Investigation into the Errors Introduced by Performing EN 61000-4-3 Test Set-up Verification Using Multiple E Field Probes

The electromagnetic compatibility (EMC) test standard EN 61000-4-3 [1] requires the equipment under test (EUT) to be illuminated by an E field having a uniform intensity across the plane occupied by the forward-facing side of the EUT. To ensure the E field is sufficiently uniform, the test standard describes a verification method using an electric field probe to measure the field intensity across the test plane at sixteen (extending to twenty for floor-mounted EUT) points on a 0.5m x 0.5m grid. Should the field intensity across these points fall outside the prescribed limit (at least 75% must be within -0/+6dB of the target level) then action must be taken to improve the field uniformity.

The test is performed in an absorber-lined screened room acting as a free-space environment (Figure 1) and, in practice, modifications to obtain field uniformity may involve simple strategies such as simply adjusting the field transmit power or the positioning of the transmit antenna and EUT. However, any remaining reflections within the interior of the room can establish interference patterns that lead to significant localized deviations in field intensity. Improvements may therefore extend to changing the characteristics of the room, for example by altering the type or arrangement of absorber material lining. This kind of work may be needed at regular intervals, where the room infrastructure and absorber are subjected to the daily wear and tear of a busy laboratory. Because the location and magnitude of any peaks and nulls in the interference pattern are affected by spatial position and source signal wavelength, adjusting the test environment to fix one problem spot may introduce a new problem at a different location/frequency combination. The verification process must therefore be repeated following any changes, meaning a fresh frequency sweep for each of the sixteen (or twenty) test points. Using a single probe, the verification process described in the test standard can take hours for a “good” test environment, while a room with “marginal” characteristics in need of adjustment can tie up time and resources with days of painstaking work. Furthermore, the test standard states that verification should take place before the facility is used to test an EUT, which in practice can be prohibitively difficult given the time constraints.

Figure 1: Field uniformity verification points for EN 61000-4-3 radiated immunity test setup.

Being able to measure all test points simultaneously would significantly reduce the time needed to perform the verification. However, leaving aside the cost of owning several probes, the problem with this approach is that each probe distorts the E field to some degree. The degree of field distortion caused by a hypothetical multiple E field probe array has been investigated to determine whether the resulting measurement error is small enough to make this a viable means of addressing the requirements of EN 61000-4-3. If the error lies within the order of a typical E field probe measurement accuracy of around 1dB [2 – 4], then this would be considered acceptable for pre-test verification of the test setup.

Sources of Measurement Error

The presence of a conducting body, such as the metallic field probe, causes distortion of the E field by scattering the incident electromagnetic wave. Depending on the surface properties of the body, such scattering may be specular (e.g. from a mirrored surface) or random (e.g. diffuse reflection from a rough surface). The re-radiated fields combine with the source field to establish interference patterns, leading to field intensity peaks and nulls whose spatial position and magnitude vary depending on the wavelength of the incident wave and the relative positions of any scattering surfaces.

Further distortion is caused where the body absorbs energy from the incident wave and then, by some mechanism, causes a new field to be radiated. For a body such as an antenna, currents induced by the incident field will themselves radiate fields, the pattern and orientation of which will be dependent on the “shape” of the surface currents.

These effects are iterative, becoming increasingly complex with the number of bodies within the field. A flowchart indicating the interaction between two such bodies has been presented previously (Figure 2) [5]. To investigate the properties of field probes, the flowchart has been expanded to include the case where the bodies contain antenna elements. The antenna elements are considered to have scattering, absorption and re-radiation properties that can be analyzed separately from those of the body itself, relating as they do to currents flowing in the antenna due to their physical structure and the connected load circuitry.

Figure 2: Flowchart for analyzing two-body scattering proposed by Elsherbeni and Harmid, expanded to include the case where bodies contain current-carrying antenna structures.

