Open-Area Test Site Measurements: Dealing with Ambients
|Figure 1. Typical trace of ambient signals in an OATS in the megahertz frequency range. (click to enlarge).
For measurements of radiated or conducted emissions generated from equipment under test (EUT), it is essential to distinguish between emissions from the EUT and ambient electromagnetic (EM) signals, i.e., signals from other EM sources. Ambient signals are the major cause of inaccurate measurements.
EM ambients can be continuous or transient, broadband or narrowband. In the frequencies up to 1000 MHz, continuous narrowband ambient signal sources include UHF radio and TV broadcast signals, as well as those from pager and mobile phone stations. Typical continuous broadband ambient sources are local industrial plants and switching energy systems, and devices such as computer networks. Figure 1 shows a typical ambient trace measured by a spectrum analyzer or EMI receiver from 9 kHz to 1000 MHz on an OATS. Broadcast, pager, and mobile phone transmissions are evident in Figure 1.
Their comparatively high levels can exceed the measurement limits by several tens of dB, and thus the ambients can render the measurement results incorrect and unusable.
Using practical electromagnetic interference (EMI) measurements, certain steps can be taken to mitigate the effects of some of these ambient signals. These steps include:
• Performing tests on the weekend or in the evening, when industrial continuous and transient broadband noise is low.
• Performing measurements with the narrowest bandwidth over a narrow span to distinguish the EUT emission from narrowband ambients.
If the radiated emission of the EUT should be tested, additional steps should include:
• Orienting the measurement range to right angles to the worst offenders, taking advantage of the antenna’s null on-axis.
• Reducing the antenna-to-EUT distance below the 3-m standard value to increase the EUT signal contribution. The emission limits should be checked, and results should be noted in the test record.
Electronic Cancellation of Ambient Signals
|Figure 2. Basic setup for ambient signal cancellation (click to enlarge).
EMI receivers allow effective cancellation of most ambient interference signals from the desired measured ones. This digital signal processor–based function is already included in some EMI receivers currently available.1,2
The basic method of ambient (i.e., undesired) signal cancellation by means of subtraction is illustrated in Figure 2. After the test site is prepared for the emission measurement, the EUT is switched off. Therefore, the input of the EMI receiver consists of the ambient (i.e., undesired) signals only.
The mode switch S in the IF block of the EMI receiver (see Figure 2) is in the normal-mode position, so that the ambients can then be checked in the required frequency range. The trace indicating the ambients as whole or at some selected frequencies is then digitized and stored in the internal memory of the receiver by switching S to the hold-mode position.
|Figure 3. Effect of the subtraction of ambient signal from the measured trace (click to enlarge).
The EUT is then switched on, and the differential mode position of switch S is selected. The input of the EMI receiver contains the measured (i.e., emitted from the EUT) signal S(t) and actual ambient signals Ad(t). In this mode, the required emission test (i.e., ambients plus EUT emission) is performed, while the differential amplifier (diff amp) in the receiver (see Figure 2) subtracts the stored noise trace from the actual measurement. Thus, the resultant output signal U(t), following the subtraction, can be written as
U(t) = [S(t) + Ad(t)] – Ah(t), (1)
where Ah(t) denotes the primary ambient signals stored in the hold mode of the switch S. Squaring both sides of Equation 1 yields
U2 = S2 + (Ad – Ah)2 + 2 · S(Ad – Ah). (2)
Taking the expectation (i.e., mean value) of both sides of Equation 2 and defining S as uncorrelated with ambient signals Ad and Ah, so that the results of S · Ad and S · Ah are zero, yields
E[U2] = E[S2] = E[(Ad – Ah)2]. (3)
From the previous equations, it can be seen that the output signal U(t) becomes equal to S(t) only if the stored ambient signal Ah(t) is identical to the actual received ambient Ad(t). In this situation, the measured signal power E[U2] = E[S2], and it will be unaffected by ambient signals.3
|Figure 4. Measurement with continuous ambient signal: a) initial trace of ambient signal; b) displayed result in differential operating mode (click to enlarge).
In reality, the actual ambient signals may be very variable: continuous or transient, often with random variation in magnitude, phase, or frequency. For this reason, the output signal U(t) will contain the desired signal S(t) and some undesired signals, although the undesired signals will be substantially reduced in magnitude.
In all events, subtracting Ad and Ah maximizes the measured output signal–to–undesired noise ratio, and most of the common ambient signals will be removed from the actual measurement.
Evaluation of the Measurement
Many EMI receivers enable simultaneous indication of all traces during each measurement step. The first measured (and stored) hold trace of the ambient Ah(t) is displayed on the receiver, which allows a comparison of the two traces Ah(t) and U(t). Figure 3 shows a trace of the hold ambient Ah(t) before selecting the differential mode (upper trace). The lower trace in Figure 3 shows the typical effect of canceling out the ambient noise by selecting the differential mode of the receiver.
|Figure 5. Measurement with square-wave continuous ambient signal: a) initial trace of ambient signal; b) displayed result in differential operating mode (click to enlarge).
