And yeah, Teradrive has a discrete YM3438, but in a Discord server where I mentioned this it was brought up that Sega also added some extra logic precisely to fake the YM2612's behavior with the status register (looks like they had already found something that broke by then). I'm not sure how much of the behavior is recreated however, would need further testing (my guess would be that they're remapping everything to $A04000 on reads, but may not have necessarily accounted for the undefined bits of the status register).
On that note, it seems that MD2 VA2/VA2.3 is basically undetectable compared to some model 1 revision (MD1 VA5/VA6, my guess?), so good luck there. Quoting from #general-2 in the Plutiedev Discord server:
I don't think you can detect the buggy ASIC (from what I recall the issues are only exposed on the output)Ace wrote:It's essentially a shrunk-down Genesis model 1.
FM BUSY only accessible from $4000 in the Z80's address space and using a YM2612. However, VA2.3s commonly have a buggy ASIC, the 315-5685, which has misplaced shadow/highlight and is sometimes prone to raster effect glitches.
btw, do all MD1 VA5 return back version %0001 in $A10001 even when TMSS is disabled? (since from what I recall some VA5 boards disable it?)
EDIT: on that note, does any revision ever return any other version values besides %0000 and %0001?