There's a common mistake about how to write up changes, if it is done properly then the process and decision making is a little clearer.
Consider you have Model XYZ configuration 1.7 (hardware, software, mechanical).
The material supplier phases out the ABS plastic, and you need to change to a new type. This then creates Model XYZ configuration 1.8 (C1.8), which is identical to C1.7 except for the plastic case.
Let's say we do some background analysis and decide to do just "ball impact" and "push"(250N) test , while skipping dielectric and ball pressure, stress relief, drop test, leakage current etc are all OK without test. We then do those two tests and write up a report saying it passed just those tests. Put the test report in the design file. End of story.
Well, not really ...
According to the law, you need at least a document (report) that clearly states configuration 1.8 complies with
all the applicable standards (as were used for C1.7).
Compliance with each individual requirement may be based on:
- new tests on C1.8
- assessment that the tests on C1.7 are representative of C1.8 (changes don't impact compliance)
- spot checks on C1.8 on the assumption that C1.7 is representative
- other rationale (e.g. material specifications indicate new design is OK without tests)
Regardless, the document needs to state that C1.8 complies with all the requirements,
including ball pressure test, dielectric strength, drop test, leakage current etc
even if the tests were not done for whatever reason.
Going back to the original request:
"[please tell me] a deterministic method to determine which IEC 60601-1 tests to redo"
The concern is that this kind of approach overlooks the records related to the tests not done. If for example you engaged an expert but give them two options:
1) analyse the difference between C1.7 and C1.8, make a list of tests to redo on C1.8, and make a report just for those differences
2) same as (1) but make a report that says C1.8 complies with all requirements (using data from tests on C1.7 as appropriate)
then you would get two different lists of tests.
By the way, RM isn't the format for this. It's too vague and the requirements in standards are too specific to be handled in an oversimplified RM table. Plus most of the decisions are simple engineering judgement (e.g. old material is ball pressure test rating of 80°C, new material is rated 85°C so obviously OK, no test required). Just make a checklist of
all requirements (compressed as appropriate, but still covering all requirements), go though and record the decision and justification. Make sure it's prepared by a qualified engineer that signs the report and takes responsibility for the decisions.