Control conditions of EMI measurements

Make design changes at your facility by correlating measurements that you take at a test lab.

Mark Briggs and Tom Parker, Elliott Laboratories, Sunnyvale, CA -- Test & Measurement World, 8/1/2001

You bring your product to a 10-m open-area test site (OATS) for compliance testing, but the product fails an emissions test. Rather than use expensive lab time to troubleshoot the emissions, you prefer to return to your facility where you can find and fix the emissions problem.

Before you can measure the effects of design changes at your facility, you need to take baseline emissions measurements at the OATS. You can then duplicate the test conditions at your facility and make design changes to solve any problems. Through careful procedures and documentation, you can return to the OATS with high confidence in your design changes.

Assume that at an OATS, you find your equipment under test (EUT)—an Ethernet switch that has four Ethernet ports and a serial port—radiates unacceptable noise levels at 120 MHz and at 140 MHz. Emissions below 300 MHz usually come from I/O cables or power cords rather than from the EUT’s enclosure.

When you return to your facility, you’ll be able to locate and fix the emissions’ sources. But before you leave the OATS, you must identify the cable or cables that radiate the unacceptable emissions. Remove all of the EUT’s I/O cables, leaving only the power cord. Turn off the peripherals, then measure the EUT’s emissions at the frequencies in question. If you still see significant emissions (within 3 dB of the original measurement), the noise may radiate from the power cord.

Wiggle the power cord. If the signal’s amplitude varies with the cable’s motion, the noise most likely radiates from the power cord. You must reduce its contribution before moving to the I/O cables. If you don’t, the emissions from the power cord will mask any emissions from the other cables.

To minimize emissions from the power cord, insert an AC line filter between the cord and the EUT’s AC input receptacle. Remember to “bond” the metal case of the line filter to the EUT enclosure with a short (less than 1 in.) strip of copper tape. Otherwise, you’ll compromise the filter’s effectiveness.

If the filter doesn’t sufficiently reduce the emissions, clamp a split ferrite core around the AC power cable at the point it enters the EUT. The ferrite or the filter (or both) should reduce emissions by 6 dB to 10 dB. If neither technique sufficiently reduces emissions, position the power cord to produce the lowest amplitude and continue to investigate the contribution from the I/O cables.

If the AC power cord doesn’t significantly contribute to the unacceptable emissions, or if you have sufficiently reduced the cord’s emissions, you can identify which of the I/O cables emit excessive EMI. Connect only one I/O cable at a time to the EUT and exercise that cable’s port. Repeat the emissions measurements at 120 MHz and 140 MHz and record the results. In most cases, you’ll probably isolate the emissions to one or two cables.

If time at the OATS permits, try to locate the particular signal line within the cables that produces the bulk of the emissions. Disconnect the cable and touch a 1-m piece of wire to each I/O pin as you exercise the port and observe the emissions level. If you see the emissions level jump, you’ve more than likely identified a pin that may significantly contribute to the overall emissions. From such pins, you can trace the emissions back to their sources inside the EUT. You’ll know where to make design changes when you return to your facility.

Remove any shielded cables and attach the 1-m wire to each cable’s mating connectors on the EUT. If you see a high emissions level at the frequency of interest, you can conclude the connector lacks a proper ground to the enclosure.

After you identify the radiating cables, move them until you find a simple orientation that produces a high signal level. Then, hold the cables in place with tape and photograph the test setup. You’ll need these photos to know how to position the cables in your facility. You’ll probably have two configurations, one for 120 MHz and another for 140 MHz.

Once you’ve identified the radiators, adjust conditions at the OATS to emulate those at your facility, and then take another set of measurements. (To be prepared to make such measurements, you should arrive at the OATS with a list of the specifications of your debug facility and its equipment.) You’ll use those measurements as a baseline when you return to the OATS after you make changes at your facility. Use the same instrument settings—peak detector vs. quasi-peak and resolution/video bandwidths—that the equipment at your facility will allow. If, for example, you have a precompliance EMI lab with a 3-m test site and a fixed-height antenna, make the final OATS measurements at that distance and height.

Figure 1. Tape cables in place and photograph the test setup at the OATS so you can duplicate it in your own lab.

If you don’t have RF antennas and a 3-m test site at your facility, you can measure emissions on cables using RF current probes (Ref. 1). Use the current probes to measure the common-mode current in each cable that produces significant emissions. Move the probe along the cable to find the point of maximum emissions. Then, mark the cable at that point, tape the cables in place, and photograph the test setup (Fig. 1).

Accurately record the EUT’s orientation and configuration, the cabling, and the operating modes of the device. Doing so will help you to create a repeatable configuration and operating conditions in your lab.

At your lab

Once you return to your facility, you will likely make relative measurements. Assume that while at the OATS, you observed emissions at 120 MHz that exceeded the test limit by 6 dB. Also assume you want the emissions to fall 6 dB below the test limit. Take another baseline measurement at your lab and make design modifications until you see a 12-dB drop in that level. The emissions should then drop by approximately 12 dB when you return to the OATS, regardless of the absolute values you measured at each location.

Make only one design change at a time. Document each change with schematics or photos and make spectrum-analyzer plots after each change. While two or more design changes between measurements may fix a problem, if you don’t work one step at a time, they’re just as likely to confuse you later on. If you make more than one change between measurements, remove the fixes one at a time, and carefully note the reverse effect. You’ll verify which modification reduced the emissions.

If your EUT has multiple ports of the same type, modify only one of the ports, and then re-measure the emissions. The new measurement will let you compare the EMI produced by the modified port to that from the unmodified ports. If you see an improvement in the modified port, make the same modification to the other ports. If you see no difference in the other ports, though, remove that change. Don’t modify ports that don’t need changes.

By following a similar approach to the one we’ve described for isolating problems in an Ethernet switch, you can take baseline emissions measurements at the OATS, then return to your lab to make design changes. If you can duplicate the lab conditions at the OATS, you can correlate measurements at your lab with the baseline measurements you obtained at the OATS. This will increase your chances of fixing the emissions problem on your own, which you can then verify upon returning to the OATS. T&MW

Reference

1. Smith, Douglas C., High Frequency Measurements and Noise in Electronic Circuits, Van Nostrand Reinhold, New York, NY, 1993.

Mark Briggs obtained his MSc degree in Electromagnetic Compatibility (EMC) from the University of York, England in 1994. Since then, he has worked in EMC test labs in both the UK and the US. He is currently the director of engineering at Elliott Laboratories in Sunnyvale, CA.