If the coil is getting hot then it’s obviously staying energised, which rules out a mechanical problem. A stuck flipper would presumably be on low power anyway if just mechanically stuck anyway, and should not get hot no matter how long it’s held like that (otherwise cradling would be dangerous)
It is certainly feasible that electrical problems manifest when components elsewhere in the machine have warmed up.
A marginal end-of-stroke switch could be not-fully-closed and periodically opening up, triggering the power coil to come back on (as per design - well, at least until the game starts noticing that it's broken and starts ignoring it)
Said EOS switch could also be binding up.
And a partially-not-fully-failed diode could be doing strange things to the transistor by not fully cutting out back EMF. (Yes this is electrical and not mechanical, but the diode lives on the flipper coil and would be grouped with 'mechanical problems' as part of the wire-swap test)
Like I said before - it's zebra-hunting in the extreme but as it stands this would be a quite an unusual failure case for a transistor anyway - one where if the transistor gets warmed-up enough, it becomes de-rated enough to start displaying avalanche breakdown behaviour at way, way lower energy than it should.
It's a little easier to have this behaviour on pre-Fliptronic systems where the EOS is normally-closed instead, and not having it open would keep the power winding on. There would still need to be some sort of failure up-stream to make the game keep the power flowing, though.
I'm used to seeing very weird problems sometimes - the latest one absolute nonsense problem with an absolute nonsense solution I've seen, is this impossible-to-diagnose bad boy:
https://forums.arcade-museum.com/sh...zxTe2FYWNdqUPoeKSMJZTysBZIbxHimCAwI7qSNO0kJmM
Wasted hours upon hours with this one, only to find the solution online and I can say that even if I put another ten hours into the thing, I would never find it.
As to why I was even suggesting such mad things - well, I'd said transistor-at-fault first, and I heard that peace-of-mind by positively identifying the fault before replacing anything was the ideal.
Because I'm me, I took that challenge to my usual extremes.