Antenna Calibration for EMC Tests and Measurements

A GTD model is proposed to discuss the normalized site attenuation (NSA) associated with an open area test site (OATS) with finite metallic ground plane. In this study, the NSA values for three OATS sites, based on GTD model and measurement, are carefully examined and agreement between calculated and measured results supports the usefulness of the proposed GTD model. The GTD model is then adopted to analyze the deviation in NSA values for several OATS structures. The GTD simulation has revealed that the key factor affecting NSA is the width of rectangle metallic plane which is essential in reducing the NSA deviation under the finite test site condition.

This paper summarizes a recent investigation into the measurement and modeling of antenna factor as a function of separation distance. Measurements were made of coupling and antenna factor (AF) for a standard gain dipole as a function of distance from an identical source. Coupling and AF were also computed using two electromagnetic computer codes to compare the numerical results with each other and with the measurements. The coupling measurements were made in an anechoic chamber using two identical standard gain dipoles at 300 MHz, both with and without a ground plane. In the case with a ground plane, both horizontal and vertical polarizations were examined. Distances were varied to cover from the very near-field to the far-field of the antenna. Measured values of the AF at 300 MHz were frequency scaled to predict the AF at 30 MHz. In addition to the measurements, the Numerical Electromagnetics Code (NEC) Method of Moments and the General Electromagnetic Model for the Analysis of Complex Systems (GEMACS) were used to calculate coupling and AF values for direct comparison with the measured values. >

The complex fit normalized site attenuation (CFNSA) method was introduced in 2000 as an enhancement to the normalized site attenuation (NSA) model used in ANSI C63.4 and C63.5. The simple dipole NSA model is insufficient to predict the behavior of complex antennas like log periodic dipole arrays (LPDAs) and horns. Thus, NSA measurements made with these antennas using free-space antenna factors can have significant deviations from that predicted by the model. The CFNSA model is a modification to the simple dipole NSA model to account for the variations in antenna pattern and phase center that are possible in complex antennas. This paper evaluates the effectiveness of the phase center/pattern matching (PCPM) method previously proposed, as well as the use of directly measured pattern and phase center information, for a variety of antenna types.

For decades, site validation measurements below 1 GHz have been framed within the concept of Normalized Site Attenuation (NSA) proposed by Smith et al. in 1982. Conceptually, measured results are compared to a theoretical model, which is derived based on a raytracing model between two antennas over a conducting ground plane. It was quickly realized that the theoretical model cannot fully encompass the complexity of real antennas in geometries needed for site validation, and the individual antenna effects cannot be “normalized” entirely from the site attenuations by a simple subtraction (in dB) of the antenna factors. For dipole antennas, correction factors have been included to amend the errors (derived from numerical simulations). For broadband antennas, there had been no simple solution until Geometry-Specific Correction Factor (GSCF) was introduced in ANSI C63.5 in 2000. It may not have been immediately clear to users of the standards that since the introduction of GSCF, the Smith model NSA has been essentially bypassed. Of the two prescribed methods for site validations, the biconical GSCF provided in the form of correction values is based on a numerical model, while the measurement GSCF is a site-to-site comparison method. ANSI C63 can benefit significantly in clarity by removing the pretence of a simple theoretical dipole NSA model, and rewrite the site validation requirements directly in terms of either the numerical or the site-to-site comparison methods.

Complex Fit Normalized Site Attenuation (CFNSA) is an improved theoretical model for transmission between broadband antennas over a conducting ground plane. The CFNSA model overcomes the limitations of the original NSA model proposed by Smith, German and Pate, by allowing variable radiation pattern and variable phase center positions as a function frequency. The CFNSA model is well suited for log periodic dipole arrays (LPDAs) from 200 MHz to 1000 MHz because the variable phase centers and antenna pattern deviations relative to dipoles are the dominant source of errors. CFNSA has been shown to be effective to reduce systematic errors in the theoretical NSA model for LPDAs by up to 1 dB per antenna factor (or 2 dB in a site attenuation measurement). Antenna calibration and site validations above 1 GHz are increasingly addressed by using fully anechoic rooms (FAR). This makes earlier efforts to improve the CFNSA model above 1 GHz to be of limited application at present. This paper concentrates on improving the model for below 1 GHz. In this study, the antenna patterns are shown to be better represented by a piecewise function versus radiation angle, as compared to the earlier proposed functions, such as cosn θ, or polynomial functions. Efforts are also made to clarify the mathematical process, so that CFNSA can be adopted by the antenna calibration standards, e.g. ANSI C63.5.

This paper discusses the following topics: 1) methods of calibration of dipole-like VHF antennas in the frequency range 30 Hz to 3 GHz; 2) Properties of antennas: antenna factor (gain), return loss, cross-polar rejection, balun imbalance; 3) Calibration of antennas in free-space and above a ground plane (i.e. including mutual coupling); 4) Traceability of antenna factor to national standards, with uncertainty budget; and 5) problems encountered with antennas for EMC testing. (11 pages)

This paper describes an analysis methodology that implements the theoretical normalized site attenuation (NSA) model into a standard electronic spreadsheet program, such as Microsoft Excel, that can calculate and plot the exact theoretical NSA value for any given frequency of interest. The theoretical NSA is referenced in ANSIC63.4(1992) and is discussed in numerous papers in the open literature. The mathematical model for the theoretical NSA consists of a set of complex equations for horizontal and vertical polarization and is best suited for calculation via computer automation. The spreadsheet analysis technique addressed in this paper has several advantages: (1) the popularity of spreadsheets today provides an easy to use software platform that can be utilized to fully automate the complex theoretical NSA calculations,( 2) the spreadsheet provides an open framework whereby all areas of the model are open for visual inspection, and (3) the spreadsheet model can be used for "what-if" scenarios whereby the test site input parameters are easily modified and the calculations re-pun for new test site configurations.

