Cables and Radiated-Emission-Measurement Uncertainty above 1 GHz

Martin Alexander and Tian Hong Loh

A study indicates that cables have less effect on the total radiated energy at higher EMC test frequencies than at frequencies below 200 MHz.

Radiated-emissions compliance testing for electromagnetic compatibility (EMC) is usually accompanied by high measurement uncertainties. It is not uncommon for uncertainty in the measured field strength to be of the order of ±10 dB. On a good-quality measurement site, with modern equipment and with good-quality antenna calibration, the uncertainty should be below ±5 dB, which is in line with the magnitude of UCISPR given in CISPR 16-4-2, a standard of the International Special Committee on Radio Interference of the International Electrotechnical Commission.¹

The main reason for the higher uncertainties is the lack of precise definition of the equipment layout, which means that reproducibility, including measurements by other test houses, is not maintained. The poor reproducibility arises predominantly from the radiation from various cables attached to the equipment under test (EUT). The magnitude and phase of this radiation are very sensitive to the precise layout of the cable in relation to the EUT and to the nature of the cable termination. The cable radiation combines with the radiation from the body of the EUT and, in the worst case, can cancel the latter.

An informative treatment of uncertainty components additional to instrumentation can be found in CISPR 16-4-1 under the heading “Compliance Uncertainty.”²

Emissions from cables are most noticeable below 200 MHz. In this frequency range, the cables act like moderately efficient antennas, a phenomenon that has been well documented.^3,4 Recently, however, there has been a surge in activity in EMC testing above 1 GHz, and with it great interest in knowing how important the influence of the cables is, especially up to 6 GHz.

The National Physical Laboratory (NPL; Teddington, Middx, UK) in 2005 commissioned a dummy EUT to which cables could be attached and which incorporated a comb generator that gave measurable output up to 10 GHz.

The measurements taken with the dummy constituted a basic investigation into the mechanism of the radiation. Researchers still have to do a lot more work using more-realistic cables and a range of terminations. But the early measurements did confirm that the radiation was emitted in an end-fire direction rather than orthogonal to the cable. The NPL experiment explores that effect.

Figure 1. Radiation pattern of a 0.5λ wire (a) contrasted with the end-fire radiation pattern of a 4λ wire (b) that is fed from the left end and unterminated on the right end. Source: FE Terman, Radio Engineering (New York: McGraw-Hill, 1951), 668.

The difference in cable radiation pattern between a low-frequency system with an electrically short cable and a higher-frequency system with an electrically long cable is displayed in Figure 1. Figure 1a shows the pattern for electrically short cables, typically those with frequencies below 200 MHz, and for cases where wavelengths (λ) are longer than 1.5 m. The pattern in Figure 1b is more typical of electrically long cables, that is, cables having a length of one or more wavelengths. At 1 GHz, λ = 0.3 m; thus, the cable length of 0.8 m that is cited in several EMC standards is 2.7λ. As the frequency increases, the electrical length increases proportionally.

In a radiated-emissions test in which the cable has the cardioid pattern seen in Figure 1a, when the cable is vertical, the maximum of the lobe points toward the antenna. Also, when the cable is horizontal and broadside to the antenna, the maximum signal is again be picked up. However, for a vertical cable with an end-fire pattern like that depicted in Figure 1b, the main lobe is directed toward the ground, and the signal picked up by the antenna is much less. This is one reason that the effect of cables above 1 GHz does not appear to be as prominent as the effect of cables below 200 MHz.

Experimental Measurements

Figure 2. EUT test setup in the GTEM cell, with parallel wires (a) and with perpendicular wires (b).

Measurements were made at NPL in a 9 × 6 × 6-m fully anechoic room (FAR) and in a MEB-1750 gigahertz transverse electromagnetic (GTEM) cell. The source was a battery-operated CGE01 comb generator (York EMC Services; Castleford, W. Yorks, UK) with 100-MHz harmonic spacing and a fundamental signal of 100 MHz, enclosed in a metal box with dimensions of 230 × 210 × 110 mm. A subminiature version A (SMA) connector in one wall of the box was fed by the CGE01 via a semirigid cable.

Attached to the outside of the SMA connector was an SMA male connector whose center pin was joined to a wire whose length, for the experiment, was varied from 0.1 to 0.8 in steps of 0.1 m.

Figure 3. EUT setup in the 9-m FAR.

In the case of the 0.3-m wire, the NPL investigators performed an additional test in which the EUT was mounted on a foam table in the GTEM cell (see Figure 2). Another 0.3-m-long wire was attached by a screw to the chassis parallel to the active wire (see Figure 2a) and measurements were taken. The wire layout was then made perpendicular (see Figure 2b).

This roughly simulates a real product that has more than one wire attached. The difference in the emissions with and without the passive wires was negligible up to 6 GHz, which showed that the radiation pattern is dominated by the active wire. However, as the frequency increased to 10 GHz, the difference increased to 5 dB.

Figure 4. Azimuthal radiation patterns plotted for a 0.3-m cable at 3 GHz (electrical length 3λ) (a) and 6 GHz (electrical length 6λ) (b).

