Communication Quality Analysis for WPT and NFC Systems via Broadband Equivalent Circuit Models

This letter presents a novel full-system communication quality analysis of a wireless power transfer (WPT) and near-field communication (NFC) system. The analysis is conducted via a thoroughly verified broadband equivalent circuit (EC) model of a WPT and NFC system prototype. Such an EC model can accelerate the iterative system design process for systems with interoperating or coexisting WPT and NFC interfaces. The EC model is verified by measurements of a prototype. EC model and measurements agree excellently by comparing impedance parameters, single-sided frequency spectra, and the signal constellation diagram of the NFC tag-to-reader communication signal. The NFC communication quality analysis shows a severe impairment of the NFC tag-to-reader communication by the WPT system.


I. INTRODUCTION
H ANDHELD devices are being equipped with more wire- less features, which are forced to coexist and interoperate nearby each other, e.g., wireless charging and mobile payments in smartphones or smartwatches [1], [2].For wireless charging, handheld devices typically use specific wireless power transfer (WPT) standards, e.g., the Qi standard [3], which transmits energy via magnetic resonant coupled devices at roughly 100 kHz.Mobile payments are typically realized via near-field communication (NFC), which transmits data via magnetic resonant coupled devices at 13.56 MHz [4].WPT and NFC systems integrated into chargers (transmitter, TX) and handheld devices (receiver, RX) might impair each other's functionalities when coexisting or interoperating [5].
Recent work on WPT and NFC systems coexistence and interoperability is given in Table I.Researchers have been investigating coexisting WPT and NFC systems via inductive decoupling Manuscript received 1 August 2023; accepted 27 August 2023.Date of publication 11 September 2023; date of current version 5 January 2024.This work was supported by the "University SAL Labs" initiative of Silicon Austria Labs (SAL) and its Austrian partner universities for applied fundamental research for electronic-based systems.(Corresponding author: Richard Fischbacher.) Richard Fischbacher, Jose Romero Lopera, and David Pommerenke are with the Graz University of Technology, 8010 Graz, Austria, and also with the Silicon Austria Labs, 8010 Graz, Austria (e-mail: richard.fischbacher@tugraz.at; romerolopera@tugraz.at; david.pommerenke@tugraz.at).
Digital Object Identifier 10.1109/LAWP.2023.3314198using novel coil shapes [6] and coil positioning [7] to reduce the inductive coupling between WPT and NFC coils.Further work used frequency separation techniques to enable interoperating WPT and communication systems.Here, researchers realized a WiFi communication link for WPT [8], [9], and NFC at higher frequencies [10].In this letter, we investigate the coexistence of WPT and NFC systems.In contrast to previous work, we present a solution that allows us to analyze the NFC system communication quality directly to improve the coexistence and interoperability of WPT and NFC systems.In previous work [6], [7], [8], [9], [10], researchers used electromagnetic (EM) simulation, measurements, and narrowband equivalent circuit (EC) models [5] to investigate the abovementioned solutions directly.Contrary to this, we use EM simulations and measurements to generate and verify a novel broadband EC model of a WPT and NFC systems prototype.This broadband EC model allows the investigation of high-order WPT harmonics, contrary to narrowband EC models.In our previous work [11], we presented the basics of physical broadband EC models for two coupled NFC coils and a behavioral EC model for a WPT coil [12].In this letter, we further expand on this research and combine physical and behavioral coil EC models to generate a WPT and NFC systems prototype EC model.Further, we use this broadband EC model to conduct the NFC system communication quality analysis.The use of an EC model has all advantages of circuit-level simulation tools [13] and can be used for a fast and efficient iterative system design process.

II. BROADBAND EC MODEL
Fig. 1(a) shows the WPT and NFC system prototype used in this work, consisting of a WPT TX, WPT RX, NFC reader, and NFC tag.The WPT system coexists and interferes with an active NFC system.The TX and RX coils are aligned in a coaxial and coplanar manner with the coil distances being d 1 = 15 mm and d 2 = 3 mm.The four subsystems are realized with a coil printed circuit board (PCB) [see Fig. 1(b)] and its respective tuning circuit PCB [see Figs.1(a) and 3].The WPT TX and RX coils were provided by Würth Elektronik [14] and the tuning circuits were custom-built.Both are in accordance with the Qi standard [3].The NFC reader and tag coils were custom-built (NFC class 5 form factor [15]) and the tuning circuits are based on state-of-the-art designs [16].
The EC models of the coupled coils are based on our previous work and are shown in Fig. 2 [11].The chosen frequency range of interest is 0.1-60 MHz.We advanced a standard coil EC model (L n,r/t,i , R n,r/t,i , R p,n,r/t,1 , and C p,n,r/t,1 ) [11], [17] for the coupled NFC coils while spreading the coupling capacitances C c over the inductance and DC resistance [11].The WPT coil was modeled using a behavioral modeling approach [12] with parallel resonant circuits (L w,rx/tx,i , R w,rx/tx,i , R p,w,rx/tx,i , and C p,w,rx/tx,i ) to align the EC model behavior with the measured prototype behavior [11].In this specific case, the chosen frequency range and used WPT coils dictate the use of three resonant circuits for the WPT coil EC models [11], which in turn requires the NFC coil inductance to be split into three parts (i = 1, 2, 3, Fig. 2).The coils interact with one another by inductive and capacitive coupling, which are implemented in the EC model using coupling coefficients k and coupling capacitances C c .

