Transfer impedance simulation and measurement methods to analyse shielding behaviour of HV cables used in Electric-Vehicles and Hybrid-Electric-Vehicles

In the power drive system of the Electric Vehicles (EVs) and Hybrid Electric Vehicles (HEVs), High Voltage (HV) cables play a major role in evaluating the EMI of the whole system. Transfer impedance (ZT) is the most commonly used performance parameter for the HV cable. To analyse and design HV cables and connectors with better shielding effectiveness (SE), appropriate measurement and simulation methods are required. In this paper, Ground Plate Method (GPM) with improvements has been proposed to measure ZT. Use of low-frequency ferrites to avoid groundloop effects has also been investigated. Additionally, a combination of analytical model with a circuit model has been implemented to simulate limitations (frequency response) of the test setup. Also parametrical studies using the analytical model have been performed to analyse the shielding behaviour of HV cables.


Introduction
The Transfer impedance (Z T ) of a cable shield is considered as a benchmark of shielding performance, which has been used by communication cable industries for many decades now.It quantifies the immunity of a communication cable.In recent years, since the shielded cables are being used in EV and HEV, it has become important to measure the shielding performance against disturbances from the inner conductor(s).Established measurement methods (IEC-62153-4-1) such as Triaxial Method (IEC-62153-4-3) and Line Injection Method (LIM) (IEC-62153-4-6) are commonly used to measure Z T .Triaxial Method requires for different dimensions of the cable, to rebuild a large partition of the test structure.Line Injection Method (LIM) is easier to apply, but Z T measurement results are sometimes sensitive to different positions of the injection line, especially in case of non-symmetrical cables and connectors.To overcome some of these limitations in the measurements, Ground Plate Method (GPM) was developed to measure Z T of high voltage cables and cableconnector systems (Mushtaq et al., 2013).Although, it had the flexibility to measure various types of shielded cables and cable-connector systems with different sizes and lengths using same test-setup, still some points had to be improved e.g.wider measurable frequency range and simple method of making the termination connections at far-end were required.Usually in the standards (e.g.IEC-62153-4) for 1 m DUT the maximum measureable frequency of Z T is required to be up to 30 MHz, but to analyse Z T of HV-cables and connectors, it is important to reach a higher cut-off frequency.FM radio reception is very important in automotive EMC and connectors can add inductances which are dominant at higher frequencies.
In this paper, further improvements in the GPM have been proposed.To increase the measureable frequency range of Z T , use of matched terminations for the inner and outer circuits in the test setup has been suggested.Additionally, the GPM measurement precision has been increased, by using a simple method to create termination connections.Comparison of all three test-methods has been made and the results are discussed.
Apart from the measurement of Z T , analytical methods to simulate Z T of shielded cables using simple braid parameters are also very useful to predict shielding behaviour and improve the design of the HV cable and connectors systems.In this paper, an approach to build a simulation model based on cable geometry, cable materials, and transmission line theory has been proposed.Additionally, the simulation models have been used to show the effect of geometric variations, i.e. braid parameters (e.g.braid-wire thickness and weave angle on Z T ).The presented investigations are necessary for in depth electromagnetic analysis, improvements in the shield designs and for better understanding of the shielding behaviour of the HV cable-connector systems.
In this paper, Sect. 2 compares the existing measurement methods and gives details of the improved GPM, Sect. 3 describes the combination of analytical and circuit models, and Sect. 4 gives suggestions to improve Z T of the HV cable and optimize GPM measurement setup, followed by concluding remarks in Sect. 5.

