Passive Power Line Communication Filter Design and Benchmarking Using Scattering Parameters

: NB-PLC (narrowband power line communication) is a method of data communication that involves superimposing a relatively high-frequency signal (9 kHz to 500 kHz), which contains data, onto the power grid’s low frequency (50 to 60 Hz) signal. While using the existing power grid as a transmission medium is convenient, the power grid was not designed for this purpose, leading to challenges such as conducted emissions and infrastructure limitations. To overcome these technical challenges, passive ﬁlters are necessary. This article presents the design, simulation (using scattering parameters), and evaluation of an NB-PLC ﬁlter by comparing it to commercially available ﬁlters. Our proposed design and benchmarking methods enable the accurate prediction of the ﬁlter’s behavior in ﬁeld conditions. After comparing our ﬁlter with commercially available ﬁlters, we observed that it exhibits superior characteristics. Speciﬁcally, our ﬁlter has the best insertion loss versus frequency, achieved three times higher attenuation at 50 kHz ( − 130 dB) compared to the best commercially available ﬁlter ( − 40 dB), and has a power consumption of 0.6 W, which is comparable to the most power-efﬁcient commercial ﬁlter (0.5 W). Additionally, our ﬁlter has the second best input and output impedance of 3.6 Ω within the frequency range of 35–95 kHz.


Introduction
As of 2020, it is estimated that 43% of electricity consumers in the EU-28 (European Union) have SMs (smart meters) installed, amounting to 123 million [1]. It is expected that by 2024, the number of installed SMs will increase to 223 million and that by 2030 it will reach 266 million. A total of 16 countries out of EU-28 have chosen NB-PLC (Narrow band power line communication) as a communication method between SM and DC (data concentrator) [2]. The study performed by the Joint Research Centre [2] estimated higher rollout rates compared to [1], while a market study performed at the end of 2020 showed that 130 million smart meters had been rolled out in Europe out of which 72% use PLC [3].
We have identified that only in the following cases do standards specify the emission limits in the 3-148.5 kHz frequency band: NB-PLC specific standard [16], lightning equipment [17], and one home appliance type (inductive ovens) from the home appliances EMC standard [18]. Figure 1 graphically represents the IEC EMC standards that cover most equipment connected to the power grid in Europe and the frequency range they cover; for most of the standards there is no emission limit below 150 kHz. standard [18]. Figure 1 graphically represents the IEC EMC standards that cover most equipment connected to the power grid in Europe and the frequency range they cover; for most of the standards there is no emission limit below 150 kHz. Mains connected switched-mode power supplies and inverters commonly utilize switching frequencies below 150 kHz, resulting in the generation of conducted emissions that impact NB-PLC [15,19]. The selection of switching frequencies below 150 kHz is motivated by several key factors [20]: • Efficiency: lower switching frequencies can improve the overall efficiency of the SMPS. At lower frequencies, the switching losses in the power semiconductors, such as transistors or diodes, are reduced. This results in lower power dissipation and higher conversion efficiency; • Electromagnetic interference (EMI): higher switching frequencies can generate more emissions due to increased harmonics and faster switching transitions. Relaxed or an absence of limits below 150 kHz contribute to this choice as well; • Thermal management: switching components in SMPS can generate heat and operating at lower frequencies can help manage thermal issues. With reduced switching frequency, the components have more time to dissipate heat between each switching cycle; • Component selection: some components used in SMPS, such as inductors and transformers, may have limitations at higher frequencies, including increased losses and size constraints.
The growing adoption of NB-PLC within smart grid applications, combined with existing standardization gaps and the presence of grid-connected devices that operate at switching frequencies below 150 kHz, necessitates the development of power line filters capable of providing significant attenuation within the Cenelec bands A, B, C, and D.
Passive filters were used for a long time [21][22][23][24] as a means of addressing conducted emissions in the power grid or to isolate different sections of a PLC network [25]. This article aims to contribute to this field of expertise by addressing several novel aspects. Firstly, it serves as a comprehensive source of information for designing and testing narrowband NB-PLC filters. Through an extensive literature review, it provides a complete understanding of the development, integration, and testing considerations associated with NB-PLC filters.
In addition, this research introduces a new approach by employing S-parameters (scattering parameters) for the design and simulation of NB-PLC filters. While the use of Mains connected switched-mode power supplies and inverters commonly utilize switching frequencies below 150 kHz, resulting in the generation of conducted emissions that impact NB-PLC [15,19]. The selection of switching frequencies below 150 kHz is motivated by several key factors [20] The growing adoption of NB-PLC within smart grid applications, combined with existing standardization gaps and the presence of grid-connected devices that operate at switching frequencies below 150 kHz, necessitates the development of power line filters capable of providing significant attenuation within the Cenelec bands A, B, C, and D.
Passive filters were used for a long time [21][22][23][24] as a means of addressing conducted emissions in the power grid or to isolate different sections of a PLC network [25]. This article aims to contribute to this field of expertise by addressing several novel aspects. Firstly, it serves as a comprehensive source of information for designing and testing narrowband NB-PLC filters. Through an extensive literature review, it provides a complete understanding of the development, integration, and testing considerations associated with NB-PLC filters.
In addition, this research introduces a new approach by employing S-parameters (scattering parameters) for the design and simulation of NB-PLC filters. While the use of S-parameters in design and simulation methods is not unprecedented [26,27], this study represents the first instance of applying this specific approach to NB-PLC filter design.
To demonstrate the efficacy of the proposed design methods described in this article, a 13th order power line filter was developed utilizing custom-made inductors. The designed filter exhibits high insertion loss characteristics below 150 kHz. It achieves a remarkable Appl. Sci. 2023, 13, 6821 3 of 21 attenuation of −130 dB at 50 kHz while maintaining a low power consumption of 0.6 W with an input and output impedance of 3.6 Ω.
Through these findings, the article highlights the potential and advantages of employing the described methods for designing NB-PLC filters and offering improved performance in terms of insertion loss, attenuation, power consumption, and impedance characteristics.
The rest of the article is structured as follows: Section 2 provides a comprehensive overview of the integration of power line filters into the power grid, highlighting the key services present in the grid and the two main types of power line filter topologies.
Section 3 presents the performance tests that ought to be conducted as an integral part of power line filter benchmarking. The relationship between testing methods and the actual conditions found in the power grid is also presented. By doing so, we aim to provide a better understanding of the factors that affect filter performance in real-world settings.
Section 4, in addition to Section 3, outlines the safety and immunity tests that should be conducted as part of benchmarking as well as the passive filter standard [28].
Section 5 presents the design, simulation, and evaluation of the proposed passive power line filter as well as a comparison with three commercially available filters. Our approach is pragmatic and uses S-parameters to facilitate the filter design process.
Section 6 presents the conclusions of our research activities presented in the article and proposes future work in this field of expertise.

