An unbalanced balun introduces common-mode current to the antenna feed cable and radiation from the vertically aligned cable interacts with the vertical antenna. In the worst case, this condition nullifies the antenna radiation (or reception by reciprocity). In April 2003, Amendment 2 to CISPR 16-1 was published, which included a procedure to test for balun imbalance. Fortunately manufacturers have now, by and large, addressed this problem, but there are still plenty of old antennas that are still in use. Because very tight mechanical tolerances are required to achieve good balance at the lowest frequencies, especially for antennas with a higher power rating, some antennas can go out of specification because of general wear and tear. Such antennas require regular checking.

Calibration of Dipole Antennas

CISPR 16-1:1999 is the international standard specifying apparatus and methods for measuring radio disturbance. Clause 5.5.4.1.1 describes the dipole antenna. This is a resonant antenna at frequencies between 80 and 1000 MHz, but for frequencies below 80 MHz, the fixed-length 80-MHz dipole is used as a short dipole to cover the frequency range 30–79 MHz. This avoids the problems caused by using a 30-MHz resonant dipole on a 3-m test site, since the dipole is 4.8 m long.

An important consideration is that the coupling of the short dipole to the ground plane has less effect on the antenna factor than for a resonant dipole. This is because the dipole self-impedance increases rapidly as the frequency is decreased. This predominates over the mutual impedance with its image. For example, the antenna factor of a 30-MHz resonant dipole varies by up to 6 dB (depending on balun design) when the horizontally polarized antenna is scanned in height from 1 to 4 m, whereas, for the same height range, the antenna factor variation of an 80-MHz dipole used at 30 MHz is less than 0.1 dB. However, the antenna factor is high, which reduces the sensitivity of an EMC test.

CISPR 16-1 concentrates on methods of site validation but has little to say about methods of antenna calibration. Clause 5.6.6.2 refers to a method in Annex G. Clause G.5 simply gives the formula for free-space antenna factor of a tuned half-wave dipole. A footnote states that a calibration procedure is under consideration. Clause 5.13.1 states that the intention is to measure free-space antenna factor. Clause 5.13 defines the specification of a site suitable for calibrating antennas, but the method of measuring antenna factors has yet to be added. This is being gradually addressed by an ad hoc group of CISPR subcommittee A, and it is hoped there will be significant progress by 2004.

Another widely used document is ANSI C63.5:1998. Its intention is to define methods for measuring free-space antenna factors. However, it stipulates a method that bears similarity to the way antennas are used for emission testing over a ground plane. Vertical polarization is ruled out because of large measurement uncertainties, so horizontal polarization is used as described below.

Clause 4.3 considers the antenna factors as near free-space antenna factors, which is a change from the 1988 version of C63.5, bringing it into line with CISPR 16-1. However, free-space antenna factor can be assumed to be a basic parameter even in the 1988 version because the tuned dipole described as a reference antenna is accompanied by a formula with which free-space antenna factor is calculated. Clause 4.4 gives a choice of calibration methods, as follows.

The method widely used in the United States is the standard site method (SSM), a version of the three-antenna method. One antenna is placed at a height of 2 m above the ground plane and the other antenna is height scanned from 1 to 4 m. The antenna separation is 10 m. Clause 5.1 states that this is the international calibration measurement geometry and is the preferred geometry. The other geometries for use in the United States are a height of 1 m and a separation of 3 m, with heights of 1 and 2 m.

