A Comparison of Radiated Emissions Testing to European Directive 95/54/EC Using an Open Field Test Site and a Semi-anechoic Chamber.

Transcription

1 1005 A Comparison of Radiated Emissions Testing to European Directive 95/54/EC Using an Open Field Test Site and a Semi-anechoic Chamber. Dr Peter Miller Motorola AIEG Tayh Road Stotfold, Hitchin, Her& SG5 4AY, England Abstract: Most automotive component level testing, and in polarisation (figure 4 - this is also apparent to a lesser degree particular European Directive requires that under some in vertical polarisation - figure 3), which was not present in (in practice most) conditions radiated emissions measurements the other measurements. A somewhat smaller notch is also from Electrical/electronic Sub-Assemblies (ESA s) arc made visible at -140MHz. 95/54 measurements in a semi-anechoic with the ESA under test mounted on an elevated ground plane chamber (here the table top was earthed by extending it Im high on a wooden table. It is shown in this paper that how backwards through the absorbers to the chamber wall) arc this ground plane is connected to the test sites ground is very shown in figure 5. important, and that this is in fact a key issue in obtaining correlation between different open field sites and also to obtaining correlation to test results obtained in a semianechoichamber. INTRODUCTION Many authors (for example [2],[3],[4],[5]) have presented results showing differences in radiated emissions results between test sites. In particular [5] raises many questions about correlation between sites used for vehicle component testing. Most vehicle component test specifications (e.g. CISPR-25, SAE J1113/41, European directives 95/54iEC and 97/24iEC and IS0 standards IS and ISO14892)[5] require that the equipmento be tested is placed upon an elevated ground plane while the measurements are made. How this elevated ground plane is earthed adds an exha degree of variability, and seems to make to the correlation behveen results obtained on open area sites and in semi-anechoic Again, evidence of two notches is seen at approximately chambers worse [5]. It should be noted that some tests defined 55MHz and 11OMH.z. It should be noted this chamber is in MIL-STD-462 also have a similar test set-up. relatively small ( m) and is lined with 20in pyramidal absorbers. The general test set-up used is shown in EXPERIMENTAL RESULTS figure 7, while figure 8 provides more detail. Figure 9 shows a photograph of the open area site during testing, the ground Measurements were made using a commercially available strap can just be seen on the back right hand comer of the comparisonoise emitter (CNE)(a wide band noise generator table. which is battery powered and fitted with a small top loaded Given the small (100mm long monopole) antenna used on the monopole antenna) (see figure 1). Using a wooden table, the CNE it was not expected that this would directly be the cause results obtained (figure 2) show a good correlation between of the observed notches, and the measured SWR plot on a the open field test site and the semi-anechoichamber. ground plane (figure 10) confvms this. However, when the ground plane was fitted to the top of the Some results have also been obtained using another open area wooden table (earthed via a vertical length of braid in the test site, these also appear to show a notch about 76MHz - open area site), a deep (>20dB) notch at just under 7OMHz these are not shown here because of space limitations. was apparent in the free field measurements for horizontal 786

2 Further measurements on both open area sites and semi- major resonance at -1SOMHz [6]. It is however, possible to anechoic chambers are given in [S], however these tests used a see further evidence of the lower frequency notch, and 1.5m long horizontal antenna on the signal source which increased site-site variability in the region of this notch, in the therefore overlaid its own resonant emissions pattern on that results presented in [5]. of the test site. Simulations of this antenna structure show a Frequency [Hz] 90 E zi :: ;: % # ID 30M 40M50M i 0M IOOM 200M 300M 400M BOOM IG Frequency Figure 2. A: CNE as measured on free field site at 10m. B: in semi-anechoic chamber at 3m, vertical polarisation. [Hz] 787

3 ._.._ _ _._-.._ _-_ _ _-_ :- _--_ cnqk printed on 24/2/1999 at IO: p 70 E xl w 2m x0 ml 700 mm Figure 3. CNE measured on a free field site at lm to 95/S, vertical polarisation. MHz -- - cnehpk printed on 24?2! at 12:03 90 Cnt hpk E p ci k #I 40 so x0 xa iom MHz Figure 4. CNE measured on free field site to 95/54 with horizontal polarisation of CNE and measuring antenna. 788