Filters Integration, Topologies, and Impact in the Power Grid
This section presents the basic types of PLC filters and their main characteristics as well as interferences between filters and other services in the power grid.

Integration of Power Line Filters in the Grid
Typically, power line communication filters are used in the smart grid in the following cases:

•
Between the SM and the fuse box for filtering the conducted noise that is generated by the devices in the house [29]. If the filter does not require a ground connection or if the main fuse is not differential, it can be installed after the main fuse without any issues. However, if the filter does require a ground connection or if the main fuse is differential, installing the filter after the main fuse may cause the fuse to trip. This filter use case is labeled as Filter Type 1 in Figure 2; • Between two sections of the power-line for separating NB-PLC devices from two areas which are connected to the same voltage transformer. The division in two areas ensures that the SM using NB-PLC from each area connects to the PLC-DC (power line communication data concentrator) they are supposed to. This type of filter installation is required when there are many SM connected to a branch and an additional DC is installed so that the data-rate is improved [25]. This filter use case is labeled as Filter Type 2 in Figure 2.

General Topolgies of Powerline Filters
In terms of the safety rating (leakage current) and insertion loss, both are two topologies of PLC filters with and without Y capacitors, namely C y , having a middle point connected to protective earth [21]; these types of filters are shown in Figure 3. All other types of passive filters are different arrangements starting from these two types of filters.
The graph in Figure 4 shows the insertion loss (with symmetrical loading) of the two filters depicted in Figure 3, as simulated using SPICE (Simulation Program with Integrated Circuit Emphasis). The optional resistor has the function to dampen oscillations at resonant frequencies and to discharge the capacitors. Appl

General Topolgies of Powerline Filters
In terms of the safety rating (leakage current) and insertion loss, both are two topologies of PLC filters with and without Y capacitors, namely Cy, having a middle point connected to protective earth [21]; these types of filters are shown in Figure 3. All other types of passive filters are different arrangements starting from these two types of filters. The graph in Figure 4 shows the insertion loss (with symmetrical loading) of the two filters depicted in Figure 3, as simulated using SPICE (Simulation Program with Integrated Circuit Emphasis). The optional resistor has the function to dampen oscillations at resonant frequencies and to discharge the capacitors.

Services Present in the Power Grid
The below-mentioned features present in the grid should not be affected if a PLC filter is fitted in the installation.

Residual Current Protective Devices
RCD (residual current protective device) is used in the installation as a protective device against faulty devices or the carelessness of users [30]. The tripping current of an RCD is Δn ≤ 30 mA, which can be an issue in the case of a PLC filter with a leakage current being installed after the main fuse. Of course, the tripping current can be adjusted in some RCDs but the leakage of the Y capacitors might change due to capacitor ageing causing the RCD to trip unreliably.

Ripple Control System
Ripple control systems are devices used by the grid operator for load switching. Ripple control receivers are components of a system of remote control permitting the simultaneous switching of many loads from a central point. For the system to function it is necessary that the signal arrives from the injection point to the receiver sites. The ripple control signal frequency band is between 110 Hz and 3000 Hz [31] so it might be easily filtered out by the PLC filter.

