Inside EMC Antennas

Ideally, EMC testing could be accomplished with one antenna, one amplifier, and one receiver. At least the receiver part generally can be achieved, and in some sports, one out of three is pretty good. Antennas and especially amplifiers perform best over limited bandwidths, a characteristic that complicates EMC test setups.

Nevertheless, recently developed broadband antennas span a wide range of frequencies such as 30 MHz to 2 GHz for the Com-Power Combi-Log Model AC-220. Typically, these types of antennas combine the mechanical features of log-periodic and biconical antennas to extend bandwidth.

The log-periodic part of the AC-220 is 37″ long x 22″ wide, approximately the same size as the company’s Model AL-100, a log-periodic antenna covering 300 MHz to 1 GHz. The large bow-tie structure at the rear of the AC-220 measures 50″ x 25″, similar to the overall dimensions of the Model AB-900, a biconical antenna with a 30-MHz to 300-MHz frequency range.

The ATR26M6G-1 from AR RF/Microwave Instrumentation, shown in Figure 1, is even larger than the AC-220 at 86″ x 29″ x 63.5″ and covers 26 MHz to 6 GHz. This antenna uses bent elements and inductive loading to minimize the size required to handle the lower frequencies.

According to the datasheet, “[The] bent element approach combined with additional innovations provides a size reduction of approximately 75% without sacrificing key electrical performance such as gain and bandwidth…. The considerable size reduction minimizes field loss resulting from room loading.”

The basic problem addressed by extending the low-frequency response is the long length of an equivalent dipole. A half-wavelength dipole at 28 MHz is more than 16-ft long and won’t fit in many anechoic chambers. Gain falloff at lower frequencies in a smaller antenna shouldn’t be unexpected, but the goal is to extend the bandwidth while avoiding extreme parameter variation. For example, the Model ATR26M6G-1 specifies maximum VSWR at 6:1 from 26 MHz to 80 MHz and 3:1 above 80 MHz.

The AC-220 datasheet includes a graph of antenna factor (AF) vs. frequency that follows the corresponding graphs for the basic AL-100 Log-Periodic and AB-900 Biconical Antennas very well (Figure 2). AF increases by more than 5 dB/m below about 80 MHz, the resonant frequency of the bow-tie element.

Antenna Parameters

A real antenna’s gain often is quoted relative to the field produced by a perfectly isotropic antenna. All physical antennas are directional to some extent so the isotropic ideal is a convenient reference, not something that actually can be constructed. Nevertheless, together with a unified azimuth and elevation coordinate system, it does make antenna comparison straightforward.

Gain is equal to an antenna’s directivity multiplied by its radiation efficiency. The large effect of directivity is easily seen when comparing a horn antenna with as much as 30-dBi gain—dBi means dB relative to isotropic—and a much less directional half-wave dipole with 2.14-dBi gain.

In contrast to the nondimensional gain value, the AF actually is a transfer function relating the applied field to the received voltage. A large AF implies the need for a more sensitive receiver. For example, the AC-220 Antenna is specified to produce a field of 3 V/m at a distance of 1 m at 200 MHz with 100 mW applied. From Figure 2, AF = 10 dB/m at 200 MHz. Being a voltage quantity, AF(dB) is defined as 20 log (AF) or

so AF = 3.16, and you should measure 950 mV from the antenna: 0.950 V x 3.16 /m = 3 V/m.

If the AF has been defined under conditions similar to those in which the antenna actually is used, you should get accurate results. On the other hand, it’s not valid to use a 50-MHz AF determined with a 1-m antenna spacing when you’re working at 10 m. The 1-m spacing is in the near field, but 10 m is almost in the far field. In contrast, it probably is valid to use a 1-GHz AF at any spacing greater than 1 m.

The wavelengths corresponding to low frequencies are very large, which means that even relatively bulky EMC antennas will be operating in the near-field region. Although the demarcation between regions is approximate, the radiating near field is assumed to extend to

For an antenna with a 2-m largest dimension, the radiating near field at 50 MHz extends to at least 1.3 m or about 4 ft. The far field generally is considered to start at two or three times the wavelength, 12 m for 50 MHz.

For a biconical or dipole antenna, the largest dimension of the structure is well defined. This isn’t the case for a log-periodic antenna in which the active region shifts along the length of the antenna depending on the signal frequency. A good example of a large log-periodic antenna is the ETS-Lindgren Model 3144 with a frequency range from 80 MHz to 2 GHz and measuring 83″ wide x 67″ long x 3.8″ deep.

