PCB Repair: Rolling Thunder

John here (first ever blog!)
A game we bought 20 years ago and, after a few failed attempts, was consigned to the cupboard of shame.
At boot it goes through the usual few pages of random tiles/sprites but then freezes.

One initial mistake was to remove the voice PCB, realise it made no difference and then decide the voice PCB was at fault – it’s not in this case.

If you scope both 6809’s, they’re not resetting (rather obvious that it’s not watchdogging). With a logic analyser, you can see the writes to 0x8000 to clear the watchdogs. So the processors are running, but waiting for something. The interrupt line is NOT being cleared by the main CPU, but running MAME in debug and observing the boot sequence shows that interrupts do stop during boot. I ain’t got Philz Skillz with MAME, otherwise I could probably tell from MAME exactly what everything is waiting for.
But, it does feel like the 2 6809s need to speak to each other, or to the sound CPUs. If it’s not the voice PCB, then it’s probably the sound CPU on the main PCB, The squealing sound does suggest something isn’t right in the sound area. A Google of Rolling Thunder repairs shows Caius traced a near-identical fault to this area too. It seems reasonable that failure of the logic ICs that handle where the main CPU communicates to the sound section could cause things to freeze.

schem

A probe around with an oscilloscope shows this waveform from an output of 8B (74LS257) in the sound section.

Whilst most TTL is well short of 5V, this signal is barely over 2.5V with sluggish rise times. Putting another IC on top, but not allowing the outputs to touch, shows a new IC reading much higher.
New IC: