respectively [54]. v u 2 tu + 2 2 1 S11 S21 z = − , (9) 2 ± 1 S2 S2 − 11 − 21 1 n h i h io η = Re ln einkl Im ln einkl , (10) kl −

in which, the S11 and S21 values are represented by its respective magnitude and phase components, Re[] and Im[] represent the real and imaginary components, respectively, k represents the wave number, l the longest metamaterial unit cell length, and einkl is deﬁned as [54]:

S einkl = 21 . (11) z 1 1 S − − 11 z+1 Sensors 2019, 19, 4255 7 of 27

4. Material and Methods The applied methodology was divided in three different stages: computational procedures, laboratory experimental tests, and field application. In the computational procedures, the software High Frequency Structure Simulator (HFSS) from ANSYS Electronics Desktop was used for all the simulation and design processes of the metamaterial (individually and positioned as superstrate). In all the simulations, the substrate is a low-cost fiberglass (FR-4) with εr = 4.4, h = 1.6 mm and loss tangent (δ) equal to 0.02. Moreover, the simulations were performed for the frequency of 200 MHz up to 1600 MHz with a discrete sweep type with 201 points. The solution frequency was centered in 1 GHz (central frequency for the designed antenna) with a maximum number of passes and Delta S equals to 6 and 0.02, respectively. The simulated PMA and metamaterial bandwidths were defined as the entire frequency range at which the reflection coefficient values were below the 10 dB threshold. In addition, the gain values − were also extracted for the individual PMA and the set PMA-metamaterial in order to evaluate the gain enhancement provided by the metamaterial application as superstrate on the bio-inspired designed PMA. Lastly, the radiation patterns were obtained through simulations in order to evaluate if the metamaterial superstrate application resulted into distortions in the original bio-inspired PMA patterns. In the experimental procedures, the designed PMA and the set PMA-metamaterial were built and subjected to comparative measurements of reflection coefficient and gain in an anechoic chamber in order to reduce the effect of external interferences and signal reflections during the tests. Then, the set PMA-metamaterial was used in a PD generator setup in order to estimate the PD detection sensitivity

Sensorsof the 2019 set, regarding19, x IEC 60270 standard method. 7 of 27 Finally, the PMA-metamaterial set was subjected to field application for the detection of partialFinally, discharges the PMA-metamaterial in a 230 kV Current set was Transformer subjected to (CT) field from application a Brazilian for the substation, detection Mussur of partialé II discharges(Chesf—Companhia in a 230 HidrokV Current Elétrica doTransformer São Francisco). (CT) from a Brazilian substation, Mussuré II (Chesf—CompanhiaIn the following subsections,Hidro Elétrica the do computational São Francisco). procedures, laboratory experimental tests, and field applicationIn the following are detailed. subsections, the computational procedures, laboratory experimental tests, and field application are detailed. 4.1. Bio-Inspired PMA Design 4.1. Bio-InspiredFrom the PMA PMA model Design presented in Figure2 and Equations (1) and (2), the bio-inspired PMA based on theFromJatropha the PMAmollissima model(Pohl) presented Baill leaf in was Figure obtained 2 and in [Equations26]. The final (1) dimensions and (2), the of bio-inspiredthe designed PMAbio-inspired based on PMA the inJatropha [26] and mollissima a photograph ( of the Jatropha mollissima (Pohl) Baill leaf are presented in Figure6.

13.2

10 1 1.6

0 3

2 3.6 1.5 1 6 1.7

7.5 0.3

20 (a) (b) Figure 6. Bio-inspired PMA: (a) Jatropha mollissima (Pohl) Baill leaf; (b) details of the final designed model Figure 6. Bio-inspired PMA: (a) Jatropha mollissima (Pohl) Baill leaf; (b) details of the ﬁnal designed (all the dimensions are in centimeters) [26]. model (all the dimensions are in centimeters) [26]. 4.2. Metamaterial Unit Cell Simulations Through Equations (3)–(6) and dimensional fine adjustments, the resonance frequencies of the designed metamaterials unit cells were adjusted to present the DNG feature for the bandwidth within the PD activity frequency range (300–1500 MHz). The final dimensions for the designed rectangular and circular SRR-CLS metamaterial unit cells are presented in Figure 7, respectively.

CLS CLS

(a) (b) Figure 7. SRR models for each metamaterial unit cell (the green strips are the CLS structures): (a) rectangular; (b) circular. In order to characterize the models presented in Figure 7 as metamaterials, each model was simulated individually before the application as superstrate on the PMA. For the unit cells evaluation in electric and magnetic terms (ε and µ), special boundary conditions were established as presented in Figure 8, in which each metamaterial unit cell model is surrounded by perfect electric conductor walls (PEC) and perfect magnetic conductor walls (PMC) in parallel and orthogonal orientations regarding the Z axis, respectively. The unit cell model excitation is performed by two Sensors 2019, 19, x 7 of 27

Finally, the PMA-metamaterial set was subjected to field application for the detection of partial discharges in a 230 kV Current Transformer (CT) from a Brazilian substation, Mussuré II (Chesf—Companhia Hidro Elétrica do São Francisco). In the following subsections, the computational procedures, laboratory experimental tests, and field application are detailed.

4.1. Bio-Inspired PMA Design From the PMA model presented in Figure 2 and Equations (1) and (2), the bio-inspired PMA based on the Jatropha mollissima (Pohl) Baill leaf was obtained in [26]. The final dimensions of the designed bio-inspired PMA in [26] and a photograph of the Jatropha mollissima (Pohl) Baill leaf are presented in Figure 6.

13.2

10 1 1.6

0 3

2 3.6 1.5 1 6 1.7

7.5 0.3

20 (a) (b) Sensors 2019, 19, 4255 8 of 27 Figure 6. Bio-inspired PMA: (a) Jatropha mollissima (Pohl) Baill leaf; (b) details of the final designed model (all the dimensions are in centimeters) [26]. 4.2. Metamaterial Unit Cell Simulations 4.2. Metamaterial Unit Cell Simulations Through Equations (3)–(6) and dimensional ﬁne adjustments, the resonance frequencies of the designedThrough metamaterials Equations unit (3)–(6) cells and were di adjustedmensional to fine present adjustments, the DNG featurethe resonance for the bandwidthfrequencies within of the thedesigned PD activity metamaterials frequency unit range cells (300–1500 were adjusted MHz). to present the DNG feature for the bandwidth withinThe the ﬁnal PD dimensions activity frequency for the designedrange (300–1500 rectangular MHz). and circular SRR-CLS metamaterial unit cells are presentedThe final indimensions Figure7, respectively. for the designed rectangular and circular SRR-CLS metamaterial unit cells are presented in Figure 7, respectively.

CLS CLS

(a) (b) Figure 7. SRR models for each metamaterial unit cell (the green strips are the CLS structures): (a) Figure 7. SRR models for each metamaterial unit cell (the green strips are the CLS structures): rectangular; (b) circular. (a) rectangular; (b) circular. In order to characterize the models presented in Figure 7 as metamaterials, each model was simulatedIn order individually to characterize before the models the application presented in Figureas superstrate7 as metamaterials, on the PMA. each model For wasthe simulatedunit cells individuallyevaluation in before electric the and application magnetic as superstrateterms (ε and on µ), the special PMA. Forboundary the unit conditions cells evaluation were inestablished electric and as magneticpresented terms in Figure (ε and 8,µ ),in special which boundary each metamaterial conditions unit were cell established model is as surrounded presented in by Figure perfect8, in electric which eachconductor metamaterial walls (PEC) unit cell and model perfect is surrounded magnetic cond by perfectuctor electricwalls (PMC) conductor in parallel walls (PEC) and andorthogonal perfect Sensorsmagneticorientations 2019 conductor, 19, regardingx walls the (PMC) Z axis, in parallel respectively. and orthogonal The unitorientations cell model regardingexcitation the is performedZ axis, respectively. by8 of two 27 The unit cell model excitation is performed by two waveports. In this way, it is possible to acquire the S11 waveports. In this way, it is possible to acquire the S11 and S21 values used for the calculation of ε and and S21 values used for the calculation of ε and µ, as presented in Equations (5)–(9). From the ε and µ µ, as presented in Equations (5)–(9). From the ε and µ values, it was possible to classify the unit cells values, it was possible to classify the unit cells as metamaterials (ε < 0 and µ < 0). as metamaterials (ε < 0 and µ < 0).

Figure 8. BoundaryBoundary conditions for the simulated SRR-CLS models. 4.3. Metamaterial Superstrate Simulations 4.3. Metamaterial Superstrate Simulations After the characterization as metamaterials, the simulated metamaterial unit cells were positioned After the characterization as metamaterials, the simulated metamaterial unit cells were in a 3 3 array as superstrates over the bio-inspired PMA as presented in Figure9. positioned× in a 3 × 3 array as superstrates over the bio-inspired PMA as presented in Figure 9.

(a) (b)

Figure 9. Metamaterial superstrates positioned over the bio-inspired PMA: (a) rectangular; (b) circular.

The superstrate height was adjusted so that the reflection coefficient values of the bio-inspired PMA were minimally impacted. The optimal superstrate height was defined as 135 and 125 mm for the rectangular and circular SRRs, respectively. In addition, the final array arrangement was also adjusted so that the gain values provided by the metamaterial superstrates were maximized. These adjustments consisted in some physical parameter variations, such as, the interconnection or not between the CLS of each unit cell, the adjustment in the horizontal and vertical spacing between the cells, the number of cells, and the position of the array regarding the PMA. Then, the horizontal and vertical spacing between the SRR unit cells were defined, respectively, as 10 and 15 mm, for the rectangular SRR superstrate, and as 8 and 14 mm, for the circular SRR superstrate. Moreover, the CLS of both unit cell designs were separated by a 5 mm gap. In addition, in order to attest that the gain improvement was actually coming from the inserted metamaterial structures and not from the dielectric material adopted as superstrate (FR-4), the bio-inspired PMA was also simulated with only the dielectric material as superstrate at similar height. After the application of the simulation settings presented at the beginning of this section in Figure 9 models, the following meshing presented in Figure 10 was obtained.

Sensors 2019, 19, x 8 of 27

waveports. In this way, it is possible to acquire the S11 and S21 values used for the calculation of ε and µ, as presented in Equations (5)–(9). From the ε and µ values, it was possible to classify the unit cells as metamaterials (ε < 0 and µ < 0).

Figure 8. Boundary conditions for the simulated SRR-CLS models.

4.3. Metamaterial Superstrate Simulations Sensors 2019, 19, 4255 9 of 27 After the characterization as metamaterials, the simulated metamaterial unit cells were positioned in a 3 × 3 array as superstrates over the bio-inspired PMA as presented in Figure 9.

(a) (b)

FigureFigure 9. 9.Metamaterial Metamaterial superstrates superstrates positioned positioned over over the bio-inspired the bio-inspired PMA: (aPMA:) rectangular; (a) rectangular; (b) circular. (b) circular. The superstrate height was adjusted so that the reflection coefficient values of the bio-inspired PMA wereThe superstrate minimally impacted.height was The adjusted optimal so superstrate that the re heightflection was coefficient defined asvalues 135 and of the 125 bio-inspired mm for the rectangularPMA were andminimally circular impacted. SRRs, respectively. The optimal In addition, superstrate the height final array was arrangementdefined as 135 was and also 125 adjusted mm for sothe that rectangular the gain values and circular provided SRRs, by the respectively. metamaterial In superstrates addition, the were final maximized. array arrangement These adjustments was also consistedadjusted inso somethat the physical gain values parameter provided variations, by the such metamaterial as, the interconnection superstrates orwere not maximized. between the These CLS ofadjustments each unit cell, consisted the adjustment in some inphysical the horizontal parameter and variations, vertical spacing such betweenas, the interconnection the cells, the number or not ofbetween cells, and the the CLS position of each of unit the cell, array the regarding adjustment the in PMA. the horizontal Then, the and horizontal vertical and spacing vertical between spacing the betweencells, the the number SRR unitof cells, cells and were the defined,position respectively,of the array regarding as 10 and the 15 PMA. mm, forThen, the the rectangular horizontal SRR and superstrate,vertical spacing and as between 8 and 14 the mm, SRR for unit the circularcells were SRR defined, superstrate. respectively, Moreover, as the 10 CLS and of 15 both mm, unit for cell the designsrectangular were SRR separated superstrate, by a 5 mmand gap.as 8 and 14 mm, for the circular SRR superstrate. Moreover, the CLSIn of addition,both unit cell in orderdesigns to were attest separated that the by gain a 5 mm improvement gap. was actually coming from the insertedIn addition, metamaterial in order structures to attest and that not the from gain the im dielectricprovement material was actually adopted coming as superstrate from the inserted (FR-4), themetamaterial bio-inspired structures PMA was and also not simulated from the with dielectric only the material dielectric adopted material as as superstrate superstrate (FR-4), at similar the height.bio-inspired After thePMA application was also ofsimulated the simulation with only settings the presenteddielectric atmaterial the beginning as superstrate of this sectionat similar in FigureSensorsheight. 92019 models,After, 19, xthe the application following of meshing the simulation presented settings in Figure presented 10 was obtained.at the beginning of this section9 of 27in Figure 9 models, the following meshing presented in Figure 10 was obtained.