From a theoretical point of view, as well as from practical experience, it is clear that successfully canceling ambient signals is due primarily to the character of the canceled (ambient) signals. To obtain the best results using the cancellation function of EMI receivers, it is important to note that:
The ambient noise signals should be continuous, i.e., time and frequency must be stable. Transient signals from aircraft transmissions, arc welders, mobile phones, etc. may not be effectively canceled out by using the subtracting function. Such transients often appear during only one of the measurements. They are not present in the same form both in ambients (normal mode), as in ambient plus EUT signals (differential mode) measurements.
For the same reason, the actual EUT emission measurement should be created immediately after the ambient noise mapping, and the subtraction of the background EM noise from the measured trace should be repeated several times.
|Figure 6. Cancellation of white-noise ambient signal in the range of 9 kHz to 1000 MHz (click to enlarge).
To verify the cancellation method for various types of ambient signals, the EMC precompliance test laboratory at the Institute of Radio Electronics at the Brno University of Technology (Brno, Czech Republic) has developed a set of practical emission measurements. An EMI receiver with a built-in subtraction function was used in the frequency range from
9 kHz to 1000 MHz.
The verification was treated as a null-emission signal measurement, i.e., with the desired emission signal S(t) = 0. From Equation 1, if the ideal cancellation condition Ah(t) = Ad(t) occurs, the displayed signal should be U(t) = 0.
First, researchers investigated a situation in which the
ambient would be considered near ideal (from the viewpoint of a successful cancellation). In other words, the ambient signal was continuous and time- and frequency-stable. A sine-wave 2-MHz signal from a laboratory signal generator was used. The displayed frequency trace (frequency
spectrum) on the EMI receiver in the range from 1 to 11 MHz is shown in Figure 4a.
|Figure 7. Cancellation of additional broadcast ambient signal in the 100 MHz frequency band (click to enlarge).
The ambient signal that was used was not perfect; in addition to the first-harmonic component (2 MHz), it also contained some higher harmonics, e.g., second, fourth, etc. The ambient trace on Figure 4a was stored in the EMI receiver memory (hold mode), and then the differential measurement was obtained. The result is shown in Figure 4b, in which the lower trace indicates the actual input signal after the cancellation effect.
From these measurements, it was evident that the cancellation of ambients may be 50 dB or more. Residual deviations from the ideal trace were very small and were caused by frequency instabilities of the measured signal and the limited digitizing accuracy and resolution of EMI receiver. The null trace level (~49 dBµV in Figure 4b) was due to adjusted amplifier gain and the receiver’s own noise level.
A second verification measurement used the ambient signal in the form of a continuous square wave at a frequency rate of 2 MHz and a duty cycle of 50%. The results were about the same as the previous (harmonic) ambient signal (see Figure 5). Due to the subtraction effect in Figure 5b, the ambient signal is nearly completely canceled, with suppression of some tens of dB.
|Figure 8. Decreasing sweep frequency range and using quasi-peak detection increases the cancellation effect of ambient signals (click to enlarge).
These conclusions can be extrapolated to apply to ambients that cause additive noise or, of course, of both deterministic and additive ambient signals. Therefore, in the next measurement, the ambient was created through a stochastic signal in the form of a broadband white-noise signal. A laboratory noise generator with an avalanche noise diode was used with an additional broadband amplifier with 30-dB gain. The resulting frequency trace is shown as the upper trace in Figure 6. In the same figure, the lower trace represents the actual input signal (i.e., zero signal in the measurement) after the difference mode of the receiver was selected.
Figure 6 shows that in the frequency band of 100 MHz, other ambient EM signals appear along with the white-noise signal. These are likely broadcast signals. Their cancellation seemed insufficient (see the lower trace in Figure 6). To check this perception, the next ambient noise signal measurement was obtained in a narrower frequency band (50–150 MHz). The result (see Figure 7) shows that the additional broadcast ambients were canceled sufficiently by using the differential mode.
Furthermore, practical application of this cancellation method has shown that the subtraction effect is more applicable for quasi-peak receiver detector mode than in peak detector mode. The accuracy of the cancellation increases with the decrease in the span of the measured frequency. Figure 8 shows the differential measurement in quasi-peak detector mode in a narrower span (95 to 105 MHz). The cancellation is nearly perfect.
A method for efficient ambient cancellation has been discussed, and through a series of practical tests, the method has been shown to effectively reduce ambient signals in an OATS. The cancellation method can be used with utilities already provided in many EMI receivers. It provides effective noise reduction, and thus eliminates the uncertainties caused in areas of high background EM noise.
By following certain steps, this method can be very effective with cancellation of deterministic as well as undesired ambients. Moreover, the differential- mode measuring techniques can be used successfully for both radiated and conducted emissions tests.
1. Making Precompliance Conducted and Radiated Emissions Measurements with EMC Analyzers (Palo Alto, CA: Agilent Technologies, 2000), [on-line] available from Internet: www.agilent.com.
2. Spectrum Receiver Manual, (Durham, UK: Seaward Electronic Ltd., 1998).
3. RN Ghose, Interference Mitigation—Theory and Application (New York: IEEE Press, 1996), 13–38.
JirŠ í SvacŠ ina is a professor of electronics
and communication at the Institute of Radio Electronics, Brno University of Technology (Brno, Czech Republic). In addition to microwave techniques, he is interested in the specialized problems of EMC. He is a senior member of IEEE and a fellow of IEE. He can be reached via e-mail at firstname.lastname@example.org.