4 vpkl prmtedon 24i2/1999 at la, a-m 240 MHz ^,,,.,_.,,.,,,,,,,.,,,,.,,.,..,. ",....,,., ~..~./..~~.~. ~-- -- i -~ ---_.-_-- --~ -.-.-"~--;T+ _--. Figure 5. CNE radiated output to 95/54 in a semi-anechoic chamber, vertxal polarlsatlon. ATTEN 1 OdB MKR -z!.4. Gr5dE3rll RL -20. bdbm 780, lmhz CENTER 515. QMHa SPAN GI7tis DMH;E -x-rbw 1 OOkH-iz VE3.W 1 CJl.JkHs: *SWP 2 fyc3esec Figure 6. CNE conducted emissions (peak). 789

5 1005, Tjy& I- fiaurr R. Detail of95lsj EC~U~. TABLE AS A TOP LOADED MONOPOLE? A possible mechanism to explain the notch is that the elevated ground plane could be acting as an antenna, the notches being the fiat and 2nd resonance s of this structure. If we consider the 95/54 open tield site set-up, then the vertical wire connecting the ground plane to earth could be acting as a top loaded monopole and resonance s of this could be causing the deep nulls. Assuming a top loaded monopole strochue, reference [l] ~~2-40 shows that a vertical wire lm long and OSin diameter would be resonant at 68MHz with a top loading of -50pF. Tbe self-capacitance of the sheet used (1 l 1.15m) was calculated as 5OpF. This close agreement in frequency strongly suggests that this is the mechanism causing the notch. SIMlJLATlONS To investigate further, both the free field and the semianechoi chamber set-ups were simulated using a version of NEC. Modelling the table top as a grid of 5.5 wires, each wire being 10 segments long gave a resonant frequency of 72Mhz 790

6 over a perfect ground plane - in good agreement with figures 3,4. The simulated impedance (looking from the bottom of the vertical wire connecting real ground to the tabletop ) is shown in figure 11 -which clearly shows the resonance. 95/54 table 0.J I I I I Frequency (MHz) Figure 11. Simulated table impedance. A similar simulation of the situation in the semi-anechoic room resulted in a simulated resonance at 54MHz, again in good agreement with the measurements in figure 5. Clearly this basic structure is difficult to change, as any reasonable connection to earth in a free field site will result in a sort of top loaded antenna structure. This leaves the option of either using the same ( non-optimum ) earthing structure in the semi-anechoic chamber, or of including the effects of the structure resonance in the correlation required by (which treats the free field site as the reference, while not defining how the plate on the table is earthed). Given that the earthing arrangements of the tabletop ground plate do currently vary between different free field test sites, the correlation approach was chosen. To date correlation s have been obtained for two NAMAS accredited open area sites, which are run by different companies. These correlation s allow approximate correction of the results measured at a specific semi-anechoic chamber to those of the 2 open area sites. CONCLUSIONS Measurements and simulations have shown that the elevated ground plane can cause notches in the measured radiated emissions. It has been conjectured that this can be explained by treating the elevated ground plane as an antenna structure - simulations based on this explanation have shown very good agreement with the measured results. Given the good explanation, and the lack of variability in the structures causing this, it is reasonable to apply a correction factor to see approximately what the measurements would be at another site. This is particularly useful when the semi-anechoic chamber is used for pre-compliance testing and open area sites are used for compliance testing. It should be noted that at present CISPR-25 requires correlation between a semi-anechoic chamber and an open area site be obtained before the semi-anechoic chamber is used for measurements to CISPR-25, but does not permit the use of a correction factor on the results. This paper has shown that to obtain such a correlation requires attention to many details not currently specified in CISPR-25. The author believes that similar simulations could also play a significant role in designing and tolerancing test set-ups so as to minimise variability between test sites. ACKNOWLEDGEMENTS The following people at Motorola inputted to this work, Bob Lempkowski and Jerry Meyerhoff re-enforced the idea that the table may be acting as an antenna which was the basis of this investigation, and Martin O Hara performed the semianechoic chamber measurements and provided some of the drawings used in this paper. REFERENCES [l] ARRL Antenna book, 17th ed., [2] Heise, E., Heise, R., A method to compute open area test site uncertainty using ANSI C63.4 Normalised site attenuation measurement data, ~~ , IEEE 1996 International symposium on EMC. [3] Heise, E., Heise, R., A method to calculate uncertainty of radiated emissions, pp , IEEE 1997 International symposium on EMC. [4] Kolb,L., Statistical Comparison of site to site measurement reproducibility, pp24 l-244, IEEE 1997 International symposium on EMC. [5] Swanson,D., Investigation of the Calibration Procedure from CISPR25 Annex B, for use with vehicle component testing, Proceedings 1998 IEEE International Symposium on Electromagnetic Compatibility, ~~ [6] Miller,P., Bulk current immunity testing, its advantages and disadvantages for Automotive EMC testing, Proceedings AutoTech