(a) (b)

FigureFigure 10.10. MeshingMeshing plotplot ofof thethe structuresstructures presentedpresentedin inFigure Figure9 :(9: (aa)) rectangular;rectangular; ( (bb)) circular. circular.

4.4.4.4. ReflectionReflection CoeCoefficientfficient and and Gain Gain Measurements Measurements TheThe reflectionreflection coecoefficientfficient values values were were obtained obtained through through an an ENA ENA E5062A E5062A network networkanalyzer analyzerfrom from AgilentAgilent Technologies.Technologies. ForFor thethe gaingain measurements,measurements, anan experimentalexperimental setupsetup composedcomposed byby aa referencereference antennaantenna (Hyperlog(Hyperlog 30100X:30100X: 380380 MHz–10MHz–10 GHz),GHz), withwith aa 4.54.5 dBidBi meanmean gain,gain, andand thethe antennaantenna underunder testtest (AUT)(AUT) positioned positioned beyond beyond the the far far field field distance distance (R) (R) as as presented presented in in Figure Figure 11 11.. The selected far field distance was defined regarding the maximum operation frequency of PD main activity range in oil, i.e., 1500 MHz [9,10]. Therefore,Anechoic Chamber from the final bio-inspired PMA dimensions and the far field equations presented in [20], the value of R was defined as 1.0 m.

Reference Antenna AUT

Support Support R

Figure 11. Schematic of the experimental setup used in the gain measurement test.

The selected far field distance was defined regarding the maximum operation frequency of PD main activity range in oil, i.e., 1500 MHz [9,10]. Therefore, from the final bio-inspired PMA dimensions and the far field equations presented in [20], the value of R was defined as 1.0 m. From the measured values of transmission and reflection coefficients and reference antenna gain (GR = 4.5 dBi), the designed bio-inspired PMA gain (GD) was calculated according to the adapted Friis equation [20]:

𝐺 =𝐺 + |𝑆| − |𝑆| + 1−𝑆 − (1−|𝑆| ) (𝑑𝐵) , (12)

where, S21aut and S21ref represent the transmission coefficient of the antenna under test and reference antenna, respectively, and S11aut and S11ref represent the reflection coefficient of the antenna under test and reference antenna, respectively. For the application of Equation (12) in the gain estimation, it was necessary to assess the performance of two measurements using the schematic presented in Figure 8. The first measurement was performed using two identical reference antennas (Hyperlog 30100X), one positioned at the AUT distance and the other at the reference antenna position. In this way, it was possible to obtain the reference transmission coefficient (S21ref). The second measurement was performed with the PMA

Sensors 2019, 19, x 9 of 27

(a) (b)

Figure 10. Meshing plot of the structures presented in Figure 9: (a) rectangular; (b) circular.

4.4. Reflection Coefficient and Gain Measurements The reflection coefficient values were obtained through an ENA E5062A network analyzer from Agilent Technologies. For the gain measurements, an experimental setup composed by a reference Sensors 2019, 19, 4255 10 of 27 antenna (Hyperlog 30100X: 380 MHz–10 GHz), with a 4.5 dBi mean gain, and the antenna under test (AUT) positioned beyond the far field distance (R) as presented in Figure 11.

Anechoic Chamber

Reference Antenna AUT

Support Support R

FigureFigure 11. 11.Schematic Schematic of of the the experimental experimental setup setup used used in in the the gain gain measurement measurement test. test.

FromThe selected the measured far field values distance of transmission was defined and regarding reflection the coe maximumfficients and operation reference frequency antenna of gain PD (GmainR = 4.5 activity dBi), the range designed in oil, bio-inspired i.e., 1500 PMAMHz gain [9,10]. (GD )Therefore, was calculated from according the final to bio-inspired the adapted FriisPMA equationdimensions [20]: and the far field equations presented in [20], the value of R was defined as 1.0 m. From the measured values of transmission and reflection coefficients and reference antenna 2 2 2 2 GD = GR + S21 S21 + 1 S 1 S11aut (dB) , (12) gain (GR = 4.5 dBi), the designed| |aut −bio-inspired| |re f −PMA11re gain f − (GD−) |was calculated| according to the adapted Friis equation [20]: where, S21aut and S21ref represent the transmission coefficient of the antenna under test and reference 𝐺 =𝐺 + |𝑆 | − |𝑆 | + 1−𝑆 − (1−|𝑆 |) (𝑑𝐵) , (12) antenna, respectively, and S11autand S11refrepresent the reflection coefficient of the antenna under test and reference antenna, respectively. where, S21aut and S21ref represent the transmission coefficient of the antenna under test and reference For the application of Equation (12) in the gain estimation, it was necessary to assess the antenna, respectively, and S11aut and S11ref represent the reflection coefficient of the antenna under test performance of two measurements using the schematic presented in Figure8. The first measurement and reference antenna, respectively. was performed using two identical reference antennas (Hyperlog 30100X), one positioned at the AUT For the application of Equation (12) in the gain estimation, it was necessary to assess the distance and the other at the reference antenna position. In this way, it was possible to obtain the performance of two measurements using the schematic presented in Figure 8. The first measurement Sensorsreference 2019, transmission 19, x coefficient (S ). The second measurement was performed with the10 PMAof 27 was performed using two identical21ref reference antennas (Hyperlog 30100X), one positioned at the positioned at the AUT distance, allowing the obtaining of the S . A photograph of the schematic AUT distance and the other at the reference antenna position.21aut In this way, it was possible to obtain positionedshown in Figure at the 11 AUT is presented distance, inallowing Figure 12the. obtaining of the S21aut. A photograph of the schematic shownthe reference in Figure transmission 11 is presented coefficient in Figure (S21ref 12.). The second measurement was performed with the PMA

Figure 12. PhotographPhotograph of of the gain measurem measurementent test experimental setup.

The PMA sensitivity was tested by checking the criteria established by [27] for PD detection applications, which recommends that PMAs must present a mean gain higher than 2 dBi for their respective operating bandwidth.

4.5. PD Detection Sensitivity Tests The experimental procedure applied for the generation and detection of PD consisted of a gradual increase of voltage, by means of a regulating transformer, until PD activity was generated in the PD source and detected by the bio-inspired PMA and the IEC 60270 standard method. The bio-inspired PMA was positioned at the selected far field distance (1.0 m) from the PD source. In order to represent typical internal partial discharge signals presented in high voltage equipment, an oil cell composed by a point-to-plane electrode configuration spaced by a dielectric (polyamide disk), was used as PD generator. The standard method is represented by a coupling capacitor (1000 pF) and a resonant circuit (LDM–5) as presented in Figure 13. 15 mH Oil Cell Metamaterial Superstrate 1000 pF

Electrodes 0-100kV Bio-inspired PMA LDM-5 Foam

0-220V Polyamide Disk

Coaxial Cable

Oscilloscope

Figure 13. Experimental setup applied for Partial Discharge (PD) measurement.

The PD pulses were acquired using a Keysight oscilloscope DSO90604A with 6 GHz bandwidth, 20 GSa/s sampling rate, 70 ps rise time, and four analog channels. In order to assess the antenna PD detection sensitivity in terms of apparent charge, all the calibration procedures established by the IEC 60270 standard were executed before the voltage

Sensors 2019, 19, x 10 of 27

positioned at the AUT distance, allowing the obtaining of the S21aut. A photograph of the schematic shown in Figure 11 is presented in Figure 12.

Sensors 2019, 19, 4255 11 of 27

Figure 12. Photograph of the gain measurement test experimental setup. The PMA sensitivity was tested by checking the criteria established by [27] for PD detection applications,The PMA which sensitivity recommends was tested that by PMAs checking must the present criteria a mean established gain higher by [27] than for 2 PD dBi detection for their respectiveapplications, operating which recommends bandwidth. that PMAs must present a mean gain higher than 2 dBi for their respective operating bandwidth. 4.5. PD Detection Sensitivity Tests 4.5. PD Detection Sensitivity Tests The experimental procedure applied for the generation and detection of PD consisted of a gradual increaseThe ofexperimental voltage, by meansprocedure of a regulatingapplied for transformer, the generation until PDand activitydetection was of generated PD consisted in the of PD a sourcegradual and increase detected of voltage, by the bio-inspired by means of PMA a regulating and the transformer, IEC 60270 standard until PD method. activity was generated in the PDThe source bio-inspired and detected PMA was by the positioned bio-inspired at the PMA selected and far the ﬁeld IEC distance 60270 standard (1.0 m) frommethod. the PD source. In orderThe tobio-inspired represent typical PMA internalwas positioned partial dischargeat the sele signalscted far presented field distance in high (1.0 voltage m) from equipment, the PD ansource. oil cell In composedorder to represent by a point-to-plane typical internal electrode partial conﬁguration discharge signals spaced bypresented a dielectric in high (polyamide voltage disk),equipment, was used an oil as PDcell generator. composed The by standarda point-to-plane method iselectrode represented configuration by a coupling spaced capacitor by a dielectric (1000 pF) and(polyamide a resonant disk), circuit was (LDM–5) used as asPD presented generator. in FigureThe standard 13. method is represented by a coupling capacitor (1000 pF) and a resonant circuit (LDM–5) as presented in Figure 13. 15 mH Oil Cell Metamaterial Superstrate 1000 pF

Electrodes 0-100kV Bio-inspired PMA LDM-5 Foam

0-220V Polyamide Disk

Coaxial Cable

Oscilloscope

Figure 13. Experimental setup applied for PartialPartial Discharge (PD) measurement.

The PDPD pulses pulses were were acquired acquired using using a Keysight a Keys oscilloscopeight oscilloscope DSO90604A DSO90604A with 6 GHz with bandwidth, 6 GHz 20bandwidth, Gsa/s sampling 20 GSa/s rate, sampling 70 ps rise rate, time, 70 andps rise four time, analog and channels. four analog channels. InIn orderorder toto assessassess thethe antennaantenna PDPD detectiondetection sensitivitysensitivity inin termsterms ofof apparentapparent charge,charge, allall thethe calibration procedures established by the IEC 6027060270 standardstandard werewere executedexecuted beforebefore thethe voltagevoltage application tests. For this, the LDC-5 calibrator from Doble Lemke was connected to the PD generator terminals and the pulses were collected by the LDM-5 system. The PD generator calibration apparent charge results are presented in Table1.

Table 1. Apparent charge calibration results for the PD generator used.

Apparent Charge (pC) Voltage (mV) 20 25.6 100 121 500 584

Since the results presented in Table1 show an approximately linear relation between the apparent charge and the measured voltage values, it is possible to correlate the apparent charge PD values, generated by the oil cell, with the voltage values measured by the IEC 60270 standard method. Then, the antenna PD detection sensitivity in terms of apparent charge can be evaluated. Calibration procedures for radiometric free-space, such as presented in [55], can also be used for this purpose. Sensors 2019, 19, x 11 of 27 application tests. For this, the LDC-5 calibrator from Doble Lemke was connected to the PD generator terminals and the pulses were collected by the LDM-5 system. The PD generator calibration apparent charge results are presented in Table 1.

Table 1. Apparent charge calibration results for the PD generator used.

Apparent Charge (pC) Voltage (mV) 20 25.6 100 121 500 584 Since the results presented in Table 1 show an approximately linear relation between the apparent charge and the measured voltage values, it is possible to correlate the apparent charge PD values, generated by the oil cell, with the voltage values measured by the IEC 60270 standard method. Then, the antenna PD detection sensitivity in terms of apparent charge can be evaluated. Sensors 2019, 19, 4255 12 of 27 Calibration procedures for radiometric free-space, such as presented in [55], can also be used for this purpose. 4.6. PD Detection in a 230 kV Current Transformer 4.6. PD Detection in a 230 kV Current Transformer The equipment selected for the ﬁeld application tests was a 230 kV Current Transformer (CT), presentedThe equipment in Figure 14 selected. for the field application tests was a 230 kV Current Transformer (CT), presented in Figure 14.

Figure 14. TheThe 230 230 kV kV Current Current Transformer Transformer evaluate evaluatedd in the ﬁeldfield application tests.

Previous routine tests performed by the maintenance personnel, such as power factor tests, Previous routine tests performed by the maintenance personnel, such as power factor tests, indicated CT insulating system degradation, pointed out the possibility of high partial discharges indicated CT insulating system degradation, pointed out the possibility of high partial discharges activity in this equipment. Therefore, this CT was selected for the field application, since a high partial activity in this equipment. Therefore, this CT was selected for the field application, since a high discharge activity was expected. Such as for the PD sensitivity tests, the PD pulses were acquired in partial discharge activity was expected. Such as for the PD sensitivity tests, the PD pulses were time domain. However, a Tektronix oscilloscope TDS5104B with 1 GHz bandwidth, 5 Gsa/s sampling, acquired in time domain. However, a Tektronix oscilloscope TDS5104B with 1 GHz bandwidth, 5 and four analog channels, was used. GSa/s sampling, and four analog channels, was used. According to [1,56], a PD signal can be identified by the pulse location regarding a reference According to [1,56], a PD signal can be identified by the pulse location regarding a reference voltage signal. Usually, the PD signals occur in the phase values corresponding to the semicycle voltage signal. Usually, the PD signals occur in the phase values corresponding to the semicycle transitions of a sinusoidal reference voltage. In addition, the PD pulses present a relative periodicity transitions of a sinusoidal reference voltage. In addition, the PD pulses present a relative periodicity with a phase displacement of 180◦ between each pulse. Therefore, in order to evaluate if the signals emitted by the 230 kV CT and detected by the antenna were originated by PD, a signal generator was used in order to produce a sinusoidal reference voltage. Moreover, the bio-inspired antenna without the metamaterial superstrate was used as reference in the field applications for comparative purposes. For this, both antennas were positioned simultaneously at the same distance from the CT. In addition, cables with the same length and attenuation were used. The signals were obtained simultaneously by triggering the oscilloscope.