An efficient method to analyze the electromagnetic field inside rectangular anechoic and semianechoic chambers from 30 to 100 MHz is shown. A modal expansion is employed to develop the homogeneous and the inhomogeneous field. This is combined with circuital techniques to compute the effect of the layers of ferrite and dielectrics in the walls of the chamber. Several numerical results of the Free Space Transmission Loss (FSTL) of an anechoic chamber and the Normalized Site Attenuation (NSA) of a semianechoic chamber are provided.

The existence of differences between different open area test sites (OATS) in antenna calibration and electromagnetic compatibility (EMC) compliance measurements is still a real issue. To minimize the differences, it is necessary to analyze the influencing factors. To examine the influence of the ground plane geometry on the site attenuation, a comparison of different open area test site geometries and sizes has been carried out. The investigation has been performed numerically with the method of moments in the frequency range 30-130 MHz for test distances of 3 m and 10 m. The deviation of the site attenuation (with receiving antenna height) in relation to the ideal test site has been calculated. The minimal plane, for example, which is suggested in the standards, shows a high deviation. Furthermore, the significant influence of on and off axis measurements is shown and the reasons are explained.

For the validation of an open area test site usually the normalized site attenuation (NSA) is considered. However, the model for the calculation of the NSA in the various standards is a point dipole. While this may be a correct assumption for biconical antennas or reference dipoles with a well-defined phase center, it is wrong for the antennas used in an automotive antenna test setup. Measurements of automotive antennas for radio and TV reception generally are done on a dedicated test range consisting of a turntable that can bear the weight of the vehicle and a test antenna that is placed in a considerable distance in order to ensure far field conditions. For broadband measurements often log-periodic antennas are used. In these antennas the phase center is not known exactly for all frequencies. The phase center of the automotive antenna is not known either. This uncertainty in the position of the phase centers in combination with the ground reflection occurring between object under test and test antenna results in a measurement error that cannot be neglected. In this work we show how this error can easily be calculated for all test site geometries and give an exemplary calculation for the test site used at our company.

An open test site located at the Telecommunication Lab in Taiwan is characterized by measuring normalized site attenuation (NSA) to determine the suitability of this open test site for antenna measurement. The measured NSA data are compared to the theoretical values. The results show a maximum deviation of less than 3 dB over the frequency range of 30 to 1000 MHz.

A new method for systematic correlation of measurements in semi-anechoic and anechoic chambers is presented. A signal measured in a semi-anechoic chamber is first processed using the matrix-pencil method which is applied sequentially on small frequency intervals. On each interval the measured signal is decomposed into its propagating wave components. The component corresponding to the wave reflected from the ground is identified and removed to fully retrieve the signal measured in an anechoic chamber. Three examples using log-periodic and Vivaldi antennas in different anechoic and semi-anechoic chambers for varying frequency ranges illustrate the method. The proposed decomposition algorithm can be utilized to characterize measurement test site imperfections.

In January 1993 an intercomparison exercise began in which 15 laboratories from nine European countries participated in the measurement of six antennas. The antenna factor and reflection coefficient were measured at a selection of frequencies in the range 30 MHz to 1 GHz. The project was sponsored by the European Commission under its BCR programme, with NPL as the pilot laboratory. The main purpose of the intercomparison was to find out whether national laboratories could calibrate EMC antennas to an uncertainty of better than /spl plusmn/1 dB at a confidence level of /spl ges/95%. The results showed that an agreement of /spl plusmn/1 dB was a reasonable target. There were 38 mandatory antenna factor measurements. All of the results of four laboratories and 92% of the results of three more laboratories were within /spl plusmn/1 dB of the averages. More than 95% of the results of three other laboratories were within /spl plusmn/1.5 dB. With the experience and information gained from participation in this intercomparison most of the laboratories should be capable of meeting the target of /spl plusmn/1 dB. Uncertainties of better than /spl plusmn/0.5 dB have been achieved by the pilot laboratory. This is a realistic future target for laboratories that offer services for the calibration of EMC antennas.

Testing electronic equipment for radiated emissions requires the accurate calibration of antennas. This in turn entails the use of a high-quality measurement site. The design of an open-field site for the calibration of antennas in the frequency range 30-1000 MHz is described. An antenna that is used as a primary standard must have a calculable antenna factor since there are uncertainties in quantifying the quality of the site and hence of setting up a known field. The design of a standard dipole antenna and the S-parameter characterization of its balun/matching network is described, followed by methods for calibrating the antenna factor of antennas.

In the ANSI C63.4 development of normalized site attenuation (NSA) for vertical polarization the E-field sine pattern generated by short dipoles was used to account for the additional attenuation that occurs as a function of the antenna patterns. A pattern dependent NSA equation is developed incorporating antenna pattern factors for any types of antennas for which mathematical expressions of the patterns exist. For resonant dipoles, biconicals and log-periodics the decibel differences from the existing standard have been found to be small, leading to increased confidence in the utility of the present standard.