A System to Simplify EMC Testing

Mike Marino

New virtual chamber technology provides ambient cancellation, as well as capabilities for EMI source localization.

Ambient signals create a significant problem for EMC. With more wireless phones and the advent of high-definition television, EMI ambient signals are increasing exponentially. That trend combined with general population growth has made using an outdoor area test site (OATS) much more difficult. Shielded anechoic chambers are an ideal solution for large manufacturers, but these may not be practical or financially feasible for smaller companies.

System for EMC ambient cancellation and emissions source localization.

CASSPER (configurable automated system for sensing and processing electromagnetic radiation), an advanced, PC-based instrumentation system for radiated-emissions testing and RF source localization, has been developed by CASSPER Instrumentation Systems Inc. (Lake Forest, CA). The system is a low-cost alternative to a shielded room. It records and isolates signals of interest from equipment under test (EUT) without any need for anechoic chambers. It delivers ambient cancellation and removes the speculation from signal identification.

The dual-channel multiport system is designed to overcome the disadvantages of spectrum analyzers by employing two simultaneous and synchronous RF measurements.

System Overview

The system has two modes of operation: ambient cancellation and source localization. Ambient-cancellation mode provides an automated method for suppressing ambient signals and identifying EUT emissions. While not every ambient may be suppressed, a large majority can be suppressed to acceptable levels. Source-localization mode provides a method for isolating signal sources, which is used to determine whether the ambients that have not been suppressed are coming from the EUT or from nearby equipment. For emissions identified as coming from the EUT, source localization can be used to determine their exact source within the equipment.

A spectrum analyzer would be able to correctly measure the EUT emissions and ambient signals. But what if there were 100 ambient signals? Each one of them would have to be investigated to ensure that it was not an EUT emission. Even if there were complete prior knowledge of the site's ambient signature, the signals would have to be examined because the ambient signals could be masking the EUT emissions. Masking occurs when the ambient and EUT signals are close in frequency and the ambients are stronger than the EUT's emissions. Typically, CASSPER in ambient-cancellation mode can automatically suppress 90% of ambient signals. Needing to investigate only 10 of our presumed 100 ambient signals would mean a dramatic savings in test time. In dealing with ambient masking of EUT emissions, ambient-cancellation mode can extract emission signals that are as much as 40 dB below the ambient level. Our remaining 10 signals could be investigated using CASSPER's source-localization mode, which would enable these signals to be either identified or rejected as coming from the EUT.

System hardware consists of a dual-channel, synchronized receiver unit with four multiplexed ports for each receiver. The multiplexers are automatically controlled via the system software for automatic band switching, which is useful in performing frequency scans. Other antennas or probes, such as near-field probes for diagnostic work, can be attached to the remaining ports to reduce the need for switching cables between tests. A Pentium III PC, running Windows NT and with a dual-channel digital signal processor (DSP) card, completes the system. The combination of two simultaneous RF measurements with advanced signal-processing techniques is the foundation of CASSPER's performance capabilities.

The heart of the system's functionality is its ability to determine if two signals are the same. Two phase-locked, frequency-synchronized receivers are used to simultaneously sample both signals. The samples are continuously transferred from the receivers to two ultra-high-speed DSPs in the PC via two gigabit-per-second data links. The data are processed by the DSP to achieve the various bandwidth and detector functions (peak, quasi-peak, average) and then perform ambient cancellation or source localization. Because the two signals are measured instantaneously, variations in their magnitude or frequency do not matter. Thus, modulated (AM, PM, PWM, etc.) or transient signals do not degrade system performance.

The user interface for the system closely resembles that of a typical spectrum analyzer. Based on National Instruments' LabView and running under Windows NT, the display provides for entry of start and stop frequencies, bandwidth setting, and several system-specific parameters. A standard Windows menu provides control over other settings, including port setup, selection of transducer factors (antenna factors, current probe transfer impedance, and so on), cable-loss tables for each port, receiver filter selection, maximum hold, data storage, and configuration storage and retrieval. Most of the screen is dedicated to a large graphic display of the spectrum from each receiver channel along with a third trace representing the processed data. Each trace can be toggled on or off independently. Finally, a suspect-frequency list is provided to allow recording or tracking of signals near or over the limit line.

Ambient Cancellation

A typical ambient-cancellation setup is shown in Figure 1. An emissions antenna is attached to channel A of the receiver, and a broadband antenna for measuring mainly ambient-only signals, such as a biconical-log hybrid, is attached to channel B and placed remotely from the test site. The system will suppress those signals that are common to both antennas and extract those that are primarily in the EUT antenna. Therefore, ambient cancellation requires all EUT signals to be at least 20 dB lower at the ambient antenna than at the EUT antenna.This corresponds to a distance at least 10 times the EUT test distance for OATS testing. For precompliance testing, it is often suitable to place the ambient antenna on top of a building or other prominent location. The additional isolation afforded by the building will typically provide the necessary 20 dB. The ambient antenna should be oriented in the same direction and polarization as the EUT antenna. It is desirable that the ambients at the ambient antenna be as high as or higher than those at the EUT antenna. Note that no gain or phase matching is required; extra signals in the ambient-only antenna do not cause any problems.

