Pre-construction Evaluation Modeling of Open Area Test Sites (OATS)

Transcription

1 Preconstruction Evaluation Modeling of Open Area Test Sites (OATS) Bruce Archambeault IBM Raleigh, N.C. Introduction The construction of Open Area Test Sites (OATS) for commercial EM1 testing for FCC and CE radiated emission requirements is at an alltime high. The demand for test facilities to meet the amount of testing required for computer, consumer, medical, and other products has forced many test laboratories to turn away business due to a lack of test capacity. Semianechoic rooms are sometimes used to perform radiated emissions testing, but these sites are very expensive and usually limited to large companies doing internal product testing. Therefore the construction of OATS facilities is rising to try and fulfill the test capacity demand. The preferred location for an OATS is close to the product development areas, and yet in a wide open, RF quiet environment, These requirements are usually in conflict with each other, since a geographical location with a high concentration of product development activity tends to be located in close proximity to a highly developed area. Land prices are high, and the ability to have large open spaces around the OATS requires often excessive property purchases. As new OATS facilities are planned, the use of existing land is preferred, even if the CISPR 22 [I] recommended space and ground plane sizes are impossible. Once a desired location is identified, and the ground plane size and nearby metal structures are identified, some determination must be made to see if the proposed test site will meet the CISPR site attenuation requirements. Since it s often desirable (or necessary) to violate the CISPR ground plane and open space recommendations, some way to give a high degree of confidence the OATS will be acceptable without extra expense after construction is needed. This paper discusses some OATS analysis using numerical modeling techniques (in this case, the Method of Moments). Comparison between various ground plane sizes and shapes, and nearby metal objects, are made to a perfect site. The Perfect OATS The definition of a perfect OATS would include an infinite ground plane with no other metal structures (fences, posts, power lines, etc.). Of course, it is impossible to achieve such a perfect OATS in reallife, but a perfect OATS can be simulated using numerical modeling techniques. For this work, both the Method of Moments (MOM) and the FiniteDifference TimeDomain techniques were used. A small dipole antenna was created (the length of the antenna was very short compared to the frequencies of interest, 30 MHz to IGHz), and the receive location was scanned from a height of 1 to 4 meters, at a distance of 10 meters from the transmit antenna. The case of a perfect (infinite) ground plane was first compared to the case of no ground plane for horizontal polarization. The deviation between perfect and no ground plane for the MOM case is shown in Figure 1. The comparison was repeated using FDTD with cell sizes of 100 millimeters (which should give reasonable results to MHz). The FDTD comparison results are shown in Figure 2. In both cases, the distance between transmit and receive antennas was 10 meters, and the receive antenna was scanned over one to four meters for the maximum. As can be seen by comparing Figures 1 and 2, the two modeling techniques agreed to within one db over the frequency range of 30 MHz to about 800 MHz. e O /97/$

2 Figure 1 MOM Deviation Between Infinite and No Ground Plane a MOM Model Difference Between Infinite Ground Plane and No Ground Plane Horizontal Polarization Receive Location Scan 14 meters 2 g F : 0.m P 5 2 a \ Frequency (MHz) Figure 2 FDTD Deviation Between Infinite and No Ground Plane 6 FDTD Model Difference Between Infinite Grohd Plane and No Ground Plane Horizontal Polarization Receive Location Scan l4 meters OOE+07 1.OOE+06 Frequency (Hz) I.OOE

3 The perfect, infinite ground plane case became the normalized case. That is, all further results are shown as a deviation from this perfect OATS case. Although most of the results are shown for the horizontal polarization, the vertical polarization can be analyzed just as easily (and is used for the nearby metal conductor examples). MOM Ground Plane Mesh The size of the CISPR recommended ground plane for a 10 meter site is 20 meters long by 17.3 meters wide. If the entire ground plane is to be enclosed in a weather proof enclosure, the size of the enclosure will be large, and therefore expensive. Space limitations will also sometimes require smaller ground planes. Typically, an opening size of about l/l O* wavelength (at the highest frequency) is used in wire mesh applications. In this case, at 1 GHz, the wavelength is.333 meters, resulting in a recommended mesh size of.033 meters. Such a mesh size over the full ground plane would result in approximately 650,000 wire segments; clearly too much for a typical MOM model to complete in a reasonable amount of time, even on a fast workstation, Therefore the initial analysis was performed with much larger mesh sizes, and the size reduced until the results converged reasonably well with the perfect OATS case. The final selected size was % meter meshes. This size is slightly larger than 10 times the typical MOM ruleofthumb, but it was expected that any negative effects would only exist at high frequencies (and be obvious). Figure. 3 shows the deviation between the perfect ground plane and various wire mesh ground planes. The mesh #6 case (l/2 meter) shows results easily within the +/ 4 db allowed by the normalized site attenuation requirements Figure 3 Effect of MOM Mesh Density on EField 25 Electric field Deviation from Perfect Ground Plane Case for Various Mesh Densities Horizontal Polarization mesh #2 mesh #3 mesh #4 mesh $5 mesh#6 I Frequency(MHz) 464

