Is a Design Change Without A Re-Test Too Risky?

EMC Re-test decision

The frustrating question often came up when I was designing electronic products whether a minor hardware change to an already tested product would mean we’d have to go through EMC testing all over again. I could never find a concise answer on the net, and the test lab never seemed to give us the full story. We didn’t know who had been fined in the past and what for. Ultimately we weren’t sure how big the risk was if we didn’t re-test. If you want to know in what situations fines are issued, how much they were and also the responsibilities and risks of re-testing, read on..

What type of equipment do you have?

Firstly, I’ll break the guidelines down a bit because the answer is different depending on the type of product you make. From the FCC’s perspective, lets just break it down to wireless and non-wireless devices. If you would like a similar post on CE testing in the future, let me know in the comments section below.

Non-Wireless Devices (FCC)

Products that don’t incorporate a wireless transmitter are typically subject to the ‘verification’ or ‘declaration of conformity’ authorization procedures. This is a fancy way of saying an EMC test lab needs to measure the radiated and conducted emissions coming from the product and your test reports most likely don’t need to be filed with a TCB. You can determine if your product is subject to these authorization procedures by checking out the table in Part 15.101 of the rules.

If your product is subject only to ‘verification’, then the paragraph you need to know is:

Guidance for devices subject to ‘Verification’

“In verifying compliance, the responsible party … warrants that each unit of equipment marketed under the verification procedure will be identical to the unit tested and found acceptable with the standards and that the records maintained by the responsible party continue to reflect the equipment being produced under such verification within the variation that can be expected due to quantity production and testing on a statistical basis.

Verified equipment shall be re-verified if any modification or change adversely affects the emanation characteristics of the modified equipment. The party designated in §2.909 bears responsibility for continued compliance of subsequently produced equipment.” (FCC CFR47 Part 2.953)

As you can see from the text in bold, it’s a bit of a chicken and egg scenario. Of course, to know 100% whether the change you made adversely affected the ’emanation characteristics’, you would need to re-verify the test results.

Guidance for devices subject to ‘Declaration of Conformity’

The FCC is a bit clearer with the guidance for DoC authorized devices:

Equipment shall be retested to demonstrate continued compliance with the applicable technical standards if any modifications or changes that could adversely affect the emanation characteristics of the equipment are made by the responsible party. The responsible party bears responsibility for the continued compliance of subsequently produced equipment.” (FCC CFR47 Part 2.1073(d))

So this leaves it up to the manufacturer to determine whether they believe the changes they made could negatively affect the emissions profile.

The ultimate answer would be to re-test (or at least do a short pre-scan), but in reality there is probably some wiggle room.

How Much Wiggle Room Do I have?

Let’s keep in mind that the measurement uncertainty of radiated emissions at an EMC lab are mandated to be within +/- 4dB from an ‘ideal’ site. This means that an emission measured at one lab may be as much as +8 dB higher than the same emission measured at another lab. That’s pretty big. If the FCC is investigating a complaint of a non-compliant device, they’re likely going to give you a bit of the benefit of the doubt since an emission that they measure could perceivably be 8 dB higher than when you had your device measured at another lab. Not to mention the natural device to device variance. Ultimately they will review the measurements compared to the actual limits, but from a risk analysis perspective it means that you essentially have some wiggle room for making minor changes if you started with a lot of margin.

Practically speaking, a test lab re-measuring the emissions from your product would not be able to tell whether an increase in a particular emission amplitude came from:

(a) Test set-up (did they recreate an identical test set-up?)

(b) Product to product variance

(c) Lab to lab measurement variance (potentially +/- 8 dB)

(d) A design modification

If I was wondering how much wiggle room I had, I would consider what the perceivable impacts were of the change I made and also how far below the limits the product was to begin with. If I changed the value of a pull up resistor for instance on a static GPIO net, I’d think the risk of adversely affecting emissions would be pretty damn low. On the other hand, if I changed a processor out for a completely different model, then I would have no idea what the new emissions profile would look like, so I would get the product re-tested. Or if I changed the termination resistors on bus signals that I knew were oscillating at a frequency listed on a previous test report as 2 dB below the limit line, I would call that risky.

Here is a non-exhaustive list of changes that I would perceive to be high risk:

  • Changing any active components to components with completely different switching/power characteristics
  • Changing the enclosure material
  • Any components around the power supply that could affect conducted emissions (switching supplies, smoothing circuits, rectifiers etc).
  • Changing LCDs
  • Increasing the port count (always test with all ports enabled)
  • PCB stack-up changes

And low risk:

  • Pull up resistors
  • Protection circuitry
  • Low speed GPIO routing
  • Identical part substitution
  • Adjusting connector placement slightly

The point is that common sense should prevail. If you seriously don’t know what impact your change might have on emissions, then get it re-tested.

What’s the Risk of Being Fined by the FCC?

FCC fines

So you decided to go ahead and make a change to your product without having it retested and you’re worried that you’re going to be busted. Read on below for a history of the fines issued by the FCC for non-compliant devices (it’s not as bad as you think)…

Take a look at this history of legal actions by the FCC on product manufacturers from 2006-2012 compiled by Fish and Richardson legal firm. You’ll see that the vast majority of the penalties were for willfully marketing and/or selling electronic products that were not tested at all. Very few (under 10, in 6 years!) were violations where previously tested equipment was later found to be non-compliant. Now, I can’t vouch that this is a complete list of violations, and it certainly doesn’t list any lawsuits brought by private companies or individuals for any resulting liabilities due to non-compliant devices. But it would be easy to see how a manufacturer could read this enforcement list and perceive that the risk of being fined by the FCC for modifying a previously tested device is low, especially for non-radiating devices.

Ultimately the risk lies with the manufacturer, so use your judgement wisely and consult a lab or legal team if in doubt.

If you want to know more about the potential legal liabilities of non-conforming equipment, I highly recommend scanning this “The Legal Aspects of Regulatory Compliance” or contacting Fish and Richardson directly. This presentation delves a bit deeper into the legal aspects of compliance and the theories of liability.

Intentional Radiators (FCC)

If your product has a transmitter in there, the rules are much clearer. These devices are generally subject to the ‘certification’ authorization procedure, which means your product will need to be tested at an EMC test lab, then you’ll file an application with a TCB. You will probably have to include your schematics and bill of materials, so it’s much easier for the FCC to track modifications made to your device after certification.

They also undertake market surveillance to ensure that (a) the original test lab performed the testing adequately and (b) to ensure ongoing compliance. The FCC mandates all TCBs to perform follow up market surveillance on 5% of all transmitter certifications.

The FCC have issued a full PDF (178919 Permissive Change Policy) outlining the different types of changes you can make to a certified transmitter. Note that any modifications to the baseband circuitry not associated with the transmitter are subject to the DoC or Verification guidance above. i.e. the information in the PDF only apply to changes specifically to aspects of the RF transmission portion of your circuit.

There is too much information within the Permissive Change Policy to cover in this post, but essentially it covers the following types of changes to your device:

• Antenna changes
• Printed Circuit Board (PCB) and Hardware changes
• Enclosure changes
• Software changes
• Miscellaneous changes

Depending on the seriousness of the change you’ve made, you may have to do one of the following things:

  • File a completely new certification application (the device will be treated as a different transmitter)
  • File a modification to your existing certification
  • Just re-verify a subset of the RF tests and keep the measurement data on file (don’t need to file anything with a TCB)

Let me know in the comments below if you’d like a clearer look at the permissive change policy!

Wrap Up

Deciding whether or not to re-test your product after a design modification is tricky. I hope you now have a better feel for the risks and responsibilities for making this decision.

What’s your experience been with re-testing? Ever had an issues crop up with a regulatory body? Any questions? Use the comments section below…

Also, if you liked this article and found it useful, please forward it to your colleagues and ask them to sign up to the mailing list so that they can keep receiving more valuable info like this!

