286 internally divides the clock input by 2. So does 386, where the input is actually called CLK2. So if your chip is running at 8MHz then it needs to get 16MHz at its clock input. Perhaps there is another clock generator, or it's that X3 (with further /2 external divider) that actually provides CPU clock.

I had a look around at the CPU board some time ago when I had the lid off, but I can’t remember there being any crystal that was specifically 16 MHz. Hence, the 32 MHz divided by 2 with an external divider and then divided again by 2 at the CPU would make sense. Can some 74 series logic do the external division?

RetroBard wrote on 2024-08-04, 19:53:

I had a look around at the CPU board some time ago when I had the lid off, but I can’t remember there being any crystal that was specifically 16 MHz. Hence, the 32 MHz divided by 2 with an external divider and then divided again by 2 at the CPU would make sense. Can some 74 series logic do the external division?

Easy enough to do with flip flops. A JK set to toggle mode and an oscillator on the clock input will only flip once (on the falling edge of the clock waveform) where the clock input flips twice (at both the falling and rising edge).

I’ve seen other 286 boards where the cpu crystal is divided by three, so that is likely the case with your board too. Replace the 24Mhz crystal with a 36mhz one for 12mhz.

Thanks for all the responses! It would probably make sense to do some probing onto the board to see which crystal is generating the clock to the CPU. Unfortunately, I haven’t been able to find any schematics of the system.

I’m actually wondering if I need to upgrade both crystals respectively. There’s also the fact that 12 MHz 286s come in PLCC package, where as the 8 MHz one is PGA. That should be easiest to overcome by adding a PLCC socket in between. How tolerant the 8MHz 286 is for overclocking? How would it cope with a 33% clock increase?