5. Results The analysis of the results is divided into three subsections. The first is about the simulation results regarding the bio-inspired PMA, metamaterial unit cell, and superstrate. The second subsection is regarding the practical results regarding reflection coefficient and gain measurements, as well the PD Sensors 2019, 19, x 12 of 27 Sensors 2019, 19, x 12 of 27 with a phase displacement of 180° between each pulse. Therefore, in order to evaluate if the signals with a phase displacement of 180° between each pulse. Therefore, in order to evaluate if the signals emitted by the 230 kV CT and detected by the antenna were originated by PD, a signal generator was emitted by the 230 kV CT and detected by the antenna were originated by PD, a signal generator was used in order to produce a sinusoidal reference voltage. used in order to produce a sinusoidal reference voltage. Moreover, the bio-inspired antenna without the metamaterial superstrate was used as reference Moreover, the bio-inspired antenna without the metamaterial superstrate was used as reference in the field applications for comparative purposes. For this, both antennas were positioned in the field applications for comparative purposes. For this, both antennas were positioned simultaneously at the same distance from the CT. In addition, cables with the same length and simultaneously at the same distance from the CT. In addition, cables with the same length and attenuation were used. The signals were obtained simultaneously by triggering the oscilloscope. attenuation were used. The signals were obtained simultaneously by triggering the oscilloscope. 5. Results 5. Results SensorsThe2019 analysis, 19, 4255 of the results is divided into three subsections. The first is about the simulation13 of 27 The analysis of the results is divided into three subsections. The first is about the simulation results regarding the bio-inspired PMA, metamaterial unit cell, and superstrate. The second results regarding the bio-inspired PMA, metamaterial unit cell, and superstrate. The second subsection is regarding the practical results regarding reflection coefficient and gain measurements, detectionsubsection sensitivity. is regarding Finally, the practical Section3 results is regarding regard theing fieldreflection application coefficien of thet and designed gain measurements, bio-inspired as well the PD detection sensitivity. Finally, the third section is regarding the field application of the PMA-metamaterialas well the PD detection set on sensitivity. the PD detection Finally, in the a 230 third kV section CT. is regarding the field application of the designed bio-inspired PMA-metamaterial set on the PD detection in a 230 kV CT. designed bio-inspired PMA-metamaterial set on the PD detection in a 230 kV CT. 5.1. Simulation Results 5.1. Simulation Results 5.1. SimulationAs previously Results mentioned, the bio-inspired PMA presented in [26] meets the size, bandwidth, and gainAs values previously requirements mentioned, for PD the detection bio-inspired through PMA dielectric presented windows in [26] meets in high the voltage size, bandwidth, equipment. As previously mentioned, the bio-inspired PMA presented in [26] meets the size, bandwidth, andHowever, gain althoughvalues requirements the PMA mean for gain PD isdetectio highern than through the 2 dBidielectric criteria threshold,windows itin canhigh be verifiedvoltage and gain values requirements for PD detection through dielectric windows in high voltage equipment.in Figure 15 However,that the bio-inspired although the PMA PMA presented mean gain maximum is higher gain than values the 2 dBi lower criteria to the threshold, recommended it can equipment. However, although the PMA mean gain is higher than the 2 dBi criteria threshold, it can bethreshold verified for in the Figure frequency 15 that values the nearbio-inspired to the lower PMA operating presented frequency maximum (487 MHz).gain values Therefore, lower in to order the be verified in Figure 15 that the bio-inspired PMA presented maximum gain values lower to the recommendedto explore the threshold dielectric for windows the frequency for the values PMA gainnear to enhancement, the lower operating metamaterial frequency unit (487 cells MHz). were recommended threshold for the frequency values near to the lower operating frequency (487 MHz). Therefore,simulated forin order the application to explore as the dielectric dielectric superstrates. windows for the PMA gain enhancement, metamaterial Therefore, in order to explore the dielectric windows for the PMA gain enhancement, metamaterial unit cells were simulated for the application as dielectric superstrates. unit cells were simulated4 for the application as dielectric superstrates. 4 ) i 3.5 B ) i

d 3.5 B (

d n

( 3 i

a n 3 i G a

G 2.5 m

u 2.5 m u im 2 x im a 2 x a M 1.5 M 1.5 1 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1 0.5 0.6 0.7 0.8 Fre0.9 quenc1 y (GHz)1.1 1.2 1.3 1.4 1.5 Frequency (GHz) Figure 15. Simulated bio-inspired PMA maximum gain. Figure 15. Simulated bio-inspired PMA maximum gain. The reflectionreflection coecoefficientfficient results results for for the the metamaterial metamaterial unit unit cells cells with with circular circular and and rectangular rectangular SRR The reflection coefficient results for the metamaterial unit cells with circular and rectangular SRRare presented, are presented, respectively, respectively, in Figure in Figure 16. 16. SRR are presented, respectively, in Figure 16. 0 0 0 0 -5 -5 -5 -5 -10 -10 -10 -10 (dB) (dB) -15 -15 11 11 (dB) S S (dB) -15 -15 11 11 -20 -20 S S -20 -20 -25 -25 -25 -25 -30 -30 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 -30 -30 0.2 0.4 0.6 Frequency0.8 (GHz)1 1.2 1.4 1.6 0.2 0.4 0.6 Frequency0.8 (GHz)1 1.2 1.4 1.6 Frequency(a) (GHz) Frequency(b) (GHz) (a) (b) Figure 16. Metamaterial unit cells reflection coefficients: (a) circular; (b) rectangular. FigureFigure 16.16. Metamaterial unitunit cellscells reflectionreflection coecoefficients:fficients: ((aa)) circular;circular; ((bb)) rectangular.rectangular. Both the designed metamaterial unit cells presented bandwidths within the PD activity BothBoth thethe designed designed metamaterial metamaterial unit unit cells presentedcells presented bandwidths bandwidths within thewithin PD activity the PD frequency activity frequency range and for the frequency values that presented lower gain results for the bio-inspired rangefrequency and range for the and frequency for the frequency values that values presented that presented lower gain lower results gain forresults the for bio-inspired the bio-inspired PMA.

The circular and rectangular unit cells presented operating bands up to 1033 and 1117 MHz, respectively. Besides the reflection coefficient, the transmission coefficient values were also obtained as presented in Figure 17. Therefore, through the magnitude and phase values ( Figure 18 and Figure 19) extracted from both coefficients (reflection and transmission), it was possible to calculate the permittivity (ε) and permeability (µ) values through Equations (5)–(9). The real values of µ and ε for both metamaterial unit cells are presented in Figure 20. Sensors 2019, 19, x 13 of 27 Sensors 2019, 19, x 13 of 27 Sensors 2019, 19, x 13 of 27 PMA. The circular and rectangular unit cells presented operating bands up to 1033 and 1117 MHz, PMA. The circular and rectangular unit cells presented operating bands up to 1033 and 1117 MHz, PMA.respectively. The circular and rectangular unit cells presented operating bands up to 1033 and 1117 MHz, respectively. respectively.Besides the reflection coefficient, the transmission coefficient values were also obtained as Besides the reflection coefficient, the transmission coefficient values were also obtained as presentedBesides in theFigure reflection 17. Therefore, coefficient, through the thetransmissi magnitudeon coefficient and phase values values were( Figure also 18; obtained Figure 19)as presented in Figure 17. Therefore, through the magnitude and phase values ( Figure 18; Figure 19) presentedextracted infrom Figure both 17. coefficients Therefore, through(reflection the andmagn transmission),itude and phase it valueswas possible ( Figure to 18; calculate Figure 19)the extracted from both coefficients (reflection and transmission), it was possible to calculate the extractedpermittivity from (ε) andboth permeability coefficients (µ)(reflection values through and transmission), Equations (5)–(9). it was The possible real values to calculateof µ and εthe for Sensorspermittivity2019, 19, ( 4255ε) and permeability (µ) values through Equations (5)–(9). The real values of µ and14 ε of for 27 permittivityboth metamaterial (ε) and unit permeability cells are presented (µ) values in through Figure Equations20. (5)–(9). The real values of µ and ε for both metamaterial unit cells are presented in Figure 20. 0 0 both0 metamaterial unit cells are presented in Figure 20. 0 0-5 0-5 -5 -5 -5-10 -5-10 -10 -10 -10 -10 (dB) (dB) -15 -15 21 (dB) (dB) 21 -15 -15 S S 21 21 (dB) (dB) -15 -15 S

S -20 -20 21 21 -20 -20 S S -20-25 -20-25 -25 -25 -25-30 -25-30 -30 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 -30 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 -300.2 0.4 0.6 Frequency0.8 (GHz)1 1.2 1.4 1.6 -300.2 0.4 0.6 Frequency0.8 (GHz)1 1.2 1.4 1.6 0.2 0.4 0.6 Frequency0.8 (GHz)1 1.2 1.4 1.6 0.2 0.4 0.6 Frequency0.8 (GHz)1 1.2 1.4 1.6 Frequency(a) (GHz) Frequency(b) (GHz) (a) (b) (a) (b) Figure 17. Metamaterial unit cells transmission coefficients: (a) circular; (b) rectangular. Figure 17. Metamaterial unit cells transmission coefficients: (a) circular; (b) rectangular. FigureFigure 17.17. MetamaterialMetamaterial unitunit cellscells transmissiontransmission coecoefficients:ﬃcients: ((aa)) circular;circular; ((bb)) rectangular.rectangular.

4 4 4 4 43 43 3 3 32 32 2 2 21 21 1 1 10 10 0 0 Phase (rad) Phase 0-1 (rad) Phase 0-1 Phase (rad) Phase -1 (rad) Phase -1 Phase (rad) Phase -1-2 (rad) Phase -1-2 -2 -2 -2-3 -2-3 -3 -3 -3-4 -3-4 -4 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 -4 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 -4 Frequency (GHz) -4 Frequency (GHz) 0.2 0.4 0.6 Frequency0.8 (GHz)1 1.2 1.4 1.6 0.2 0.4 0.6 Frequency0.8 (GHz)1 1.2 1.4 1.6 Frequency (GHz) Frequency (GHz) (a) (b) (a) (b) (a) (b) Figure 18. Reflection coefficient phases for the metamaterial unit cells: (a) circular; (b) rectangular. Figure 18.18. ReﬂectionReflection coecoefficientﬃcient phases forfor thethe metamaterialmetamaterial unitunit cells:cells: (a)) circular;circular; ((b)) rectangular.rectangular. Figure 18. Reflection coefficient phases for the metamaterial unit cells: (a) circular; (b) rectangular.

4 4 4 4 43 43 3 3 32 32 2 2 21 21 1 1 10 10 0 0 Phase (rad) Phase Phase (rad) Phase 0-1 0-1 Phase (rad) Phase Phase (rad) Phase -1 -1 Phase (rad) Phase Phase (rad) Phase -1-2 -1-2 -2 -2 -2-3 -2-3 -3 -3 -3-4 -3-4 -4 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 -4 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 -40.2 0.4 0.6 Frequency0.8 (GHz)1 1.2 1.4 1.6 -40.2 0.4 0.6 Frequency0.8 (GHz)1 1.2 1.4 1.6 0.2 0.4 0.6 Frequency0.8 (GHz)1 1.2 1.4 1.6 0.2 0.4 0.6 Frequency0.8 (GHz)1 1.2 1.4 1.6 Frequency (GHz) Frequency (GHz) (a) (b) (a) (b) Sensors 2019, 19, x (a) (b) 14 of 27 FigureFigure 19. 19. Transmission coefficient phases for the metamaterial unita cells: (ab) circular; (b) Figure 19.Transmission Transmission coe ﬃcoefficientcient phases phases for the for metamaterial the metamaterial unit cells: unit ( ) circular;cells: (a () )circular; rectangular. (b) Figurerectangular. 19. Transmission coefficient phases for the metamaterial unit cells: (a) circular; (b) rectangular. 2 rectangular. 5

0 0

-2 -5

-4 -10

Re(ε ) Re(ε ) -6 eff -15 eff μ Re(μ ) Re( eff) eff Permeability and Permittivity -8 Permeability and Permittivity -20 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 Frequency (GHz) Frequency (GHz) (a) (b)

Figure 20.20. Metamaterial unit cells µ and εε values: (a) circular; (b) rectangular.rectangular.

