exxos's Atari, Amiga & retro forum

exxos wrote: 18 Sep 2025 13:59

Badwolf wrote: 18 Sep 2025 13:53 ...whereas the TUNE resistor will be 100% in sync as it's the actual signal.

Right. I somehow assumed the RAM CLK would be higher than 16MHz somehow.. That's what confused me.

Nothing to do with RAM. It's the the CPU and FPU clocks that goes through those TUNE resistors south-east of the CPU. Looking at a photo, I think it's R19 for the CPU.

But anyway, I scoped both TUNE resistor and P92, make no odds either way really.

That's good, but I think you're on the wrong TUNE resistor if you're talking about the RAMCLK. There are four, I think.

BW

Badwolf wrote: 18 Sep 2025 14:20 That's good, but I think you're on the wrong TUNE resistor if you're talking about the RAMCLK. There are four, I think.

I was on R20. It said TUNE :P


So R19..


BLUE is R19, YELLOW P92.


I found some GAL sources for the Falcon. 2 files look wrong kind of format, missing pin numbers. The Other GALs, don't look like CLK is used even though its connected.

exxos wrote: 18 Sep 2025 14:32

Badwolf wrote: 18 Sep 2025 14:20 That's good, but I think you're on the wrong TUNE resistor if you're talking about the RAMCLK. There are four, I think.

I was on R20. It said TUNE :P

Yeah, it looks like your silkscreen has lost the 'TUNE' by R19/18

BLUE is R19, YELLOW P92.

So about one nanosecond. Not bad, but we're trying to rule out as much as possible here, so worth it, I think.

I found some GAL sources for the Falcon. 2 files look wrong kind of format, missing pin numbers. The Other GALs, don't look like CLK is used even though its connected.

Definitely some of the GALs use the clock. There's a delayed AS for starters.

This is what I've been working to. BW

Badwolf wrote: 18 Sep 2025 14:37 Definitely some of the GALs use the clock. There's a delayed AS for starters.

I was just looking at that..


My only thought is maybe the DFB1X /AS being delayed by 7ns, is adding to much delay in the GAL logic, that maybe it screws up something down the line which is close to a clock edge, which further aggravates the problem.

But this would also mean there is actually a flaw in the GAL logic anyway, as a stock Falcon has the same issue, needing a delayed SDMA clock to solve it.

:dizzy:

Did the patch as suggested, probably missed something as I'm not getting anything activity on the DFB1X lights .
What did I miss?

whomper wrote: 18 Sep 2025 14:49 Did the patch as suggested, probably missed something as I'm not getting anything activity on the DFB1X lights .
What did I miss?

You may not have a CLK output on P92, it may only be the latest firmware V105 which does that, I dont remember.

You could move the P92 wire to the left side of R19 as its the same clock.

BTW The delay off pad on the V3 does not look like its had solder on it ? So can you confirm delay off was tried on stock falcon as you mention it only works with delay on ?

exxos wrote: 18 Sep 2025 14:46 But this would also mean there is actually a flaw in the GAL logic anyway, as a stock Falcon has the same issue, needing a delayed SDMA clock to solve it.

I stand to be corrected, but I don't think the GALs will have much to do with things.

That said, I don't think it's entirely obvious the mechanism by which clock skew causes these audio problems anyway, so I am willing to be corrected.

BW

exxos wrote: 18 Sep 2025 14:59 BTW The delay off pad on the V3 does not look like its had solder on it ? So can you confirm delay off was tried on stock falcon as you mention it only works with delay on ?

Yes, I started with delay off.

Badwolf wrote: 18 Sep 2025 15:07 That said, I don't think it's entirely obvious the mechanism by which clock skew causes these audio problems anyway, so I am willing to be corrected.

I think this is why there is so many versions of clock patches with various delays and inversions and god knows what else because the problem was never truly diagnosed.