Figure 1. A typical ambient-cancellation setup.

In ambient-cancellation mode, the system display looks almost identical to that of a typical spectrum analyzer. In fact, it behaves like a two-channel spectrum analyzer, showing received channels A and B simultaneously. Channel A is defined as the EUT channel and channel B is used for ambient detection. Examination of the traces from the two channels reveals traces identical to those a spectrum analyzer would produce, with channel A showing all signals received in the vicinity of the EUT and channel B displaying all signals received at the ambient detecting antenna. Traditionally, an experienced EMC engineer or technician might go through several steps to try to isolate the signals in the vicinity of the EUT that were actually from the EUT. One common method has been to turn the EUT on and off and compare the traces. The dual-channel operation of the system provides these two traces in real time, one from the EUT antenna and one from the ambient antenna.

CASSPER's ambient-cancellation mode advances beyond current methods by displaying a third trace that is derived from the first two. Clearly, the first thing this trace should show is all the signals that are visible only on the EUT channel; those are EUT emissions. Also, the trace should remove signals that are not coming from the EUT. This is where the system improves on traditional techniques. Normally, given the A and B traces, a tester would not be able to comment on EUT signals that might be hidden in the ambient. It might be possible, after extensive probing with near-field probes or other devices, to show that the EUT did not emit at the ambient frequencies; however, more often than not, signals buried in ambients are ignored. Because the system is capable of determining if two signals are the same, it applies cancellation whenever a signal detected by the ambient antenna is the same as one detected by the EUT antenna. When only part of the detected signal is the same, the system can separate a measurement into its ambient and EUT components, displaying only the latter.

The system is capable of canceling up to 40 dB worth of ambient signal, although the amount of cancellation is proportional to the magnitude of the signals above the noise level. For smaller signals, 20–25 dB is common. The recovered EUT signals are typically within a few decibels of their actual value, although, again, recovery quality is proportional to both the EUT and ambient signal magnitudes. Because the process of cancellation is time-consuming, and since only signals near the limit line typically are of interest, the system provides a threshold mode in which ambient cancellation is applied only when a signal from the EUT antenna is within a user-defined tolerance from the limit line. This can greatly increase the speed of testing. Of course, the unit will try to cancel the noise floor of the receivers if that is what the user desires.

Data can be stored in comma-separated-variable format to be loaded into a spreadsheet or other software for processing.

Source Localization

Figure 2 shows a typical source localization setup using the system. Channel A is used to detect the problem signal, ordinarily employing the same antennas involved in the emissions test, although other transducers such as current probes may be used. Channel B is assigned to probing the EUT for the same signal.

Figure 2. A typical source localization setup using CASSPER.

In this mode the display is split into two graphs. The first shows the power spectrum (amplitude versus frequency) for channels A and B, while the second graph shows coherence versus frequency.

Coherence is a measure of phase stability and synchronization between two measurements. Absolute phase is not important. This is why two simultaneous and synchronized receivers are needed. A coherence value of 1 (100%) indicates that two measured signals are perfectly phase-locked to each other, which implies that the two measurements are from the same source. A coherence value of 0 indicates that two measurements, even though they might be at the same frequency, are independent of each other. This implies that the measured signals are from different sources. A coherence value between 0 and 1, for example 0.6 (60%), indicates that two measurements are somewhat related but that there is another more-dominant signal source at that frequency.

The coherence function enables emissions to be source localized. Signals that are present in an antenna can be traced to either the EUT or some nearby equipment. This is especially useful for in situ or laboratory benchtop testing situations where nearby facility equipment cannot be turned off. Source localization requires simply that some type of probe be used. When the coherence moves to a value near 1, then the source has been found. Even when two identical EUTs are operating next to each other, if one is emitting and the other is not, the coherence can be used to quickly identify which equipment is the emissions source. A spectrum analyzer cannot provide this capability. Merely having two signals at the same frequency carries no implication that they are related, but having a high coherence value does imply a relationship.

The user can manipulate three parameters: the resolution bandwidth; the averaging level, where the quality of the correlation can be improved by making more measurements; and the amplitude threshold, below which signals from the channel B probe are ignored. The last of these is useful for spatially localizing a source. If it requires larger amplitude, the probe is probably closer to the source.

Conclusion

The increase in ambient signals has rendered radiated-emissions testing at OATSs extremely difficult. Efforts to eliminate those ambients prior to introduction of the CASSPER system were incomplete and led to inaccuracies. The system's innovative technology can deliver true ambient cancellation and eliminate guesswork from signal identification. The system also facilitates EMI source localization, helping engineers to quickly locate hard-to-identify radiation sources.

Optionally, CASSPER can be used for in situ testing on the factory floor without requiring production interruptions for equipment shutdowns. The versatile system has application for OATS testing, as an alternative for preproduction testing, for sample testing, and for performing overflow testing when on-site shielded chambers are unavailable.

Mike Marino is vice president of engineering at CASSPER Instrumentation Systems Inc. (Lake Forest, CA), where he is responsible for the company's R&D efforts. His accomplishments include BS and MS degrees from the Massachusetts Institute of Technology and 12 years of design experience in the commercial and military electronics industries.