Matti, a couple of comments:
1) Keep in mind that using the wrong power plug with a too-large inner conductor hole, may well have caused dozens of very short power interruptions, perhaps some every time you used it.? So it wasn't necessarily a single power interruption that caused the failure.
2) Hans has described several times why he believes the 1ms scan has very little to do with sub-millisecond power failures, so there may be something else going on here.
3) The 3.3V supply has nothing to do with the PA transisitors, so I wouldn't assume that they are the problem.? The IC501/502 xor gates that drive the clock into the transmitter are powered by the 3.3V supply, so you may want to check those.? Check first to see if there is a transmit-frequency clock on CLK2 out of the 5351, then see if it also appears on the output pins of IC501 and 502.
4) When it is determined what really causes the failure, then we can tell whether adding a large-value electrolytic cap to the input or 12V rail would help - but it probably wouldn't hurt.? Certainly using the correct power connector will help.? That one is easy.
Stan