From the datasheet’s AF graph, the variation caused by operating at 80 MHz in the near or far field is apparent. At a 1-m distance, AF = 9 dB/m. At 3 m, AF = 5 dB/m; at 10 m, AF = 3 dB/m. As you would expect, the AF value becomes more consistent at higher frequencies, varying between 25 and 23 dB/m at 1 GHz regardless of distance because far-field conditions start at less than
1 m.

The relationship between AF and gain often is given as

where G₀^‘ is the antenna gain adjusted for mismatch and radiation efficiencies. This expression only holds for far-field conditions in a 50-Ω system with matched antenna and incident field polarization. At low frequencies and short distances, it’s more accurate to use the antenna manufacturer’s data because it’s likely the far-field restriction isn’t being satisfied.

Test Considerations

The ETS-Lindgren Model 3144 datasheet includes a useful technical tip regarding the different measurement references used by various standards: “Log periodic antennas are calibrated at different distances from the source depending on the standard. For example, calibration for MIL-STD immunity testing is performed at a distance of 1 m as measured from the tip of the antenna to the source. However, calibration for commercial immunity testing is performed at a distance of 3 m…. Calibration for MIL-STD emissions testing is performed at a distance of 1 m as measured from the center of the antenna to the source. Calibration for commercial emissions testing is performed at a distance of 3 m and 10 m…from the center of the antenna….”

Using the antenna tip as a reference not only causes the greatest error at low frequencies, it also ignores the inexact relationship between the location of the antenna tip and that of the active receiving elements. In a paper that examined the role of EM simulation in log-periodic antenna design, the authors discussed the dependence of radiation pattern on frequency:

Ideally, the pattern should be nearly independent of frequency, but in an initial design, a strong correlation was shown. By adding several more elements to extend the high-frequency range, the radiation pattern became constant over the bandwidth of interest. This means that, depending on the antenna model, the tip may be spaced a different distance from the active antenna elements if the design is more conservative or less conservative.¹

More generally, the accuracy with which EMC emissions testing can be performed is limited by the antenna calibration accuracy. This is not the case for immunity testing where these measurements rely on the calibration of the field probes. If the probe reading is accurate, the required field strength can be achieved by increasing or decreasing the antenna power as required.

For emissions testing, several antenna calibration procedures are available, some more appropriate than others depending on the tests to be performed. The most general calibration determines the AF as though the antenna were in a free-space environment (AFfs). In practice, this is often measured in a full anechoic chamber or at an extreme height above a ground plane.

With AFfs, test-site imperfections have not influenced the calibration. For example, the very informative Good Practice Guide No. 73 from the National Physical Laboratory (NPL) states, “CISPR 16-1-5 gives a specification for a calibration test site with the criterion that the site has to be within ±1 dB of a specified value. Unless the site itself is closer to ±0.5 dB, an antenna calibrated on it by the three-antenna method is likely to have an uncertainty greater than ±1 dB. On the other hand, the standard antenna method is less demanding of the quality of the site because if the standard antenna has similar dimensions to the antenna under test, the unwanted site effects can mostly cancel.

“In practice, a good approximation to free space can be achieved by an anechoic chamber, particularly if the antennas are directional and therefore they see less of the chamber walls. The corollary to this is the omnidirectional dipole antenna, especially at frequencies below 150 MHz, where it is not practical to get far enough away from chamber walls or the ground if outdoors: the solution for which is to quantify the ground reflection by means of a large, flat, well-conducting ground plane.”²

For large antenna calibration, NPL recommends TEM and GTEM cells as alternative methods. Between the cell plates, plane-wave conditions exist, and calibration accuracy as low as ±1 dB has been achieved using a TEM cell with 0.915-m plate separation for a 0.6-m dia antenna.

The high-precision ground plane at the NPL antenna calibration facility is estimated to contribute less than 0.05 dB of the 0.1-dB difference between theoretical and measured insertion loss values for two horizontally polarized resonant dipole antennas. This level of performance does not come cheaply: The ground plane consists of welded 8-mm thick mild steel sheets and measures 30 m x 60 m with 95% of the area being flat to within ±6 mm.

Dr. Vince Rodriguez, senior principal antenna design engineer and antenna product manager at ETS-Lindgren, discussed the company’s welded steel 50-m x 80-m OATS. “We can perform geometry-specific AF calibrations which provide a more accurate normalized site attenuation (NSA) measurement. The antennas used for NSA are usually calibrated in pairs, and a combined AF for the pair is determined. Above 1 GHz, the new methods such as the site VSWR (SVSWR) measurement are more of a reference method, so the calibration of the antenna is not as important since you are just looking at peaks and valleys on a given standing wave at a given frequency,” he explained.

