Whether they own a timeshare apartment or a modest cottage near the beach, many people enjoy the change of scene that a second residence offers. Others may rent a small place for rest and relaxation during the vacation season. In either case, the anticipation of new experiences in a different location justifies making the trip.
EMC engineers also need to get away from it all, but for them it’s a matter of necessity. Traditionally, open air test sites (OATS) have been used for radiated emissions tests. However, the time spent traveling to a remote, low-noise OATS and the chance that a day’s testing may be rained out have encouraged engineers to find a better alternative.
According to James Press, chief EMC scientist at National Technical Systems, “Many years ago, international committees such as IEC, ANSI, and CISPR dictated that the OATS measurement was the standard. However, recent developments in ferrite technology have made 3-meter chambers quite cost-effective. Even the 10-meter chamber now is viable for companies with deep pockets, and a few are presently in operation.
“As the technology moves further in this direction, the OATS eventually will be retired,” he continued. “Until that time, an OATS is necessary for larger equipment where the antenna distance of 3 meters is not practical and the units may be too large to fit in a 3-meter chamber.”1
What options do you have if your company doesn’t really need the space afforded by a large chamber and requires only occasional use of a test site? Have you considered buying one of the many transverse electromagnetic (TEM) or gigahertz TEM (GTEM) waveguides or small shielded chambers now available? It could be the answer to your problem.
Alternative Test Cells
TEM cells, sometimes referred to as Crawford cells, are two-port, rectangular cross-section, 50-W transmission lines (Figure 1). There is sufficient space between the inner conductor, termed the septum, and the outer conductor to insert a DUT for EMC testing. However, most manufacturers recommend maximum DUT dimensions of one-third this distance and only one-third the cell width. This restriction means that a TEM cell with a 1-meter square center section cannot accommodate a DUT larger than about 165 mm × 330 mm in cross section.
Internally, the 377-W field impedance of the center section supports TEM mode operation, but TEM cells are not perfect. Without modification, a basic TEM cell exhibits multiple waveguide modes at high frequencies, partially because of its parallel surfaces. A conservative cutoff frequency is defined as corresponding to a wavelength 50% greater than the height of the cell. For a cell height of 437 mm, a first resonance appears at 450 MHz or l = 660 mm, as reported by Lorch and Mönich.2 This means that only small TEM cells can be used at high frequencies, obviously limiting the size of the DUT.
Above 450 MHz, Lorch and Mönich observed that higher-order modes of transmission were excited. To extend the frequency range of TEM cells, these modes must be suppressed without adversely affecting field uniformity.
By modifying the septum to include absorber material, Lorch and Mönich raised the first resonance to 670 MHz. Wall absorbers alone extended the useful range to about 1 GHz, but with attendant amplitude ripple of up to ±6 dB. Because some transmission modes exhibit zero amplitude at the walls, wall-mounted absorbers are limited in the improvement they can achieve.
Other modifications to Crawford TEM cells have been undertaken commercially. Wayne Kerr’s TEMCell has two septums and is used in a variety of industries for pre-, post-, and compliance testing to the IEC1000-4-3 standard—26 MHz to 1 GHz at 10 V/m with over-testing to 20 V/m. “The automotive industry is requiring much higher V/m on the order of 100 to 200 V/m,” said Owen Wiseman, the U.S. regional sales manager at Wayne Kerr. “With simple internal modifications to the design, our TEMCell can produce 100-V/m fields over smaller volumes.”
The dual-septum cell has been described by John Whitehouse, Ph.D., of the University of Reading.3 Basically, the cell is derived from the early work of Crawford and Workman in which they offset the septum toward the bottom of a conventional TEM cell to increase the usable DUT test volume. The KeyTek G-Strip™ and the Schaffner-Chase emCell also have a single, offset septum.
Dr. Whitehouse’s improvement adds another septum toward the top of the cell to create a symmetrical design. That way, the field is zero along the central plane of symmetry, so the problem of accounting for the effects of cabling attached to the DUT can be alleviated. However, the cell must be driven differentially and exhibits a 25-W impedance.
Separately, Fischer Custom Communications has extended the usable range of a small, conventional TEM cell to 2 GHz. Joe Fischer, chief engineer, said, “Instead of impedance matching right at the terminations, we recognized that there’s a constant impedance transformation required along the entire flared section at each end of the cell. In our design, we use a tapered transmission line. We keep transforming the impedance as we go.” This cell is recommended by the SAE J1752-3 test specification for IC emission measurement, and the design has been patented.
