For frequencies above 1 GHz, a new site-qualification technique has been needed to take the place of the normalized site attenuation (NSA) test. This article provides a detailed description of this new technique, noting areas for caution and care necessary to ensure test repeatability. Testing was conducted to validate this technique. The results are discussed in detail.

The new procedure uses two identical horn antennas aimed at each other. Test site is located at ETS-Lindgren (Round Rock, TX).

The methods presented have been developed under the auspices of the American National Standards Institute (ANSI) Accredited Standards Committee (ASC) C63, Subcommittee 1, Working Group 1-13.2. This work applies to site requirements for radiated measurements above 1 GHz of radio-frequency (RF) compliance test sites.

The objective of any site-qualification test is to determine whether the site has sources of reflection that will cause errors in product measurements. The traditional NSA test described in ANSI C63.4 and CISPR 16.1 fails to perform this function as frequency increases because of the natural beam forming that occurs in antennas above 1 GHz.^1–3 Figure 1 illustrates the pattern of a log-periodic dipole array at 7 GHz.

It has also been shown that as the patterns of the transmitting and receiving antennas narrow, only the space between the antennas is being evaluated in a traditional NSA test.⁴ Obviously, this is a case of diminishing returns—although the log-periodic dipole array has a wide enough pattern to illuminate a sidewall or an object, this pattern narrows with increasing frequency.

Figure 1. Log-periodic dipole array pattern (5 dB/div.) and 3-dB beam width (46°) at 7 GHz, horizontal polarity.

If the methods used at lower frequencies fail to meet the objective of identifying reflecting objects at higher frequencies, a fresh perspective must be taken. The site qualification test should be as representative as possible of the actual measurement techniques used to evaluate products. Product testing requires rotating the product through 360° to find the emissions. During this process, the site should not introduce errors to the measurement from reflection of emissions at other azimuth angles of the product. Figure 2 illustrates this concern.

For the latter half of the twentieth century, the location of aircraft has been identified in aviation using radar. A radar method, or rotational illumination technique, seems at first blush to be analogous to the measurement methods currently used in EMC. It would, therefore, appear to be appropriate for identifying reflecting objects in EMC testing.

Figure 2. Illustration of errors from multilobe sources and site reflections.

New Procedure

The proposed procedure has been thoroughly researched and is deceptively simple.² First, two identical horn antennas are bore-sighted or aimed at each other at the desired measurement distance, usually 3 m. The transmitting antenna is located on a turntable with the aperture over the center of the turntable. The ground plane between the antennas is covered with absorbing material to ensure an anechoic environment.

Using a network analyzer saves significant time and ensures adequate dynamic range. The analyzer should be calibrated, and the maximum output level should be set. Before beginning the measurements, a through measurement and a simple dynamic-range test is conducted. The through measurement is intended only to verify that the cables being used are in good working order and do not have any unusual power holes or narrow frequency ranges at which very high losses occur. The dynamic range of the system is evaluated by terminating the transmitting cable and taping the cable to the transmitting antenna, near the antenna terminal. With this setup, sweeping the analyzer over the frequency range of interest displays the system dynamic range. Figure 3 illustrates typical results.

Figure 3. Typical measurement system dynamic range.

Once the dynamic range has been determined to be adequate, the antennas can be connected to the cables and the bore-sight voltage measured as a function of frequency. The site measurement can be conducted starting with the transmitting antenna rotated 90°, or perpendicular to the receiving antenna. The measurements begin at 90° because the receiving antennas normally have sufficiently narrow beam widths such that any testing at angles less than 90° would require the signal to have multiple reflections in order to be measured by the main lobe of the receiving antenna. The frequency is swept, for example, from 1 to 18 GHz, and the received voltage is recorded. The azimuth is then incremented 6°, and the measurement is repeated. As a result, voltage is obtained as a function of frequency and azimuth. This procedure is conducted at both a calibration site and at a test site. It is, of course, critical that the calibration site be free of reflections (a subject of further study). Once this test has been conducted at both sites, sufficient information is available to compare the sites. Any variations in the patterns would be the result of reflections and can be represented in linear terms by Equation 1:

Acceptance Criteria

The question that must now be answered is how much error is acceptable in a product measurement from a site reflection? The answer is under study now. However, there is some guidance in ANSI C.63.6.^5,6 This standard describes the derivation of the ±4 dB criteria used for the NSA test. That criterion assumes 1 dB of error for the site. An expression of 1 dB as the maximum acceptable error contribution from sites above 1 GHz is shown in Equation 2:

In linear terms, these criteria would be expressed as Equations 3 and 4:

Taking the log of Equation 4 defines the pass-fail criterion as: the ratio of the reflected voltage to the bore-sight voltage should not exceed –18.3 dB. This criterion as applied to the proposed procedure is expressed in Equation 5:

This criterion allows for 1 dB of error from one reflection source. A product under measurement may have multiple emission lobes at one frequency, and the site may have multiple reflection sources. However, there is also a question as to the probability of the reflections and sources aligning in a measurement. Also, consideration still must be given to the uncertainty of the measurement. Piloting (beta testing) the proposed criterion is the preferred method for evaluating how stringent this criterion is.

Be Alert for Specific Hazards

The first few attempts to apply this method and these criteria reveal numerous issues that need to be considered. Most laboratories have certain cables used for high-frequency measurements. These cables often have narrow frequency ranges in which the cables and fittings have very high losses. These abrupt changes in insertion loss are often referred to as power holes. Power holes are far more common than most laboratories suspect. After setting up the equipment, a through measurement of the cables reveals any power holes. Obviously, such cables should be replaced. Repeated measurements of the through connection after making, breaking, and remaking the connection will be required to ensure that intermittent problems are identified.

Figure 4. Log-periodic dipole array pattern (5 dB/div.) and 3-dB beam width (46°) at 15 GHz, horizontal polarity.

Another important hazard to be avoided is antenna patterns exhibiting large back lobes. Those antennas that have a large slope (dB/degree of azimuth) on the back lobes will cause very large uncertainties associated with the positioning of the transmitting antenna. An example of such an antenna is shown in Figure 4.

Validation Results

Initial validation testing was conducted at the National Institute of Standards and Technology (NIST) in Boulder, CO, and at ETS-Lindgren in Round Rock, TX. These initial tests involved the measurement of the antenna patterns followed by repeating those measurements with an intentional scattering object in the test environment. Figure 5 illustrates one such measurement setup.

Figure 5. Photo shows setup for pattern measurement with a foil-backed foam sample material.

This test setup uses a 3-m measurement distance between the antennas, and the scattering objects were located 3 m behind the transmitting antenna. A second location for the scattering object was 3 m from the bore-sight path and perpendicular to the center of that bore-sight measurement path. Scattering objects were selected from typical building construction materials. Applying Equation 5 to the measured data and assigning a floor to the plot of –18.3 dB provides a good illustration of the results. Figure 6 shows one such plot of results when foil-backed foam was used as the scattering object. This foam is commonly used as a vapor barrier in building construction and would be an extreme example. Note in Figure 6 that the maximum voltage displayed coincides with the path loss difference between the bore-sight path, 3 m, and the reflected path, 9 m.

Figure 6. Graph provides example results of foil-backed foam scattering object, horizontal polarity.

A more realistic material test was made using a sheet of plywood. Figure 7 illustrates the typical results from a 2 X 2-ft sheet of plywood. In this test, it is clear that the plywood was located directly behind the antenna (180°). The resonant frequencies of the plywood are narrower than originally thought. Reflections from the plywood exceeded the established criterion only in narrow frequency ranges.

Figure 7. Graph shows example results with plywood scattering object, vertical polarity.

Seven materials were tested in this fashion. Those materials were particleboard, fiberglass, plywood, foil-backed foam, roofing (plywood with shingles), treated plywood, and chipboard. Testing these materials demonstrated that the procedure worked well. Validating the procedure and the criteria also requires testing real test sites. To date, three semianechoic chambers have been tested using this procedure, and as many as six more sites, including more chambers, open-area test sites, weather-protected open-area test sites, and even gigahertz transverse electromagnetic (GTEM) cells will be tested.

Figure 8. Graph shows example results of a 10-m semianechoic chamber, vertical polarity.

Figure 8 shows the results of testing of a 10-m semianechoic chamber with full absorber treatment on the walls and ceilings. The only reflections found were at the 1-GHz frequency range and directly behind the transmitting antenna. This is most likely the result of the measurement axis being perpendicular to the rear wall. Improvements could probably be realized by changing the measurement angle.

Figure 9. Graph shows example results for a 10-m semianechoic chamber with partial absorber treatment, vertical polarity.

Testing of a 10-m chamber of the same size employing absorber only in the spectral regions (partial treatment) yielded very different results, as shown in Figure 9. These data were taken at –10° from the centerline of the chamber. It is also important to note that the turntable was not located in the center of the width of the chamber, but rather a few meters to one side of the centerline of the chamber. The results clearly show that the corner of the chamber is reflecting as a result of the lack of absorber treatment in that corner area.

Figure 10. Graph shows example results for a 5-m semianechoic chamber, vertical polarity.

The third semianechoic chamber tested was a partially treated 5-m chamber. This chamber also had an offset turntable, and the measurement angle was also approximately 10° from the centerline. Here again, it appears that the corners are a source of reflection that exceeds the 1-dB criterion.

The three semianechoic chambers tested revealed an obvious problem. Partially absorber-lined chambers may need modification to be able to meet the proposed qualification test. Conversely, consideration can be given to increasing the allowable reflections while recognizing that the allowable error decided upon must be taken into account in determining measurement uncertainty.

Conclusion

Caution must be made in conducting these tests to ensure that the cables are free of power holes and that the antennas do not have significant back lobes. Work remains to be done to evaluate the performance of existing test sites. In addition, procedures need to be developed to evaluate the adequacy of the ground-plane absorber applied and the calibration site selected.

The proposed procedure appears from all the validation tests to work quite well. These results show that objects that can be a source of reflection will be measured in this procedure. How significant the reflections and the resulting errors in measurements can be will probably be a topic of significant discussions in the EMC community and certainly in ANSI ASC C63.

References

1. M Windler and D Camell, "Measuring Antennas above 1 GHz," in Proceedings of the 14th International Zurich Symposium on Electromagnetic Compatibility (Zurich: Swiss Electrotechnical Association, 2001), 91.

2. ANSI C63.4-2000, "American National Standard for Methods and Measurement of Radio-Noise from Low-Voltage Electrical and Electronic Equipment in the Range of 9 kHz to 40 GHz," American National Standards Institute, New York.

3. CISPR 16-1:1999-10, "International Special Committee on Radio Interference—Specification for Radio Disturbance and Immunity Measuring Apparatus and Methods–Part 1: Radio Disturbance and Immunity Measuring Apparatus," International Electrotechnical Commission, Geneva.

4. R Johnk et al., "NIST Time-Domain Scattering Experiment Using Dual-Ridged Horn Antennas and an Ultrawideband Measurement System," CISPR A, Working Group 1, Ad Hoc Group on > 1 GHz, report dated January 2002.

5. M Windler and D Camell, "Research on Site Qualifications above 1 GHz," in Proceedings of the 15th International Zurich Symposium on Electromagnetic Compatibility, (Zurich, Swiss Electrotechnical Association, 2003).

6. ANSI C36.6-1996, "American National Standard Guide for the Computation of Errors in Open-Area Test Site Measurements," American National Standards Institute, New York.

Michael J. Windler is head of international EMC and NEBS services for Underwriters Laboratories Inc. (Northbrook, IL). He can be reached via e-mail at michael.j.windler@us.ul.com.