Remote Connection of the Disconnect Unit
Some SMs detect the state of the main fuse when the power supply is disconnected from the SM disconnect unit. This feature is used to remote close the disconnect unit of the SM by cycling the main fuse off and on so that the energy supply to the customer is reestablished. The feature can be affected by the PLC filter in case it is fitted between the

Services Present in the Power Grid
The below-mentioned features present in the grid should not be affected if a PLC filter is fitted in the installation.

Residual Current Protective Devices
RCD (residual current protective device) is used in the installation as a protective device against faulty devices or the carelessness of users [30]. The tripping current of an RCD is I ∆n ≤ 30 mA, which can be an issue in the case of a PLC filter with a leakage current being installed after the main fuse. Of course, the tripping current can be adjusted in some RCDs but the leakage of the Y capacitors might change due to capacitor ageing causing the RCD to trip unreliably.

Ripple Control System
Ripple control systems are devices used by the grid operator for load switching. Ripple control receivers are components of a system of remote control permitting the simultaneous switching of many loads from a central point. For the system to function it is necessary that the signal arrives from the injection point to the receiver sites. The ripple control signal frequency band is between 110 Hz and 3000 Hz [31] so it might be easily filtered out by the PLC filter.

Remote Connection of the Disconnect Unit
Some SMs detect the state of the main fuse when the power supply is disconnected from the SM disconnect unit. This feature is used to remote close the disconnect unit of the SM by cycling the main fuse off and on so that the energy supply to the customer is reestablished. The feature can be affected by the PLC filter in case it is fitted between the SM and fuse box [32][33][34]. There is no description of the way this feature is implemented but it requires signal injection and measurement, which may be affected when using a PLC filter.

Filter Performance Measurement
This section presents the performance tests that should be performed as part of the benchmarking: insertion loss, PLC signal filtering, and impedance measurement.

