Manufacturers must ensure that the testing services they are paying for are being performed correctly and accurately. If a test laboratory does not perform a test correctly, it could lead to erroneous data that result in product recalls, fines, or the loss of time and money in the final emissions-measurement stage.

Radiated-emissions testing can be one of the most critical issues in EMC compliance testing. It is a complex measurement that relies not only on expensive instrumentation but also on experienced personnel to ensure that measurements are being performed properly.

Test labs perform radiated-emission prescans on products to identify suspect frequencies that are close to or greater than the specified limits. Once these frequencies are identified, the lab remeasures the equipment under test (EUT) on a final measurement site that meets specific criteria known as normalized site attenuation. A pass-fail determination is made at the final measurement site. In most cases, test labs perform prescans with a peak-voltage detector. Final measurements are then made with either average or quasi-peak voltage detectors, depending on the frequency and the measurement standards used.

Radiated-emissions prescans can be performed in many ways and in many locations. Prescan measurements can be completely automated, semiautomated, or completely manual. Each measurement technique could be employed in several types of measurement facilities. Common test facilities include shielded enclosures, semianechoic chambers, fully anechoic chambers, and open area test sites (OATS).

Depending on the technology an EMC test lab possesses, a prescan can be fairly comprehensive, and the data obtained can represent final qualification data. Conversely, a lab with minimal equipment might conduct prescans on only one side of the EUT or at a single measurement height.

When conducting a prescan, it is important to refer to the ANSI C63.4-2000 interim standard.¹ This standard provides the measurement procedure for making emissions measurements. It states: "The full frequency spectrum (for the range to be checked for meeting compliance) shall be investigated. This investigation is performed with the EUT rotated 360º, the antenna height scanned between 1 and 4 m, and the antenna rotated to repeat the measurements for both the horizontal and vertical antenna polarizations.

During the full frequency spectrum investigation, particular focus should be made on those frequencies found in exploratory testing that were used to find the final test configuration, mode of operation, and arrangement (associated with achieving the least margin with respect to the limit). This full spectrum test constitutes the compliance measurement." It is logical to conclude that if the prescan technique was not important, then these criteria would not have demanded such extensive final measurements.

Shielded Enclosure. Shielded enclosures are probably among the most difficult environments in which to make an effective prescan. Shielded enclosures provide for the most uncertain prescan results; however, they eliminate the problem of measuring among existing radio-frequency (RF) signals, also known as ambient signals. This uncertainty is due to reflections that add and subtract at the receiving antenna during measurements.

As signals propagate in the shielded enclosure, they reflect off the shield and, depending on the phase of the reflected and the incident signals, add or subtract at the measuring antenna.

This propagation creates the illusion that a signal is significantly below or above the specified limit. For example, if emissions at a given frequency are the result of the direct signal and one or more reflected signals subtracting (i.e., out of phase), then the EUT could in fact be failing and would never be checked at the final measurement site.

The opposite holds true as well. If the direct and reflected signals are added (i.e., in phase) during the prescan and the measured value is substantially greater than the limit, it could take longer to determine, during final measurements, that the emission level was really not as significant as originally thought.

Figure 1 shows a typical arrangement for a prescan in a shielded enclosure and the reflections that occur. Note that Figure 1 only illustrates reflections in a single plane and does not show the expected wall reflections that further complicate the measured results. In addition, Figure 1 shows only two reflected signals as an illustration. The actual number of reflected signals is large and unknown.

Figure 1. Arrangement for a prescan in a shielded enclosure, showing reflections that occur.

Semianechoic Chamber. One type of measurement site that is currently popular is the semianechoic chamber. This type of chamber also yields ambient-free RF environments and reduces unwanted reflections that are encountered in shielded enclosures.

Typically, five of the six sides of a semianechoic chamber are treated with a material that attenuates RF energy as it passes through the material.

Three options are available for room construction. The first is to line five walls with ferrite tiles. The second uses both ferrite tiles and carbon-impregnated cones. The third option uses just carbon-impregnated cones. Typically known as the ground plane, a reflective floor simulates a standard OATS. The theoretical ideal OATS is the model for final measurements, and all alternative final test sites must correlate to the theoretical model.

Making precompliance measurements in a semianechoic chamber lessens the possibility of missing frequencies to be measured at the final test site. The semianechoic chamber reduces the amplitude of wall- and ceiling-reflected signals at the receiving antenna to a negligible level. Only two significant signals can induce voltages at the receiving antenna: the incident (direct) signal, and a single reflected signal from the ground plane (see Figure 2).

Figure 2. Arrangement for a prescan in a semianechoic chamber.

Emissions measurements performed in a semianechoic chamber produce repeatable prescans, significantly reducing the uncertainties associated with prescans. It should be noted that if a semianechoic chamber has a 10-m separation distance and meets volumetric site attenuation, it can be used as both the final measurement site and the prescan site. It must, however, be constructed with the necessary structure height to perform scans at antenna heights between 1 and 4 m.