TABLE I RELATED WORK OF WPT COEXISTENCE AND INTEROPERABILITY INVESTIGATIONS
The tuning circuits of the coupled coils are shown in Fig. 3.The WPT TX and NFC reader circuits are either connected to a vector network analyzer (VNA, R&S ZVL) or a 50 Ω arbitrary waveform generator (AWG, Keysight 33600 A) for system analysis.The WPT tuning circuits are discussed in [5].The NFC reader tuning circuit topology is given in [16].It uses an inductance (modeled via a coil standard EC model R n,r,f , C n,r,p,f , R n,r,p,f , and L n,r,f ) and the capacitance C n,r,f as a low-pass filter, the capacitances C n,r,1/2 to set the series resonance frequency of the NFC reader to 13.56 MHz (while only being coupled with the NFC tag), and the resistance R n,r to adjust the quality factor.Usually, one would design a parallel resonance, but it is not required for EC model verification purposes.In addition, we added the resistance R n,a to adjust the input impedance at port NFC (see Fig. 3) at 13.56 MHz, to align the quality factor between measurement and simulation.The NFC tag tuning circuit uses the resistance R n,t to set the quality factor, the capacitance C n,t as combined tag chip and tuning capacitance to set the resonance frequency of the NFC tag to 13.56 MHz, and the resistance R n,chip to model the tag chip resistance [18].The load modulation switch (LM Switch) is realized via two transistors [19], [20].

III. EC MODEL VERIFICATION
Vector network analyzer measurements are required to determine scattering (S) and impedance (Z) parameters of the single coils and prototype [11].Each coil was measured with no other coils in close proximity to determine the lumped elements for the single-coil EC models.These measurement results need to be adjusted for the prototype coil arrangement shown in Fig. 1(a), due to the WPT ferrites.The inductance of each coil has to be re-evaluated by conducting one port measurements of each coil while being assembled in the prototype coil arrangement (without tuning circuits).The DC resistances were not adjusted.The introduced loss of the ferrite and other material might be the cause why R n,a is required to adjust the magnitude at 13.56 MHz.
We determine the coupling coefficients between the coils via a two-port measurement [11], where each port of the VNA is connected to a coil.The resulting transmission impedance parameter Z 12 between two coils is used to calculate the mutual inductance M = Im{Z 12 }/ω and the coupling coefficient where ω is the angular frequency and L 1/2 are the coil inductances.It is important to conduct these measurements within the prototype setup (without tuning circuits) while the not measured coils are terminated with an open circuit.The coupling capacitances C c are determined by connecting each Balun output to a coil, forming a capacitance via the two coils, represented by the imaginary part of its Z-parameter (Z 11 ).The capacitance is then calculated via C c = −1/(ω • Im{Z 11 }) [11].In contrast to previous work [11] due to the coupling of behavioral and physical EC models, we did not split the inductances L n,r/t,i , resistances R n,r/t,i , and coupling capacitances C c,1..6,i evenly.We divide them according to the inductance ratios of the WPT coil inductances.The values are given in the Fig. 2 description, leading to the overall best EC model and prototype impedance parameter agreement.Fig. 4 shows the impedance parameter magnitude of the prototype with port 1 being connected to port WPT and port 2 being connected to port NFC (see Fig.     [21], and its S-parameters were extracted and applied to the same tuning circuits as used in the EC model.We used annealed copper for metals, FR-4 for PCBs, and modeled the ferrite according to its datasheet.The EC model and measurement results agree very well between 1 and 60 MHz.|Z 12 | M and |Z 22 | M show measurement inaccuracies below 1 MHz due to dynamic range issues caused by the measurement devices.The upper-frequency limit of 60 MHz could be raised by introducing more WPT coil parallel resonant circuits (see Fig. 2).We chose three as a compromise between bandwidth and EC model complexity.The EM simulation agrees with the EC models below 1 MHz and shows good agreement with measurements up to 20 MHz, but at higher frequencies uncertainties are caused by material parameters [22] and simplifications of the 3-D model.