It is important to note that horizontally polarized resonant dipoles can couple strongly with their images in the ground plane, causing antenna factor variations of up to 7 dB (see Figure 1). This is in direct contrast to the ANSI statement in the fifth paragraph of clause 5.1 that "this measurement is relatively insensitive to site variations for horizontal polarization, and it yields free-space antenna factors even though the reflecting plane will not create a free-space environment during calibration." In practice, using broadband antennas, the variation does not exceed 2 dB, and it is shown that across most of the frequency band, the deviation from free-space antenna factor is within ±1 dB.⁸ Bear in mind that this height-related uncertainty has to be added to the uncertainty of the calibrated antenna factors. A welcome addition to EMC uncertainty considerations is CISPR 16-4.⁹

ANSI assumes that a resonant dipole antenna will be used below 80 MHz, in contrast to the short dipole specified in CISPR 16-1 (CISPR 16-1 takes precedence unless clear instructions are given to use tuned dipoles below 80 MHz). Although the property of a short dipole that makes AF insensitive to height above ground is desirable, the magnitude of AF rises sharply with reduction in frequency. For an 80-MHz dipole with a 100-Ω balun, AF at 30 MHz is 24 dB greater than the AF of a dipole resonant at 30 MHz. If the increase in AF can be tolerated, it is preferable to use the shorter dipole. The 80-MHz dipole (or better, a well-calibrated biconical antenna) is very desirable for vertically polarized measurements, because a 30-MHz resonant dipole cannot be placed at a height less than 2.5 m, and the signal maximum is below this height for vertical polarization.

The other recommended method is the reference antenna method, clause 6.² This could give significantly different results, but ANSI does not make a distinction. In this method, the antenna under test is placed at a height greater than 2.5 m above the ground plane and is calibrated by substitution with a reference antenna. A 30-MHz dipole calibrated at a height of 3.5 m by this method could differ in result by 5 dB compared with that calibrated at a height of 1 m by SSM.¹⁰ This assumes that the antenna factor of the reference antenna, including coupling to its image, is known. However, ANSI assumes that the antenna factor is its free-space value calculated by a formula. It is notable that ANSI C63.5 does not include the use of a quasi-free-space test site.

Free-space antenna factor, AF_FS, is specified in the relevant standards. If a horizontally polarized resonant dipole is more than 1.4 wavelengths in height above the ground plane, the antenna factor, AF, will deviate from AF_FS by less than ±0.4 dB due to mutual coupling with the ground plane. The maximum height recommended for antenna calibration is 4 m, because most laboratories have 4-m-high masts used for EMC height scanning. If a resonant dipole is 4 m above the ground plane, the deviation from AF_FS could be more than ±0.4 dB for frequencies below 105 MHz. A deviation of ±0.4 dB should probably not be exceeded if the total uncertainty in measured AF is to be less than ±1 dB.

Because the deviation of AF from AF_FS decays sinusoidally with increase in height, there will be a height relatively near to the ground plane at which AF = AF_FS (see Figure 1). This height is approximately 0.2l, which is 2 m for a 30 MHz dipole, but this is on a steep gradient and the uncertainty may be greater than ±0.4 dB.

The next height is 0.5l, which would be better to use. However, although it may be time-consuming to adjust the height for each frequency to achieve AF_FS, this method is relatively inexpensive to achieve low uncertainties compared with building a quasi-free-space site; the exact height depends on the dipole impedance and the balun to which it is connected. But the error will be less than ±0.4 dB if a height of 0.5l is used, so this could be a useful method for obtaining AF_FS.

Figure 1. Deviation in decibels of antenna factor from free-space antenna factor with height above ground plane in wavelengths, of a horizontally polarized resonant dipole.

Figure 1 is calculated for the dipole design given in Annex C of ANSI C63.5. The balun uses 50-Ω transmission lines. The maximum deviation of AF is 7.03 dB at 30 MHz. However, for the 100-Ω balun design given in Annex S of CISPR 16-1, the maximum deviation at 30 MHz is 3.65 dB.

Recommended Methods for Measurement of Antenna Factors of a Tuned Dipole Antenna Used as a Reference for Field-Strength Measurements. First the decision needs to be made whether the short dipole is to be used below 80 MHz. This may not be desirable for EMC testing at a 10-m distance because of the increase of 24 dB in antenna factor at 30 MHz. However, it may be acceptable for EMC testing at 3 m because there should be an increase of 9.5 dB in signal strength. It is desirable to use the short dipole on a 3-m site because there will be Fresnel errors in using a 4.8-m long dipole antenna to measure the field from a product 3 m away.

