Compliance Testing Strategies

Question

How do people / companies deal with compliance testing (CE, UL, FCC, etc.) what I would call "platform-oriented and modular" products?

The idea being that a family of products might exist, where a lot of the electronics and firmware are shared among them, but each variant has different modular sub-systems integrated, and corresponding differences in firmware. Let's assume the product is an intentional radiator, using an FCC registered Wi-Fi module to keep things concrete, and that the product is powered by a UL marked 5V AC/DC USB wall-wart.

What strategy do you use to certify for compliance all modular combinations of the product family of products. It feels like modularity is a good design goal, but if it implies hundreds of variations each having to be independently tested for compliance, it seems untenable financially for real products. Is there a nuance I'm missing that makes the economics of modular design work from a compliance standpoint?

In 2018 it's also not uncommon to have field upgradeable firmware. This gives rise to the question of what exactly is tested for compliance? Does updating firmware in the field invalidate compliance marks?

Do the institutions that govern product safety and electrical compliance have allowances for these types of concerns, or are they just becoming dated? I want there to be a method to say: I tested the product in these configurations, and they represent a superset of all possible configurations, and they all met the regulatory standards; therefore other hardware configurations and firmware updates in the same family are compliant, no more testing required. Is that possible? Do product test labs support that sort of thing?

Answer 1

You're definitely on the right track with:

I tested the product in these configurations, and they represent a superset of all possible configurations, and they all met the regulatory standards; therefore other hardware configurations and firmware updates in the same family are compliant, no more testing required.

Having worked at UL for a few years, I can speak to it from the safety side, although I don't have experience on the EMC end. This wasn't an uncommon way of testing these types of modular designs. What you'll want to make clear is how these systems can be put together; i.e. "The end product will use one of these 5 power supplies, one of these 4 I/O modules, 2 of these XYZ widgets, and up to 4 of these ABC widgets."

Then, what you try to do is put together a "worst-case" end-product; the one that deals with the highest voltage, current, and overall power. Maybe this is one combination, maybe its a couple.

Then, the testing is representative of the entire set. Again, this was very common. My advice to you would be to propose your own test plan, explaining why the ones you chose are worst-case. Make it so the testing engineer only has to say yes, rather than try to choose on their own.

Hope this helps on the safety (fire and shock) side.

Answer 2

Product testing laboratories support manufacturing in a lot of ways. I have worked at two places that had an in-house engineer that served as the main contact person with UL and advised the other engineers about UL requirements. They worked with UL to develop in-house testing procedures. The UL inspector usually just reviewed the testing documentation at the development location and visited the manufacturing facilities periodically to make sure that routine testing requirements etc. were being performed properly. A consultant could probably be found to assist you if you don't have an in-house person who can learn to do it. The UL inspector would probably assist with getting an in-house person up to speed, but I don't think they do much if anything without being paid.

You might also look into Intertek ETL Semko. They are not as well known by the general public as UL, but their mark is accepted as equal to UL by all US jurisdictions.

I have no experience with intentional radiators. There are certainly consultants that can help with unintentional radiators and probably intentional radiators also.

The modularity concept should not be much of a problem for UL standards. You may need to do something for each product, but product family similarity will definitely make things a lot easier.

Answer 3

Is there a nuance I'm missing that makes the economics of modular design work from a compliance standpoint?

Its not economical to test every iteration of a product, the testing time cost would be exorbitant. Some iterations need to be tested, such as 220/110V. But not every function of the product needs to be tested under these modes, it's all about satisfying the requirements. One way to satisfy the requirements is to test X,Y and Z modes with A and B modules, do a worst case. Maybe A module and B module are similar and B needn't be tested.

In some cases accessories could be added to a product later, but they can't change the requirements that have been tested. I may be able to add a module that has (lets say a) heater. But if I add a microprocessor to that module with an oscillator, that will need to be tested for an unintentional radiator. If the module has wifi, it will need to be tested for an intentional radiator. If the heater has a voltage higher than the low voltage requirement (if I remember right ~60V) then it will need to be tested because it has a mains voltage, and possibly for saftey. If the heater can go over a temperature that makes it unsafe, it will need to be tested again.

In 2018 it's also not uncommon to have field upgradeable firmware. This gives rise to the question of what exactly is tested for compliance? Does updating firmware in the field invalidate compliance marks?

If you make any changes to a product with an unintentional radiator, an intentional radiator, anything that's listed under safety requirements (high temp, sharp edges, exposed rotating parts, AC mains, ect) or is listed as a critical component will need to be retested if you make a change to that aspect of your product (which will cost money) so design the product so that anything on that list will not need to be changed. (ie, if you have a power supply, or fan, it you will test again if you change one of those. So either test with two power supplies, or find a power supply that can be sourced over the lifetime of the product).

Changes to firmware don't need to be retested, however, if you say change the frequency of an intentional or unintentional radiator (like a DC DC converter) that may need to be retested. But if you tested under all frequencies through regulatory testing, then your good to go.

The idea is to minimize your testing time. The best way to do this is through precompliance (where you can buy lab equipment and test the product as much as you can in house.) Make sure there is a plan for failure (if you think a product may fail an unintentional radiator test because of a component X, build into the design the option for shielding X and have the option ready to install when testing for regulatory).

In the products I have passed through regulatory it is customary to test worst case. For example, if I have a product that needs to meet conducted emissions, and has a range of power it can draw, then test it in normal and max current operations.

This may be confusing, but remember product regulatory testing is an art, not a science. It may be advantageous to get a consultant which will save you money on testing (I've got one in-house).