From Figure 20, it can be verified that both metamaterial unit cells presented double negative characteristic of ε and µ for almost the entire range of PD activity, with the exception of the bands corresponding to 1190–1350 MHz for the rectangular SRR and 1100–1300 MHz for the circular SRR, where the values of µ assume positive magnitudes. Therefore, both the SRR unit cells were classified as metamaterials applicable as superstrate for the designed bio-inspired PMA. After the classification as metamaterials, the SRR unit cells were positioned as superstrates, as presented in Figure 9 and described in the Material and Methods section. The first analysis regarding the set PMA-Metamaterial is concerning the influence on the reflection coefficient due the application of the superstrate over the bio-inspired PMA. The results for this analysis are presented in Figure 21.

-5

-10

(dB) -15 11 S -20 Bio-inspired PMA Circular SRR -25 Rectangular SRR FR4 Superstrate -30 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 Frequency (GHz)

Figure 21. Comparison between the reflection coefficients of the simulated structures.

From Figure 21, it can be verified that the superstrate application, with or without the presence of the metamaterial unit cells, resulted in shifts in the lower operating frequency of the bio-inspired PMA. For the FR4 superstrate (without metamaterial), a displacement of 42 MHz can be observed, resulting in a 5.2% operating bandwidth reduction regarding the reference bio-inspired PMA (487– 1497 MHz). For the metamaterial unit cells, the displacements verified for the lower operating frequency were 56 and 77 MHz for the rectangular and circular SRRs, respectively. In addition, the metamaterial application also resulted in an increase of the upper frequency for both SRR geometries, 22 MHz for the rectangular and 64 MHz for the circular, resulting in a 3.9% and 1.8% operating bandwidth reduction, respectively, regarding the reference bio-inspired PMA. Moreover, the application of superstrates resulted in higher reflection coefficient values regarding the reference bio-inspired PMA for the frequency range of 1050–1300 MHz, resulting in an entire operating bandwidth with reflection coefficient values under the –10 dB threshold. Therefore, the superstrate addition had a positive influence, since the coefficient reflection values were improved, and no significant bandwidth changes were observed.

Sensors 2019, 19, x 14 of 27

2 5

0 0

-2 -5

-4 -10

Re(ε ) Re(ε ) -6 eff -15 eff μ Re(μ ) Re( eff) eff Permittivity and Permeability and Permittivity -8 Permeability and Permittivity -20 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 Frequency (GHz) Frequency (GHz) (a) (b)

Sensors 2019, 19, 4255Figure 20. Metamaterial unit cells µ and ε values: (a) circular; (b) rectangular. 15 of 27

From Figure 20, it can be verified that both metamaterial unit cells presented double negative From Figure 20, it can be verified that both metamaterial unit cells presented double negative characteristic of ε and µ for almost the entire range of PD activity, with the exception of the bands characteristic of ε and µ for almost the entire range of PD activity, with the exception of the bands corresponding to 1190–1350 MHz for the rectangular SRR and 1100–1300 MHz for the circular SRR, corresponding to 1190–1350 MHz for the rectangular SRR and 1100–1300 MHz for the circular SRR, where the values of µ assume positive magnitudes. Therefore, both the SRR unit cells were classified where the values of µ assume positive magnitudes. Therefore, both the SRR unit cells were classified as metamaterials applicable as superstrate for the designed bio-inspired PMA. as metamaterials applicable as superstrate for the designed bio-inspired PMA. After the classification as metamaterials, the SRR unit cells were positioned as superstrates, as After the classification as metamaterials, the SRR unit cells were positioned as superstrates, presented in Figure 9 and described in the Material and Methods section. The first analysis as presented in Figure9 and described in Section4. The first analysis regarding the set PMA-Metamaterial regarding the set PMA-Metamaterial is concerning the influence on the reflection coefficient due the is concerning the influence on the reflection coefficient due the application of the superstrate over the application of the superstrate over the bio-inspired PMA. The results for this analysis are presented bio-inspired PMA. The results for this analysis are presented in Figure 21. in Figure 21.

-5

-10

(dB) -15 11 S -20 Bio-inspired PMA Circular SRR -25 Rectangular SRR FR4 Superstrate -30 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 Frequency (GHz)

FigureFigure 21.21. ComparisonComparison betweenbetween thethe reflectionreflection coecoeffifficientscients ofof thethesimulated simulatedstructures. structures. From Figure 21, it can be verified that the superstrate application, with or without the presence of From Figure 21, it can be verified that the superstrate application, with or without the presence the metamaterial unit cells, resulted in shifts in the lower operating frequency of the bio-inspired PMA. of the metamaterial unit cells, resulted in shifts in the lower operating frequency of the bio-inspired For the FR4 superstrate (without metamaterial), a displacement of 42 MHz can be observed, resulting PMA. For the FR4 superstrate (without metamaterial), a displacement of 42 MHz can be observed, in a 5.2% operating bandwidth reduction regarding the reference bio-inspired PMA (487–1497 MHz). resulting in a 5.2% operating bandwidth reduction regarding the reference bio-inspired PMA (487– For the metamaterial unit cells, the displacements verified for the lower operating frequency were 1497 MHz). For the metamaterial unit cells, the displacements verified for the lower operating 56 and 77 MHz for the rectangular and circular SRRs, respectively. In addition, the metamaterial frequency were 56 and 77 MHz for the rectangular and circular SRRs, respectively. In addition, the application also resulted in an increase of the upper frequency for both SRR geometries, 22 MHz for the metamaterial application also resulted in an increase of the upper frequency for both SRR rectangular and 64 MHz for the circular, resulting in a 3.9% and 1.8% operating bandwidth reduction, geometries, 22 MHz for the rectangular and 64 MHz for the circular, resulting in a 3.9% and 1.8% respectively, regarding the reference bio-inspired PMA. operating bandwidth reduction, respectively, regarding the reference bio-inspired PMA. Moreover, the application of superstrates resulted in higher reflection coefficient values Moreover, the application of superstrates resulted in higher reflection coefficient values regarding the reference bio-inspired PMA for the frequency range of 1050–1300 MHz, resulting regarding the reference bio-inspired PMA for the frequency range of 1050–1300 MHz, resulting in an in an entire operating bandwidth with reflection coefficient values under the 10 dB threshold. entire operating bandwidth with reflection coefficient values under the –10 dB threshold.− Therefore, Therefore, the superstrate addition had a positive influence, since the coefficient reflection values were the superstrateSensors 2019, 19 , xaddition had a positive influence, since the coefficient reflection values15 of 27 were improved, and no significant bandwidth changes were observed. improved, and no significant bandwidth changes were observed. In orderIn order to assess to assess the superstrate’s the superstrate’s influence influence on theon the bio-inspired bio-inspired PMA PMA radiation radiation patterns, patterns, the the lower, central,lower, and central, upper and frequency upper frequency values were values simulated were simulated and presented and presented in Figures in Figure22–24 .22 Figure 23 Figure 24.

(a) (b) (c) (d) Figure 22. E-plane co-polarization (red) and cross-polarization (purple) radiation patterns for 487 MHz: Figure 22. E-plane co-polarization (red) and cross-polarization (purple) radiation patterns for 487 MHz: (a) reference bio-inspired PMA; (b) FR4 Superstrate; (c) rectangular SRR; (d) circular SRR. (a) reference bio-inspired PMA; (b) FR4 Superstrate; (c) rectangular SRR; (d) circular SRR.

(a) (b) (c) (d)

Figure 23. E-plane co-polarization (red) and cross-polarization (purple) radiation patterns for 992 MHz: (a) reference bio-inspired PMA; (b) FR4 Superstrate; (c) rectangular SRR; (d) circular SRR.

(a) (b) (c) (d) Figure 24. E-plane co-polarization (red) and cross-polarization (purple) radiation patterns for 1497 MHz: (a) reference bio-inspired PMA; (b) FR4 Superstrate; (c) rectangular SRR; (d) circular SRR. From Figures 22–24, it can be verified that the superstrate addition without SRRs (only FR4) did not significantly change the radiation patterns regarding the reference bio-inspired PMA. However, radiation pattern distortions were verified for the upper operating frequencies for both SRR superstrates, as presented in Figure 24c,d. Moreover, from the obtained radiation patterns, it can be verified that the antenna with metamaterial is suitable for dielectric window application, since the most of maximum radiation values are concentrated in the direction of coupling of the PMA to the dielectric window, i.e., 0° direction as presented in the Figure 25. In addition, considering that the PMA coupled to the dielectric window is back shielded against external interferences, the symmetrical back lobes presented in the radiation patterns also represent an advantage in the application for PD detection, since the back lobe allows the detection of the PD radiated signals reflected by the shielding, enhancing the PMA sensitivity of detection [26].

Sensors 2019, 19, x 15 of 27 Sensors 2019, 19, x 15 of 27 In order to assess the superstrate’s influence on the bio-inspired PMA radiation patterns, the lower,In central,order to and assess upper the frequency superstrate’s values influence were simulated on the bio-inspired and presented PMA in radiation Figure 22patterns, Figure the23 Figurelower, 24.central, and upper frequency values were simulated and presented in Figure 22 Figure 23 Figure 24.

(a) (b) (c) (d) (a) (b) (c) (d) Sensors 2019,Figure19, 4255 22. E-plane co-polarization (red) and cross-polarization (purple) radiation patterns for 487 MHz: 16 of 27 (Figurea) reference 22. E-plane bio-inspired co-polarization PMA; (b )(red) FR4 Superstrate;and cross-polarizat (c) rectangularion (purple) SRR; radiation (d) circular patterns SRR. for 487 MHz: (a) reference bio-inspired PMA; (b) FR4 Superstrate; (c) rectangular SRR; (d) circular SRR.

(a) (b) (c) (d) (a) (b) (c) (d) Figure 23. E-plane co-polarization (red) and cross-polarization (purple) radiation patterns for 992 FigureMHz:Figure 23. E-plane( a23.) reference E-plane co-polarization bio-inco-polarizationspired (red)PMA; (red)and (b) andFR4 cross-polarization Superstrate;cross-polarization (c) rectangular (purple) (purple) radiation SRR;radiation (d) patternscircular patterns SRR. for for992 992 MHz: (a) referenceMHz: (a bio-inspired) reference bio-in PMA;spired (b )PMA; FR4 Superstrate;(b) FR4 Superstrate; (c) rectangular (c) rectangular SRR; SRR; (d) circular (d) circular SRR. SRR.

(a) (b) (c) (d) Figure(a) 24. E-plane co-polarization(b (red)) and cross-polarization (purple)(c) radiation patterns for( d1497) MHz: Figure 24. E-plane co-polarization (red) and cross-polarization (purple) radiation patterns for 1497 (Figurea) reference 24. E-plane bio-inspired co-polarization PMA; (b) (red)FR4 Superstrate; and cross-polarizat (c) rectangularion (purple) SRR; radiation(d) circular patterns SRR. for 1497 MHz: MHz: (a) reference bio-inspired PMA; (b) FR4 Superstrate; (c) rectangular SRR; (d) circular SRR. From(a) reference Figures bio-inspired 22–24, it can PMA; be (verifiedb) FR4 Superstrate; that the superstrate (c) rectangular addition SRR; ( dwithout) circular SRRs SRR. (only FR4) did not significantlyFrom Figures change 22–24, the it can radiation be verified patterns that rethegarding superstrate the reference addition bio-inspired without SRRs PMA. (only However, FR4) did From Figures 22–24, it can be verified that the superstrate addition without SRRs (only FR4) radiationnot significantly pattern change distortions the radiation were verified patterns fo rergarding the upper the referenceoperating bio-inspired frequencies PMA. for bothHowever, SRR did notsuperstrates,radiation significantly pattern as presented distortions change in theFigure were radiation 24c,d. verified patterns for the upper regarding operating the referencefrequencies bio-inspired for both SRR PMA. However,superstrates,Moreover, radiation as patternpresentedfrom the distortions inobtained Figure 24c,d.radiation were verified patterns, for theit can upper be operatingverified that frequencies the antenna for bothwith SRR superstrates,metamaterialMoreover, as presented is suitablefrom the in for Figureobtained dielectric 24 c,d.radiation window patterns, application, it can since be theverified most ofthat maximum the antenna radiation with Moreover,valuesmetamaterial are concentrated from is suitable the obtained infor the dielectric direction radiation window of couplin patterns, application,g of the it canPMAsince be tothe verifiedthe most dielectric of thatmaximum window, the antenna radiation i.e., 0° with metamaterialdirectionvalues are isas suitableconcentrated presented for dielectricin in thethe Figuredirection window 25. of In application,couplin addition,g of considering the since PMA the to most that the ofdielectricthe maximum PMA window,coupled radiation toi.e., the 0° values direction as presented in the Figure 25. In addition, considering that the PMA coupled to the are concentrateddielectric window in the directionis back shielded of coupling against of theexte PMArnal interferences, to the dielectric the window,symmetrical i.e., back 0◦ direction lobes as presentedpresenteddielectric in the inwindow Figurethe radiation 25is .back In patterns addition, shielded also consideringagainst represent exte an thatrnal advantage theinterferences, PMA in coupledthe applicationthe symmetrical to the for dielectric PD back detection, windowlobes is presented in the radiation patterns also represent an advantage in the application for PD detection, back shieldedsince the againstback lobe external allows interferences,the detection of the the symmetrical PD radiated backsignals lobes reflected presented by the in shielding, the radiation enhancingsince the backthe PMA lobe sensitivallows itythe of detectiondetection [26].of the PD radiated signals reflected by the shielding, patterns also represent an advantage in the application for PD detection, since the back lobe allows enhancing the PMA sensitivity of detection [26]. the detection of the PD radiated signals reflected by the shielding, enhancing the PMA sensitivity of detection [26]. Finally, from the extracted radiation patterns, the maximum gain values along the operating frequency for each simulated structure were obtained, as presented in Figure 26. The application of both designed superstrates resulted in maximum gain values almost increased in double regarding the reference bio-inspired PMA for the frequency range of 500–1000 MHz. In addition, the circular SRR superstrate also demonstrated significant gain improvement for the frequency range of 1300–1500 MHz, probably due to the strong DNG feature presented for this frequency range, as presented in Figure 14. Before the superstrate application, the reference bio-inspired PMA presented gain values higher than 2 dBi only for frequencies above 600 MHz. With the metamaterial addition, the gain values were higher than 2 dBi for the entire antenna operating frequency (500–1500 MHz). The calculated mean gain values for each simulated structure are presented in Table2. Sensors 2019, 19, 4255 17 of 27

Table 2. Simulated mean gain values for each simulated model.

