Hello Peter Selvey,
Thanks a lot for the detailed explanation. It looks very plausible.
So, to sum up, I can distinguish two critical components lists (ССL): "live" CCL and CCL "for a testing agency".
"Live" CCL is based on risk analysis and involves evaluating risks of appearance HAZAROUS SITATIONS i.e. if malfunction of a component can lead to electrical shock, fire, violation of EMI requirements or deviations from ESSENTIAL PERFORMANCE etc., so, this component should be recognized as Critical component. It's important to make your own "live" CCL to ensure your device is really safe and reliable.
CCL "for a testing agency" is a formalized table, which depends on a particular agency and is based rather on an established practice than on genuine risk analysis. It can be reasonable. CB test (at least ours) doesn't involve evaluation of a risk analysis file, therefore it's easier to use forms "to fill in" than try to evaluate reasonability of including each component to CCL and actually to accomplish risk analysis one more time.
So, an algorithm of creating CCL "for a testing agency" can be like this:
1. To accomplish risk analysis and create your own "live" CCL.
2. Basing on information from the Internet (special thanks to the members of this thread) about "typical" critical components create rough CCL "for a testing agency".
3. In cooperation with the particular test agency clarify CCL "for a testing agency"
I think, it's will be correct to use both CCLs for a production process.