High Frequency Measurements Web Page
Douglas C. Smith

 page header graphic



Technical Tidbit - December 2014
EMC and Other Test Lab Errors

Suspect conducted emissions plot

Figure 1.
Suspect Data from a Conducted Emissions Test

See my one hour webinar on this topic. Click here to see a description of the webinar. This webinar is a real "eye opener." Lots of information and checklists for both test lab customers and test labs to protect oneself. "Trust but verify."

Abstract: EMC testing is a required part of bringing a product to market. Passing the test itself can be a real headache for developers and marketers, but what if the test was performed in error and the test results are not valid! This is a real problem in the more general field of lab testing as well as specifically EMC testing. Some thoughts are shared on the topic.

Figure 1 shows a conducted emissions plot that was supposed to represent the emissions form a product on one of the two power leads (neutral and line are the two leads) but seems to just show the noise floor of the instrumentation. No product is or should be this quiet. If it is, the designer spent way too much money on the design. In this case, the other power lead had the expected emissions so it is just about impossible for the plot above to be correct. I have a similar plot of a radiated emissions test where the data was close to the noise floor of the instrumentation. Given results like these, a test engineer or technician should immediately check to see the EUT (equipment under test) is operating as expected and that the measurement equipment is also operating as expected. In fact, at the beginning of each new product test, the measurement equipment should be checked for proper operation in some manner.

I have personally witnessed many errors made by EMC test labs over the years. For the most part, they did not affect me or my company at the time because the errors were spotted and corrected before they caused problems. Sometimes there was push-back from the test lab to my suggestions, however I prevailed in 100% of the cases. But test labs should not expect their clients to be able to do this or in fact have any EMC experience at all.

Currently I am helping a company who may face significant additional cost because of a combination of miscommunication between the test lab and the company (the test lab must have assumed the company had significant EMC knowledge/experience) and two egregious test errors that yielded bad data upon which the company based business commitments.

Here are just a few of many examples from my experience:
  1. Every test lab where I have been present during a conducted immunity test performed the test in error that resulted in an over test averaging about 6 dB! They tested according to the standard, but apparently did not read the whole standard. Just a few words missed in a test standard can result in bad testing that persists for years.
  2. There are a number of ESD test tables of one design placed in companies and test labs in Silicon Valley that do not meet the requirements of IEC 61000-4-2. They initially do in one respect, but after a month fail in a way that can apply much more ESD energy to the EUT! In another respect, these test tables never complied with the IEC standard to begin with. The effect is to affect the test outcome. I have also seen similar problems elsewhere in test labs across the US, both private and commercial.
  3. I have encountered several labs that made errors in radiated emissions measurements as a result of human error or equipment failure. Many labs take a short cut on the radiated emissions test setup that can increase apparent emissions by a few dB causing the EUT to fail when it should have passed.
  4. Just having a current calibration sticker on test equipment is not adequate and is no guarantee of good test data. All of the bad data I have seen was generated on equipment with current calibration stickers. I tested eight ESD simulators a few years ago. All had current calibration stickers and two of the simulators were defective.
  5. Having current lab certifications does not avoid the above problems either.
So what is a company with a new product to do? What should test labs do since the normal accreditation process does not seem to catch these errors? Using a large, well known EMC test lab does not necessarily avoid these problems either. And although experienced test personnel are a good thing for a lab to have, experience can actually cause certain types of errors. Just because a test has always been done a certain way does not make it correct! The situation is sort of like SCUBA diving. The experienced divers are often the ones that get in the most trouble.

At the least, ask a lot of questions of the test personnel and make sure they understand how your equipment works and possibly what to expect during the test. Have them explain at each stage in the test what they are doing and ask what they expect to see. Ask what measures the lab takes to insure the accuracy of their measurements and test procedures. Did they check their equipment today? Ask the test lab to let you see a copy of the standard they are testing to and, while the test is in progress, read it instead of sitting around waiting for results.

In any event, don't just take a report with data as being good, try to validate the data if you can by pre-compliance measurements or having a second test lab take the same data. Failing a good product is bad enough, but passing a product that in fact should fail is even worse in the long run and can cause lots of business problems.

I am compiling more information and will have more to say at some point, including recommendations for both test labs and designers needing to have their products tested. For obvious reasons, I cannot go into detail on some of the above examples but if you want to discuss your issues either as a designer with a product to test or a test lab, feel free to contact me.

By the way, if you want to file a complaint against a test lab, this A2LA.org link may be useful. I am not aware of a link for a test lab to file a complaint against an unruly customer though.

Summary: Product developers must be on the lookout when bringing their product in for EMC testing. Asking a lot of questions and making sure the test personnel understand how your equipment works can help to get an accurate test. Having someone else look over the results is not a bad idea either. "Trust but verify."

See my one hour webinar on this topic. Click here to see a description of the webinar. This webinar is a real "eye opener." Lots of information and checklists for both test lab customers and test labs to protect oneself. "Trust but verify."

Links in this article:
A2LA.org A2LA test lab complaint page

NEW: Click here, to see descriptions and registration information on my one hour webinars than can solve difficult design problems quickly. Touching on topics like signal integrity, EMC, ESD, and general analog and digital design, these webinars are a quick way to get your design working. There are public presentations every week as well as private presentations on demand for as low as $40/person. New webinars are added periodically.

Check out my public seminar offering in Boulder City, NV. This is one of the best seminar values around because the industry typical fee includes more than just the seminar and lunch. It includes airport transportation in NV, lodging in the historic Boulder City Dam Hotel and Museum, and breakfast and lunch each day. Click here for more details.

The combination of the material presented, the venue, and the passionate presentation will likely make for the best technical short course/seminar you have ever taken! Didn't think a technical presentation can be passionate,? This one is. The result is this short course/seminar will likely be the most informative, practical, and FUN seminar you can attend.

If you like the information in this article and others on this website, much more information is available in my courses. Click here to see a listing of upcoming courses on design, measurement, and troubleshooting of chips, circuits, and systems. Click here to see upcoming seminars in Boulder City, NV.

Boulder Dam Hotel and Museum
Our office, laboratory, and classrooms are located in the
----------
Historic Boulder City Dam Hotel and Museum
1305 Arizona Street, Boulder City, Nevada 89005
Phone: (702) 293-3510
----------
Come for a technical seminar, design review/troubleshooting, or just for a visit and mix a little history of the Old West with your work!


Hangar One
Our Silicon Valley associate office and lab are located at
----------
NASA-Ames Research Center
RMV Technology Group
Bldg. 19, Suite 1073, M/S 19-46C
Moffett Field, CA  94035


Is your product failing ESD, EMC, or performance testing? Attend one of my webinars. Contact me at doug@dsmith.org for more details or click here.

Need help with a design or additional training on technical subjects? Click on the image below to go to CircuitAdvisor.com, an engineering resource for training, news, and fun. The site is being reorganized for group memberships and should be back up soon.

Click here for a description of one of my seminars titled (now also available online as a WebEx seminar):

EMC Lab Techniques for Designers
(How to find EMC problems and have some confidence your system will pass EMC testing while it is still in your lab).



Top of page
Home

Questions or suggestions? Contact me at doug@dsmith.org
Copyright 2014 Douglas C. Smith