Model Maximum Mean Gain (dBi)

Sensors 2019, 19, x Bio-inspired PMA 2.81 16 of 27 FR4 Superstrate 2.83 Rectangular SRR 3.96 5 5 Sensors 2019, 19, x Circular SRR 4.07 16 of 27 0 0

-5 -5 5 5 -10

0 -100 -15 -5 -20 -15-5 -10

-25 (dBi) Gain Gain (dBi) Gain -20-10 -15 -30 -25-15 -20 Bio-Inspired PMA -35 Bio-Inspired PMA FR4 Superstrate FR4 Superstrate

-25 (dBi) Gain Gain (dBi) Gain -30-20 -40 Rectangular SRR Rectangular SRR -30 Circular SRR Circular SRR -45 -35-25 -150 -100 -50 Bio-Inspired0 PMA50 100 150 -150 -100 -50 0 50 100 150 -35 Bio-Inspired PMA ThetaFR4 (Degree) Superstrate ThetaFR4 (Degree) Superstrate -30 -40 Rectangular SRR Rectangular SRR Circular SRR Circular SRR -45 (a) -35 (b) -150 -100 -50 0 50 100 150 -150 -100 -50 0 50 100 150 Theta (Degree)5 Theta (Degree) (a) 0 (b)

5 -5

0 -10

-5 -15 Gain (dBi) Gain

-10 -20 Bio-Inspired PMA -15 -25 FR4 Superstrate Gain (dBi) Gain Rectangular SRR -20 Circular SRR -30 -150 -100 -50 Bio-Inspired0 PMA50 100 150 Theta (Degree) -25 FR4 Superstrate Rectangular SRR Circular SRR -30 (c) -150 -100 -50 0 50 100 150 Theta (Degree) Figure 25. Rectangular plots of the E-plane radiation patterns for the frequencies of (a) 500 MHz; (b) 900 MHz; (c) 1500 MHz. (c)

Finally,FigureFigure 25.25. from RectangularRectangular the extracted plots plots of ofradiation the the E-plane E-plane patterns, radiat radiationion the patterns patterns maximum for for the the gainfrequencies frequencies values of along ( ofa) (500a) the 500 MHz; MHz;operating (b) frequency(900b) 900 MHz; for MHz; each(c) (1500c )simulated 1500 MHz. MHz. structure were obtained, as presented in Figure 26.

5.5 Finally, from the extracted radiation patterns, the maximum gain values along the operating 5 frequency for each simulated structure were obtained, as presented in Figure 26. 4.5 5.5 4 5 3.5 4.5 3 4 2.5 3.5 Maximum GainMaximum (dBi) 2 Bio-inspired PMA 3 Circular SRR 1.5 Rectangular SRR 2.5 FR4 Superstrate 1 Maximum GainMaximum (dBi) 0.52 0.6 0.7 0.8 0.9Bio-inspired1 PMA1.1 1.2 1.3 1.4 1.5 FrequencyCircular SRR (GHz) 1.5 Rectangular SRR FR4 Superstrate 1 Figure 26. ComparisonComparison0.5 between 0.6 0.7 the 0.8 maximum 0.9 1 gain 1.1 values 1.2 of 1.3 the 1.4 simulated 1.5 structures. Frequency (GHz) The application of both designed superstrates resulted in maximum gain values almost increased inFigure double 26. regardingComparison the between reference the maximum bio-inspired gain PMA values for of thethe simulated frequency structures. range of 500–1000 MHz. In addition, the circular SRR superstrate also demonstrated significant gain improvement for The application of both designed superstrates resulted in maximum gain values almost the frequency range of 1300–1500 MHz, probably due to the strong DNG feature presented for this increased in double regarding the reference bio-inspired PMA for the frequency range of 500–1000 frequency range, as presented in Figure 14a. MHz. In addition, the circular SRR superstrate also demonstrated significant gain improvement for the frequency range of 1300–1500 MHz, probably due to the strong DNG feature presented for this frequency range, as presented in Figure 14a.

Sensors 2019, 19, x 17 of 27

Before the superstrate application, the reference bio-inspired PMA presented gain values higher than 2 dBi only for frequencies above 600 MHz. With the metamaterial addition, the gain values were higher than 2 dBi for the entire antenna operating frequency (500–1500 MHz). The calculated mean gain values for each simulated structure are presented in Table 2.

Table 2. Simulated mean gain values for each simulated model.

Model Maximum Mean Gain (dBi) Bio-inspired PMA 2.81 FR4 Superstrate 2.83 Rectangular SRR 3.96 Sensors 2019, 19, 4255 Circular SRR 4.07 18 of 27 From Table 2, it can be noticed that the metamaterial application as superstrate resulted in a maximumFrom Table mean2, it gain can beimprovement noticed that thehigher metamaterial than 1 dBi, application regarding as the superstrate reference resulted bio-inspired in a maximum PMA, for meanboth gainSRR improvementgeometries chosen. higher Although than 1 dBi, the regarding circular SRR the referencestructure bio-inspiredpresented a PMA,slightly for higher both SRRmean geometriesgain value chosen. (4.07 dBi) Although in comparison the circular to the SRR rectangular structure presentedgeometry a(3.96 slightly dBi), higher the reflection mean gain coefficient value (4.07results dBi) for in comparisonthe rectangular to the SRR rectangular presented geometry lower (3.96bandwidth dBi), the impacts reflection for coefficientPD application results than for the the rectangularcircular one. SRR In presented addition, lower the rectangular bandwidth SRR impacts pres forented PD applicationlower radiation than thepatterns circular distortions one. In addition, than the thecircular rectangular one for SRR the presented upper frequencies, lower radiation mainly patterns in the back distortions lobe direction, than the as circular presented one for in the Figure upper 24; frequencies,Figure 25. mainlyTherefore, in the since back lobesymmetrical direction, back as presented lobes are in Figuresdesirable 24 andfor 25PD. Therefore,detection sincethrough symmetrical dielectric backwindows, lobes are the desirable rectangular for PD SRR detection was throughconsidered dielectric the fittest windows, structure the rectangular for the experimental SRR was considered PD tests thepresented fittest structure in the following for the experimental section. PD tests presented in the following section.

5.2.5.2. Laboratory Laboratory Experimental Experimental Results Results AsAs mentioned mentioned in thein previousthe previous subsection, subsection, the simulated the simulated reflection reflection coefficients, coefficients, radiation patterns,radiation andpatterns, mean gainand appointedmean gain the appointed rectangular the SRR rectangular metamaterial SRR superstrate metamaterial as the superstrate fittest structure as the for fittest the experimentalstructure for PD the tests. experimental Therefore, PD from tests. the modelsTherefor presentede, from the in Figuresmodels 6presentedb and9a, the in bio-inspiredFigure 6b and PMAFigure without 9a, the and bio-inspired with the metamaterial PMA without superstrate and with were the built, metamaterial as presented superstrate in Figure 27were. built, as presented in Figure 27.

(a) (b)

FigureFigure 27. 27.Manufactured Manufactured bio-inspired bio-inspired PMA: PMA: ( a()a without) without metamaterial metamaterial superstrate superstrate [26 [26];]; (b ()b with) with metamaterialmetamaterial superstrate. superstrate.

TheThe first first experimental experimental analysis analysis for the for manufactured the manufa antennasctured antennas is regarding is theregarding reflection the coe ffireflectioncients. Incoefficients. Figure 28 aIn comparisonFigure 28 a comparison between the between measured the andmeasured simulated and simulated reflection reflection coefficients coefficients for the bio-inspiredSensorsfor the2019 ,bio-inspired 19, PMA x with andPMA without with and the rectangularwithout the SRR rectangular metamaterial SRR superstrate metamaterial is presented. superstrate18 of 27 is presented. 0

-5

-10

-15

(dB) 11

S -20

-25 Measured PMA without Meta

-30 Measured PMA with Meta Simulated PMA without Meta Simulated PMA with Meta -35 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 Frequency (GHz)

Figure 28. Measured and simulated reﬂectionreflection coecoefficientsﬃcients forfor thethe bio-inspired PMAPMA withwith andand without metamaterial superstrate.superstrate.

Both measured and simulated reflection coefficient results presented similar behavior. For the bio-inspired PMA without metamaterial, the measured operating bandwidth started at 472 MHz up to values above the upper PD frequency of interest (1500 MHz). With the metamaterial, the lower frequency was displaced in 68 MHz (close to the 77 MHz simulated value) regarding the reference bio-inspired PMA, starting from 540 MHz until values above 1500 MHz. In addition, better reflection coefficient performance for the frequency range of 1.1–1.5 GHz was verified for the measured curve with metamaterial, as predicted by the simulations. The slight differences between the simulated and measured reflection coefficient behaviors can be attributed to PMA constructive aspects, such as minor discrepancies between the simulated FR4 and the one used in practice, as well the non-consideration of the SMA connectors during the simulation. Through the procedures described in the Material and Methods section, the bio-inspired PMA gains with and without the metamaterial superstrates were measured. The comparison between the measured and simulated gains are presented in Figure 29. 5.5

4.5

4 )

Bi 3.5 (d

3 ain ain G 2.5

2 Measured without Meta Measured with Meta 1.5 Simulated without Meta Simulated with Meta 1 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 Frequency (GHz)

Figure 29. Measured and simulated gain for the bio-inspired PMA with and without metamaterial superstrate.

From Figure 29, it can be observed that, although there is a discrepancy in the gain values, the behavior of the measured and simulated curves are in agreement. For both measured and simulated curves for the bio-inpired PMA without metamaterial, an increase in the gain values up to frequencies around 1.1 GHz can be verified, followed by a gain decrease up to 1.2 GHz and, again, a gain increase up to 1.4 GHz, followed by a new decrease

Sensors 2019, 19, x 18 of 27

-5

-10

-15 (dB) 11

S -20

-25 Measured PMA without Meta

-30 Measured PMA with Meta Simulated PMA without Meta Simulated PMA with Meta -35 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 Frequency (GHz) Sensors 2019, 19, 4255 19 of 27 Figure 28. Measured and simulated reflection coefficients for the bio-inspired PMA with and without metamaterial superstrate. Both measured and simulated reflection coefficient results presented similar behavior. For the bio-inspiredBoth measured PMA without and simulated metamaterial, reflection the measuredcoefficient operating results presented bandwidth similar started behavior. at 472 MHz For the up tobio-inspired values above PMA the without upper metamaterial, PD frequency the of interest measured (1500 operating MHz). bandwidth With the metamaterial, started at 472 the MHz lower up frequencyto values above was displaced the upper in PD 68 frequency MHz (close of to interest the 77 MHz(1500 simulatedMHz). With value) the metamaterial, regarding the the reference lower bio-inspiredfrequency was PMA, displaced starting in from 68 MHz 540 MHz(close until to the values 77 MHz above simulated 1500 MHz. value) In addition, regarding better the reflectionreference coebio-inspiredfficient performance PMA, starting for from the frequency 540 MHz rangeuntil valu of 1.1–1.5es above GHz 1500 was MHz. verified In addition, for the measuredbetter reflection curve withcoefficient metamaterial, performance as predicted for the frequency by the simulations. range of 1.1–1.5 The slight GHz di wasfferences verified between for the the measured simulated curve and measuredwith metamaterial, reflection coeas predictedfficient behaviors by the cansimulations. be attributed The toslight PMA differences constructive between aspects, the such simulated as minor discrepanciesand measured between reflection the coefficient simulated behaviors FR4 and thecan onebe attributed used in practice, to PMA as constructive well the non-consideration aspects, such as ofminor the SMAdiscrepancies connectors between during the the simulation. simulated FR4 and the one used in practice, as well the non-considerationThrough the proceduresof the SMA describedconnectors in during Section the4, thesimulation. bio-inspired PMA gains with and without the metamaterialThrough the superstratesprocedures weredescribed measured. in the TheMateri comparisonal and Methods between section, the measured the bio-inspired and simulated PMA gains arewith presented and without in Figure the metamaterial 29. superstrates were measured. The comparison between the measured and simulated gains are presented in Figure 29. 5.5

4.5

4 )