The Obligatory Disclaimer:

I am not a lawyer or a spokesperson for any regulatory agency. This article is not advice. Do not rely on this opinion to make a decision whether to test your product or not. You’d literally be crazy to make a choice based on an article you read on some blog on the internet by some guy you don’t even know. Always contact an accredited lab or a regulatory agency authorized to give legal advice if in doubt.

 

13 Comments

  • Nusreta

    Reply Reply May 13, 2014

    Hi Andy,
    I want to thank you for this article and your blog. Ofcourse I want to recommend it for reading to my colleagues. It was very knowledgeable. In my opinion I should retest exept the changes you had in your list (pull up resistors …), but I have a lab and want to be sure that hardware/software changes does not influence the performance of testing product.

    A risk analysis is also possible to do before retesting just to decide if you need to retest or not with all involved in the product changes.

    BR
    Nusreta

  • Andy Eadie

    Reply Reply May 13, 2014

    Hi Nusreta, thanks for the feedback. A risk analysis is a good idea!

  • Ben Hewson

    Reply Reply October 8, 2014

    Another great article Andy, it breaks down this often asked question very well, and you make this it easy understood and follow for all levels – and examples are terrific and I especially enjoy how that triggers others to leave other examples in the comments – your writing is very engaging.

    Kind regards,
    Ben

    • Andy

      Reply Reply October 14, 2014

      Thanks Ben!

  • Kausalya

    Reply Reply December 9, 2014

    Dear Andie,

    The article is good. It creates some awareness that Engineers should not proceed with assumptions. It is always better to clarify with Labs and then make decisions. Company reputation is very important!!!

    Regards,
    Kausalya.

  • Claudio

    Reply Reply December 31, 2014

    Hello (from Brazil). Nice article, Andy. It has always been a difficult question, sometimes tough for engineers with their managers, production and the whole company pushing for the release of the modified product.

    Every company, engineer or design group will have different attitudes before this situation depending on natural responsibility of people, company culture, resources availability, emergency situations, fear of reputation damage, EMC knowledge/awareness. In spite of your disclaim note at the end of your article, I think it may (for convenience) affect an important variable in the risk analysis of those who read it.

    Anyhow, my only (obvious) advice/reminder is: if you decide not to go through re-certification process, make a good technical report describing the analysis, considerations, calculations, partial tests if any (e.g. measurements with o-scope showing that key waveforms did not change), previous similar experience etc. A report showing that the decision passed through a proper analysis. It may be helpful in the future. In addition, it can be used to show to the rest of the company how difficult and technical such decision can be.

    Regards
    Claudio

    • Andy

      Reply Reply January 3, 2015

      Thanks for the feedback Claudio and the additional advice. Many companies would do well to approach the problem in a similar systematic way that you do. As you say, the backup paperwork will be very handy to show authorities if you do run into problems.

      That reminds me that the FCC allows measurements from GTEMs in some circumstances which may suit relative emissions measurements of small design changes. See KDB 40830 and 54796 for more details.

  • Shamsi

    Reply Reply August 18, 2015

    Thank you, Andy.

    Very useful article, because :
    Time by time we are making a necessary changes in our product design, specially due of component EOL.
    And it is always a big question to answer: test or do not test for particular standards?

    Sincerely,
    Shamsi

  • Krunal Patel

    Reply Reply September 1, 2015

    Hey Andy,

    I was just asked the same question this morning by my manager. Great timing to read this article.
    I am glad to have read your article and I liked the comment from Cluadio very much.
    I think there should be enough analysis done before we decide whether to re-test or not, alas there is no standard available for the analysis and it is yet again judgmental.

    Thanks Andy, great article!

  • Juan

    Reply Reply October 4, 2016

    Hi Andi!
    Thank you very much for the post and the blog in general. It gives practical examples really helpful for the developers, like me, that are starting on this EMC world!

    I am working for a company developing products for UE market that’s why it would be great for me if you could make a post about CE regulatories. Although from what I read I guess it is very similar.

    Thanks again and keep going with the blog!

Leave A Response

* Denotes Required Field