Falcon clock patch V4 series

[Swissfalcon]

I would suggest unplugging netusbee as people have had all sorts of strange problems with that..

Ok, I can confirm that now, with the V4 patch, letting the netusbee plugged lead often the system to hang at boot on the DHCP resolving. It was not the case before the patch.

[Steve]

The netusbee issue is not anything to do with a clock-patch... it's a well known issue with 030 & 060 cpu Ataris. It was recently investigated by ChainQ and fixed by Lotharek. I do not know if he's actively producing the new models yet.

https://www.atari-forum.com/viewtopic.p ... 31#p439631

[Swissfalcon]

The netusbee issue is not anything to do with a clock-patch... it's a well known issue with 030 & 060 cpu Ataris. It was recently investigated and fixed by Lotharek. I do not know if he's actively producing the new models yet.

https://www.atari-forum.com/viewtopic.p ... 31#p439631

good link, thanks.

The only thing I can say now is that my Netusbee worked well before the patch and hang now when booting 50% of the time. Strange.

But you are totally right, I redo the stability testing without the netusbee, it's logical.

[Swissfalcon]

I just did the test without the Netusbee, Nxplayer worked for 2 an half hour, then the computer freezed.

Then I started the Falcon without netusbee with the most basic TOS, only cacheon.prg and NVDI launched, tried Falcamp, hanged after 2 seconds.
Now I launched the exact same test (Basic TOS, cacheon.prg and NVDI, Falcamp on the same song) in 030 mode, and let it loop while I'm going to work. Case is open, regular ATX power supply is used, nothing is hooked on the falcon beside the mouse. If this work, to me it's clear that the problem occurs only when the CT60e is activated.

I'm still looking for a resistor to solder it on the V4 signal path that's lead to the expansion port.

[Swissfalcon]

I have some updates (and a big new problem, but I open another thread for this one).

First, I soldered a 66R on the V4 R216 point (couldn’t find a 68R). This is the signal I got:

IMG_3549.jpeg

Very nice. But the system still crashed after some minutes under MINT. Then I had the idea to measure the SDMA signal after the resistor and compare it with the signal on the CT60e, and this is what I saw:

IMG_3557.jpeg

The signals aren’t synchronised

I removed the 66R, and it’s still the case:

IMG_3558.jpeg

I I do the same measure just before the resistor on the SDMA patch, this is the result:

IMG_3559.jpeg

it’s better, So, it’s look like the SDMA patch is delaying the signal. I will now try the V3 patch instead the V4. I have the feeling the the delay can be the cause of my problems.

Now the bad thing: while measuring on the CT60e, the CT60e shutted down. I didn’t do a shortcut. It was when i triggered the auto calibration button on the scope that the shuttdown happened. Since this, my CT60e don’t start at all :-( I open a new thread to ask some help about this issue.

[exxos]

First, I soldered a 66R on the V4 R216 point (couldn’t find a 68R). This is the signal I got:

Looks a lot better.

Very nice. But the system still crashed after some minutes under MINT. Then I had the idea to measure the SDMA signal after the resistor and compare it with the signal on the CT60e, and this is what I saw:

I made the measurement just and saw about 5ns difference. Though the delay is inherent with the length of the track to the SDMA anyway. Also it is slowed a little by the capacitor and resistor, which there to help with ringing problems etc.

In this case I don't think the V3 would help, that adds even more more delay to the SDMA clock , on top of what you have already. Generally on the Phantom clock patch it adds around 14-22ns delay IIRC. The V3 allowed 3 delay values. So if you think the delay is the problem, then adding more delay with the V3 is likely not going to work. The V3 would have better signal sync as the capacitors were not used on the V3. The capacitors were used to solve Steves FPU problems, and the capacitor was the only fix for him. It's why the V4 moved in that direction.

You can certainly try the V3 anyway, but I don't currently see how it would change anything. The V3 does work differently, but if things were that bad, JOY demo would be failing I would assume. But again, it does not explain why I cannot replicate your issue on my CT60e.. Or on the DFB1X..

I think really we need a another person to try the CT60e with V3 and maybe V4 to see if its something more common or its just limited to your machine. Like I said previously, in the V3 doesn't work, then I have no ideas left. You would have to send me your CT60e to try in my falcon to see if it fails in mine also.

[Swissfalcon]

Thanks. To be more specific, I don’t think that the delay itself is the problem. I think that the fact that the expansion slot and the SDMA don’t have the same delay is. But it’s only a guess, I don’t know. I will try the V4 and report. But first, I have to deal with the CT60e problem :-(

[exxos]

@Swissfalcon You mentioned in your other thread, that your friends ct60e works in your falcon, but does that work reliably in your falcon with the V4 ?

[Swissfalcon]

no, it crash the exact same way, when I play music with Falcam or Mxplayer after some minutes. I tried too.

[exxos]

no, it crash the exact same way, when I play music with Falcam or Mxplayer after some minutes. I tried too.

That's a shame then :( Something must be different between your machine and mine but no idea what. Again id need to try your CT60 in my falcon to see if it's CT60 related or motherboard related issue.

What timings does your friends machine have between the expansion clock and sdma clock ?