Integrated circuit manufacturers now are being asked to design ICs that meet certain EMC specifications, according to Jim Muccioli, development engineer at Jastech. For many years, some ICs have used waveform shaping to control the slew rates of output signals. This technique avoids reflections, but it also reduces emissions because it limits high-frequency energy.
Emissions can increase as the result of a die shrink that reduces internal dimensions and capacitances of the IC. Even though the chip still may operate at the same clock rate, the signal rise and fall times will be faster.
To test for emissions, ICs are mounted on a 4″ × 4″ PCB with a ground plane on the top surface. Control signals and connectors are confined to the opposite side of the board. The TEM cell has an opening cut in it so that the IC can protrude into the cell while the PCB ground plane makes contact with spring finger stock surrounding the opening. In this way, the radiated and common-mode emissions from the IC can be measured without concern for contributions from control cables or additional support circuitry.
Most automotive IC testing is done at 1 GHz or below, but even a 2-GHz cutoff isn’t sufficient to allow more than the fourth harmonic to be measured from an IC running at 500 MHz. The solution adopted by Mr. Muccioli and the task force that originally addressed IC emissions measurement was to move to a GTEM cell, especially for fast telecomm devices.
Both TEM and GTEM cells are reciprocal devices; that is, they work equally well for emissions and immunity testing. The basic GTEM concept was invented by Dr.-Ing. Diethard Hansen while at ABB and is licensed to several manufacturers. GTEM cells also are known as wideband TEM cells. The work in which Mr. Muccioli participated resulted in definition of the Lindgren Omni GTEM that accepts 4″ × 4″ PCBs for IC testing.
“The advantages of GTEM over other TEM waveguides include better field predictability; a larger, established experience base; and a mathematically rigorous OATS correlation algorithm,” said Tim Harrington, senior electromagnetics engineer at EMC Test Systems. “Currently in the commercial electronics business, TEM waveguide testing still is used for product precompliance evaluation. A few GTEMs have been approved by the FCC for compliance testing on specific product classes.”
The construction of a GTEM is radically different from that of a TEM cell (Figure 2). Because there is no transition from a flared to a rectangular cross section, field- boundary conditions are much simpler. In contrast, Lorch and Mönich have shown that to satisfy boundary conditions near the flared-to-rectangular transition in a TEM cell at least one additional non-TEM mode is required.
Also, because the GTEM is a single-port device with both line and field terminations, few reflections are created, and field uniformity is good beyond 5 GHz. The upper frequency limit is not related to cell size, and you can achieve higher field strength by moving the DUT closer to the cell’s apex.
Those reflections that do travel toward the apex are damped by the tapered structure. The septum is positioned close to the ceiling and terminated in a distributed 50-W load. The open end of the cell is covered with absorbing material, such as foam pyramids, arranged to match the spherical shape of the impinging wavefront.
In one of many papers on the subject, Hansen and Ristau reported that improvements in the performance of the absorber material can produce better field uniformity. A combination of ferrite tiles and foam absorbers was suggested. In fact, any GTEM improvements are relative to a very good starting level of performance. Figures 3a and 3b show the field strength vs frequency for both a TEM and a GTEM cell.
The Eurotem antenna is Dr.-Ing Hansen’s latest development. It consists of four 200-W stripline antennas arranged along the edges of a GTEM-shaped volume. Because the elements are driven differentially in pairs, the E-field can be directed vertically or horizontally by switching connections.
As Dr.-Ing. Hansen commented, the single most important advantage of the Eurotem is that you don’t have to invert the DUT. It simply is rotated on a turntable as though it were in a conventional OATS. The antenna can be supplied as a complete system with driving balun and an absorber-lined chamber, or it can be retrofitted to an existing anechoic chamber.
To be sure, there are a few other varieties of TEM waveguide structures. In his paper entitled “Field Homogeneity in Different TEM Waveguides,” which predates Eurotem, Dr.-Ing H. Garbe of The University of Hannover noted that both GTEM and classic Crawford TEM cells provide good homogeneity. In spite of his attempts to dampen the influence of higher modes in other variants of TEM cells, he found their useful range was limited to a few hundred megahertz. His paper concludes with the statement, “These [variant] cells would only, at best, be suitable for precompliance testing.” 5
There also are test cells with a similar appearance to TEM cells, but they are really miniature anechoic or semi-anechoic chambers. The ARcell from Amplifier Research is an example of such a cell. The field is created or received by a log periodic antenna mounted at one end of the chamber. Small models of the ARcell also use balanced parallel-plate TEM transmission lines.
Correlation
Regardless of which type of cell is used, test engineers must relate their findings to the results that would have been obtained in OATS testing. Considerable work has been done to correlate both TEM and GTEM cells to OATS, but as a minimum, GTEM readings require three orthogonal views of the DUT.
