Four ways to remove ambient noise
These techniques let EMC labs separate ambient signals from EUT emissions so you can get reliable results from compliance tests.
Gil Bassak, Contributing Technical Editor -- Test & Measurement World, 8/1/2002
EMC compliance standards commonly specify an open-area test site (OATS) for radiated-emissions measurements. An OATS eliminates any reflections from walls that can affect emissions measurements in an anechoic chamber or shielded room, but it can also subject your tests to ambient RF signals from TV, FM, or other broadcast signals.
If an ambient RF signal falls at or near the frequency of an EMI emission, the ambient signal may either mask the emission or add to it, so you won't get a true measurement of the emission.
 |
| Figure 1. An open-area test site (OATS) uses a wood table that holds the EUT. The table sits on a turntable, and an antenna mounts on a mast. A roof protects the equipment from weather. Courtesy of Acme Testing. |
To separate ambient signals and noise from equipment under test (EUT) emissions, testing labs draw on four techniques. All four methods start with the same OATS setup. An OATS that complies with regulatory standards includes a metal ground plane, a turntable, and an antenna mounted on a mast (Figure 1). The ground plane creates a more uniform emissions environment. The turntable, which sits at one end of the ground plane, holds the EUT. The antenna mast lets EMC engineers change the height of the antenna and rotate the antenna between the horizontal and vertical planes.
The antenna must reside 3 m, 10 m, or 30 m from the EUT on the turntable, though not all labs have an OATS large enough to provide 10-m and 30-m measurement distances. Shielded electrical cables carry power and control signals from a control room to the turntable and EUT. Other cables carry signals from the receiving antenna to an EMI receiver or spectrum analyzer in the control room. In some cases, a nonconducting roof protects the EUT from the weather without altering the emissions environment.
During a compliance test, a test lab's EMC engineer rotates the EUT and the antenna to find the orientation that produces the highest level of emissions. If all peaks in a spectral scan fall at least 6 dB below the compliance limit, then the EUT will comply with emissions standards. If, however, any peaks—resulting from the combination of ambient signals, noise, and EUT emissions—exceed the 6-dB margin or exceed the compliance limit, the engineer must remove the effects of the ambient signals and noise. Using one of the available methods, the engineer can measure the true level of the EUT's emissions.
1. Tune out ambient signalsThe easiest technique, and therefore the first to consider, is off-tuning the EMI receiver. With off-tuning, an engineer adjusts the frequency or bandwidth, or both, of the test setup's EMI receiver or spectrum analyzer so it rejects the ambient signal but not the EUT's emissions. According to David W. Bare, chief technical officer at Elliott Laboratories (Sunnyvale, CA), this method requires that the frequency of the ambient signal differ from the center frequency of the emission in question by more than the instrument's resolution bandwidth (Ref. 1).
The off-tuning method works best for the case of ambient signals and EUT emissions that have narrow bandwidths, and whose frequencies sit far enough away from each other. In most cases, a tuned EMI receiver (in particular, one that meets CISPR 16 requirements), rather than a spectrum analyzer, will provide the needed high selectivity (Ref. 2). Table 1 provides the spectral ranges and resolution bandwidths for CISPR 16.
EMI receivers produce better selectivity than the spectral sweeps that you get with spectrum analyzers. Tuned receivers look only at the resolution cell plus its skirts, where a spectrum analyzer sees more spectral content. With a tuned EMI receiver, an EUT's emission need only be separated from an ambient signal by slightly more than the receiver's resolution bandwidth for you to accurately make the measurement.
Tuned notch filters, used with EMI receivers and especially with spectrum analyzers, can also help block near-frequency ambient signals and noise. But the filter must preserve the instrument's resolution bandwidth and avoid distorting the EUT's emission signal bandwidth, shape, and amplitude.
2. Substitute signalsIn cases where the ambient signal plus noise has a narrow band but falls too close in frequency to the EUT emission signal to be separated by off-tuning, EMC engineers can use the more time-consuming but highly accurate technique of signal substitution.
The signal-substitution method has two variations. The first variation takes about three times longer to perform than the off-tuning method. In this variation, an EMC engineer locates the turntable rotation, antenna orientation, and antenna height to find the highest emission levels. Then, the engineer replaces the EUT with a signal generator and dipole antenna or a horn antenna. The antenna must have a known radiation pattern and a precisely calibrated gain and impedance so that it produces a known signal strength at a given distance. After setting the signal generator to the emission's frequency, the EMC engineer increases its output until the amplitude of the radiated signal that the EMI receiver or spectrum analyzer measures—a signal that includes the ambient signal plus noise—equals that obtained from the EUT.
At that point, the engineer can determine the level of the EUT's emitted power by reading the voltage (or power) output of the signal generator and, knowing the antenna impedance and gain, make a simple calculation:
Received power (dB)
= transmit power(dB) –cable losses (dB)
+ transmitter antenna gain (dB)
While some measurements use units of power, others use voltage current, or magnetic field. So, you could also see units of µV dB/m, nW dB/m, or µA/m.
In a faster variation of the signal-substitution method, the engineer connects the signal generator directly to the EMI receiver or spectrum analyzer through a coaxial cable and adjusts the signal generator's frequency and amplitude to duplicate the peak emission signal. Then, the engineer can read the emission level on the EMI receiver's or spectrum analyzer's numerical display.
This variation emulates the emitted signal from the EUT with the signal generator, but the receiver still receives the ambient signal through the receive antenna. The EMC engineer can determine the emissions by measuring the power from the signal generator and then subtracting the cable loss. This method takes about half the time of the first signal-substitution method, or about 1.5 times as long as the off-tuning method.