It is recommended that the reference antenna method be used to measure the free-space antenna factors of resonant dipoles in the frequency range of 30 to 1000 MHz. These antenna factors are valid for measuring EMC emissions at a distance of 10 m or greater, and in the frequency range of 80 to 1000 MHz at a distance of 3 m. Between 30 and 79 MHz, it is recommended that the free-space antenna factor of the short dipole be measured; this would be used for EMC measurements at a range of 3 m. The short dipole is defined as having a half-wave resonant length at 80 MHz.

Ideally AF_FS is measured on a quasi-free-space site. In practice, this is not economical to achieve for omnidirectional antennas below about 150 MHz. However, by understanding the behavior of the antenna it is possible to make accurate measurements of AF_FS above a ground plane. The aim of this procedure is to achieve an uncertainty of less than ±1 dB for the measurement of free-space antenna factor. To this end, it is recommended that the antenna be placed at least 1.4l above the ground plane. For antennas above 200 MHz, a height of 2 m will suffice, being an accessible height for adjusting the dipoles. As the frequency is reduced, the height will be increased to >= 1.4l until a height of 5.25 m is reached, i.e., at 80 MHz. Below 80 MHz, a height of 0.5l can be used because the AF is equal to AF_FS at this height within an estimated uncertainty of ±0.4 dB. For an 80-MHz dipole used as a short dipole, a frequency-stepped measurement can be made with the dipole at a fixed height of 5.25 m.

The ANSI procedure specifies only a height of greater than 2.5 m for the above measurements. ANSI suggests that the height is adjusted up to 4 m in order to be near a signal maximum, in particular to avoid a null reading. If this method is adhered to, the deviation from AF_FS could be much larger than ±0.4 dB at frequencies below 200 MHz, and it will not be possible to achieve the aim of ±1 dB overall uncertainty (except by chance at some frequencies where the DAF crosses the zero line in Figure 1).

Recommended Methods for Measurement of Antenna Factors of Broadband Antennas for Field-Strength Measurements. Broadband antennas include biconical, log-periodic dipole arrays (LPDAs), and biconical-log hybrid antennas. The National Physical Laboratory (NPL; Teddington, UK) measures the free-space antenna factor of LPDA antennas by creating a quasi-free-space site and using the three-antenna method.

First, the lengths of the LPDA elements are measured along with their distances from the tip of the antenna. These data are used to estimate the phase center of the antenna, which is known to be better than ±0.05 m for frequencies above 150 MHz. The antennas are placed at a height of 2.5 m above the ground plane and with a separation of 2.5 m between their 300-MHz elements. The area on the ground between the antennas is covered with 1-m-high pyramidal absorbers. Alternatively, the measurements are performed in a fully anechoic chamber that has been shown to give similar or lower uncertainties of measurement.

NPL measures the free-space antenna factor of biconical antennas by supporting them vertically polarized at a height of 2 m above the ground plane, because the coupling to the ground plane is negligible at this height. The operator should evaluate the uncertainties caused by the amplitude taper in the incident wavefront.

The biconical antennas are calibrated by the standard antenna method with the site configured as a ground reflection range. The standard antenna is the broadband calculable dipole antenna, that has been demonstrated to be highly accurate.^11–13 Where it is necessary to obtain the AF of a biconical antenna horizontally polarized above a ground plane, a fixed-height variant of the three-antenna method is used, or the standard antenna method is used. The AF of a biconical can vary by 1.8 dB with height.⁸ AF_FS is approximately the mean of the AF over a range of heights and, therefore, AF_FS is the value that overall gives the lowest uncertainty for a height-scanned measurement.

The two methods described above measure biconical-log hybrid antennas as if they were separate biconical and LPDA antennas. The two data sets are knitted together at the appropriate overlap frequency. The methods described in this section give a very good approximation of free-space conditions and can be recommended for measuring the free-space antenna factor of broadband antennas—these methods are relatively straightforward to implement.