Insertion Loss
In laboratory environments, the most straightforward method of testing the filters is by measuring its insertion loss (transfer function) using a VNA (vector network analyzer) [22] or an FRA (frequency response analyzer). In general, an FRA will allow lower frequency measurements and are robust to use in conjunction with power line couplers. Figure 5 is an offline insertion loss measurement setup [35] and Figure 6 is an online impedance measurement setup which uses line couplers [36], AC power supply, and load impedance.
The practical implementation of the test setup from Figure 6 can be challenging due to non-standardized power supply impedance Z S and load impedance Z L values at PLC communication frequencies.
Passive impedance matching circuits can be used to measure the insertion loss of the filter when the impedance is different to 50 Ω. The main problem of passive impedance matching circuits is that the impedance matching is achieved only at a certain frequency; hence, to obtain a proper insertion loss with impedances different to 50 Ω, several matching circuits should be used. Another way of performing non-50 Ω measurements is to use matching transformers also known as baluns [37,38]. Some of the FRA models available on the market have separate ports for measuring and injecting the signal to the DUT (device under test) [39] so that an impedance matching circuit is not needed and the results with loading impedances different to 50 Ω are properly measured.
In laboratory environments, the most straightforward method of testing the filters is by measuring its insertion loss (transfer function) using a VNA (vector network analyzer) [22] or an FRA (frequency response analyzer). In general, an FRA will allow lower frequency measurements and are robust to use in conjunction with power line couplers. Figure 5 is an offline insertion loss measurement setup [35] and Figure 6 is an online impedance measurement setup which uses line couplers [36], AC power supply, and load impedance. The practical implementation of the test setup from Figure 6 can be challenging due to non-standardized power supply impedance ZS and load impedance ZL values at PLC communication frequencies.
Passive impedance matching circuits can be used to measure the insertion loss of the filter when the impedance is different to 50 Ω. The main problem of passive impedance matching circuits is that the impedance matching is achieved only at a certain frequency; hence, to obtain a proper insertion loss with impedances different to 50 Ω, several matching circuits should be used. Another way of performing non-50 Ω measurements is to use matching transformers also known as baluns [37,38]. Some of the FRA models available on the market have separate ports for measuring and injecting the signal to the DUT (device under test) [39] so that an impedance matching circuit is not needed and the results with loading impedances different to 50 Ω are properly measured.  In order to provide a controlled input impedance and noiseless power supply, a line impedance stabilization network (LISN) and two PLC couplers were used. The test setup from Figure 7 is derived from PLC and EMC standards [17,40] and it has the following features: • Isolated AC power supply (isolation transformer); • Symmetrical V-LISN that allows symmetrical measurements on phase and neutral; the unused half of the LISN is terminated with 50 Ω; In order to provide a controlled input impedance and noiseless power supply, a line impedance stabilization network (LISN) and two PLC couplers were used. The test setup from Figure 7 is derived from PLC and EMC standards [17,40] and it has the following features: • Isolated AC power supply (isolation transformer); • Symmetrical V-LISN that allows symmetrical measurements on phase and neutral; the unused half of the LISN is terminated with 50 Ω; • Z in and Z L can be adjusted in order to obtain the required impedance at the port of the filter. In order to provide a controlled input impedance and noiseless power supply, a line impedance stabilization network (LISN) and two PLC couplers were used. The test setup from Figure 7 is derived from PLC and EMC standards [17,40] and it has the following features: • Isolated AC power supply (isolation transformer); • Symmetrical V-LISN that allows symmetrical measurements on phase and neutral; the unused half of the LISN is terminated with 50 Ω; • Zin and ZL can be adjusted in order to obtain the required impedance at the port of the filter.  Power line impedances encountered in real situations are 4 to 15 times lower than those of commercially available LISN. Figure 8 graphically presents the impedance from 20 kHz to 150 kHz of an 50 Ω/50 µH + 5 Ω LISN [40] and the power line impedance measured experimentally in five field condition cases [41][42][43][44][45]. Appl Power line impedances encountered in real situations are 4 to 15 times lower than those of commercially available LISN. Figure 8 graphically presents the impedance from 20 kHz to 150 kHz of an 50 Ω/50 μH + 5 Ω LISN [40] and the power line impedance measured experimentally in five field condition cases [41][42][43][44][45]. The adaptive impedance circuit from Figure 9 can be used to introduce Zin and ZL closer to values encountered in field conditions [46]. Figure 10 plots the impedance | | and phase shift φ of half of the circuit (between L' and PE or N' and PE) from Figure 9; ideal components were used in the SPICE simulation. The adaptive impedance circuit from Figure 9 can be used to introduce Z in and Z L closer to values encountered in field conditions [46]. Figure 10 plots the impedance |Z| and phase shift ϕ of half of the circuit (between L' and PE or N' and PE) from Figure 9; ideal components were used in the SPICE simulation. The adaptive impedance circuit from Figure 9 can be used to introduce Zin and ZL closer to values encountered in field conditions [46]. Figure 10 plots the impedance | | and phase shift φ of half of the circuit (between L' and PE or N' and PE) from Figure 9; ideal components were used in the SPICE simulation.  The power consumption at the mains frequency f = 50 Hz of the adaptive impedance circuit from Figure 9 is P ≈ 1 kW and thus much lower compared to the power consumed The adaptive impedance circuit from Figure 9 can be used to introduce Zin and ZL closer to values encountered in field conditions [46]. Figure 10 plots the impedance | | and phase shift φ of half of the circuit (between L' and PE or N' and PE) from Figure 9; ideal components were used in the SPICE simulation.  The power consumption at the mains frequency f = 50 Hz of the adaptive impedance circuit from Figure 9 is P ≈ 1 kW and thus much lower compared to the power consumed The power consumption at the mains frequency f = 50 Hz of the adaptive impedance circuit from Figure 9 is P ≈ 1 kW and thus much lower compared to the power consumed if a purely resistive 2 Ω load is used, P = 26.45 kW. The power consumption computations are detailed in Equations (1)- (11).
X L = 2 π fL = 1.5 mΩ (3) cos ϕ = 0.9 (7) The online insertion loss test setup version 2 with Z in and Z L close to power line impedance replicates realistic power grid conditions, thus offering an accurate way of evaluating the PLC filters [47,48].

PLC Signal Filtering
In some use cases such as the filter, by separating two areas of the PLC system ( Figure 2) the filter must block the PLC signal.
In the absence of an FRA or VNA there are two pragmatic ways of testing the PLC filters: • Use PLC modems connected before and after the filter and measure the frame error rate, as shown in Figure 11; • Configure a PLC modem in transmit mode and measure the signal level with a spectrum analyzer before and after the filter, as shown in Figure 12.
The online insertion loss test setup version 2 with Zin and ZL close to power line impedance replicates realistic power grid conditions, thus offering an accurate way of evaluating the PLC filters [47,48].

PLC Signal Filtering
In some use cases such as the filter, by separating two areas of the PLC system ( Figure  2) the filter must block the PLC signal.
In the absence of an FRA or VNA there are two pragmatic ways of testing the PLC filters: • Use PLC modems connected before and after the filter and measure the frame error rate, as shown in Figure 11; • Configure a PLC modem in transmit mode and measure the signal level with a spectrum analyzer before and after the filter, as shown in Figure 12.

