We can also be pretty sure that Harry didn't do any mistakes.
@222Lab_Test222, when the time permits check our measurements.
Measure directly on the chip pin. Start with non-working board IC16 data.
The datas are same, no change at all
I'm not sure that you are as accurate as you should.
If you're not changing components yet do ROM data measurements again twice.
First take chips out and measure if you can find any rational connections between ROM socket and inside socket ICs.
(skip pins that you already know are not connected)
Any of socket pins 10, 11, 12, 22 or 24
to
any of F32 or HC125 pins 3, 6, 8 or 11.
Then put ROMs to their correct positions and measure data lines as you did earlier.
Compare old and ne results with strict and pedantic way.
Don't concentrate to timing, concentrate to width of down period,
If your earlier measurement was something else than CPU pins then measure data lines again from CPU pins.
Are they same or just similar?
Then put ROMs to their swapped positions and measure data lines again.
Do old and new comparison using the same strict and pedantic way.
This time concentrate to shape of high parts after a rising curve, like D4 and its lower noise parts and D0 and its repetition of changes.
For D6 check timing between later drop and earlier rising, and active shape after that drop.
And do it again from CPU pins if earlier measurement was from some other position.
Are they same or just similar?
If data measurement result is actually same all the time we can be sure that ROMs have something to do with measuring results.
Reasons can be many, but we must be sure, or it continues to be a wild goose chase.