The ANSI C63.5 method involves height scanning the antenna above a conducting ground plane, using a 10-m separation and with the fixed antenna at a height of 2 m. An experienced operator with the best site and test equipment should be able to measure antenna factor to within ±0.5 dB of the antenna factor measured by the quasi-free-space methods just described.

For a 3-m separation, the results differ significantly; however, it can be argued that because the ANSI method mimics the EMC test, the 3-m ANSI AF (AF_ANSI) is better. It probably takes care of some uncertainty terms, especially antenna height and directivity. Strictly speaking, however, the ANSI AF is only better for the EMC test when the product is at the same height and of a similar size to the stationary antenna in the ANSI calibration. For the more general product of different size and height, there is probably no advantage in using AF_ANSI over AF_FS.

Another advantage of using AF_FS is that it can be simpler to measure. The fact that height scanning is not required means the uncertainties can be lower. The appeal of using AF_FS is that it covers a wide variety of measurement conditions and is a good baseline from which to build antenna-related uncertainties.

Recommended Methods for Calibration of Tuned Dipole Antennas for Use in the Evaluation of Test Sites. It is normal to evaluate test sites using both horizontal and vertical polarization (HP and VP). It is also normal practice to use broadband antennas so that narrowband features of the site can be shown up in a swept frequency plot. Some users, however, may prefer to evaluate their site using tuned dipole antennas. These antennas will more fully illuminate the site because their directivity is 2.15 dBi compared with the 6 dBi typical of an LPDA antenna.

It is recommended that the SA between a pair of dipole antennas be measured on a reference site, giving SA_R. The same pair of dipoles is set up with the same measurement geometry on the customer's site and SA_C is measured on-site. The difference, SA_C – SA_R, is compared with the criterion for NSA that is specified for the site. The measurements are performed in both horizontal and vertical polarities with the antennas at fixed heights. The antenna heights are chosen to avoid signal nulls caused by the destructive interference of the direct and reflected rays. It is recommended that an 80-MHz short dipole be used instead of tuned dipoles below 80 MHz for both 3- and 10-m ranges.

Note that at 30 MHz on a 10-m site, the transmit signal strength may be greater than permitted by the regulatory authority, necessitating the use of resonant dipoles. The most-used NSA criterion is ±4 dB given in Annex L of CISPR 16-1. By adopting this procedure, the fewest uncertainty terms are involved and the uncertainty of SA on the customer's site relies on the uncertainty of SA on the reference site. The NPL ground plane contributes less than ±0.1 dB to the uncertainty budget in the frequency range of 30 MHz to 600 MHz.¹¹ There is a small additional uncertainty to allow for reflections from the antenna support and the coaxial cable. The uncertainty allowed for the receiver or tracking generator is further reduced if the same equipment is used on both sites.

Recommended Methods for Calibration of Broadband Antennas for Use in the Evaluation of Test Sites. It is recommended that the SA between a pair of broadband antennas be measured on a reference site, giving SA_R. The same pair of antennas is set up with the same measurement geometry on the customer's site and SA_C is measured. The difference, SA_C – SA_R, is compared with the criterion for NSA that is specified for the site. The measurements are performed in both horizontal and vertical polarities. The transmit antenna is set at a height of 1 m (to simulate the typical height of a product in an emission test), unless the customer requests alternate heights. The receive antenna is scanned in height from 1 to 4 m and the maximum SA is recorded. Above 200 MHz, it is only necessary to scan from 1 to 2.5 m because the maxima are below 2.5 m. It is not essential to height-scan the antenna, but for fixed heights, a second measurement may be required with one antenna changed to fill in any nulls.