Impedance Measurement
The impedance of the filter should not affect the PLC modem performance. In some cases, the PLC filter is installed very close to the SM with a PLC modem, which can severely affect the transmitted and received signal. Higher filter input impedances can be achieved using higher order filters (multi-stage) such as the filter designed by us.
The most effective ways of measuring the impedance of the filter at PLC frequencies are: • Offline measurement using the RLC meter, impedance analyzer, or VNA, Figure 13a; • Online measurement using the impedance analyzer measuring method [49], Figure  13b; • Online measurement using the voltage ratio method [49], Figure 13c; • Offline measurement and monitoring of the voltage and current consumption of the PLC modem while transmitting carrier frequencies, Figure 13d.
Single port measurements with a VNA are not reliable at extreme impedance values (i.e., <1 Ω). The single port measurement limitation (S11) comes from the fact that the VNA is designed to measure the incident wave ( + ) entering the DUT (device under test) and the reflected wave( − ) on impedances close to 50 Ω [50][51][52]. The ratio of these two waves is called the reflection coefficient (Γ) and it is used to compute the impedance of the DUT ZL. The resolution of a VNA is not high enough for single port measurements at extreme impedance values: when ZL is very low, Γ becomes close to −1, and when ZL is very high, Γ becomes close to +1.

Impedance Measurement
The impedance of the filter should not affect the PLC modem performance. In some cases, the PLC filter is installed very close to the SM with a PLC modem, which can severely affect the transmitted and received signal. Higher filter input impedances can be achieved using higher order filters (multi-stage) such as the filter designed by us.
The most effective ways of measuring the impedance of the filter at PLC frequencies are:

•
Offline measurement using the RLC meter, impedance analyzer, or VNA, Figure 13a; • Online measurement using the impedance analyzer measuring method [49], Figure 13b; • Online measurement using the voltage ratio method [49], Figure 13c; • Offline measurement and monitoring of the voltage and current consumption of the PLC modem while transmitting carrier frequencies, Figure 13d.
Single port measurements with a VNA are not reliable at extreme impedance values (i.e., <1 Ω). The single port measurement limitation (S11) comes from the fact that the VNA is designed to measure the incident wave (E + ) entering the DUT (device under test) and the reflected wave (E − ) on impedances close to 50 Ω [50 -52]. The ratio of these two waves is called the reflection coefficient (Γ) and it is used to compute the impedance of the DUT Z L . The resolution of a VNA is not high enough for single port measurements at extreme impedance values: when Z L is very low, Γ becomes close to −1, and when Z L is very high, Γ becomes close to +1.
In order to overcome the limitations of the single port measurements, the 2-Port VNA measurement technique can be used which is the equivalent of the 4-wire Kelvin DC technique for measuring resistance. Port 1 is used to inject to the DUT and port 2 is used to measure the voltage drop on the DUT.

Additional Testing
This section presents the safety and immunity tests that should be performed as part of benchmarking in addition to the passive filter standard [28]: heating, surge, overcurrent, and short circuit.

Additional Testing
This section presents the safety and immunity tests that should be performed as part of benchmarking in addition to the passive filter standard [28]: heating, surge, overcurrent, and short circuit.

Heating
Under maximum rated operating conditions, the housing of the filter shall not reach a temperature, which might cause a fire hazard or influence the filter and devices surrounding the filter.
The test can be performed by inspecting the filter with the thermal view camera [23] while maximum current passes through the filter. Given that the filter may be installed under similar conditions as the SM, it is important to test it accordingly [53]:

•
The filter should carry the maximum rated current and voltage for 2 h; • The temperature rise of the external surface should not exceed 65 • C, with an ambient temperature of 40 • C.

Surge
The test should be carried out according to the relevant surge testing standard [54] under the following conditions:

•
Filter operating condition: circuits should be energized with reference voltage +5% without any load connected to the output of the circuit; • Cable length between the surge generator and filter: 1 m; • Tested in differential mode (line to line) and, if ground connection is available, common mode (line/neutral and ground); • Phase angle: pulses to be applied at 60 • and 240 • relative to the zero crossing of the mains supply; • Test voltage on the mains lines: 4 kV; • Generator source impedance: 2 Ω; • Number of pulses: five positive and five negative; • Repetition rate: maximum 1/min.

Overcurrent
It is recommended to conduct this test if there is no fuse fitted before the filter; this is the case when the filter is separating the two areas of the PLC grid as shown in Figure 2. The filter shall be able to carry a short time overcurrent of 30 times the maximum rated current for one half-cycle at a rated frequency [55]. No degradation of the filter characteristics should be observed after this test.

Short Circuit
The short circuit condition can be simulated by short-circuiting the output of the filter and feeding the input with two times the maximum rated current until the filter will result in an open circuit. During testing, the filter should not be in an unsafe condition and after the test it should result in an open circuit.

Overvoltage
This test simulates the overvoltage that may occur in the grid due to faults such as earth fault, phase reversal, and voltage unbalances. All of the overvoltage conditions translate in an increased line or phase voltage. The most extreme case that can occur in a system having V L−N = 230 V as the phase voltage is described below [55,56]:

PLC Filter Design, Simulation, and Evaluation
This section presents the design, simulation, and evaluation of the passive PLC filter.