Transfer impedance measurement methods
Due to the complex structure of the braided shield, analytical models are not sufficient to describe Z T (Qi et al., 2006).The most accurate method for determining Z T is the measurement of complex braided shields (Sali, 2004).Commonly used test methods to measure Z T as mentioned earlier are Triaxial Method and Line Injection Method (LIM) (ref.IEC-62153-4-1, 4-3, 4-6, 4-7 and 4-15).To use the Triaxial Method for measuring Z T of the DUTs with different sizes and shapes, large test structures have to be rebuild (i.e.variable diameters and shapes of cables and connectors require variation in the tube size or cell size), which makes it a bit complex.Whereas the LIM is comparatively simpler method to apply, but due to variable positioning of the injection line (parallel wires) especially in case of non-symmetrical DUT (cables and connectors), inaccuracy could be a problem.Martin and Mendenhall (1984) proposed to use an additional braid (i.e.milked on braid method) to make outer conductor of the outer coaxial system instead of using an outer tube (IEC 62153-4 Annex C).But it is not applicable to large and non-symmetrical cable-connector system as the construction would be difficult and not easy to repeat.Among other research articles, Korovkin et al. (2003) and Hofmeister et al. (2013), proposed slight variation or improvements in Triaxial method, which may be used for variable lengths of cable (DUT).But the main issue of complex structure and termination connections for variable size of cables and connectors remains the same as in the standard Triaxial Method (IEC-62153-4-1, -4-3, -4-6, -4-7 and -4-15).Among other alternative methods, Krauthäuser et al. (2005) proposed a method to determine Z T up to 1 GHz using a loop-method and adequate Green's function (analytical expressions) for the description of loop in the field inside the GTEM cell.This method can be used to measure Z T for braided shield cables and semi-rigid cables but application to large and nonsymmetrical automotive cable connector systems is difficult as Green's functions for each DUT needs to be defined in order to measure Z T .To measure Z T without the GTEM cell or Triaxial Tube for both HV cable and HV cable connector systems the Ground Plate Method, (GPM) was proposed (Mushtaq et al., 2013).For validation, improved GPM has been compared with Triaxial Method and Line Injection Method.The three methods have similar inner-circuits.As shown in Fig. 1, they differ mainly in the construction of the return path.Triaxial Method uses cylinder/tube (Fig. 1a), LIM uses parallel wires/injection line (Fig. 1b) and GPM uses groundplane (copper-plate) as return path (Fig. 1c).

Ground Plate Method (GPM)
The two-port measurement setup of GPM can be seen in Fig. 1c.The cable shield is fixed to metal brackets, which are connected to the ground plane (copper-plate).It is important that all connections have very low impedance.The source port is connected to the HV-cable on the left-side (referred as "Near-End") and the receiver port is connected to the HVcable on the right-side (referred as "Far-End").To have maximum possible measureable frequency i.e. cut-off frequency (f CUT-OFF ) for Z T measurements, matched-matched configuration has to be used (IEC 62153-4-3, Annex E).This means the terminations R 1F and R 2F in Fig. 1c, should be selected to match the characteristic impedance of the source circuit Z 01 and the receiver circuit Z 02 respectively (i.e.R 1F = Z 01 and R 2F = Z 02 ).For GPM, in case of a symmetrical shielded cable (and in-line connectors), Z 01 and Z 02 may be calculated using the analytical formulae (Tesche et al., 1997), i.e.Eqs. ( 1) and ( 2), respectively. where where where d 1 , d 2 and h 2 are the diameter of the inner conductor, diameter of shield, and height of shield above ground plane respectively.Z 01 and Z 02 of the inner and the outer circuits may also be measured using  (Bradley and Hare, 2009), an extra piece of braid from the same DUT is cut (3-4 cm) and used (as shown at the bottom of Fig. 2).Leads of the resistor (R 2F ) are cut to reduce the series inductance in the external loop.
Z T has been calculated from measured S-parameters using Eq. ( 3): where R 0 = 50 , is the port impedance of the VNA (Agilent E5061B).In order to get a higher frequency range in the measurements, DUT with shorter length may be used, here l shield = 50 cm has been used for all measurements.Whereas, for better sensitivity at lower frequency (as Z T = R DC shield ), longer cables may be used to limit the dominance of the connectors (Morriello et al., 1998).
To show the effect of using matched terminations, Triaxial Method is shown twice, i.e. (1) Triaxial Method TUDO with matched-matched configuration and (2) Triaxial Method from the company Bedea with matched-short configuration using equipment described in Halme and Mund (2013).
Figure 3 shows, at low frequency measured Z T is equal to the DC resistance of the shield (R DC shield ∼ = 3.6 m /m) up to f DC , i.e. the frequency at which the ratio of the thickness of the braid-shield ( ) and skin depth (δ) is much less than 1, i.e. /δ 1 (Vance, 1975).Above f DC there is a decrease in Z T up to minimum  optical coverage (Sali, 1991).Minimum Z T is achieved by reducing the difference between the hole (L HOLE ) and the braid (L BRAID ) inductances (Benson et al., 1992).After f MIN , Z T depends mainly on the braid inductances i.e.Z T ∼ = j ω(L HOLE ± L BRAID ) and Z T rises again with 20 dB decade −1 .Over the entire measured frequency range, the results from all three methods show qualitatively equal trend with little differences at higher frequencies except for Bedea Triaxial method (which has f CUT-OFF ∼ = 10 MHz due to matched-short configuration).With matched-matched configurations, measurement results for HV cables up to f CUT-OFF = ∼ 300 MHz are achievable, above this frequency resonances start to occur due to physical dimensions of the test setup.As mentioned by Breitenbach et al. (1998) and Hohloch et al. (2010), mismatches in the outer circuit result in resonances at the receiver port.To overcome these, it is recommended to design the mechanical dimensions of the outer circuit such that, Z 02 = R port (Breitenbach et al., 1998).With a simplified test setup maximum measurement is achieved up to f CUT-OFF =∼ 300 MHz.Alternatively, method proposed by Krauthäuser et al. ( 2005) can be used, to estimate Z T at higher frequency using the combination of field measurements in GTEM cell and green functions (analytical expressions).

