exxos's Atari, Amiga & retro forum

OPTION2?

If that works, it's definitely the clock.

I don't think it likes that oscillator.

BW

Got it working again..I am using the same 40MHz OSC as before.. But ironically started working when I removed your clock hold off :lol: :hide:

exxos wrote: 17 Nov 2022 16:10 Got it working again..I am using the same 40MHz OSC as before.. But ironically started working when I removed your clock hold off :lol: :hide:

That might break EmuTOS.

BW

More investigation with the 50MHz not being stable. Spraying the freezer spray on the CPU makes it stable. I'm not using the FPU at all currently (it is not plugged in). There is no loading on the oscillator currently either.

I know these CPUs will work on the H5.. But the Falcon 5V is actually slightly low around 4.88V (It is varying slightly as well difficult to tell exactly). so maybe this is just enough to not have it work stable at 50MHz. Of course more caps all over the place may help but it isn't indeed a consideration to make if the Falcon power supplies are not going to be at a consistent voltage across various machines etc.

So I think it is definitely something to consider in all this.. That possibly a 40MHz OSC just be used (maybe 45MHz) or run the 5V on the booster from his own regulator powered from the 12V rail.

Now I know it is simply not going to work at 50MHz. I will revisit the FPU tests at 40MHz again.

exxos wrote: 17 Nov 2022 17:10 So I think it is definitely something to consider in all this.. That possibly a 40MHz OSC just be used (maybe 45MHz) or run the 5V on the booster from his own regulator powered from the 12V rail.

Now I know it is simply not going to work at 50MHz. I will revisit the FPU tests at 40MHz again.

Mmm. I’ll measure my 5V line when I next get a chance.

So it’s possible that fangs trace was actually caused by the CPU failing and not the CPU failing because of the fangs trace?

I’m all for shipping 40MHz osc and letting the user have at it if they want. Makes the soft FPU clock option actually align even better.

(I’ve always been BYO Osc, but the above makes sense to me)

BW

Badwolf wrote: 17 Nov 2022 17:38 So it’s possible that fangs trace was actually caused by the CPU failing and not the CPU failing because of the fangs trace?

The clock switching signal looks perfectly fine.. And then when it crashes it goes to the fangs signal..

But that is what I was saying, that I don't get how that could happen because I would assume the PLD would be directly feeding 50MHz from the OSC to the CPU basically. But I guess it really does not matter anyway. I just thought it was a little odd.

Now I looked at the clock switching waveform more , the duty cycle is actually skewed. But the logic levels are correct.. Not totally nuts like on the fang image.

So now we are hit with another curveball that did the oscillator skew make the FPU start working for some reason. :roll:

Badwolf wrote: 17 Nov 2022 17:38 I’m all for shipping 40MHz osc and letting the user have at it if they want. Makes the soft FPU clock option actually align even better.

Yeah just feeding everything with 40MHz might be the safest bet. Of course you could still have a /2 for problem FPUs.

exxos wrote: 17 Nov 2022 17:55 Now I looked at the clock switching waveform more , the duty cycle is actually skewed. But the logic levels are correct.. Not totally nuts like on the fang image.

Ties in with the cpu throwing a wobbler.

BW

@Badwolf

I think this seems to be a Catch-22 thing here..

When the caps are on the oscillator it seems to upset the CPU more because of the duty skew.. BUT.. the FPU clock is "out" by about 5% ish.. Then when the caps are there.. its pulled closer to 50:50 duty.. then the FPU works..

I wonder if running the OSC on 5V is aggravating things a little bit with it driving a 3.3V IO port basically..

The 40MHz OSC seems to run at 46% duty cycle without caps.. then pulled to 48% with caps.. Be interesting to know what your oscillator is directly putting out...

EDIT:
The 50MHz one puts out 55% duty :lol: :lol: You really couldn't make the shit up :lol: :roll:

As a couple people mentioned DC-biasing.. I did not think it was there but no stone unturned..



New crash screen :excited: :excited: :excited: :excited:


TL;DR

Made no odds.

exxos wrote: 17 Nov 2022 18:08 @Badwolf

I think this seems to be a Catch-22 thing here..

When the caps are on the oscillator it seems to upset the CPU more because of the duty skew.. BUT.. the FPU clock is "out" by about 5% ish.. Then when the caps are there.. its pulled closer to 50:50 duty.. then the FPU works..

I wonder if running the OSC on 5V is aggravating things a little bit with it driving a 3.3V IO port basically..

The 40MHz OSC seems to run at 46% duty cycle without caps.. then pulled to 48% with caps.. Be interesting to know what your oscillator is directly putting out...

EDIT:
The 50MHz one puts out 55% duty :lol: :lol: You really couldn't make the shit up :lol: :roll:

Could the voltage either before or after the oscillator be changed?