PLC Filter Design
The filter was designed using the following requirements: • The filter should reach −120 dB attenuation at 50 kHz because the maximum PLC signal level is 120 dBuV [16], thus blocking PLC signal passing through the filter; • Filter power consumption at the mains frequency should be kept as low as possible; • The filter saturation current should be at least 30 A, making it suitable for installation between SM and electrical installation of the building (i.e., main fuse); • The filter input/output impedance in the NB-PLC frequency band must be as high as possible so that it does not affect the communication. As a minimum requirement, the impedance must be above 2 Ω because the PLC modems are designed and tested to transmit on 2 Ω loads [16]; • The filter design and component selection shall be conducted in order to ensure compliance with the tests described in Section 4 for additional testing.
We have designed the 13th order filter from Figure 14 in compliance with the requirements from the beginning of this section. The compliance and performance have been assessed in Section 5.3.
Electronic component selection has an important role in the implementation of the filter, Table 1 is the bill of materials explaining our selection.

TMOV
The filter incorporates a TMOV (thermally protected metal oxide varistor) from Little Fuse™ TMOV20RP460E [57]. The TMOV compared to traditional metal oxide varistors provides surge protection while ensuring that it will not conduct at mains voltage, thus overheating and ultimately catching fire [58]. The selected TMOV will start conducting at 460 V, thus safe operation of the filter is ensured during overvoltage V (L−L)max = 438 V.

R1, R2 and R3
An array of three series connected resistors is used for passive damping of oscillations that might occur in the grid [59]. Surge safety resistors SSR300J10K0TKZTB500 from Firstohm™ [60] were chosen. Series connection is preferred in order to ensure that there is no arching due to insufficient clearance and creepage distances [61].

C1 and C5
Polyester film capacitors capable of withstanding V (L−L)max = 438 V were used. Specifically in our design, JGGC series from JB ® [62] were used.

L1 to L8
The inductors were designed and built using a toroidal core made out of Sendust MS-184075-2 [63] manufactured by Micrometals™. Sendust is appropriate to be used in applications requiring high permeability, low coercivity, high resistivity, and high magnetization, such as inductors used in filtering applications [64]. In order to obtain L = 100 µH and a current rating of 40 A, N = 26 turns of 4 mm 2 stranded wire are used.

C2, C3, C4
Polyester film capacitors capable of withstanding V (L−L)max = 438 V were used. Specifically in our design, JGGC series from JB ® [62] were used. The value of the capacitors was adjusted to C = 2.2 µF based on simulation and calculation values in order to have a trade-off between power consumption, attenuation, and input/ output impedance.
tering applications [64]. In order to obtain L = 100 μH and a current rating of 40A, N = 26 turns of 4mm 2 stranded wire are used.

C2, C3, C4
Polyester film capacitors capable of withstanding V ( − ) = 438 V were used. Specifically in our design, JGGC series from JB ® [62] were used. The value of the capacitors was adjusted to C = 2.2 μF based on simulation and calculation values in order to have a trade-off between power consumption, attenuation, and input/ output impedance.   RFSim99 is a linear S-parameter-based circuit simulator which solves matrix equations for two-port devices. Figure 15 is the graphical representation of a two-port network [66,67]:

PLC Simulation Using the S-Parameters Simulator
• a 1 and a 2 are the incident waves for Port1 and Port2; • b 1 and b 2 are the reflected waves for Port1 and Port2; • S 21 is the forward transmission coefficient (from port 1 to port 2) and S 12 is the reverse transmission coefficient (from port 2 to port 1); • S 11 is the input reflection coefficient and S 22 is the output reflection coefficient.

PLC Simulation Using the S-Parameters Simulator
S11 parameters for each component were measured using a VNA and the obtained Touchstone files [65] are used in simulating the filter in the RFSim99 S-Parameters simulator. RFSim99 is a linear S-parameter-based circuit simulator which solves matrix equations for two-port devices. Figure 15 is the graphical representation of a two-port network [66,67]: • 1 and 2 are the incident waves for Port1 and Port2; • 1 and 2 are the reflected waves for Port1 and Port2; • 21 is the forward transmission coefficient (from port 1 to port 2) and 12 is the reverse transmission coefficient (from port 2 to port 1); • 11 is the input reflection coefficient and 22 is the output reflection coefficient. Below is the matrix algebraic representation of two port S-parameters [66,67] which is also used by the simulator: For the measurements, PLC filter online insertion loss test setup version 2, shown in Figure 7, was utilized. The actual implementation of the test setup is depicted in Figure  16. Below is the matrix algebraic representation of two port S-parameters [66,67] which is also used by the simulator: For the measurements, PLC filter online insertion loss test setup version 2, shown in Figure 7, was utilized. The actual implementation of the test setup is depicted in Figure 16.
For the measurements, PLC filter online insertion loss test setup version 2, shown in Figure 7, was utilized. The actual implementation of the test setup is depicted in Figure  16.   Measurements and graphical representation were performed with the following settings and conventions: The measured S21 [dB] parameter is considered to be the insertion loss of the filter [68]; • Due to the dynamic range and noise floor of the VNA, measurements can be performed starting low at −130 dB [69]. Although the simulated S21 magnitude values are as low as −300 dB, all magnitude values ≤ −140 dB were scaled to −140 dB in order to facilitate graphical representation.
Appl. Sci. 2023, 13, 6821 15 of 23 Figure 17 presents the results of the simulated and measured insertion loss for the proposed filter.
Measurements and graphical representation were performed with the following settings and conventions: • Measurement equipment reference impedance: 0 = 50 Ω; • The measured S21 [dB] parameter is considered to be the insertion loss of the filter [68]; • Due to the dynamic range and noise floor of the VNA, measurements can be performed starting low at −130 dB [69]. Although the simulated S21 magnitude values are as low as −300 dB, all magnitude values ≤−140 dB were scaled to −140 dB in order to facilitate graphical representation.