Bi 3.5 (d

3 ain ain G 2.5

2 Measured without Meta Measured with Meta 1.5 Simulated without Meta Simulated with Meta 1 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 Frequency (GHz) Figure 29. Measured and simulated gain for the bio-inspired PMA with and without metamaterial superstrate. Figure 29. Measured and simulated gain for the bio-inspired PMA with and without metamaterial Fromsuperstrate. Figure 29, it can be observed that, although there is a discrepancy in the gain values, the behavior of the measured and simulated curves are in agreement. From Figure 29, it can be observed that, although there is a discrepancy in the gain values, the For both measured and simulated curves for the bio-inpired PMA without metamaterial, an increase behavior of the measured and simulated curves are in agreement. in the gain values up to frequencies around 1.1 GHz can be verified, followed by a gain decrease up For both measured and simulated curves for the bio-inpired PMA without metamaterial, an to 1.2 GHz and, again, a gain increase up to 1.4 GHz, followed by a new decrease tendency up to increase in the gain values up to frequencies around 1.1 GHz can be verified, followed by a gain frequency values above 1.5 GHz. For both measured and simulated curves for the bio-inspired PMA decrease up to 1.2 GHz and, again, a gain increase up to 1.4 GHz, followed by a new decrease with metamaterial, the same trend was verified, except for the frequencies above 1.4 GHz, in which the bio-inspired PMA with metamaterial presented a gain increase trend up to frequencies above 1.5 Ghz. Besides the constructive aspects mentioned before, the differences between the measured and simulated values can be attributed to the conception of the simulated environment, which neglects some practical limitations faced during the measurements, such as the cables attenuation, the alignment between the reference antenna and the test antenna, and the losses in the far-field measuring system. Even so, the gain enhancement provided by the metamaterial superstrate was according to that predicted by the simulations, presenting significant gain improvement for mostly the bandwidth of the bio-inspired PMA. The measured mean gain for the bio-inspired PMA without metamaterial was equal to 2.92 dBi, close to the simulated mean gain value presented in Table2 (2.86 dBi). In addition, the bio-inspired PMA without metamaterial only presented gain values above the 2 dBi threshold for the frequencies above 700 MHz. With the metamaterial superstrate, the calculated mean gain was 3.61 dBi, resulting in an enhancement of 0.7 dBi regarding the bio-inspired PMA. Moreover, the application of the metamaterial superstrate resulted in an antenna with gain values above 2 dBi from 575 MHz up to 1.5 GHz, resulting in an antenna with high sensitivity for PD detection. Sensors 2019, 19, x 19 of 27

tendency up to frequency values above 1.5 GHz. For both measured and simulated curves for the bio-inspired PMA with metamaterial, the same trend was verified, except for the frequencies above 1.4 GHz, in which the bio-inspired PMA with metamaterial presented a gain increase trend up to frequencies above 1.5 Ghz. Besides the constructive aspects mentioned before, the differences between the measured and simulated values can be attributed to the conception of the simulated environment, which neglects some practical limitations faced during the measurements, such as the cables attenuation, the alignment between the reference antenna and the test antenna, and the losses in the far-field measuring system. Even so, the gain enhancement provided by the metamaterial superstrate was according to that predicted by the simulations, presenting significant gain improvement for mostly the bandwidth of the bio-inspired PMA. The measured mean gain for the bio-inspired PMA without metamaterial was equal to 2.92 dBi, close to the simulated mean gain value presented in Table 2 (2.86 dBi). In addition, the bio-inspired PMA without metamaterial only presented gain values above the 2 dBi threshold for the frequencies above 700 MHz. With the metamaterial superstrate, the calculated mean gain was Sensors 2019, 19, 4255 20 of 27 3.61 dBi, resulting in an enhancement of 0.7 dBi regarding the bio-inspired PMA. Moreover, the application of the metamaterial superstrate resulted in an antenna with gain values above 2 dBi from 575From MHz up the to measured1.5 GHz, resulting reflection in an coe anteffinnacient with and high gain sensitivity results for presented, PD detection. the bio-inspired PMA-metamaterialFrom the measured set can bereflection considered coefficient fit for PD and detection. gain results Therefore, presented, it was the submitted bio-inspired to high voltagePMA-metamaterial PD detection sensitivityset can be considered tests. fit for PD detection. Therefore, it was submitted to high voltage PD detection sensitivity tests. Through the setup presented in Figure 13, the PD activity was detected simultaneously by the Through the setup presented in Figure 13, the PD activity was detected simultaneously by the bio-inspired PMA-metamaterial set and the IEC 60270 standard method for the application of 13.4 kV bio-inspired PMA-metamaterial set and the IEC 60270 standard method for the application of 13.4 on the oil cell. Figure 30 presents one sample of the simultaneously detected PD pulses. kV on the oil cell. Figure 30 presents one sample of the simultaneously detected PD pulses.

40 PMA-Metamaterial Set 30

-10 Voltage (mV) Voltage

-20

-30

-40 -2 0 2 4 6 8 10 12 14 16 Time (μs) 80 IEC 60270 60

-20 Voltage (mV) Voltage

-40

-60

-80 -2 0 2 4 6 8 10 12 14 16 Time (μs) Figure 30. PD pulses sample detected by the bio-inspired PMA-metamaterial set and IEC 60270 Figure 30. PD pulses sample detected by the bio-inspired PMA-metamaterial set and IEC 60270 standard method, respectively, for the application of 13.4 kV on the oil cell. standard method, respectively, for the application of 13.4 kV on the oil cell. From Figure 30, it can be verified that the bio-inspired PMA-metamaterial set was able to detect all the PD pulses detected by the IEC 60270 standard method. In addition, the voltage measured at the antenna’s terminals presented values reduced by one-half compared to the standard method. This result is expected, since the radiated waves are susceptible to higher losses, such as reflections and propagating attenuations, than those in the standard method, directly connected with the oil cell. Nonetheless, the antenna sensitivity could be attested, since, according to Table1, the detected PD pulses presented apparent charge values between 15 and 60 pC. This range of apparent charges is similar to the pC values detected in practice for the PD low intensity pulses related to the inception of an insulating problem in transformers oil, for example in [57–59]. In order to evaluate the frequency range of the detected PD pulses, a spectrogram was extracted from the PMA detected PD pulses shown in Figure 30 and presented in Figure 31. From Figure 31, it can be verified that the four PD pulses observed in Figure 23 were highlighted in the spectrogram, presenting higher concentrations of spectral energy for frequencies up to 1500 MHz. The obtained frequency range is in agreement with the values reported in [9,10,60] as the main range of interest for the monitoring of partial discharges in oil (300–1500 MHz). Sensors 2019, 19, x 20 of 27

From Figure 30, it can be verified that the bio-inspired PMA-metamaterial set was able to detect all the PD pulses detected by the IEC 60270 standard method. In addition, the voltage measured at the antenna’s terminals presented values reduced by one-half compared to the standard method. This result is expected, since the radiated waves are susceptible to higher losses, such as reflections and propagating attenuations, than those in the standard method, directly connected with the oil cell. Nonetheless, the antenna sensitivity could be attested, since, according to Table 1, the detected PD pulses presented apparent charge values between 15 and 60 pC. This range of apparent charges is similar to the pC values detected in practice for the PD low intensity pulses related to the inception Sensorsof an2019 insulating, 19, 4255 problem in transformers oil, for example in [57–59]. 21 of 27 In order to evaluate the frequency range of the detec

-125

-130

-135 16 14

-140 12 10 -130 -145 8 Energy -140 6 Time (µs) -150 -150 4 0 2 1 2 3 0 -155 4 5 6 7 8 9 10 Frequency (GHz) Figure 31. Spectrogram of the PD pulses sample detected by the bio-inspired PMA-metamaterial set Figure 31. Spectrogram of the PD pulses sample detected by the bio-inspired PMA-metamaterial set presented in Figure 30. presented in Figure 30.

Therefore,From Figure besides 31, the it sensitivitycan be verified (gain), thethat antenna the four operating PD pulses bandwidth observed was alsoin Figure attested 23 in were the highhighlighted voltage PDin the tests, spectrogram, since the bio-inspired presenting PMA-metamaterial higher concentrations set was of spectral able to successfullyenergy for frequencies detect all theup main to 1500 frequency MHz. The range obtained of energy frequency concentration range is for in PDagreement pulses. with the values reported in [9,10,60] as theHence, main the range bio-inspired of interest PMA-metamaterial for the monitoring set wasof partial approved discharges in the laboratory in oil (300–1500 PD experimental MHz). tests, being ableTherefore, to be applied besides in the PD sensitivity measurements (gain), on the substations, antenna asoperating presented bandwidth in the following was also subsection. attested in the high voltage PD tests, since the bio-inspired PMA-metamaterial set was able to successfully 5.3.detect PD Detectionall the main in a frequency 230 kV Substation range of energy concentration for PD pulses. ApplyingHence, the methodologybio-inspired presentedPMA-metamaterial in Section 4.6set, PDwas measurements approved in on the a 230 laboratory kV CT were PD performedexperimental and thetests, comparison being able between to be applied the PD in pulses PD measurements detected by the on bio-inspired substations, PMA-metamaterial as presented in the Sensorssetfollowing and 2019 by, 19 the subsection., x reference bio-inspired PMA are presented in Figure 32. 21 of 27

5.3. PD Detection in a 230 kV Substation 60 Reference Voltage Applying the methodology presented in Subsection 4.6, PD measurements on a 230 kV CT were 40 Bio-Inspired PMA-Metamaterial performed and the comparison between the PD pulses detected by the bio-inspired PMA-metamaterial20 set and by the reference bio-inspired PMA are presented in Figure 32.

Voltage (mV) Voltage -20

-40 0 5 10 15 20 25 30 35 40 Time (ms)

Reference Voltage 40 Reference Bio-inspired PMA

Voltage (mV) Voltage -20

-40 0 5 10 15 20 25 30 35 40 Time (ms) Figure 32. Comparison between the 230 kV Current Transformer (CT) PD pulses detected by the Figure 32. Comparison between the 230 kV Current Transformer (CT) PD pulses detected by the bio-inspired PMA with and without the metamaterial superstrate, respectively. bio-inspired PMA with and without the metamaterial superstrate, respectively.

From Figure 32, it can be verified that the detected signals presented PD features, since they occurred in the half-cycle transitions of the sinusoidal reference voltage and with a phase displacement approximately equal to 180° between each other. In addition, the bio-inspired PMA-metamaterial set presented higher detection sensitivity than the reference bio-inspired PMA for all of the detected PD pulses, presenting a signal with amplitudes almost doubled regarding the reference bio-inspired PMA. As expected, according to previous measurements performed by the substation maintenance personnel, a considerable level of PD was detected in the 230 kV CT, attesting the sensitivity of the developed antenna in the practical detection of PD. Although the PMA presented high sensitivity for PD detection, it was not possible to estimate the PD apparent charge levels inside the equipment and detected by the antennas, since the calibration procedure on the 230 kV CT could not be performed due to practical limitations (coupling capacitor parallel connection). Therefore, at this point of the research, the CT operational condition diagnosis (low, medium, or critical level of PD) could not be estimated. Nonetheless, the application of the developed bio-inspired PMA-metamaterial set in the CT PD measurement was done in order to attest its practical sensitivity in the PD detection. Hence, as future works, calibration, classification, and localization techniques will be applied over the proposed antenna in order to fully characterize it as a sensor applicable in the PD continuous monitoring and diagnosis, specially through dielectric windows improved by the application of metamaterials on its surface.

Sensors 2019, 19, 4255 22 of 27

From Figure 32, it can be veriﬁed that the detected signals presented PD features, since they occurred in the half-cycle transitions of the sinusoidal reference voltage and with a phase displacement approximately equal to 180◦ between each other. In addition, the bio-inspired PMA-metamaterial set presented higher detection sensitivity than the reference bio-inspired PMA for all of the detected PD pulses, presenting a signal with amplitudes almost doubled regarding the reference bio-inspired PMA. As expected, according to previous measurements performed by the substation maintenance personnel, a considerable level of PD was detected in the 230 kV CT, attesting the sensitivity of the developed antenna in the practical detection of PD. Although the PMA presented high sensitivity for PD detection, it was not possible to estimate the PD apparent charge levels inside the equipment and detected by the antennas, since the calibration procedure on the 230 kV CT could not be performed due to practical limitations (coupling capacitor parallel connection). Therefore, at this point of the research, the CT operational condition diagnosis (low, medium, or critical level of PD) could not be estimated. Nonetheless, the application of the developed bio-inspired PMA-metamaterial set in the CT PD measurement was done in order to attest its practical sensitivity in the PD detection. Hence, as future works, calibration, classiﬁcation, and localization techniques will be applied over the proposed antenna in order to fully characterize it as a sensor applicable in the PD continuous monitoring and diagnosis, specially through dielectric windows improved by the application of metamaterials on its surface.