IV. COMMUNICATION QUALITY ANALYSIS
Further, the EC model is verified by comparing the NFC tag-to-reader communication to evaluate how well the EC model can be used to conduct NFC communication quality analysis.
Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.Fig. 5. Single-sided frequency spectrum of V com (see Fig. 3): The different components of the V com spectrum (lower sideband, upper sideband, WPT harmonics, noise, and carrier) are illustrated.Fig. 6.WPT signal at NFC reader: The signal V com is shown while only the WPT source is active (i.e., NFC reader source is not active and replaced by a 50 Ω resistance).The right-hand figure has a zoomed-in x-axis to show how well the measurement and EC model waveforms agree.
Here, we supply square waves V w,pp = 6 V with 105.9375 kHz to port WPT and V n,pp = 4 V with 13.56 MHz to port NFC by two 50 Ω sources of the AWG via a coaxial cable and Balun [23].In addition, the LM switch is actuated with 2 • 847.5 kHz to carry out LM with 847.5 kHz [15], [24].The NFC tag-to-reader communication signal V com is usually determined, as shown in Fig. 3.We determine this signal in measurement via an oscilloscope (R&S RTO 2044) and a differential probe (R&S ZD-30) and in circuit-level simulation using the previously described EC model.This EC model includes S-parameter descriptions of the cables and Baluns to include their behavior in the simulations.
The single-sided frequency spectrum of the measured (M) and simulated (EC) V com time signal is calculated via a fast Fourier transform (FFT) and shown in Fig. 5.The chosen FFT boundary conditions lead to frequency bins at integer multiples of the WPT square wave frequency.The figure shows the NFC carrier, odd upper and lower sideband harmonics, and WPT signal versus frequency.The upper and lower sidebands are NFC tag-to-reader communication signals caused by LM.These show excellent agreement between measurement and simulation for the lower sideband and slight errors at the upper sidebands.Since the WPT harmonics are difficult to compare while the NFC source is active (spectral leakage and limited oscilloscope resolution), we show measured and simulated V com in Fig. 6, with only the WPT source being active.The figure shows excellent agreement between measurement and simulation, verifying that the EC model accurately describes the WPT TX and NFC reader interaction.
In addition, we conduct ideal in-phase/quadrature-phase (I/Q) demodulation of the communication signal V com [25].The corresponding signal constellation diagram is shown in Fig. 7, showing mean values and the accompanying error bars for the logical states "0" and "1." Measurement and simulation show excellent agreement (EC and M, see Fig. 7).Realistic WPT and NFC systems would use different source impedances and might have to use different tuning setups when tuning the NFC reader with a strongly coupled (SC) tag in mind.
Further, we conducted a WPT and NFC systems coexistence analysis for a SC (3 mm NFC coil distance, k 4,i = 0.455, and C c,4,i = [6.14, 0.045, 0.185] pF for i = 1, 2, 3) and weakly coupled (WC, 15 mm NFC coil distance, k 4,i = 0.098, and C c,4,i = [0.84,0.006, 0.025] pF for i = 1, 2, 3) NFC coils, while keeping all other coupling interactions constant (see Fig. 2).As a first step, we adjusted the NFC reader tuning circuit via L n,r,f = 107 nH, C n,r,f = 590 pF, C n,r1 = 820 pF, and C n,r2 being replaced by the inductance L n,r2 = 350 nH, resulting in an NFC tuning setup with a parallel resonance of the NFC reader at 15 MHz, while being SC to the NFC tag.In addition, we adjusted the inner resistance of the WPT source (0.8 Ω) and the NFC source (5 Ω).The SC results show a significant distance between the logical "0" and "1" states (see Fig. 7).Detuning the system by applying a WC interaction between NFC coils leads to a significant overlap of logical states, indicating possible issues regarding the NFC communication quality.The main reason is the signal created by the WPT source during switching operations (see Fig. 6).

V. CONCLUSION
We generated a broadband EC model to investigate the coexistence of WPT and NFC systems.VNA measurements verify the EC model, showing an excellent agreement of the impedance parameters in the frequency range from 1 to 60 MHz.NFC tag-to-reader communication shows an excellent agreement for measurement and EC model.Further, we investigated how NFC tag-to-reader communication could be impaired by WPT via a signal constellation diagram.In addition, the misalignment of RX and TX has to be investigated.Coil misalignment leads to a change in the coupling coefficient [26], which can be easily modeled by the presented solution [4].However, misalignments will also lead to a change in capacitive coupling between the close-by coils.Future work has to focus on calculating capacitive coupling between the coils for a specific misalignment, which can then be modeled in the presented solution.

Fig. 1 .
Fig. 1. and NFC prototype setup including coils and tuning circuits: The prototype is shown in (a) and the separate coils are shown in (b)[11].

Fig. 3 .
Fig. 3. Tuning circuits EC models: It shows the actual schematics except for R n,a and the filter inductance EC model.The + and − nodes indicate the nodal connections with the tuning circuits shown in Fig. 2.

Fig. 4 .
Fig. 4. Magnitude of the impedance parameters Z 11 , Z 12 , and Z 22 of the prototype: Measurement results show mean values of 100 measurement runs, including minimum and maximum values with error bars.

TABLE II COIL
LUMPED ELEMENT VALUES FOR THE EC MODEL SHOWN IN FIG. 2