Debate on the CISPR Reference Dipole

For a number of years, CISPR/A has been mulling over the status of the CISPR reference antenna. Since at least 1987, the reference antenna for electric field strength in the frequency range of 30 to 1000 MHz has been the tuned dipole antenna. The 1999 version of CISPR 16-1 has the tuned dipole as the reference between 80 and 300 MHz. The dipole tuned to 80 MHz is used as a fixed length from 30 to 80 MHz (the shortened dipole). From 300 MHz to 1 GHz, one of two antennas can be used: the tuned dipole or a broadband antenna that meets the beam width characteristics of Clause 5.5.5.2.

In June 2001, a questionnaire was circulated among the working-group members of CISPR subcommittee A, which put forward options for the future reference antenna. The final version of this questionnaire is currently out for comments, due to be returned by national committees by September 2003. The rationale for this questionnaire was to invite experts to make a choice between two fundamental approaches for the ultimate arbiter as to whether an emission has exceeded the specified limit.

One approach presented was to continue with the CISPR reference antenna, which involves a standard design of antenna used to make a measurement in a particular way. In this method, it no longer seems to matter whether the antenna factor is affected by the ground plane. It also does not matter whether the antenna is at an excessive height (in the case of a vertically polarized tuned dipole). Rather, what matters is that everyone does the measurement in exactly the same way (i.e., the measurement is reproducible). In this method, it is not necessary to know the antenna factor (knowing the voltage output of the antenna is sufficient) to be compared with a voltage limit based on the reference antenna design. The voltage limit does not yet exist but could be derived from the existing field-strength limit. For at least the past 15 years, the CISPR reference antenna has been the balanced dipole, and it has been assumed that the purpose of this is to measure field strength. The rules are being changed if it is now suggested that the output voltage is sufficient, because the relationship of voltage to field strength depends on the antenna impedance, which in turn depends on coupling of the antenna to its surroundings.

The other approach presented was to use a calibrated antenna to measure the emitted field strength, which could be directly compared with the field-strength limit already specified in the standard. The antenna factor is traceable to national standards (in the normal way for physical parameters). This approach allows users their choice of antenna, but CISPR has to constrain the radiation pattern if the method of emission testing involves a ground plane, because the two ray paths to the EUT are not necessarily in the antenna bore-sight direction. Such constraint is not necessary for testing in fully anechoic rooms because very limited height scan, if any, is involved. Antenna-related uncertainties, such as the effect of the ground plane, are dealt with in a transparent way. Their magnitudes are no greater than the uncertainties associated with using the reference antenna.

Regarding the reference antenna approach, there is the problem with making the method too prescriptive. If the ultimate arbiter is not the actual field strength emitted by the device, but rather the voltage output of an artifact used in a particular way, this will make it more difficult to correlate the result with alternative methods, such as in a FAR, a gigahertz transverse electromagnetic (GTEM) cell, or a reverberation chamber. The tuned dipole is rarely deployed because the length has to be adjusted for every frequency. The other problem is that there can be significant coupling with the ground plane. The main argument used for this method is that the measurement is reproducible (and the antenna is simple to construct). However, the intention is still to convert the output voltage to a field strength level and, until the advent of CISPR 16-4, no account was taken of the effect of the environment, which can typically be 2 dB, on the antenna. A more suitable broadband standardized antenna could be introduced; however, this would require that all laboratories buy another antenna.

The calibrated-antenna approach has presented other difficulties. In the past, there was a problem that calibration was not always reliable. Over the last 15 years, this has totally changed. Most antenna calibration laboratories and antenna manufacturers offer calibrations to uncertainties of less than ±1 dB. This means that test laboratories are free to choose their antennas from the market.

In actual fact, the most common antennas are based on designs so similar, such as the biconical elements based on MIL-STD 461, that these antennas can be considered already standardized. In addition, LPDA antennas made by different manufacturers have a similar footprint. The hybrid designs that cover the whole frequency range of 30 to 1000 MHz are causing some concern, but limits of length and radiation pattern can be mandated. This means that concerns over measurement reproducibility have been met, because the errors in using this approach are no greater than the errors using the reference antenna approach. An advantage of measuring the field strength is that it can be more easily related to the field strength measured by alternative methods, and it is a more useful quantity when considering the undesirable EMI influence on the performance of nearby electronic products.