Fully Anechoic Chamber. Fully anechoic chambers provide for an ambient-free RF environment. A fully anechoic chamber does not have any reflecting surfaces exposed during measurements (see Figure 3). No appreciable (depending on the quality of the RF-absorbing material) reflected signal can be measured at the receiving antenna. This simulates emissions in free space.

Figure 3. Arrangement for a prescan in a fully anechoic chamber.

OATS. It might be thought that an OATS is the best place to make a prescan because the product would not have to be moved to make final measurements. However, many factors must be taken into account before making a prescan at an OATS. The first major factor is the ambient conditions of the local area. Ambient signals can cause many measurement problems such as high noise floors or masked product emissions. One problem area, for example, is the FM radio band. Between 88 and 108 MHz, there is a frequency of significant amplitude every 400 kHz.

If the EUT is radiating anywhere in this band, it becomes very difficult to measure the proper amplitude accurately. A preamplifier does not solve this problem because it does not improve the signal-to-noise ratio in the presence of an ambient signal. Figure 4 depicts a typical OATS ambient. The number of ambient signals throughout this band makes it difficult to perform swept or discrete prescanning. However, with experience and proper measurement-detection techniques, such prescanning is possible.

Figure 4. Typical example of OATS ambient. Representation of typical RF ambient conditions present at most open area test sites.

Measurement Factors

Other important factors contribute to making an accurate prescan. These include frequency-span error, sweep times, and antenna distances.

Frequency Span. Prescans are typically performed in peak-detection mode with wide frequency spans to decrease testing time. However, large frequency spans introduce measurement errors. For a typical spectrum analyzer, the span error depends on the center frequency.

Span error is calculated using the following formulas:

Span error is the difference between the actual frequency of a signal and the measured frequency. Span error creates a shift in frequency during the prescan; therefore, when final measurements are made, the suspect frequency might not be measured. This error can lead to products that fail to comply with the specified emissions limits. Examples of span error are shown in Tables I and II, in which the measurement span is set equal to the bandwidth for two popular antennas. The first is a log-periodic array antenna, and the second is a biconical/log-periodic dipole hybrid antenna.

200–1000 MHz

Full span Accuracy error = 12.12 MHz

Broken-down spans (6 segments) Accuracy error = 2.12 MHz

Table I. Span error examples for log-periodic array antenna.

 

30–1000 MHz

Full span Accuracy error = 14.67 MHz

Broken-down spans (6 segments) Accuracy error = 2.545 MHz

Table II. Span error examples for hybrid antenna.

As shown in Tables I and II, depending on the selected receiver span, it is very easy to create a prescan scenario that misses frequencies at the final measurement. If the prescan results in a measured frequency of 500 MHz, then based on the example for the log-periodic array antenna, an 800-MHz span will yield an estimated error of ±12.12 MHz.

The result is that the actual frequency of interest can be anywhere from 487.88 to 512.12 MHz. If the test lab is not aware of this error or does not investigate this entire range, the suspect frequency could be missed, which could cause a product to be noncompliant. Formulas for determining the receiver or analyzer span error can be found in documentation provided by instrument manufacturers.

In addition to errors associated with receiver span, the sweep time of the analyzer and the rotational speed of the turntable must be such that the sweep rate of the analyzer allows for sufficient sampling when making fully automated prescan measurements. If the span is too large, the analyzer sweep rate could be too slow for the software to acquire a sufficient number of samples. This can lead to missed frequencies at the final test site. Equation 3 defines the analyzer sweep time.

If this calculated time is much slower than the rotational speed of the turntable, then either the analyzer span is too large or the turntable is rotating too fast. In both cases, adjustments are necessary to ensure an accurate prescan.

Another factor associated with receiver span is the system noise floor. Wide spans result in a greater inherent receiver noise floor and a reduced ability to resolve emissions at or near the limit. Furthermore, the amplitude uncertainty is significantly increased when there is a low ratio of narrowband signal plus noise to broadband noise.

Receiving Antenna Height. Antenna height plays an important role in determining the maximum emissions that an EUT produces. During final radiated-emissions measurements from 30 to 1000 MHz, the antenna is scanned between 1 and 4 m over the ground plane. If a prescan is performed at only one antenna height in a semianechoic environment, the maximum amplitude of a measured frequency could be missed.

It is not absolutely necessary to obtain the maximum emission during prescanning; however, if the criterion a test lab uses does not take this into consideration during the selection of test frequencies, it is possible to miss these crucial frequencies. For example, if a test lab requires that any emission within 5 dB of the specified limit must be remeasured at the final test site—and the prescan is insufficient—the maximum emission level of that particular frequency may not be measured at the final test site. In this case, a noncompliant product could be marketed as compliant.