6. Discussion Although there are more competitive antenna types that can provide higher gain than PMA, such as hornet and Vivaldi antennas, their use in PD detection is compromised due to their constructive aspects (mainly their length), since they are not suitable for installation in dielectric windows in power transformers or GIS. Despite the fact that PMA are not as efficient as another types of antennas, they still represent a good option for PD detection, since its large bandwidth, gain, radiation pattern and dimensions are still suitable for PD detection, as verified and shown on other studies in the literature [25–27,61–66]. As presented in this paper, the distance between the antenna and the metamaterial superstrate (dielectric window) was adopted in order to reduce the impact on the antenna's reflection coefficient. Still, even with the distance between the bio-inspired antenna and the metamaterial superstrate (dielectric window), the final structure has a volumetric dimension of 200 mm 200 mm 135 mm × × to accommodate an antenna with an operating bandwidth of 500–1500 MHz and average gain of 3.61 dBi, which represents an acceptable size for PD application and good PD detection sensitivity features as demonstrated experimentally during the paper and through comparisons with other results obtained in the literature [67–73]. If another type of antennas were used for this same bandwidth, their higher length would require a much larger dielectric window, becoming an impractical application. The dimensions of some more directional antennas (commercial and from other papers) with lower operating frequency equivalent to the proposed antenna (500 MHz) are presented in the Table3.

Table 3. Comparison between the proposed PMA-Metamaterial set and other type of antennas.

Antenna Bandwidth (GHz) Gain (dBi) Size (mm) Proposed PMA-Metamaterial Set 0.5–1.5 3.61 (mean) 200 200 135 (depth) × × [74] (Hornet Antenna) 0.5–8 3–17 418 318 645 (length) × × [75] (Hornet Antenna) 0.5–3 4–11 440 290 350 (length) × × [76] (Vivaldi Antenna) 0.5–6 3.07–7.7 120 (width) 225 (length) × [77] (Vivaldi Antenna) 0.5–4 2–11 173 (width) 299 (length) × [78] (Reference Antenna–Hyperlog 0.4–10 4.5 (mean) 360 (width) 640 (length) 30100X) ×

As can be seen from Table3, although the good results of bandwidth and gain for PD detection application, all the other antennas present a much larger length than the proposed PMA-dielectric Sensors 2019, 19, 4255 23 of 27 window set (135 mm). Moreover, by coupling the presented antennas over dielectric windows, we would have even larger dimensions for the antenna-dielectric window set than those presented in Table3, becoming impractical for the application on PD detection in power transformers or GIS. Therefore, this reinforces the advantage of using PMA for the achievement of dielectric windows with smaller dimensions and higher ease of coupling on the structure of high voltage equipment.

7. Conclusions In this paper, the representation of dielectric windows as a metamaterial superstrate over a bio-inspired antenna (based on the Jatropha mollissima (Pohl) Baill leaf) was proposed in order to improve the detection sensitivity of UHF sensors designed for PD measurement. During the simulation tests, both the selected SRR shapes (rectangular and circular) proved themselves eﬃcient in the enhancement of the antenna detection sensitivity, presenting a mean gain improvement higher than 1 dBi compared to the structure without metamaterial. However, the rectangular SRR was selected for the laboratory tests due to the better relation gain/bandwidth/radiation pattern achieved. The designed bio-inspired PMA-metamaterial set presented potential to be used as a UHF sensor applied through dielectric windows for PD detection, since it meets the bandwidth (300–1500 MHz) and gain (>2 dBi) requirements according to the laboratory tests carried out in an anechoic chamber. This potential was attested in the PD detection tests, in which the developed sensor was able to detect several PD signals generated in laboratory, by the point-to-plane electrode conﬁguration, and in practice, by a 230 kV CT. Although the PD apparent charge levels could not be estimated in practice through the signals detected by the antenna, the evaluation of the induced voltage in the antenna terminals can provide information about the PD activity evolution inside the equipment. Therefore, the proposed bio-inspired PMA-Metamaterial set can provide continuous, non-invasive, and relatively low-cost monitoring of PD activity by means of dielectric windows in high voltage insulating systems. The conclusions obtained from the results of this article can be summarized as follows:

(1) Both the SRR-CLS designed unit cells presented the double negative behavior (µ < 0, ε < 0) for almost all the main frequency range of PD activity (300–1500 MHz). Therefore, both the SRR unit cells were classified as metamaterials applicable as superstrate for the designed bio-inspired PMA; (2) The application of a metamaterial superstrate resulted in displacements in the bio-inspired PMA operating frequency range, resulting in a 3.9% and 1.8% bandwidth reduction for the circular and rectangular SRRs, respectively; (3) Although there was a bandwidth reduction regarding the bio-inspired PMA, the application of the both metamaterial superstrates resulted in higher reflection coefficient values for the frequency range of 1100–1500 MHz, in which some reflection coefficient values were lower than the threshold of 10 dB, improving the antenna power transmission/reception capacity for this − frequency range; (4) Radiation patterns distortions were observed at the upper bio-inspired PMA operating frequencies for both metamaterial superstrates, in which the rectangular SRR presented lower radiation patterns distortions than the circular one for the upper frequencies, mainly at the back lobe direction; (5) The application of both designed superstrates resulted in maximum gain values almost increased in double regarding the reference bio-inspired PMA for the frequency range of 500–1000 MHz, resulting in a mean gain improvement higher than 1 dBi; (6) The practical application of the rectangular SRR superstrate resulted in a UHF sensor with an operating bandwidth (540–1500 MHz) that covers 80% of the main frequency range of PD activity (300–1500 MHz); (7) The measured mean gain for the bio-inspired PMA-metamaterial set was equal to 3.61 dBi, resulting in an enhancement of 0.7 dBi regarding the bio-inspired PMA (2.92 dBi) and representing a good detection sensitivity for PD application; Sensors 2019, 19, 4255 24 of 27

(8) In PD laboratory tests, the bio-inspired PMA-metamaterial set was able to detect apparent charges above 15 pC, generated in a point-to-plane electrode configuration in an oil cell; (9) The bio-inspired PMA presented measured voltage values with half of the magnitude obtained from the IEC 60270 standard method, resulting in a high PD detection sensitivity for a radiometric based method; (10) Lastly, the bio-inspired PMA-metamaterial set presented itself as effective in the practical PD detection application, since it was possible to detect a significant level of PD activity in a 230 kV substation CT.

As future works, the sensor will be improved through the application of calibration, classification, and localization techniques in order to fully characterize it as a PD UHF sensor applicable for equipment diagnosis.

Author Contributions: Writing, G.V.R.X.; Bio-inspired antenna design and simulation, G.V.R.X. and L.A.M.M.N.; Metamaterials design and simulation, G.V.R.X.; Laboratory tests, G.V.R.X.; A.C.d.O. and A.J.R.S.; Methodology, G.V.R.X.; A.C.d.O.; A.J.R.S.; E.G.d.C. and L.A.M.M.N.; Writing—Review and Editing, A.C.d.O.; A.J.R.S.; E.G.d.C. and L.A.M.M.N.; Substation field tests, G.V.R.X.; A.C.d.O.; V.C.d.S. and L.A.M.M.N.; Conceptualization, G.V.R.X.; A.J.R.S.; E.G.d.C. and L.A.M.M.N. Funding: This research received no external funding. Acknowledgments: The authors want to acknowledge the Postgraduate Program in Electrical Engineering (PPgEE) of the Federal University of Campina Grande (UFCG), the Coordination for the Improvement of Higher Level Education Personnel (CAPES), the National Council for Technological and Scientific Development (CNPq), and the Companhia Hidro Elétrica do São Francisco (Chesf). Conflicts of Interest: The authors declare no conflict of interest.

References

1. Natrass, D.A. Partial Discharge Measurement and Interpretation. IEEE Electr. Insul. Mag. 1988, 4, 10–23. [CrossRef] 2. Kreuger, F.H.; Gulski, E.; Krivda, A. Classiﬁcation of partial discharges. IEEE Trans. Electr. Insul. 1993, 28, 917–931. [CrossRef] 3. International Electrotechnical Comission. High Voltage Test Techniques: Partial Discharge Measurements; Technical Repost No. 60270; International Electrotechnical Comission: Geneva, Svitzerland, 2000. 4. Stone, G. Partial discharge diagnostics and electrical equipment insulation condition assessment. IEEE Trans. Dielectr. Electr. Insul. 2005, 12, 891–904. [CrossRef] 5. Meira, M.; Ruschetti, C.R.; Álvarez, R.E.; Verucchi, C.J. Power transformers monitoring based on electrical measurements. IET Gener. Transm. Distrib. 2018, 12, 2805–2815. [CrossRef] 6. Stone, G.C. Condition monitoring and diagnostics of motor and stator windings—A review. IEEE Trans. Dielectr. Electr. Insul. 2013, 20, 2073–2080. [CrossRef] 7. Refaat, S.S.; Shams, M.A. A review of partial discharge detection, diagnosis techniques in high voltage power cables. In Proceedings of the 2018 IEEE 12th International Conference on Compatibility, Power Electronics and Power Engineering (CPE-POWERENG), Ad Dawhah, Qatar, 4–10 December 2018. 8. Sahoo, A.; Subramaniam, A.; Bhandari, S.; Panda, S.K. A review on condition monitoring of GIS. In Proceedings of the 2017 International Symposium Electrical Insulating Materials (ISEIM), Toyohashi City, Japan, 12–15 September 2017. 9. Tenbohlen, S.; Denissov, D.; Hoek, S.M.; Markalous, S.M. Partial discharge measurement in the ultra high frequency (UHF) range. IEEE Trans. Dielectr. Electr. Insul. 2008, 15, 1544–1552. [CrossRef] 10. Judd, M.; Pryor, B.; Kelly, S.; Hampton, B. Transformer monitoring using the UHF technique. In Proceedings of the 11th International Symposium on High Voltage Engineering (ISH), London, UK, 23–27 August 1999. 11. Hampton, B.F.; Meats, R.J. Diagnostic measurements at UHF in gas insulated substations. IEEE Proc. 1988, 135, 137–144. [CrossRef] 12. Masaki, K.; Sakakibara, T.; Murase, H.; Akazaki, M.; Uehara, K.; Menju, S. On-site measurement for the development of on-line partial discharge monitoring system in GIS. IEEE Trans. Power Del. 1994, 9, 805–810. [CrossRef] Sensors 2019, 19, 4255 25 of 27

13. Judd, M.; Farish, O.; Pearson, J. UHF couplers for gas-insulated substations: A calibration technique. IEEE Proc. Sci. Meas. Technol. 1997, 144, 117–122. [CrossRef] 14. Judd, M.; Yang, L.; Hunter, I. Partial discharge monitoring for power transformer using UHF sensors. Part I: Sensors and signal interpretation. IEEE Electr. Insul. Mag. 2005, 21, 5–14. [CrossRef] 15. Lee, C.H.; Lin, Y.C.; Chiu, M.Y.; Huang, C.H.; Yen, S.S.; Haeng, C. The study on diagnostics for aging trend of cable termination. In Proceedings of the International Conference on Condition Monitoring and Diagnosis, Beijing, China, 21–24 April 2008. 16. Judd, M.; 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] 17. Judd, M. Locating Partial Discharges in Power Transformers. In Proceedings of the 10th Euro TechCon, Warwick, UK, 7–9 November 2011. 18. Hoshino, T.; Nojima, K.; Hanai, M. Real-time PD identification in diagnosis of GIS using symmetric and asymmetric UHF sensors. IEEE Trans. Power Del. 2004, 19, 1072–1077. [CrossRef] 19. Ju, T.; Zhongrong, X.; Xiaoxing, Z.; Caixin, S. GIS partial discharge quantitative measurements using UHF microstrip antenna sensors. In Proceedings of the Annual Report on Electrical Insulation and Dielectric Phenomena, Vancouver, BC, Canada, 14–17 October 2007. 20. Balanis, C. Antenna Theory–Analysis and Design; Wiley: Hoboken, NJ, USA, 2005. 21. Ahmed, O.M.H.; Sebak, A.R. A Novel Maple-Leaf Shaped UWB Antenna with a 5.0–6.0 GHz Band-Notch Characteristic. Prog. Electromagn. Res. 2009, 11, 39–49. [CrossRef] 22. Ebnabbasi, K. A bio-inspired printed-antenna transmission-range detection systems. IEEE Antennas Propag. Mag. 2013, 55, 193–200. [CrossRef] 23. Silva, P.F., Jr.; Freire, R.C.S.; Serres, A.J.R.; Silva, P.H.F.; Silva, J.C. Wearable Textile Bioinspired Antenna for 2G, 3G and 4G Systems. Microw. Opt. Technol. Lett. 2016, 58, 2018–2023. [CrossRef] 24. Serres, A.J.R.; Serres, G.K.F.; Silva, P.F., Jr.; Freire, R.C.S.; Cruz, J.N.; Albuquerque, T.C.; Oliveira, M.A.; Silva, P.H.F. Bio-Inspired Microstrip Antenna. IntechOpen 2017.[CrossRef] 25. Cruz, J.N.; Serres, A.J.R.; Oliveira, A.C.; Xavier, G.V.R.; Albuquerque, C.C.R.; Costa, E.G.; Freire, R.C.S. Bio-Inspired Monopole Antenna Applied to Partial Discharge Detection. Sensors 2019, 19, 628. [CrossRef] 26. 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] 27. Yang, F.; Peng, C.; Yang, Q.; Luo, H.; Ullah, I.; Yang, Y. An UWB Printed Antenna for Partial Discharge UHF Detection in High Voltage Switchgears. Prog. Electromagn. Res. C 2016, 69, 105–114. [CrossRef] 28. Veselago, V.G. The Electrodynamics of Substances with Simultaneously Negative Values of ε and µ. Sov. Phys. Uspekhi 1968, 10, 509–514. [CrossRef] 29. Pendry, J.B.; Holden, A.J.; Stewart, W.J.; Youngs, I.I. Extremely Low Frequency Plasmons in Metallic Mesostructures. Phys. Rev. Lett. 1996, 76, 4773–4776. [CrossRef][PubMed] 30. Pendry, J.B.; Holden, A.J.; Robbins, D.J.; Stewart, W.J. Magnetism from Conductors and Enhanced Nonlinear Phenomena. IEEE Trans. Microw. Theory Tech. 1999, 47, 2075–2084. [CrossRef] 31. Shelby, R.A.; Smith, D.R.; Schultz, S. Experimental Verification of a Negative Index of Refraction. Science 2001, 292, 77–79. [CrossRef][PubMed] 32. Enoch, S.; Tayeb, G.; Sabouroux, P.; Guerin, N.; Vincent, P. A Metamaterial for Directive Emission. Phys. Rev. Lett. 2002, 89, 213902. [CrossRef][PubMed] 33. Ziolkowski, R.W.; Erentok, A. Metamaterial-Based Efficient Electrically Small Antennas. IEEE Trans. Antennas Propag. 2006, 54, 2113–2130. [CrossRef] 34. Islam, M.T.; Islam, M.M.; Samsuzzaman, M.; Faruque, M.R.I.; Misran, N. A Negative Index Metamaterial-Inspired UWB Antenna with an Integration of Complementary SRR and CLS Unit Cells for Microwave Imaging Sensor Applications. Sensors 2015, 15, 11601–11627. [CrossRef] 35. Soffiati, A.; Max, Y.; Silva, S.G.; Mendonça, L.M. Microwave Metamaterial-Based Sensor for Dielectric Characterization of Liquids. Sensors 2018, 18, 1513. [CrossRef][PubMed] 36. Majid, H.A.; Rahim, M.K.A.; Masri, T. Microstrip Antenna’s Gain Enhancement of Circular Microstrip Antenna with Left-Handed Metamaterial Structure. Prog. Electromagn. Res. 2009, 8, 235–247. [CrossRef] Sensors 2019, 19, 4255 26 of 27