Proposed Filter Evaluation and Comparison to Other Commercially Available Filters
The filter proposed in Figure 14 is evaluated and compared to other filters advertised as for PLC and with similar characteristics to the proposed filter. The evaluation is performed according to a part of the test methods described above. In order to prevent product biasing, the filter manufacturers and manufacturer part numbers are anonymized; however, the schematic of the benchmarked filters together with the component values are shown in Section 5.3.1.

PLC Filter Insertion Loss and Power Consumption Measurements
Insertion loss measurements were performed in the following conditions:

•
The test setup used is the PLC filter online insertion loss test setup 2 shown in Figure  7; • The first step is to validate the test setup by directly connecting the PLC couplers to one another. The setup validity check is named "PLC Couplers" in the below figures; • The subsequent actions involve conducting measurements while the filter is installed, Figure 17. Comparison between measured and simulated insertion loss.

Proposed Filter Evaluation and Comparison to Other Commercially Available Filters
The filter proposed in Figure 14 is evaluated and compared to other filters advertised as for PLC and with similar characteristics to the proposed filter. The evaluation is performed according to a part of the test methods described above. In order to prevent product biasing, the filter manufacturers and manufacturer part numbers are anonymized; however, the schematic of the benchmarked filters together with the component values are shown in Section 5.3.1.

PLC Filter Insertion Loss and Power Consumption Measurements
Insertion loss measurements were performed in the following conditions:

•
The test setup used is the PLC filter online insertion loss test setup 2 shown in Figure 7; • The first step is to validate the test setup by directly connecting the PLC couplers to one another. The setup validity check is named "PLC Couplers" in the below figures; • The subsequent actions involve conducting measurements while the filter is installed, considering three different states: no mains voltage named "Unenergised", mains voltage named "Energised", and a 2000 W load with the mains voltage named "Energised with 2000 W load". The naming convention utilized in the figures below, which display the results of insertion loss measurements, remains consistent.  Figure 19 shows the schematic extracted from the datasheet of Filter 1. During measurements, the ground connection of the Y-capacitors between C2 and C3 was left floating and thus the filter was connected in the same way as an eighth-order passive filter or a two-stage filter from Figure 3b. Figure 20 presents the insertion loss of filter 1.   Figure 21 shows the reverse-engineered schematic of Filter 2 (the schematic was not available in the datasheet). The filter is a resonant type of filter with a resonance frequency  Figure 19 shows the schematic extracted from the datasheet of Filter 1. During measurements, the ground connection of the Y-capacitors between C2 and C3 was left floating and thus the filter was connected in the same way as an eighth-order passive filter or a two-stage filter from Figure 3b. Figure 20 presents the insertion loss of filter 1.  Figure 19 shows the schematic extracted from the datasheet of Filter 1. During measurements, the ground connection of the Y-capacitors between C2 and C3 was left floating and thus the filter was connected in the same way as an eighth-order passive filter or a two-stage filter from Figure 3b. Figure 20 presents the insertion loss of filter 1.   Figure 21 shows the reverse-engineered schematic of Filter 2 (the schematic was not available in the datasheet). The filter is a resonant type of filter with a resonance frequency between 20 kHz and 30 kHz, this is also visible in Figure 22 where the attention S21 [dB]  Figure 19 shows the schematic extracted from the datasheet of Filter 1. During measurements, the ground connection of the Y-capacitors between C2 and C3 was left floating and thus the filter was connected in the same way as an eighth-order passive filter or a two-stage filter from Figure 3b. Figure 20 presents the insertion loss of filter 1.   Figure 21 shows the reverse-engineered schematic of Filter 2 (the schematic was not available in the datasheet). The filter is a resonant type of filter with a resonance frequency between 20 kHz and 30 kHz, this is also visible in Figure 22 where the attention S21 [dB]  Figure 21 shows the reverse-engineered schematic of Filter 2 (the schematic was not available in the datasheet). The filter is a resonant type of filter with a resonance frequency between 20 kHz and 30 kHz, this is also visible in Figure 22 where the attention S21 [dB] has its minimum value. These types of filters have the disadvantage that the attenuation is highly dependent on the input and output resistance. Proof of this statement entails the huge difference in attenuation (10 dB to 40 dB) from Figure 22 between the attenuation when the filter is energized (no load) and the case when the filter has a 2000 W load. Figure 22 presents the insertion loss of filter 2. Figure 23 shows the schematic extracted from the datasheet of Filter 3 which is a sixth order passive filter. Figure 24 presents the insertion loss of filter 3.
Appl. Sci. 2023, 13, 6821 17 of 23 when the filter is energized (no load) and the case when the filter has a 2000 W load. Figure  22 presents the insertion loss of filter 2.   Figure 23 shows the schematic extracted from the datasheet of Filter 3 which is a sixth order passive filter. Figure 24 presents the insertion loss of filter 3.  when the filter is energized (no load) and the case when the filter has a 2000 W load. Figure  22 presents the insertion loss of filter 2.   Figure 23 shows the schematic extracted from the datasheet of Filter 3 which is a sixth order passive filter. Figure 24 presents the insertion loss of filter 3.  when the filter is energized (no load) and the case when the filter has a 2000 W load. Figure  22 presents the insertion loss of filter 2.   Figure 23 shows the schematic extracted from the datasheet of Filter 3 which is a sixth order passive filter. Figure 24 presents the insertion loss of filter 3.    Table 2 summarizes the results of the insertion loss measurements and power consumption measurements while comparing them to datasheet values. The rated current value is written in the table as a similarity reference between the filters.
The data from Table 2 are obtained from below methods and sources: • Power consumption measurements at 230V 50Hz; • Insertion loss extracted from measurement results is shown in Figures 18, 20, 22, and 24; • Insertion loss gap measurement using PLC signal filtering capability test setup version 1 is shown in Figure 11. These measurements reflect the gap between the filter attenuation and the attenuation at which the communication stops. The filter developed in this study exhibits exceptional features, including the best insertion loss versus frequency, a three-fold increase in attenuation at 50 kHz (−130 dB) in comparison to the best commercially available filter (−40 dB), and a power consumption of 0.6 W, similarly to the most power-efficient commercial filter (0.5 W). Furthermore, it has the second-best input and output impedance of 3.6 Ω within the frequency range of 35 kHz-95 kHz. Table 3 summarizes the PLC modem current consumption in transmit mode (sending frequency beacons in 35 kHz-95 kHz) while being connected in turns at the input and output of the filter. The filter impedance estimation method is shown in Figure 13d.