4 FDTD Modeling of Ground Plane Size convenient to use. The deviation from a Although either MOM or FDTD could be perfect ground plane for both the horizontal used to model the reduced ground plane and vertical cases are shown in Figures 4 and sizes, FDTD was considered more 5 respectively. Figure 4 FDTD Model with Various Ground Plane Sizes 4.5 m 4 FDTD Model of Deviation from Perfect Ground Plane Case Horizontal Polarization OxlOm 15x6m l.ooe+06 l.ooe+07 Frequency (Hz) l.ooe+06 l.ooec09 Nearby Conductors Surrounding Walls The presence on metal conductors in the Since it is often desirable to put a weathernearby vicinity of an OATS is sometimes a proof structure around an OATS to allow fact of life that is unavoidable. For example, yearround use, the question of how much a number of cases could be used to determine conductivity can be allowed in the walls the effect of allowing a metal fence either often arises. The FDTD method is alongside the OATS or behind the receive particularly useful for materials with limited antenna at various distances. An example is conductivity or even a relative dielectric shown here of a simulation of a metal constant of other than one. Figure 7 shows light/utility post at two different distances the results of using FDTD to model an from the EUT side of the OATS ground OATS with external walls that were partially plane. As shown in Figure 6, the effects of conductive. As can be seen from this figure, the different setback distances for the posts the walls had a definite effect at lower are clear. frequencies. 465

5 Figure 5 Effect of Various Ground Plane Sizes on FDTD EFields FDTD Model of Deviation from Perfect Ground Plane Case Vertical Polarization i.ooe+o7 l.ooe+ob Frequency (Hz) l.ooe+09 Figure 6 Effect on MOM Model for NearBy Metal Post MOM Model of Deviation from Perfect Ground Plane Case Due to Nearby Metal Post Vertical Polarization Frequency (MHz)

6 Figure 7 Effect on FDTD EFields from Surrounding Walls FDTD Model of Deviation from Perfect Ground Plane Case for Partially Conducting Walls 6 1.OOE+07 1.ooE+08 Frequency (Hz) 1.OOE+09 Summary The design of OATS facilities sometimes require a deviation from the recommended ground plane sizes or nearby conductor spacing. Since such deviations can result in serious cost penalties if the OATS can not pass the normalized site attenuation certification tests after construction, normal practice is to avoid any deviation even though it will result in extra cost to the facility during construction. However, numerical modeling techniques have been shown to be able to help analyze these nonstandard OATS designs and provide engineers with a risk assessment in terms of expected normalized site attenuation error vs cost of design options. The various design parameters which can be analyzed include ground plane size, ground plane shape, and distance to nearby conductors, such as fences, metal poles, etc. Both FDTD and MOM have been shown to be useful, depending upon the type of problem. MOM requires a wire frame mesh (or a surface patch plane) which must be fine enough to be useful at high frequencies. MOM is particularly useful for problems involving long wires, poles, etc. FDTD requires a fine grid, again for high frequency accuracy, and it is especially useful for problems involving partially conducting walls, intervening dielectrics, etc. Although the analysis shown here was limited in frequency to about MHz, this could be easily extended by finer wire grids and cell sizes. Since practical OATS calibration has shown most of the problems due to site construction to be apparent at low frequencies rather than high frequencies, finer mesh models may not be necessary. References [l] CISPR22 (1993) Limits and Methods of Measurement of Radio Disturbance Characteristics of Information Technology Equipment, IEC 467