Transfer impedance simulation methods
Due to complex structure of the braided shield, the measurements are the most reliable method of determining the Z T (Sali, 2004), but the analytical models of Z T are also useful for shielding analysis.Based on braid parameters, various analytical models have been developed for Z T , like Tyni (1976), Demoulin et al. (1981), Sali (1991), Kley (1993), and Beatric Model (Schippers et al., 2011), etc. to predict the shielding characteristics of coaxial cables.These models have slight variations and modifications, based on the construction of the shield and mathematical simplifications.Modelling of Z T may be divided into low and high frequency parts.At low frequencies, the mechanism for the linkage between the fields inside and those outside are due to diffusion of the magnetic currents induced in the shield and can be modelled as diffusion impedance (Z d ) (Vance, 1975;Sali, 1991;also shown in Eq. 5).With the increase in frequency, braid inductances become dominant.Based on the physical parameters of the shield hole and porpoising inductances can be calculated using analytical expressions (Eqs.6 and 7, respectively) given by Tyni (1976) and Demoulin et al. (1981).It has been found in previous investigations (Mushtaq et. al., 2013) that comparatively Demoulin model (Demoulin et al., 1981;Demoulin andKone, 2010, 2011), as per Eqs.( 4)-( 7), gives good approximation for HV shielded cable.
Porpoising inductance Analytical models show an ideal behavior for Z T i.e.Z T increases at 20 dB decade −1 as Z T = j ω • (L HOLE ± L BRAID ) at higher frequencies without any limits, whereas in practice at higher frequencies, due to mismatches and higher modes of propagations, resonance is measured at the receiving port (summarized in coupling equations IEC-62153-4-1 and 4-3).Z T may be estimated up to very high frequencies as proposed by Demoulin andKone (2010, 2011) using mathematical expressions to add the effect of multiple wave propagation into analytical expressions.In this paper, circuit model has been proposed to simulate the variation in Z T due to measurement setup at higher frequencies.

Circuit models to simulate measurement limitations
Transfer impedance analytical models are combined with circuit models.It simulates the variation in measured results due to measurement setup.The combined circuit model for shielded cable is implemented in the circuit simulation program QUCS (Quite Universal Circuit Simulator, http://qucs.sourceforge.net/).The circuit model shown in the Fig. 4, simulates the inner and outer circuits of the GPM test setup.For the inner circuit, the connecting cables and connectors have been added as lumped T-models.The value of inductances (L CC and L COAX ) and capacitances (C CC and C COAX ) for N-type-connectors and the coaxial cable are calculated using Eq. ( 1).For the external circuit the voltage at outer-shield is calculated using an analytical expression for Z T from Demoulin et al. (1981) and the inner circuit current (I SHIELD ), i.e.V T = I SHIELD •Z T .The value of the inductance (L SG ) and (2) for the shield over ground (Tesche et al., 1997).Additionally, errors in the measurement setup due to imperfect connections can also be modelled using proposed circuit model.
Comparison of ideal analytical model, combined circuit model from 10 kHz to 1 GHz and measured results from 10 kHz to 300 MHz are shown in Fig. 5. Above f CUT-OFF = 100 MHz, resonance starts to appear in the measured and combined circuit simulation model.The benefit of combined circuit simulation model is to identify the limiting factors and physical effects of the test setup in measuring the Z T .In future works, effects of HV connectors may be simulated using combined circuit model.

Analysis and optimization
In this section, use of analytical model and measurements to do shielding analysis for improving shielding designs are discussed.