PLC Filter Impedance Estimation
The measured modem's power consumption is 0.42 A when transmitting over a resistive impedance of 1 Ω.  Table 2 summarizes the results of the insertion loss measurements and power consumption measurements while comparing them to datasheet values. The rated current value is written in the table as a similarity reference between the filters.
The data from Insertion loss gap measurement using PLC signal filtering capability test setup version 1 is shown in Figure 11. These measurements reflect the gap between the filter attenuation and the attenuation at which the communication stops. The filter developed in this study exhibits exceptional features, including the best insertion loss versus frequency, a three-fold increase in attenuation at 50 kHz (−130 dB) in comparison to the best commercially available filter (−40 dB), and a power consumption of 0.6 W, similarly to the most power-efficient commercial filter (0.5 W). Furthermore, it has the second-best input and output impedance of 3.6 Ω within the frequency range of 35-95 kHz. Table 3 summarizes the PLC modem current consumption in transmit mode (sending frequency beacons in 35-95 kHz) while being connected in turns at the input and output of the filter. The filter impedance estimation method is shown in Figure 13d.

PLC Filter Impedance Estimation
The measured modem's power consumption is 0.42 A when transmitting over a resistive impedance of 1 Ω.

Conclusions and Future Work
In conclusion, this paper has presented comprehensive benchmarking, design methods, and actual measurements for the development and evaluation of power line filters, with a specific focus on the CENELEC A communication band.
We have designed a power line filter using an S-parameters simulator and real passive components and detailed information regarding the component selection and filter construction has been provided. Practical validation of the simulated values was achieved through actual measurements. Benchmarking against three commercially available filters shows that our filter is the best suited for filtering below 150 kHz.
Our filter has superior characteristics, such as the best insertion loss versus frequency, three times higher attenuation at 50 kHz (−130 dB) compared to the best commercially available filter (−40 dB), and a power consumption of 0.6 W which is comparable to the most power-efficient commercial filter (0.5 W). Additionally, our filter has the second-best input and output impedance of 3.6 Ω within the frequency range of 35-95 kHz.
The proposed filter has the maximum rated current limit of only 30 A, which makes it suitable as a type 1 filter but unsuitable as a type 2 filter (separating two sections of the power grid) as defined per Section 2.1. It is of interest to design a type 2 filter using the design methods presented in this paper.
As future work, we propose to benchmark additional power line filter using the test methods presented in Sections 3 and 4. In addition, it is of interest for us to design and evaluate a hybrid between resonant and nth order filters with the scope of reducing the order of the filter while maintaining the attenuation below 150 kHz.