In summary, the use of a calibrated antenna would treat the antenna like most equipment, which has to be traceably calibrated. Antenna factor uncertainties of better than ±1 dB are routinely achieved. However, EMC measurements are notorious for their large uncertainties—±10 dB not being atypical—and there is a body of opinion that the ultimate arbiter of an emission test is to perform it in a prescribed way using prescribed antennas. Currently, these are a short dipole up to 80 MHz, a tuned dipole up to 300 MHz, and a broadband antenna up to 1 GHz. Even a calibrated antenna can be used in a prescribed way, if necessary: it must be judged to determine whether the use of a calibrated antenna is sufficient to ensure reproducibility or whether the antenna type has to be mandated.

References

01. CISPR Publication 16, "Specification for Radio Disturbance and Immunity Measuring Apparatus and Methods, Part 1: 1999 Apparatus," International Electrotechnical Commission (IEC), Geneva.

02. ANSI C63.5:1998, "American National Standard for Electromagnetic Compatibility—Radiated Emission Measurements in Electromagnetic Interference (EMI) Control—Calibration of Antennas (9 kHz to 40 GHz)," (The methods of antenna calibration are unchanged from C63.5:1988), ANSI, Washington, DC.

03. CENELEC Report R210-010, "Electromagnetic Compatibility—Emission Measurements in Fully Anechoic Chambers," European Committee for Electrotechnical Standardization, Brussels, June 2002.

04. W Müllner and H Garn, "From NSA to Site-Reference Method for EMC Test Site Validation," in Proceedings of the IEEE EMC Symposium, (Montreal: IEEE EMC Society, 2001), 948.

05. MD Foegelle, "Site Validation Theory 101: Techniques and Methods," Compliance Engineering 17, no. 5, (2000): 42–53.

06. MJ Alexander, "Using Antennas to Measure the Strength of Electric Fields Near Equipment," Compliance Engineering Annual Reference Guide 17, no. 3, (2000) 66–83.

07. Def. Stan 59-41 Part 5, "Performance Specification for Specialized EMC Test Equipment," Available on the Internet: http://www.dstan.mod.uk. Select "Defence Standards," select "59," select "59-41," then find "Part 5 Section 5."

08. MJ Alexander et al., "Getting the Best Out of Biconical Antennas for Emission Measurements and Test Site Evaluation," in Proceedings of the IEEE EMC Symposium, (Austin, TX: IEEE EMC Society, 1997), 84.

09. CISPR 16-4 (2002-05), "Specification for Radio Disturbance and Immunity Measuring Apparatus and Methods, Part 4: Uncertainty in EMC Measurements," IEC, Geneva.

10. MJ Alexander and MJ Salter, "Corrections to Antenna Factors of Resonant Dipole Antennas Used Over a Ground Plane," NPL Report DES 131, November 1993, National Physical Laboratory,
Teddington, UK.

11. MJ Alexander et al., "Broadband Calculable Dipole Reference Antennas, IEEE EMC Transactions 44, no. 1, (2002): 45–58. (Recently, the upper frequency has been extended to 2 GHz. See RF & Microwave News 15. Available on the Internet: http://www.npl.co.uk/electromagnetic/cem-publications/#newsletters.

12. MJ Alexander, "European Intercomparison of Antenna Factors in the Frequency Range 30 MHz to 1 GHz, IEE Proceedings—Science, Measurement and Technology 143, no. 4 (1996), 229–240.

13. MJ Alexander et al., "International Comparison CCEM.RF-K7.b.F of Antenna Factors in the Frequency range 30 MHz to 1 GHz," Metrologia 39, no. 3 (2002): 309–317.

Martin J. Alexander is principal research scientist at National Physical Laboratory (Teddington, UK).