When is product certification not required?

In the UK it should be tested and a CE/UKCA  declaration made even for trade samples provided free of charge. In reality between organizations that trust each other, especially  for small quantities transferred say between different bits of the same company, this is not practically enforceable.

Components and kits of parts for self assembly are exempt, but trying to fiddle that by saying its a kit but all assembled apart from 2 screws or something trivial is not acceptable.

From UK to USA is looked at much more carefully than  within the USA but the on paper it is similar.

Mike.

The only thing I'd slightly disagree with is "should be tested", if it can be shown to comply by argument then that would be acceptable (this comes up when e.g. you build a piece of equipment including an off-the-shelf CE marked wireless device - it's not a trivial argument, but it is often possible to make). Personally I'd turn this sentence round and say that a CE/UKCA declaration must be made, and that this may / is likely to need testing to allow this to be made.

I mention it because sometimes companies do nothing rather than pay the, not insignificant, cost of a test house; legally (and I'd suggest professionally) that is not acceptable. However they do sometimes miss the point that, with a bit of expert knowledge, it might not be hugely difficult to build an EMC CE compliance case and documentation at minimal cost, particularly where the project is integrating CE marked sub-systems. You have to do something, but it might be simpler than you think it is. (Disclaimer: I do work for a consultancy that provides such expert knowledge for one specific market, I even still VERY occasionally do it myself for simple cases, but there are plenty of other consultancies available, and their advice may not be needed at all if there are experienced/competent EMC staff in the organisation. This is not a sales pitch :) )

Otherwise fully agree that it doesn't matter even if you're e.g. loaning it free of charge, you are still  "[making it] available on the market as a single functional unit, intended for the end-user and liable to generate electromagnetic disturbance, or the performance of which is liable to be affected by such disturbance". But as Mike suggests, in practice this is generally a complaints led process, it's unlikely to be picked up unless you cause interference to something. However if you ship the product overseas there is a possibility / probability that customs will pick it up.

P.S. I can't comment on the US as it's been over 30 years since I was last involved in certifying products for that market.

Good luck,

Andy

Sorry yes of course, if you are brave enough to claim compliance by desk study alone you may do so unless it is medical or some other safety critical kit.  I'd be chary about an untested original design, and I am seeing it through the eyes of my own experience, which is stuff designed from a clean slate. Testing does not have to be external, but not many places do harder things like EMC in house.
Actually though the paper trail approach is very common, and sensible, when there is a very small change - the Mk I passes by miles, and the Mk II is the same except the supplier of the plastic buttons has changed or something. It may be less safe to substitute a similar rated switch mode supply for an originally linear one and expect to always be OK.

Mike

mapj1 said:

or some other safety critical kit

I have seen it done (picking my words carefully ) for safety critical railway signalling equipment, but of course only when there is a very, very good argument and by people who know exactly what they are doing - and who get the argument checked and double-checked! (Not, as e.g. I once saw, "if we run the motor at half speed that will produce less EMC than if we run it at full speed so it's fine" No, no no!!!!!!)

And of course there's another whole can of worms since testing to an EMC standard doesn't demonstrate that the equipment will actually function safely (or reliably) in the the real world (or allow other equipment to function safely or reliably)...but of course it's a good starting point. Unfortunately as an ISA I do regularly see hazard logs which identify the sole mitigation of EMC hazards as compliance to the relevant standard (by test or otherwise), and I have to gently explain that that's not good enough if the failure consequences are high - and that often I know this because I was one of the authors of the relevant standard and so know very well how limited it is! Personally when I write a safety case (I do occasionally still act as poacher rather than gamekeeper) I work on the assumption that real world EMC could cause pretty much anything to happen to electronics, and that safety protection needs to be by diversity - the chances of EMC creating two different bit patterns to exactly occur in a specific way does become very unlikely if you're careful in your design. Then the EMC testing is to demonstrate reliability, rather than safety, as far as possible. But there's still a good chance that the first time your system goes into the real world it will fall over (safely) on day one because the real world doesn't meet the EMC standard

Apologies that's all rather off topic, just strayed into an area where I have a particular nerdy interest

Cheers,

Andy

Indeed and it is RF /EMC where I too see most chance of coming unstuck for the unwary - two pieces of kit, both meeting EMC emissions, and also immunity may still interfere with each  other if placed side by side. The levels are such that things probably won't interfere if they are next door to each other, unless the victim one is a communications receiver of some sort, but there is enough gaps between the permitted levels that kit that only just makes it to the worst case levels in the standards will struggle if co-located.   As a lot of what I get involved in is either to DEFSTAN or MILSPEC, so we have another 30dB or more  to go in terms of filtering and shielding to get a clear pass, so that tends to lead to over-design on the commercial products, but we still get the odd radiating mains lead of an unfortunate length and that kind of thing.

Mike.