Recommendations

Jason Smith, applications engineering supervisor at AR, advised, “The antennas that are the best fit for immunity testing depend on the frequency range, field level, and test standard being used. Test standards will dictate the antenna beam width needed and if a uniform field area must be established.

“In most cases,” he continued, “the broader the beam width, the lower the gain and therefore the more power that is required. The use of an antenna that includes a balun is not suggested for immunity testing because the balun restricts the amount of input power that can be applied, creates a high VSWR that reduces the power actually radiated, and contributes its own inefficiency.”

The subject of baluns is dealt with in depth in the NPL Good Practice Guide, particularly with respect to biconical antenna imbalance. If a balun is not perfectly balanced, an unwanted signal will appear on the outer of the feed coax cable. The field developed by this signal can destructively interfere with the field measured by the antenna, especially in the vertical polarization direction. As a result, the antenna will give different readings if one end is oriented skyward rather than the other.

NPL suggests extending the feed cable horizontally for at least 10 m before dropping to ground to minimize this effect. Of course, this is impractical if the test procedure calls for height scanning. So the best approach is to thoroughly test an antenna for balun imbalance before use.

While on the subject of biconical antennas, it’s worth repeating some of NPL’s qualifying comments. Most of these antennas are similar in size and construction to the reference design described in MIL-STD-461, being 1.35-m long and 0.52-m wide at the broadest dimension.

This design is resonant at about 75 MHz. Unfortunately, the six-wire conical cages have a secondary resonance at about 287 MHz. This problem was solved by attaching a metal bar to one of the six wires, which extended the resonant frequency above 300 MHz.

That’s good, but it means that the AF slope may increase sharply near 300 MHz. Also, a small deviation in antenna gain is associated with the metal bar, and the bar position should be noted relative to the direction in which the field is being measured if the most precise calibration is required.

Com-Power’s President Shirish Shah said, “Above 80 MHz, generally the Combi-Log Model AC-220 is used; above 1 GHz, various horn antennas are used. The horn antenna series AH-118 through AH-640 extends performance to 40 GHz.

“The amount of power required depends on the gain or AF of the antenna, he commented. “You can reduce the power requirements by using the standard gain antennas. They have narrower bandwidths but generally higher gains.”

Although horn antennas inherently are directional, an apparent disagreement has existed between ANSI C63.5, which considers a height of 2 m above a ground plane sufficient to eliminate ground reflection effects, and SAE ARP958 which does not. NPL avoids this problem by calibrating horn antennas for AFfs in a fully anechoic chamber.

The antenna calibration method described in ARP958 is problematic because:
• It relies on the Friis far-field transmission formula but is being used in the near field.
• The effect of multiple reflections between the antennas will depend on the antennas used as will the result.
• Identical types of antennas will not have completely identical characteristics.
• Calibration is valid when two antennas are being used. When the EUT is substituted for one of them, the proximity of antenna and EUT will have a different effect on the antenna performance.

Using two antennas to cover the required frequency range necessitates switching from one to the other. It is possible to set up both antennas so that one will only minimally interfere while the other is being used. This approach means that you still have to switch the feed cable connection, but you don’t physically have to substitute antennas.

ETS-Lindgren’s Dr. Rodriguez explained, “Most amplifiers do not cover the entire range of a given standard, or they do not match with the range of the antennas. An example is MIL-STD-461, which calls for a dual-ridge horn to be used from 200 MHz to 1 GHz. Many vendors offer dual-ridge horn antennas, but some amplifiers start at 250 MHz, leaving a band break for the amplifier in the coverage of the antenna.

“Also, for a given antenna, more power usually is needed at the lower part of the band than at the upper part,” he continued. “It is a waste to have very high power amplifiers covering the entire band of a broadband antenna. It takes less time to switch between a high-power amplifier for the lower part of the band and a low-power one at the upper end than entering the chamber to physically swap antennas.”

Conclusion

AR’s Mr. Smith said, “Ideally, it would be great to cover the whole frequency range with one antenna to reduce test time and setup changes. However, the cost of the required amplifiers and equipment availability limit this option. For standard commercial testing from 80 MHz to 6 GHz, two antennas are recommended, requiring one antenna change at 1 GHz. If multiple amplifiers cover the antenna bands, RF switch matrices can be added to further automate the setup.”

References

1. Ergül, Ö. and Gürel, L., “Log-Periodic Antenna Design Using Electromagnetic Simulations,” Department of Electrical and Electronics Engineering, Bilkent University, 2007.
2. Alexander et al., “Calibration and Use of Antennas, Focusing on EMC Applications,” A National Measurement Good Practice Guide, No. 73, National Physical Laboratory, 2004.

February 2010