This can be a big problem for bulky equipment attached to one or more cables. The DUT comprises all radiating or susceptible parts of the equipment. This means that the cables also must be viewed in three positions and that they must retain their original relationship to the main equipment as the entire assembly is rotated and tilted.
For large equipment, especially equipment that is sensitive to gravity such as photocopiers, a hyper-rotated version of the GTEM is available within which only a simple DUT polar rotation is required to achieve all three views. For most smaller products, a 3-D positioning stage can accomplish the necessary changes of attitude relative to the TEM field. A Eurotem cell may be a lower-cost alternative but only recently has become available.
“Two-port TEM cell vs OATS correlation routines were extensively developed and verified by NBS/NIST,” said Mr. Harrington of EMC Test Systems. “A reduced one-port correlation method based on three orthogonal EUT positions is widely used in GTEMs. Enhanced six-, nine-, 12-, and 15-position correlation methods have been proposed. We and other parties involved in the TEM waveguide international standardization efforts are studying the relative uncertainties and conditions of these alternative schemes. Typically, the average difference over frequency between GTEM and OATS readings is 3 dB with a 4-dB standard deviation.”
Conclusion
A number of viable alternatives to OATS and semi-anechoic chambers exist for EMC testing. Although OATS is the recognized standard by which all other approaches are judged, there may be a difference of several decibels between readings taken at separate OATS. Results from tests repeated at the same OATS will be within only a few decibels of each other.
GTEMs, TEMs, and Eurotems are available with ±1-dB nonuniformity within the center third of the test volume. If you use such a cell, your test results should be repeatable even if the DUT is not located in exactly the same place during each test. When you know the field is uniform, other factors such as cable alignment or changes in control signals can more readily be associated with inconsistencies in results.
Before making any EMC measurements, especially any involving an alternative test chamber, review the assumptions you have made. As an example, the simplified far-field algorithm used to correlate GTEM results to OATS is applicable only if:
All radiating elements are short with respect to a wavelength.
TEM is the only coupling mode present.
All dipole moments are in-phase.
All radiating elements have gains less than a dipole.
Consider using a comb generator as a sanity check. A comb generator is a stable broadband noise source that produces a repeatable signal. It may help you to understand how your OATS and non-OATS results are related. If you are confident that you can account for differences between readings and sites or test environments, then you are ready to start testing the DUT.
Because there will be variations among readings and because test results depend upon frequency, exact device positioning, and the ambient noise level, statistical data reduction usually is used. In spite of the prevalence of automated testing and the importance accorded to measurement distribution, you still need to understand what the measurements actually mean.
If you are in doubt about the test results you have obtained, consider taking the advice Hansen and Ristau offer in their paper: Instead of trying to improve the emission correlation to OATS by statistics, find the sources of deviation inside the cell.
References
1. Press, J., et al, “How a Semi-Anechoic Chamber Stacks up With an OATS,” EE-Evaluation Engineering, July 1999, p. S-5.
Note: This article can be accessed on EE Online at www.evaluationengineering.com. Select EE Article Archives and use the key word search.
2. Lorch, R., and Mönich, G., “Mode Suppression in TEM Cells,” IEEE Electromagnetic Compatibility Symposium Proceedings, Santa Clara, CA, 1996, pp. 40-42.
3. Whitehouse, J., and Loader, B., “A ‘Dual Septum’ Transverse Electromagnetic Field Cell for Electromagnetic Compatibility Testing,” The University of Reading, Department of Engineering, Whiteknights, Reading.
4. Hansen, D. And Ristau, D. et al, “Sources of Problems in the GTEM Field Structure and Recommended Solutions,” IEEE Electromagnetic Compatibility Symposium Proceedings, Santa Clara, CA, 1996, pp. 48-51.
5. Garbe, H., “Field Homogeneity in Different TEM Waveguides,” presented at Symposium on Electromagnetic Interaction With Complex Systems and Protection Measures (EMC), University of Magdeburg, 1997.
TEM/GTEM Cells
Offset Septum TEM Cell
The emCell is suitable for pre- and post-compliance testing from 100 kHz to 2 GHz. It produces a uniform field in accordance with IEC1000-4-3, and the offset septum design suppresses all resonant modes up to 2 GHz. The cell is lined with both ferrite and carbon-loaded polyurethane foam absorber material. The most uniform fields are provided for test volumes up to 0.9 ft3, but DUT footprints up to 19″ × 19″ can be accommodated. CIS 9942 software is included for fully or semi-automatic testing and failure monitoring. Schaffner-Chase EMC, (973) 379-7778.