Equation 4 for Ed_max from ANSI C63.5 can be used to better understand the effects of height scanning in a semianechoic environment.² Ed_max is the magnitude of signal received by a tuned dipole antenna from both the direct and reflected signals, assuming a far-field condition.

where d_r = ,

Figure 5 shows the effects of using fixed receiving antenna heights during prescanning at a 3-m measurement distance, with transmitting and receiving antennas set at a 1-m height. Note that without multiple height adjustments during prescanning, signals at 495 and 990 MHz would result in potential emissions that might not be measured at the final test site. There is also a ±7-dB variation in the measured signal as a consequence of having only one reflecting surface. To illustrate the need for height scanning, suppose a test was conducted on an EUT with a 500-MHz processor. In this semianechoic environment, the processor clock and the first harmonic would appear at least 7 dB down from their actual value.

Figure 5. Variations in Edmax using a fixed receiving antenna height during prescanning at a measurement distance of 3 m.

Measurement Distance. The distance between the EUT and the receiving antenna should be greater than one wavelength of the lowest measured frequency. For example, if the measured frequency range is from 30 to 1000 MHz, then the shortest distance between the EUT and the receiving antenna should be 10 m. The wavelength in meters can be quickly estimated using Equation 5:

If the measurement antenna is less than one wavelength from the EUT, the measured signals are in the near field. Measurements made in the near field are erroneous due to nonplane wave illumination and unwanted antenna-to-EUT coupling. The radiation field of interest can be measured accurately only in the far-field condition. Maxwell's equation for E-fields at a predetermined distance is contingent on the distance between the antenna and the EUT.³ The E-field at distance r is given by Equation 6.

where Y equals (2pf/l) – wt.

When r is greater than l/2p, the cubed and squared terms become negligible. The E-field is defined as a function of free-space impedance (@ 377 W). Radiated-emissions prescan measurements made under near-field conditions could lead to data that do not represent the actual emissions from a product. Near-field effects can be reduced by conducting prescan measurements at the distances shown in Table III.

Frequencies <300 MHz are in the near field. l(300 MHz)@ 1 m

Frequencies <100 MHz are in the near field. l(100 MHz)@ 3 m

Frequencies <30 MHz are in the near field. l(30 MHz)@ 10 m

Table III. Conducing prescan measurements at the indicated distances can reduce near-field effects.

Examples

The following examples illustrate the effects of these factors when performing prescans. Figures 6 and 7 indicate the need for height scanning and EUT rotation at a 10-m distance. Height scanning and rotation are applicable to both semianechoic chambers and OATS. A significant number of suspect frequencies would not be measured or would be missed, even when using a 12-dB margin from the limitation to choose final measurement points.Figures 8, 9, and 10 show the differences between using a shielded enclosure, a 3-m fully anechoic chamber, and a 10-m semianechoic chamber, respectively. Each case uses the same emitter.

Figure 6. Sample prescan measured at multiple heights and full turntable rotation, horizontal polarity shown.

 

Figure 7. Sample prescan measured at 1-m height, no EUT rotation, horizontal polarity shown.

 

Figure 8. Example of 3-m measurements in a shielded enclosure, vertical polarity shown.

 

Figure 9. Example of 3-m measurements in a fully anechoic chamber, one antenna height, no rotation of the EUT, vertical polarity shown.

 

Figure 10. Example of 3-m measurements in a semianechoic chamber, one antenna height, no rotation of the EUT, vertical polarity shown.

Figures 9 and 10, representing anechoic and semianechoic chambers, respectively, illustrate the need for either multiple heights or azimuth measurements owing to ground-plane effects. Figure 8 shows that the shielded enclosure can create an emissions profile that could lead to both missed frequencies and frequencies that seem much worse than they really are.

Figure 11 is provided for comparison purposes. Comparing Figures 8 and 11 (the latter a 10-m semianechoic chamber with multiple antenna receive heights and EUT azimuth rotation) at 32 MHz, the shielded-enclosure prescan shows an emission with an amplitude of approximately 40 dBµV/m. However, in Figure 11, the emission is far below the specification. By making a prescan measurement in the shielded enclosure, this point would unnecessarily be investigated at the final measurement site, which could lead to wasted resources. At 110 MHz, the figures show the opposite effect. In the shielded enclosure, the emission appears relatively low and, depending on the test lab's procedures, this point may not be measured at the final site. However, Figure 11 shows that this point would need to be measured at the finalmeasurement site.

Figure 11. Sample prescan measured at multiple heights and full turntable rotation, vertical polarity shown.

A comparison of Figures 9 and 11 demonstrates that prescanning can be performed in a fully anechoic chamber and produce repeatable results that will provide valid data to compare to the final measurement site. However, precautions must be taken against unwanted antenna-to-EUT coupling and near-field effects. A good margin to the specified limit ensures that proper data are measured at the final measurement site.

From the data presented in this article, it is evident that prescanning is not a simple process. It takes sophisticated equipment, software, and good electromagnetic compatibility practices. Final emissions measurements are only as accurate and useful as the prescans that determine the test frequencies. The test environment, antenna placement, antenna height, and EUT azimuths all play significant roles in determining the necessary frequencies to be measured at the final test site.