Scattering Experiment Method

Comparing the measurement of the E field intensity in a typical radiated immunity (RI) test setup (Figure 3) using a single probe, with measurements made in the presence of nearby scattering objects (Table 1), shows the error introduced by field scattering. The objects were chosen to give a spread of sizes around that of the example 1GHz field probe, being roughly a 50mm metal cube. The measurement using 100mm cube scattering bodies was repeated with them covered in ferrite tile material to alter their reflective properties. The tests were performed inside a fully anechoic chamber compliant with the test standard and, to minimize their effect on the E field, low εr polystyrene stands were used to mount the source E field generator, probe and test objects.

Figure 3: Test setup for evaluating the effect of nearby objects on E field measurements with the fixed distance between the field generator and test point probe position being d = 3m and the variable distance between probe and neighbors being s m.

Table 1: Scattering bodies and their separation distances from the probe.

The E field was generated by a 30MHz – 2GHz broadband noise source fitted with a monopole antenna [6], so that sweeping the source signal was unnecessary. Measurements were taken at a point d = 3m away using a dipole antenna fitted with two 20mm long elements.

Adequate measurements were achieved using this setup to 2GHz, being limited by the upper frequency limit of the dipole probe. Further measurements to 6GHz are planned.

Experimental Results

Figures 4 to 6 show the deviation introduced by the scattering bodies from the measurement made with the probe in isolation. In this case the source and test probe antennas are vertically polarized (i.e. the scattering bodies broadside to the field probe antenna), and the results are presented with increasing separation between the test probe and the scattering bodies. Where the source and test probes are horizontally polarized (i.e. the scattering bodies are end-on the field probe antenna) a lesser degree of distortion is noted, providing the distance between the probe antenna and scattering body does not become too small (<= 0.3m as measured). Given that E field probes typically employ electrically short antennas of only a few millimetres in length, this should not be a problem using the EN 61000-4-3 verification grid of 0.5m.

Figure 4: Deviation of the E field strength due to the presence of nearby test objects, where the separation between probe and objects s = 0.3m. Traces shown are: 100mm3 cube (cyan), 120mm3 cube with ferrite (blue), 50mm3 cube (green), 15mm dia. × 100mm (red).

Figure 5: Deviation of the E field strength due to the presence of nearby test objects, where the separation between probe and objects s = 0.5m.

Figure 6: Deviation of the E field strength due to the presence of nearby test objects, where the separation between probe and objects s = 1m.

The results support the intuitive notion that a larger body leads to a greater disturbance and the further it is from the test probe the weaker the disturbance. In addition, it is clear that the separation affects the periodicity of the constructive/destructive interference pattern at the measurement point, i.e. the closer the scattering objects, the shorter the effective disturbance wavelength and hence a larger periodicity between maxima/minima in the frequency domain.

A simplified example of the interference causing this error is shown in Figure 7, which considers the test setup using vertically polarized antennas as an approximation of the multipath propagation between two horizontally polarized dipoles over a reflecting ground plane. The only ray paths considered are ones with reflections from the corner of the adjacent object (Figure 7). The phase inversion associated with ground-plane reflections of horizontally polarized waves leads to maxima (where the waves at the observation point are in-phase) when the difference between path lengths is a half-integer number of wavelengths (n_max). Similarly, minima (where the waves at the observation point are in anti-phase) occur when the path length difference equals an integer multiple of wavelengths (n_min).

Figure 7: Calculations for predicted frequencies of maxima/minima in received signal due to multipath propagation (due to reflections from the leading edge only). The distance between the field generator and test point probe position is d = 3 m.

Equation 1a

Equation 1b

In reality, the effect will be the sum of all the rays reflecting from points on the body’s surfaces. A fully descriptive closed-form analytical solution is therefore difficult to achieve, although formulae for simplified models involving pairs of scattering bodies with basic geometries have been presented [7, 8].

These results indicate that, up to a frequency of 2GHz, a disturbance of < 1dB could be achievable using probes 50mm x 50mm x 50mm in size, or smaller, on the 0.5m spaced grid defined by EN 61000-4-3. The selection of field probes for this kind of application should therefore favor a small footprint facing the source, with a shape designed to scatter any reflections away from the neighboring probes.