IC RF GTEM Test Cell
The KuTEM Omni-cell supports IC RF emissions testing to SAE J1752-3 and draft standard IEC 61967-2. The PCB on which the IC is mounted attaches to the Omni-cell, allowing the IC to protrude through an opening in the outer surface. Emission and immunity testing can be conducted on ICs and small telecomm products, automotive modules, and medical devices from DC to 16 GHz. Omni-cell is an adaptation of the GTEM 250, one of a family of five GTEMs with septum heights from 250 to 1,750 mm. Lindgren RF Enclosures, (630) 307-7200.
Cell Switches Polarization
The Eurotem family of broadband TEM waveguides requires only polar rotation of the DUT and its connecting cables. Two pairs of stripline antennas are arranged symmetrically and driven differentially from a 50-W to 200-W balun. By switching the stripline connections, the electric field polarization can be changed from vertical to horizontal. Overall dimensions are 1.0 × 1.0 × 1.3 m for Eurotem 2 and 2.0 × 2.0 × 3.5 m for Eurotem 3. DUTs can be tested up to a 0.3-m cube or a 1.0-m cube, respectively. Euro EMC Service, (01149) 3328-430-141.
Shielded TEM Cells
Two S-LINE Test Cells have 50-W input impedance, are intended for use from 150 kHz to 1 GHz, and can handle a maximum of 100-W CW input power continuously at 40°C. The Model 700 has external dimensions of 1,062 × 815 × 790 mm and produces a uniform field over a 350 × 350-mm area; the Model 1000 measures 1,512 × 1,192 × 1,121 mm and provides a 500 × 500-mm uniform field. The cells include a large door and a shielded window through which the EUT can be observed during test. The S-LINE cells are available as complete test systems for EMS, EMI, and production test applications. Tektronix, (800) 422-2200.
20-GHz GTEM Test Cell
The Model 5405 GTEM!™ has a frequency range from DC to >20 GHz with three-position OATS correlation demonstrated from 30 MHz to 5 GHz and 9-position from 9 kHz to 5 GHz. From 9 kHz to 5 GHz, the VSWR is £ 1.75:1, and the maximum CW power is 250 W. The overall dimensions are 9.8’ × 5.2’ × 5.6’ high with a castor base or 3.7’ high without a base. The maximum septum height is 21.7″ at the terminating resistor board junction, and the door dimensions are 18.1″ × 15.2″. The field strength variation within an 11.8″ × 11.8″ area is £ 3 dB. EMC Test Systems, (512) 835-4684.
Family of Cells
Four ARcell test cells range in size from the benchtop 74.8″ × 27.6″ × 27.6″ TC1000 to the stand-alone 224.4″ × 102.4″ × 94.5″ TC4000. They handle DUTs measuring from 11.8″ to 39.7″ on each side. The smaller TC1000 and TC2000 cells incorporate balanced, parallel-plate TEM transmission lines for lower frequency work and built-in log-periodic antennas for higher frequencies. The larger TC3000 and TC4000 cells use log-periodic antennas to cover the IEC 80- to 1,000-MHz range. Test cells are available separately or as part of test systems with a controller and software, a field probe, a power meter, a printer and interface, cabling, accessories, and a preamplifier. Amplifier Research, (215) 723-8181.
High-Frequency TEM Cells
The FCC-TEM-JM1 TEM Cell has a maximum VSWR of 1.2:1 over the frequency range of DC to 1,200 MHz. The FCC-TEM-JM2 supplies a maximum VSWR of 1.5:1 from DC to 2,500 MHz. Both cells measure 15.2 × 9.9 × 33.8 cm, accept EUTs with maximum dimensions of 6 × 6 × 0.5 cm, have 9.1 × 9.1-cm access ports and type N connectors, and handle up to 500-W input power. A 10-V/m field requires <3.7-mW input power or 37 W for 1,000 V/m. The cells are intended for semiconductor emissions and immunity testing. Fischer Custom Communications, (310) 891-0635.
Immunity Test System
The TEMCell uses a vertically mounted, dual-septum design to reduce floor space while testing EUTs measuring up to 50 cm on a side. The unit includes a PSG1000B/F Signal Generator, an RF amplifier, and a standard filter box, all mounted within a 80 × 80 × 197 cm rack. A 26-MHz to 1-GHz frequency range, 100% sine or square modulation from 10 Hz to 10 kHz, and programmable 1- to 20-V/m field strength are featured. A PC controls and monitors the TEMCell. A door with a shielded viewing window and safety interlock is standard. Wayne Kerr, (617) 938-8390.
Copyright 1999 Nelson Publishing Inc.
September 1999