37. Srivastava, K.; Kumar, A.; Chadhary, P.; Kanaujia, B.K.; Dwari, S.; Verma, A.K.; Esselle, K.P.; Mittra, R. Wideband and High-Gain Circularly Polarized Microstrip Antenna Design Using Sandwiched Metasurfaces and Partially Reflecting Surface. IET Microw. Antennas Propag. 2018, 13, 305–312. [CrossRef] 38. Das, S.; Mitra, D. A Compact Wideband Flexible Implantable Slot Antenna Design with Enhanced Gain. IEEE Trans. Antennas Propag. 2018, 66, 409–4314. [CrossRef] 39. Singh, A.K.; Abegaonkar, M.P.; Koul, S.K. High Gain and High Aperture Efficiency Cavity Resonator Antenna Using Metamaterial Superstrate. IEEE Antennas Wirel. Propag. Lett. 2017, 16, 2388–2391. [CrossRef] 40. Venkatesan, R.; Singaravelu, R. An Overview of Metamaterials in Biomedical Applications. In Proceedings of the Progress in Electromagnetics Research Symposium (PIERS), Stockholm, Sweden, 12–15 August 2013. 41. Ramakrishna, S.A. Physics of negative refractive index materials. Rep. Prog. Phys. 2005, 68, 449. [CrossRef] 42. Ojaroudiparchim, N.; Sheng, M.; Pedersen, G.F. Low-Profile Fabry-Pérot Cavity Antenna with Metamaterial SRR Cells for Fifth Generation Systems. In Proceedings of the 2016 21st International Conference on Microwave, Radar and Wireless Communications (MIKON), Krakow, Poland, 9–11 May 2016. 43. Singh, A.K.; Abegaonkar, M.P.; Koul, S.K. A Negative Index Metamaterial Lens for Antenna Gain Enhancement. In Proceedings of the 2017 International Symposium on Antennas and Propagation (ISAP), Phuket, Thailand, 30 October–2 November 2017. 44. Paleti, S.; Bhattacharjee, A. Design of PIFA Antenna with Metamaterial Superstrate at 2.4 GHz Using HFSS. In Proceedings of the International Conference on Communication and Signal Processing, Gold Coast, Australia, 13–15 December 2017. 45. Priyadharisini, S.G.; Rufus, E. A Double Negative Metamaterial Inspired Miniaturized Rectangular Patch Antenna with Improved Gain and Bandwidth. In Proceedings of the Progress in Electromagnetics Research Symposium (PIERS), Singapore, 19–22 November 2017. 46. Alhawari, A.R.H.; Ismail, A.; Mahdi, M.A. Compact Ultra-Wideband Metamaterial Antenna. In Proceedings of the Asia-Pacific Conference on Communications (APCC), Yogyakarta, Indonesia, 25–27 August 2016. 47. Kumar, U.; Upadhyay, D.K.; Shahu, B.L. Improvement of performance parameters of rectangular patch antenna using metamaterial. In Proceedings of the IEEE International Conference on Recent Trends in Electronics Information Communication Technology, Bangalore, India, 20–21 May 2016. 48. Smith, D.R.; Padilla, W.J.; Vier, D.C.; Nemat-Nasser, S.C.; Schultz, S. Composite Medium with Simultaneously Negative Permeability and Permittivity. Phys. Rev. Lett. 2000, 84, 4184–4187. [CrossRef][PubMed] 49. Ziolkowski, R.W. Design, Fabrication, and Testing of Double Negative Metamaterials. IEEE Trans. Antennas Propag. 2003, 51, 1516–1529. [CrossRef] 50. Islam, M.T.; Ashraf, F.B.; Alam, T.; Misran, N.; Mat, K.B. A Compact Ultrawideband Antenna Based on Hexagonal Split-Ring Resonator for pH Sensor Application. Sensors 2018, 18, 2959. [CrossRef] 51. Pandit, S.; Mohan, A.; Ray, P. A Low-Profile High-Gain Substrate Integrated Waveguide Slot Antenna with Suppressed Cross-Polarization using Metamaterial. IEEE Antennas Wirel. Propag. Lett. 2017, 16, 1614–1617. [CrossRef] 52. Zheng, Y.; Gao, J.; Zhou, Y.; Cao, Z.; Yang, H.; Li, S.; Li, T. Wideband Gain Enhancement and RCS Reduction of Fabry-Perot Resonator Antenna with Chessboard Arranged Metamaterial Superstrate. IEEE Trans. Antennas Propag. 2017, 66, 590–599. [CrossRef] 53. Hashemi, S.K. Microwave Devices and Techniques for Ultra Wideband (UWB) Communication Systems. Ph.D. Thesis, University of Essex, Colchester, UK, 2011. 54. Smith, D.R.; Schultz, S.; Markos, P.; Soukoulis, C.M. Determination of effective permittivity and permeability of metamaterials from reflection and transmission coefficients. Phys. Rev. B 2002, 65, 195104. [CrossRef] 55. Jaber, A.A.; Lazaridis, P.I.; Moradzadeh, M.; Glover, I.A.; Zharis, Z.D.; Vieira, M.F.; Judd, M.D.; Atkinson, R.C. Calibration of Free-space Radiometric Partial Discharge Measurements. IEEE Trans. Dielectr. Electr. Insul. 2017, 24, 3004–3014. [CrossRef] 56. Bartnikas, R.; McMahon, E. Engineering Dielectrics 1: Corona Measurement and Interpretation; ASTM International: West Conshohocken, PA, USA, 1979. 57. Ogihara, H. Detection and location of coronas in oil immersed transformers with corona detector. Electr. Eng. Jpn. 1964, 84, 12–21. 58. International Electrotechnical Commission (IEC). Power Transformer–Part 1: General; IEC: Geneva, Switzerland, 2011. 59. Wang, M.; Vandermaar, A.J.; Srivastava, K.D. Review of condition assessment of power transformers in service. IEEE Electr. Insul. Mag. 2002, 18, 12–25. [CrossRef] Sensors 2019, 19, 4255 27 of 27

60. Jaber, A.; Lazaridis, P.; Saeed, B.; Zhang, Y.; Khan, U.; Upton, D.; Ahmed, H.; Upton, D.; Mather, P.J.; Atkinson, R.C.; et al. Frequency spectrum analysis of radiated partial discharge signals. In Proceedings of the EUROEM 2016, London, UK, 11–14 July 2016. 61. Chai, H.; Phung, B.T.; Zhang, D. Development of UHF Sensors for Partial Discharge Detection in Power Transformer. In Proceedings of the 2018 Condition Monitoring and Diagnosis (CMD), Perth, WA, Australia, 23–26 September 2018. 62. Zhang, Y.; Lazaridis, P.; ABD-Alhameed, R.; Glover, I. A compact wideband printed antenna for free-space radiometric detection of partial discharge. Turk. J. Electr. Eng. Comput. Sci. 2017, 25, 1291–1299. [CrossRef] 63. Mukhtar, S.M.; Isa, M.; Al-Hadi, A.A. Design of UHF Antenna Sensor for Partial Discharge Detection in High Voltage Substation. IOP Conf. Ser. Mater. Sci. Eng. 2018, 318, 1–6. [CrossRef] 64. Wang, Y.; Wu, J.; Chen, W.; Wang, Y. Design of a UHF Antenna for Partial Discharge Detection of Power Equipment. J. Sensors 2014, 1–8. [CrossRef] 65. Zhang, Y.; Glover, I. Design of an Ultrawideband VHF/UHF Antenna for Partial Discharge Detection. In Proceedings of the 2014 IEEE International Conference on Signal Processing, Communications and Computing (ICSPCC), Guilin, China, 5–8 August 2014 2014. 66. Marchal, A.; Monedero, M.; Le Thuc, P.; Staraj, R. Ultra-wide band antenna for partial discharge detection inside switchgear for on-line monitoring. In Proceedings of the 2018 IEEE Conference on Antenna Measurements & Applications (CAMA), Vasteras, Sweden, 3–6 September 2018. 67. Bin, L.; Jingle, A.; Weidong, Z.; Youlin, X. A Design of Multi-band UHF Sensor for Partial Discharge Detection. Trans. Tech. Publ. Appl. Mech. Mater. 2013, 394, 435–440. 68. Liu, J.; Zhang, G.; Dong, J.; Wang, J. Study on Miniaturized UHF Antennas for Partial Discharge Detection in High-Voltage Electrical Equipment. Sensors 2015, 15, 29434–29451. [CrossRef] 69. Ye, H.; Qian, Y.; Dong, Y.; Sheng, G.; Jiang, X. Development of multi-band ultra-high-frequency sensor for partial discharge monitoring based on the meandering technique. IET Sci. Meas. Technol. 2014, 8, 327–335. 70. Li, J.; Wang, P.; Jiang, T.; Bao, L.; He, Z. UHF stacked Hilbert antenna array for partial discharge detection. IEEE Trans. Antenn. Propag. 2013, 61, 5798–5801. [CrossRef] 71. Yao, C.; Chen, P.; Huang, C.; Chen, Y.; Qiao, P. Study on the Application of an Ultra-High-Frequency Fractal Antenna to Partial Discharge Detection in Switchgears. Sensors 2013, 13, 17362–17378. [CrossRef] 72. Yongqiang, W.; Zhuang, W.; Jianfang, L. UHF Moore Fractal Antenna for Online GIS PD Detection. IEEE Antennas Wirel. Propag. Lett. 2017, 16, 852–855. 73. Lopez-Roldan, J.; Tang, T.; Gaskin, M. Optmisation of a sensor for onsite detection of partial discharges in power transformers by the UHF method. IEEE Trans. Dielect. Electr. Insul. 2008, 15, 1634–1639. [CrossRef] 74. FT-RF 500 MHz to 8 GHz Double Ridged Broadband Waveguide Horn Antenna. Model: HA-05M08G-NF. Available online: http://www.ft-rf.com/front/bin/ptdetail.phtml?Part=500MHz-8GHz-Double-Ridge- Broadband-Antenna-N-Female&Category=359789 (accessed on 29 September 2019). 75. EC Microwave Broadband Horn Antenna Operating From 500 MHz to 3 GHz Double Ridged. Available online: http://www.ecmicrowave.com/m_product/96-Broadband-Horn-Antenna-Operating-from-500MHz- to-3GHz-Double-Ridged.html (accessed on 29 September 2019). 76. Bancroft, R.; Chou, R.C. Vivaldi Antenna Impedance Bandwidth Dpendence on Stripline to Bilateral Slotline Transition. Microwave Opt. Technol. Lett. 2013, 55, 937–941. [CrossRef] 77. Biancheri-Astier, M.; Diet, A.; Le Bihan, Y.; Grzeskowiak, M. UWB Vivaldi Antenna Array Lower Band Improvement for Ground Penetrating Radar Applications. Radioengineering 2019, 28, 92–98. [CrossRef] 78. Aaronia AG Active Highhend LogPer Antenne HyperLOG 30100X. Available online: https://www.aaronia. com/products/antennas/HyperLog-30100-X/ (accessed on 29 September 2019).

© 2019 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/).