Antenna Interaction Modeling Method

The interaction between antenna elements is also iterative, with the net effect being the sum of transmitted fields from all parts of the antennas involved. A closed-form analytical solution is once again therefore difficult to achieve, although formulae for simplified models involving pairs of half-wave dipoles have been proposed [9]. Because of this, numerical modeling has been used to investigate the contribution to measurement error made by the antenna elements of a typical field probe.

One way of describing the interaction between antennas is by their mutual impedance characteristic. Consider two antennas in the presence of a common incident electromagnetic field described by two coupled equivalent circuits (Figure 8). For Antenna 2, the voltage induced by the common incident field (V_E2) causes current I₂ to flow through load impedance Z₂. This current flowing in Antenna 2 generates its own radiating field, which couples with Antenna 1 in addition to the incident field, inducing a voltage that is proportional to the current I₂. This secondarily induced voltage in Antenna 1 can be defined as Z₁₂I₂, where Z₁₂ is the term of mutual impedance.

By examination of Figure 8, it can be seen that increasing load resistances Z_L1, Z_L2 will reduce I₁ and I₂ respectively. This will reduce voltages Z₁₂I₂ and Z₂₁I₁ and hence the effect of mutual impedances Z₁₂, Z₂₁, thereby lessening the error in the intended measurement of voltages V_E1 and V_E2. For the field probes being investigated, the antenna loads are essentially Schottky detector diodes whose sensitivity and impedance characteristics are bias-current tuned. Since the impedance of the detector circuit within a given field probe is unknown to the user, a range of values needs to be investigated in order to find out whether this characteristic contributes significantly to the measurement error in the probe array.

Figure 8: Equivalent circuit of mutually coupled antennas in a common incident E field. Where: V_E1 and V_E2 are the voltages induced by an incident wave on antenna 1 and 2 respectively. Z₁ and Z₂ are the impedances of Antennas 1 and 2 respectively. Z_L1 and Z_L2 are the load impedances. Z₁₂I₂ and Z₂₁I₁ are the voltages induced by mutual coupling.

The interaction between the probe antennas was modeled using the widely available NEC-2 software. This uses the Method of Moments (MOM) technique, which is particularly suited to problems involving elementary wires, currents and fields in homogenous environments encountered by the interaction between antennas in a free-space environment. Again, the response to an incident E field of a single probe antenna in isolation is compared with that obtained when the probe is the central object in the presence of the other nearby probe’s antennas on the 0.5m test grid array (Figure 9).

Figure 9: Model arrangement of source, receive and array dipoles. NEC E field results are presented later for a plane cutting this figure centrally containing the source and central probe dipole.

The field probes are modeled as electric dipoles, comprising perfectly conducting rods with Shottky detector diode circuit equivalents at the central load points. This method has previously been used successfully to analytically and numerically study and predict the operation of electric dipole field probes [10]. Realistic values for dipole length and load are used to determine the error introduced in an actual antenna/detector pair.

Simplified equivalents of the diode detector loads have been implemented in the NEC-2 model as single-segment parallel R/C networks (Figure 10) with variable values of resistance Rj.

Figure 10: Simplified diode model and the equivalent load circuit consisting of a parallel R/C circuit.

For a typical Schottky detector diode [11], the values given are R_s = 6Ω, Cj = 0.18pF and R_j = 8.33 x 10^-5 nT / (I_b + I_s), where n = 1.08, T = temperature (K), Is = 5×10^-8A (saturation current) and I_b = bias current. From the characteristics given, a bias current of 1μA gives R_j = 26kΩ, zero-bias gives R_j = 540kΩ and 1mA bias gives R_j ≈ 27Ω. Simulations were performed using R_j values of 50Ω and 250kΩ.

The NEC-2 model uses a current-driven dipole at a distance d = 3m from the probe array to generate the incident E field (Figure 9). Comparing the voltages generated across the central probe dipole’s load in the presence of this E field, in isolation and when surrounded by eight identical antennas in a grid array, gives the error due to antenna coupling.

