Not sure how anyone else is getting on but I found 2102A to be just as much of a nightmare as I think sklpl did with 2101A. Restoring MEMCH or SCHBK from a 1912A firmware version didn't help. Then it froze several times. When i reverted back to 1912A then it wouldn't backup system.csv properly either, it set several memory channels to PASS, and even at one stage started randomly scanning a priority channel that wasn't even in the memory banks every 4 secs.
Luckily I tried sklpl's method of loading 2102A, then over writing one of the memory channels. Thanks for posting that sklpl. After that, the rest of the .csv settings magically worked. I just had to re-write that single memory channel to get back to where I wanted to be. So I strongly recommend you make lots of backups at each stage.
>Some users reported following issues.
Maybe when AOR say 'some users' they actually mean 'everyone who doesn't work for AOR'?
I've worked with 100's of consumer-grade and technology complex items before, and never known any bit of electronics to have such a flaky firmware upgrade path: I'm not sure who AOR used test this upgrade, but they must have a pretty low IQ to have passed it (and no I don't mean I/Q).