Anechoic chambers: Design and demonstration
If you've ever designed a product, you've probably had to test it for EMC compliance, which includes radiated emissions and susceptibility. Chances are, you've taken that product to an in-house or third-party EMC lab for testing, probably in an anechoic or semi-anechoic chamber. How do these chambers work and what are the procedures for verifying their performance? That's what a gathering of some 35 engineers learned from RF engineer Zubiao Xiong of ETS-Lindgren on November 16. Xiong gave a presentation, "Insights into EMC Chamber Design: How to achieve an optimized chamber for accurate EMC Measurements," at National Technical Systems (NTS) in Boxborough, Mass.
[Download the presentation slides.]
Anechoic chambers are shielded rooms used for EMI and RF testing. The shielding attenuates outside RF energy, simulating an open-area test site (OATS). A chamber's inner walls are lined with absorbing material which, for radiated emissions, minimizes reflections emitting from the equipment under test (EUT). For radiated susceptibility tests, an antenna emits signals intended to disrupt the EUT's operation.
The chamber shown in Figure 1 is semi-anechoic because the floors reflect RF energy as where a fully-anechoic chamber has absorbers on its floor, walls, and ceiling. Partially-lined chambers have absorbers on certain areas only. This chamber at NTS is what's called a "3-m chamber," meaning that the distance between the EUT and the transmit or receive antenna is 3 m. Smaller chambers typically use 1 m distance while larger chambers are large enough for 10 m distances.
Xiong explained the materials that line chamber walls, starting with ferrite tiles, which can absorb signals at frequencies below 1 GHz. For higher frequencies, the chamber has absorbers called cones. The wall in Fig. 1 shows absorbers, some with their caps removed. These hybrid absorbers are, according to Xiong, the preferred types to use because they balance the need for a broad frequency range versus size.
Figure 1. Chambers walls are lined with absorbing materials to minimize signal reflections.
Chambers require characterization and calibration before any EMI test result can have meaning. The procedures vary based on which EMC standard an EUT must comply—commercial, military, automotive, or telecom. Requirements can also vary by country.
Standards require normalized site attenuation (NSA) measurements, among others. Figure 2 shows a diagram of the test. A transmit antenna mounted on a mast needs an adjustable height from 1 m to 4 m. The antennas on the circle represent five locations at two different heights from the floor, to measure the transmitted signal. You need only one receive antenna, but you need measurements in five locations and two heights. The transmit antenna must move so the projected distance from the transmit to the receive antenna is constant for all five locations. The test is then repeated with the antennas oriented horizontally. For frequencies from 30 MHz to 1 GHz, the measured signals must be within a range of ±4 dB. From 1 GHz to 18 GHz, the received signal power must be ≤6 dB based on site voltage standing wave ratio (sVSWR). Xiong explained how CISPR 16-1-4 (2010) specifies how to make sVSWR measurements.
Figure 2. To perform a normalized site attenuation, a receive antenna needs to be mounted on a movable mast. A transmit antenna moves among five locations for measurements within the test volume. Courtesy of ETS-Lindgren.
Referencing IEC 61000-4-3 (2010), Xiong discussed the procedure for verifying field uniformity. For this test, you transmit a signal from an antenna to an electric field probe that you move to each of 16 positions in a plane that's 1.5 m² and 3 m from the transmit antenna. Of the 16 measurements, 12 must fall within a range of 6 dB.
The above chamber qualifications are defined in IEC standards, but there are other standards to contend with as well. Before moving onto chamber design, Xiong wrapped up his discussion on chamber qualifications by referencing MIL-STD-461G (military), CISPR 25 (automotive), and ETSI EN 300 328 (telecom) as they relate to absorbers inside the chamber.
In the chamber design portion of his presentation, Xiong covered methods for simulating a chamber to give you an idea of how it might perform when built. For example, you need to calculate the length of and width of the chamber based on the distance from your antenna to EUT, the size of the absorbers, the radius of the turntable under the EUT, and other factors. "The distance from the antenna to the EUT should usually be at least 1 m," said Xiong. "You can tilt the measurement axis so reflections don't arrive at the EUT at the same time because if the do, the reflections will add to each other in phase." As for height, the chamber needs to be tall enough to allow for the antenna to reach 4 m above the floor (Figure 3). The reason for 4 m has to do with catching reflective peaks in signals transmitted from the antenna.
Figure 3. Design a chamber so the antenna height can range from 1 m to 4 m above the floor. Courtesy of ETS-Lindgren.
Xiong then moved into simulation. Simulating a chamber is complicated because of the many factors you must consider. "You may need to run a simulation multiple times to get a complete picture of a chamber's performance," he said.
Because simulating a complete chamber is complex and time consuming, techniques that simplify simulation have arisen. One such method—the discrete complex image method—essentially decomposes the chamber wall response into virtual images at varying complex-valued depths underneath the wall. The characteristics of each horizontal and vertical layer are then combined to simulate the entire chamber.
Simulations require validation. Xiong compared simulations using several methods against measurements for several locations: back, front, left, center, and right. Figure 4 shows a plot comparing three simulation methods against measured values.
Figure 4. Comparison from 30 MHz to 200 MHz shows that the new method with complex images performs much better that the conventional ray-tracing algorithm at the lower frequencies. Courtesy of ETS-Lindgren.
Using simulation software, Xiong showed the engineers in the audience how they can perform simulations on chamber upgrades as well as on new chambers. Why upgrade? Absorbing cones can degrade or deform over time. Plus, materials have improved allowing for better absorbency and smaller size, which reduces reflections in the chamber and can result in smaller new chambers, or larger interior space for testing in older chambers. To see the difference that can occur when a chamber is refurbished, download the presentation and see slides 39 and 40.
The final part of Xiong's presentation covered debugging tools. Here, Xiong discussed what he called a "time-domain" response plot. This isn't time domain in the traditional sense of voltage versus time, but rather what happens in the frequency domain because of signal reflections. Recall that reflections can use signals to arrive at their destination at different times. From the arrival times, you can calculate the location of the reflection. "We use a VNA to measure S21 between two antennas (in the frequency domain), and then convert the S21 measurement to the time domain for the time gating process. The signal is a continuous wave sweeping between the specified start frequency and stop frequency."
Figure 5 shows small peaks that result from reflections. The information derived from this measurement can point to deficiencies in the chamber design that need additional absorption. See slides 43 through 46 in the presentation to see how using this time-domain technique improved a chambers response in the frequency domain.
Figure 5. Time-domain peaks represent the time that a signal needs to reach the receive antenna. Courtesy of ETS-Lindgren.
Inside an anechoic chamber