Modeling Results

An E field plot for the NEC-2 numerically generated results is shown in Figure 11. This shows a section defined by the plane containing the source and the three probe dipoles comprising the middle column of the array (Figure 8). The source antenna is visible at the top of the plot, with the cross-section through the generated E field clearly showing the characteristic toroidal pattern of intensity for a radiating electric. The individual antennas comprising the receive array are visible at the bottom of the plot, with all dipoles aligned along the y axis. Those antennas in the plane of the plot clearly show interaction with the E field as a localized disturbance in the field intensity.

Figure 11: Plot of E field across the central plane of the dipole antenna array containing the source and central probe dipoles (Figure 8 for probe array geometry).

The responses shown are of the voltage produced by the detector, normalized to the incident E field intensity generated by the source dipole. Figure 12 shows the result using a 40mm dipole probe (i.e. a pair of 20mm elements) with a 250kΩ diode detector, in isolation and in the presence of the probe array.

Figure 12: Normalized transfer function against frequency of the 40mm (two 20mm elements) dipole, with a 250kΩ detector diode model.

In Figure 13 the difference between the single probe and array responses for load resistance values of 50Ω and 250kΩ are plotted, and thus gives the array-measurement error value. This indicates that the array introduces a maximum error approaching +/-0.8dB around a frequency region close to the 40mm dipole’s half-wave resonance (assuming effective dipole length is 0.9 of the actual length, λ/2 = 4.17GHz). From this, it can be seen that increasing detector load resistance does not significantly reduce the array error.

Figure 13: Deviation versus frequency due to the presence of the array of 40mm (two 20mm element) dipoles, with 50Ω and 250k Ω detector diode models.

Field probe designs often employ electrically short antennas, which trade a flatter frequency response against lower sensitivity. At 6GHz, a 5mm antenna is still less than λ/10, thereby satisfying the criteria for an electrically short antenna. The simulation was repeated using a 10mm dipole (two 5mm elements) in place of the 40mm dipole.

Figure 14 shows the expanded difference between the single probe and array responses for both 50Ω and 250kΩ detector diode models, giving the array-measurement error. This indicates that the array now introduces a maximum error approaching +/-0.008dB, a significant improvement over the longer dipole model.

Figure 14: Deviation versus frequency due to the presence of the array, for 10mm (two 5mm element) dipoles, with 50Ω and 250k Ω diode models.

This improvement indicates a reduced degree of mutual coupling, as the shorter 10mm dipoles are less efficient receivers and transmitters than the 40mm dipoles, which are operating around their λ/2 point. The mutual coupling between them is greatly reduced and consequently the error introduced is also reduced.

The diode load impedance again does not appear to have much effect. This is because the 10mm dipoles remain electrically short across the frequency range examined and so exhibit very low antenna resistance R_r. As a result, increasing R_L beyond a few 10’s of ohms does not significantly reduce the error due to the mutual impedance between the array antennas. The probe manufacturer’s configuration of the detector circuit does not therefore appear to be significant in choosing probes to be used in the array. However the type of antenna used might be more significant, favoring those that remain electrically short across the full working frequency range.

Conclusions

This study has shown that it is possible to use an array of field probes to measure E field intensity at many points simultaneously, thereby significantly reducing the measurement time needed, whilst keeping the naturally resulting errors due to probe antenna coupling within acceptable levels for pre-test verification checks. With regards to EN 61000-4-3 test setup verification, careful selection of the probe size and antenna configuration can keep the errors to an order similar to the inherent accuracy of a typical commercially available probe.

Given the cost involved in such an array, and that the fully populated array of sixteen probes at 0.5m spacing would produce the largest errors due to scattering, a partial solution involving fewer probes could be an attractive compromise. For example, rotating four probes around the grid so as to keep 1m clearance between them would still significantly reduce the test time compared with a single probe, whilst lowering the error factor and implementation cost compared with the full sixteen probe solution.