The following experimental results pertain to tests of the EUT mounted on the foam table in the FAR and with no passive wire attached. Figure 3 shows the setup in the FAR, where an amplifier was needed to counter the attenuation of the 30-m cable to the spectrum analyzer.

Other work carried out by the same investigators but not yet published has shown that radiation patterns measured in a GTEM cell are similar to those measured in a FAR.

Results

Figure 5. Azimuthal radiation patterns plotted for a 0.1-m cable (a) and an 0.8-m cable (b) at 1 GHz. Electrical lengths are 0.33λ and 2.7λ, respectively.

A selection of azimuthal radiation patterns measured in the FAR is presented here. The data are the raw decibels referred to 1 mW (dBm) received into the spectrum analyzer, the amplitude being indicated by the vertical scale of 10-dB steps on the pattern plots in Figures 4 through 8.

In the 0° direction, the tip of the wire is pointing toward the antenna. Notable features of the radiation patterns include the end-fire lobes just to either side of 0°. As the electrical length of the wire increases from 0.1 to 0.8 m, the radiated amplitude hardly increases. Another feature is that the onset of the end-fire shape occurs at a lower frequency as the wire length is increased. That is, it can be expected that a roughly similar end-fire pattern should occur for each cable of a given electrical length. (As an aid, the figure captions include the electrical length.)

Figure 6. Azimuthal radiation patterns plotted for a 0.1-m cable (a) and an 0.8-m cable (b) at 3 GHz. Electrical lengths are λ and 8λ, respectively.

Figure 4 shows the progression of the end-fire effect on a 0.3-m wire as the frequency is increased. The signal in the forward hemisphere is enhanced by 5 dB and 12 dB at 3 GHz and 6 GHz, respectively, compared with the rear hemisphere.

These plots are also to be compared with those for the 0.1- and 0.8-m wires. Figures 5 through 8 show the end-fire lobes gradually forming as the frequency is increased from 1 to 10 GHz. The decrease in amplitude with an increase in frequency is a function of the cable attenuation and the comb generator’s output power. However, the important point is that, at a given frequency, the maximum amplitude is similar regardless of wire length. It is mainly the directivity that changes.

Conclusion

Figure 7. Azimuthal radiation patterns plotted for a 0.1-m cable (a) and an 0.8-m cable (b) at 6 GHz. Electrical lengths are 2λ and 16λ, respectively.

Measurements on a short length of wire that is fed at one end and unterminated at the other show that the radiation pattern becomes end-fire as the wire length increases beyond two wavelengths (2λ). This means that a cable that drops vertically from an EUT will tend to direct its radiation toward the ground and that, therefore, the antenna will measure side lobes of lesser magnitude.

For an electrically short antenna, the efficiency increases proportionally with frequency, which means that the electric field strength doubles as the wire length doubles. However, the results of the NPL investigation on electrically long wires reveal that the maximum magnitude of radiation is remarkably constant regardless of the increase in wire length from 0.1 to 0.8 m. One plausible interpretation of this effect is that most of the radiation above 1 GHz leaves the wire in the first 0.1 m or so.

Figure 8. Azimuthal radiation patterns plotted for a 0.1-m cable (a) and an 0.8-m cable (b) at 10 GHz. Electrical lengths are 3.3λ and 27λ, respectively.

Whatever the cause, what happens to the wire beyond this early length appears to have less effect on the radiated-emission result. This investigation is a basic study on short, straight unterminated wires. More work is needed on more-realistic terminated wires to verify these findings.

Acknowledgment

The authors thank Richard Marshall of Richard Marshall Ltd. (Harpenden, Herts, UK), who suggested that the authors explore the end-fire effect.

References

1. CISPR 16-4-2:2004, “Specification for Radio Disturbance and Immunity Measuring Apparatus and Methods—Uncertainties, Statistics, and Limit Modelling—Uncertainty in EMC Measurements” (Brussels: International Electrotechnical Commission, 2004).
2. CISPR 16-4-1:2003, “Specification for Radio Disturbance and Immunity Measuring Apparatus and Methods—Uncertainties, Statistics, and Limit Modelling—Uncertainties in Standardised EMC Tests,” technical report, (Brussels: International Electrotechnical Commission, 2003).
3. L van Wershoven, “The Effects of Cable Geometry on the Reproducibility of EMC Measurement,” in Proceedings of the IEEE EMC Symposium, vol. 2 (Seattle: Institute of Electrical and Electronics Engineers, 1999): 780–785.
4. R Marshall, “Reducing Errors due to Resonances in Radiated and Conducted EMC Testing,” in Proceedings of the 15th International Zurich Symposium on EMC, ed. G Meyer (Zürich, Switzerland: Communi-cation Technology Laboratory, ETH Zürich, 2003): 267–272.

Martin Alexander is principal research scientist at National Physical Laboratory (NPL; Teddington, Middx, UK). Tian Hong Loh, PhD, is higher research scientist at NPL.