Standards issues dominate panel discussion at EMC Symposium

IEEE International Symposium on Electromagnetic Compatibility, Detroit, MI, August 18-22

Martin Rowe, Senior Technical Editor -- Test & Measurement World, 8/26/2008 8:02:00 AM

DETROIT, MI—About 75 EMC engineers gathered on August 20, 2008, for breakfast and to hear panelists discuss several EMC issues. Panel moderator Roland Gubisch opened by saying, “In order to solve the world’s EMC problems, we’ll cover standards, designing for compliance, and accreditation.”

Gubisch: With more and more interference with DTV and wireless, are OATS [open-area test sites] only for horses?
Walton: When you can’t see anything anymore, it’s time to stop testing. [Walton was referring to the fact that many OATS that were once far from ambient signals are no longer, as cell phones and other sources of electromagnetic radiation have moved closer.]
Schaefer: OATS are still used as a reference from a standardization standpoint. There are better tools today [for measuring EMI]: software, standards, and CISPR 16. Methods to verify an OATS are still related to an OATS model. You can only get no reflections in an OATS.
Hill: I was excited to find that a former employer was going to build three OATS, but then built them on top of a hill in sight of TV towers. We had to work only when Channel 11 went off the air. Now, some clients have anechoic chambers and you can use a chamber to do troubleshooting and debug. With the tools today it’s possible to troubleshoot in a chamber, not just on the bench.
Gubisch: OATS are still important for historical purposes.

Gubisch: Speaking of chambers, can you make all the measurements in a 3-m chamber?
Schaefer: Different facets. We prefer 3 m because the chamber is smaller and more economical [than a 10-m chamber]. Making field-strength measurements at 3 m has problems and standards don’t address that. You’re measuring in the near field, which you don’t see at 10 m. Someone measures at 3 m and tries to correlate them to 10 m, but at 10 m, you get different results. You can’t just apply a mathematical formula. It doesn’t work. But, the standard tries to unify the approach so that everyone makes the same mistakes but at least the data is consistent.

Walton: At 3 m, the interferer and receptor can be less than 10 m apart. Testing at 10 m, the nature of your tests don’t work because that’s not how users use a product. No house has a ground plane and using a ground plane for testing places differences [in EMI performance]. Should we go to a 5-m chamber? Most electronic products go into confined area and you should test at shorter distances than 10 m.
Schaefer: Standards are for protection from radio interference. Suppose you have two products that work near each other. The standard we have for interference aren’t suitable. My computer and printer are 20 cm apart, but the test is at 3 m. An immunity standard that’s coming states that you must test at 10 in. apart. We had a Cisco IP (internet protocol) phone and put a cell phone next to it and had acoustical interference. We tested at 3 m with 50 V/m and the product passed. Testing doesn’t test for real applications. Standards need overhaul.

Gubisch: Standards are still evolving.
Audience participant: In the 21 years that I’ve worked at GM, we’ve migrated away from IEEE and ANSI EMC standards. We test in centimeters and millimeters today. We put away the standards and made our own. Standards take a long time to change because they involve compromise. Is there an opportunity for standards to change?
Gubisch: Industry standards need to address unique issues within that industry. International standards are slow to evolve. You must satisfy the needs of your industry. It’s a challenge to the automotive industry.

Gubisch: Regarding design for compliance, is it possible to design to meet all requirements? Are there enough design rules?
Hill: Reminds me of a president who said, “Trust, but verify.” A system was going to be put into operation without testing. We turned down the work because we didn’t think that was possible. EMC engineers and researchers still don’t have the tools to verify EMC performance without testing. Researchers are getting better at simulating a worst-case EMC, but you could overdesign. We’re not getting away from testing any time soon.
Some medical devices you have to model because you can’t measure a person’s exposure to EMC for FCC accepts modeling.

Gubisch: Are standards bodies going too far?
Walton: I was amazed that we could get planes to fly and we put a man on the moon without testing. Over time, we’ve replaced knowledgeable people with those who follow procedures and push buttons. Assessment bodies require collecting a lot of data, but without people that have the knowledge to make products work. I think we’re placing too much emphasis on standards. Assessment is not a bad thing, though.
Fischer: I don’t represent an accreditation body. I now work for the office of law enforcement standards. But I used to work in EMC with NVLAP [National Voluntary Laboratory Assessment Program]. I think that accreditation bodies don’t go far enough. Accreditation bodies are not for bringing knowledge to labs, but to verify that labs can verify performance of a product.
Schaefer: When you decide to accredit a lab, why do you do it? There is a lot of paperwork. It takes away time from technical discussions. We have accredited labs at Cisco. In the regulatory world, sometimes you need accredited labs.

Audience participant: Is the accreditation bar set too low?
Walton: Many labs that have talented people aren’t accredited. In fact, they’re better than some accredited labs. Accreditation places too much emphasis on unimportant things such as logos. We need to get to technical issues.
Fischer: Last year, I saw someone in a lab place a cover over a turntable [used for spinning an EUT at a test site]. People are using different test methods depending on the industry.
Hill: I don’t test for a living. We work with test labs, though. We expect competency in EMC engineers, but the test labs have a wide range of products that they test: military, commercial, and automotive. There is room for improvement by placing more emphasis on training the people who do the testing.
Schaefer: We need to keep in mind what accreditation means. To me, it means the minimum capability that a lab must meet. I don’t believe that just because a lab is accredited that it means the people there know everything.

Gubisch: Moving to automotive EMC standards, can the laws of physics really change country by country? Even within the FCC, different scaling factors apply when a test frequency exceeds 30 MHz. Will automotive EMC standards ever align?
Walton: I used to work in the automotive industry. In fact, I was hired to be on the automotive EMC standards committees. I could tell then that things would never be harmonized. We wrote EMC documents, but we still did our own thing. Experience and expertise warrant doing EMC testing your own way.
Audience participant: The automotive industry get closer to consistent EMC testing, but there are still differences among companies.
Walton: Sometimes, we have to take backwards steps.
Audience participant: About 70% of automotive EMC testing is the same among automakers. We’ll never get to 100%. The light at the end of the tunnel is the end of the tunnel.

Gubisch: Automotive EMC standards don’t address electric vehicles, either.
Audience participant: High-voltage testing has always been there. A raidio doesn’t care if the vehicle is powered by gasoline, diesel fuel, or electrons.
Gubisch: Capturing transients is difficult.
Audience participant: Simulations help.
Hill: Regarding sensor design, EMC immunity was not high enough. They have self interference. Hybrid vehicles are different.

Gubisch: Is using a FFT (fast Fourier transform) good enough for EMC testing?
Schaefer: CISPR includes FFT in a current project. There is a working group addressing the issue. FFT-based measurements are faster than traditional swept-receiver measurements. I expect a standard to include FFT measurements within about 2 1/2 years. It needs to reach a consensus and that takes time.

Gubisch: Regarding testing above 1 GHz, what are the issues?
Schaefer: Site VSWR is for determining reflections in a test volume. We have chambers designed for NSA requirements and we wanted to adapt them for testing to 18 GHz. The chambers would not meet the requirements. We developed a method to find where reflections came from. There is a lot of work to do when adapting a chamber to work above 1 GHz.
Gubisch: There is another method using a network analyzer. It’s a time-domain method. It will not go into ANSI C63.4 just now. The two methods can result in a chamber passing with one method, but not with another. Plus a network analyzer can cost $80,000.

Gubisch closed the session by saying, “These verbal fistfights raise many questions. You can see how standards are important.”