References

1. BS EN 61000-4-3:2006 +A1:2008, Electromagnetic Compatibility (EMC) – Part 4-3: Testing and Measurement Techniques – Radiated, Radio Frequency, Electromagnetic Field Immunity Test. Includes Appendix I, concerning the calibration and performance of E-field probes used.
2. Holaday 10kHz-1GHz battery powered isotropic field probe, 64x64x64mm cube plus antenna radomes. www.ets-lindgren.com/manuals/HI-6053.pdf
3. ETS-Lindgren HI 6005 isotropic E field probe Freq 100kHz to 6GHz. www.ets-lindgren.com/page/?i=HI-6005
4. Amplifier Research FL7006/kit 100kHz-6GHz isotropic E field probe. www.ar-worldwide.com/html/18200.asp?id=662
5. A.Z. Elsherbeni and M. Hamid, “Novel Cylindrical-wave Spectrum for Analysis of Scattering by Multiple Bodies”, IEE Proceedings, Vol 134, Pt H, No 1, February 1987 pp. 35-44.
6. CNEV+,9kHz to 3.5GHz York EMC Services Ltd battery-powered broadband radiating reference noise source. www.yorkemc.co.uk/instrumentation/cnev+/
7. O.E. Allen, “Modeling General Purpose Antennas as Minimum-Scattering Antennas”, IEEE 0-7803-8883-6/05, 2005,l pp. 47-50.
8. H.T. Hui, “A New Definition of Mutual Impedance for Application in Dipole Receiving Antenna Arrays”, IEEE Antennas and Wireless Propagation Letters, Vol. 3, 2004, pp. 364-367.
9. A. Kazemipour and X. Begaud,“A Simple Closed-form Formulafor the Mutual Impedance of Dipoles”, Microwave and Optical Technology Letters, Vol. 34, No 5, September 2002.
10. M. Kanda and L.D. Driver, “An Isotropic Electric-Field Probe with Tapered Resistive Dipoles forBroad-Band Use, 100kHz to 18GHz”, IEEE Transactions on Microwave Theory and Techniques, Vol. MTT-35, No 2, February 1987.
11. Agilent HSMS-286x datasheet. www.avagotech.com/docs/AV02-1388EN

This article has not been reviewed by the ICM Editorial Advisory Board.

	Dave Cullen, MIET, MIEEE Dave Cullen is the Test Instrumentation Manager of York EMC Services Ltd. and holds BEng (Hons) and MSc degrees in electronic engineering and product development respectively. He has 30 years experience in the manufacture and design of electronic test equipment, focussing with York EMC Services on reference signal generators for verifying and validating EMC test environments.
	Gerard Edwards Gerard Edwards is a Senior Lecturer in Electrical and Electronic Engineering at University of Bolton, UK and has taken many courses in physics/computing/electronics from undergraduate to MSc level. His research background is in theoretical semiconductor physics and solid state electronics. In recent years his primary research focus has shifted to the area of Signal Integrity and Electromagnetic Compatibility (EMC). Gerard has an MA in Physics from the University of Oxford (UK) and a Ph.D. in Theoretical Semiconductor Physics from the University of Exeter (UK). Gerard pursued postdoctoral work at the Centre for Solid State Electronics Research, Arizona State University (USA) and the Department of Electrical Engineering, University of Notre Dame (USA), in his late twenties. Electronics Subject Area, Faculty of Advanced Engineering and Science, Deane Building, University of Bolton, Deane Road, BL3 5AB, UK
	Professor Andy Marvin, FREng Fellow IEEE Andy Marvin is Technical Director of York EMC Services Ltd and Professor of Applied Electromagnetics, leading the Physical Layer Research Group at the University of York Department of Electronics. He received his BEng, MEng and PhD degrees from the University of Sheffield between 1972 and 1978. He is a Fellow of the Royal Academy of Engineering and an IEEE Fellow. He is an elected Member of the IEEE EMC Society Board of Directors, Vice-Chairman of the IEEE Std-299 Working Group on Shielding Effectiveness Measurement, an Associate Editor of IEEE Trans EMC and a Faculty Member of the IEEE EMCS Global University. His main research interests are EMC measurement techniques and shielding.