Shielding design improvements using simulations
Based on the analytical model of Demoulin et al. (1981, i.e. Eqs. 4-7), effects of varying shield parameters (i.e.braid wire diameter (d) and weave angle (ψ)) on the shielding behaviour of the cables have been investigated individually.Braid thickness is kept small in order to keep the weight and material costs low, whereas weave angle is varied usually when optimizing the optical coverage of the shield.Both parameters are very important to get the optimized trade-off between cost and performance.To perform parametrical analysis, analytical model as per Eqs.( 4)-( 7), have been used for a shielded cable with D 0 = 11.4 • mm; n = 8; and N = 24 as constant and with variable braid-wire thickness and weave angle (as d = 0.1 to 0.2 mm and ψ = 20 to 40 • , respectively).
As shown in Fig. 6a, variation of the single braid wire thickness affects the low frequency region, i.e. the resistive part of Z T .With the increase in the thickness of braid-wire, Z T decreases linearly.At higher frequencies, as the inductances are dominant, variation in braid thickness affects Z T as per Eq. ( 6), (i.e.L HOLE ∝ d 2 ).Whereas variation in weave angle ψ, has a non-linear effect on Z T , because weave angle varies the resistive as well as the inductive properties of the shield, (Eqs.5-7).Variation in weave angle means difference in optical coverage of the shield and difference in length of individual braid wires used to build a shield.Usually, it is adjusted to give lowest dip in the Z T curve (Sali, 1991), as shown in Fig. 6b, minimum Z T is achieved when weave angle of 30 • is used (i.e.L HOLE − L BRAID ≈ 0),

Optimization of Z T measurement
As discussed in current standards (IEC 62153-4-3, Annex-F), while using VNA (or when both source and receiver have the ground at same point), presence of ground loops affect the measurements results especially at low frequency (f < 100 kHz).In GPM, to overcome these measurement errors while using VNA (Agilent E5061B), low-frequency ferrites (i.e.Epcos B64290L40X830; see http://de.tdk.eu/inf/Use of three windings through two ferrites lowers the starting measurable frequency down to 600 Hz.It was also observed that use of ferrites on either source or receiver side of the connecting cables (i.e.only on one side) is sufficient.

Conclusions
An alternative method, called Ground Plate Method (GPM), has been proposed with improvements for Transfer impedance measurements and is compared with Triaxial and Line Injection methods.Triaxial Method works well but with variable size of cables and connectors different sizes of measurement tubes or Triaxial cells are required.On contrary, Line Injection Method (LIM) gives a simple test setup but is not well suited for measuring Z T of non-symmetrical cables and connectors.GPM can overcome the existing limitations.It could be shown that the GPM provides good measurement results up to approximately 300 MHz.Afterwards setup resonances appear due to the outer circuit structure.In order to cover the important FM frequency range a cut-off frequency above 100 MHz was needed.With consequent usage of matching circuits for the terminations, a cut-off frequency of close to 300 MHz could be reached.
For low-frequency measurements the influence of groundloops could be reduced by using low-frequency ferrites.Apart from measurements an analytical Z T model has been applied and was combined with a circuit model.It predicts the measurement setup limitations.Furthermore it has been shown, that the analytical model may be used for improving the shielding designs.
For future works and integration into existing standard like e.g.IEC 62153-4, additional steps to characterize and finalize physical dimensions of the test setup and measurement of different types of DUTs are required.
Edited by: F. Gronwald Reviewed by: R. Rambousky and two anonymous referees

Figure 1 .
Figure 1.Comparison of test setups for (a) Triaxial Method, (b) Line Injection Method and (c) Ground Plate Method (GPM).

Figure 2 .
Figure 2. GPM: Improved way of making Far-end connection.

Figure 3 .
Figure 3.Comparison of Z T measurement methods for HV-cable.

Figure 4 .
Figure 4. Combined analytical and circuit model for a shielded cable.

Figure 5 .
Figure 5. Verification of shielded cable model.

Figure 7 .
Figure 7. Improvements in the low frequency measurements using low-frequency ferrites.
After determining the required values of Z 01 and Z 02 , the termination loads R 1F and R 2F are connected physically as shown in Fig.2.First R 1F is realised by connecting the Surface Mounted Device (SMD) resistors in parallel as they have less inductance.Individual value of the SMD resistor is: R SMD = m • Z 01 ; where m is the number of SMD resistors used in parallel.Five SMD resistors have been used to realize R 1F in GPM measurements as shown at the top-left corner on Fig.2.After soldering and isolating the R 1F resistor, to make a low impedance contact VNA (Agilent E5061B, available at: http://cp.literature.agilent.com/litweb/pdf/5965-7917E.pdf)for Open/Short S-parameters measurements or TDR (Reflection measurements) (ref.IEC-62153-4-3, Annex A).