3. Shorten measurement distanceShortening the measurement distance offers a relatively simple way to increase the measured emission's signal strength. Moving the receiving antenna closer to the EUT increases the EUT's emission strength relative to the ambient signals and noise by the square of the difference in distance from the EUT to the antenna.
For example, at frequencies of 30 MHz and above, where electric fields dominate the measurement, a 20-dB-per-decade-of-distance rule applies. Thus, cutting the measurement distance from 30 m to 10 m will boost the EUT's emissions by nominally 10 dB. Similarly, path reductions from 10 m to 3 m and 3 m to 1m will each yield additional 10-dB gains. Below 30 MHz, where magnetic fields dominate, a 40-dB-per-decade-of-distance rule applies, so that respective reductions in the measurement distance would each yield 20-dB gains.
EMC engineers often reduce measurement distance when measuring a very "quiet" EUT. Of course, if the receiver detects only ambient noise—and no EUT emissions—as the antenna gets closer, then the EUT's emission are so low that it easily will meet regulatory emissions requirements.
If you move the receiving antenna to a distance less than 1 m from the EUT, you'll introduce new measurement problems caused by nonlinear near-field effects. Near-field effects make it difficult, if not impossible, to extrapolate the emission signal's strength to the original measurement distance.
4. Subtract noiseThe fourth method, linear subtraction, attempts to mathematically strip the ambient signal plus noise from the total signal that comprises the ambient signal, noise, and EUT emissions. CISPR 11, Annex C describes defines the emitted signals as follows (Ref. 3):
(emission)1.1 = (noise + emission)1.1– (noise)1.1
According to Roland Gubisch, chief engineer of EMC and telecom at ETL Semko (Boxboro, MA), this equation gives at least a first order approximation of the EUT's emissions. Gubisch points out that while linear subtraction is adequate for most compliance tests, it isn't a desirable technique.
Harry H. Hodes, president and principal EMC engineer of Acme Testing (Acme, WA), explains why EMC labs shy away from linear subtraction. According to Hodes, if you try to make a simple subtraction of a broadband FM radio signal from a narrowband clock pulse, you'll get an incorrect answer. Hodes claims that the very strong ambient signal will "capture" an EMI receiver's mixer when you try to subtract a strong ambient signal plus noise from the EUT emission. That mixer capturing will invalidate the equation given in Annex C of CISPR 11, because the equation represents the sum of two small signals in the mixer.
For a valid equation, the receiver must operate in its linear range, so both signals must be in the same order of signal strength. Unfortunately, that rarely occurs. One signal—say the ambient—is usually much larger in amplitude than those of the emissions or the noise. As a result, the mixer locks onto the ambient signal and ignores the intended signal, invalidating the equation.
A proper subtraction process requires vector processing. Such an approach uses five antennas (four antennas plus the main receiving antenna) and a multichannel vector-coherent preprocessing receiver placed ahead of an approved EMI receiver.
A test lab's EMC engineer will arrange four of the antennas in a square pattern, each with the same cable length to keep cable losses equal. The cables will connect to a series of mixers whose gains the engineer will adjust to the same level. Then, the EMC engineer will tune the receiver to a particular narrowband signal and get a measurement from each antenna.
A signal that radiates relatively far away from an antenna will vary very little in amplitude and more in angle or phase of arrival. In contrast, a local signal will have a great deal of difference in both amplitude and angle of arrival. If the signal varies greatly when you rotate the turntable, then it must come from an emission. If not, then it must be an ambient signal. If it's an ambient signal, then the engineer can subtract it from the measurement using an EMI receiver.
Spectrum analyzers, which use swept-tuning rather than matched filters, are even less suited for properly subtracting strong ambient signals that directly mask EUT emissions. Still, Hodes says, a test engineer can make the calculations using either instrument by taking into account the EMI receiver's or spectrum analyzer's impulse bandwidth effects
When all else failsIf none of these methods adequately strips ambient signals plus noise from the EUT's emission signals in an OATS setup, EMC engineers have two other options for making EMC compliance measurements: either move the test to an OATS that has significantly lower ambient signal levels, or abandon the OATS altogether and run the test in a suitably designed and approved anechoic chamber.
Not every compliance lab may offer the option of an approved anechoic chamber, as such chambers can cost more than $1 million. Building a 30-m anechoic chamber isn't feasible, and most anechoic chambers work for 3-m measurements only. Moreover, if the compliance specification calls for using an OATS to measure emissions, the test engineer needs to correlate the results obtained in an anechoic chamber or shielded enclosure to an equivalent for an OATS.
To help correlate the results between the OATS and a chamber, the test engineer should measure one or two signals of similar frequency to the EUT's emission at the OATS and compare those measurements to similar emissions in the chamber or shielded room. The comparison will then provide some direction for translating the results received in the chamber to what would be obtained in an OATS unperturbed by ambient noise.
| FREQUENCY RANGE | RESOLUTION BANDWIDTH |
| Below 9 kHz | No testing required |
| 9 kHz to 150 kHz | 200 Hz |
| 150 kHz to 30 MHz | 9 kHz |
| 30 MHz to 1 GHz | 120 kHz |
| Above 1 GHz | 1 MHz |
| References |
|
| Author Information |
| Gil Bassak is a freelance technical writer and former electrical engineer who lives in Ossining, NY. E-mail: gil@word-warrior.com. |
