Falcon clock patch V4 series

[exxos]

Just a quick note here as I'm not at home today so can't study everything in detail...

The waveform on patch is to be expected. It's documented in my investigation.

The only change which may be needed, is the 68R or maybe the 47pf might need to be a lower value as the rise and fall still isn't sharp like in my tests.

If you have a 22pf or change the resistor to 33R and it should clear up the sdma end. But also note changes there while may make signals better, you also have to recheck the fpu clock as making things better one end can worsen the other end of the clock trace.

[Swissfalcon]

ok, thanks for the feedback. I think that I will wait until you are coming back and check in detail.

About the FPU measure, referring as you explained me before on how to measure it:

I assume the FPU ? That is tricky. You have to do it on the back of the cap on the back of the motherboard with some flyer leads probably. Or on top of the FPU socket clock pin. But if people are not seeing the clock as expected on the SDMA, then all the other clocks are pretty much irrelevant at this point anyway.

It may be a little bit too complex for me to make it.

[exxos]

If you move the PSU out of the way, you can probe the fpu clock pins from above as shown in my investigation.The ct60e might be in the way though..

The left clock pad on the V4 IIRC goes to the expansion port clock. It looks ok. But it's what's on the destination which is important.

As to why falcamp crashes I don't know. I assumed it ran in TTram on the CT60 . I assume your running nvdi ? Are you over clocking the CPU ? But if you can zip me files to test I can try my CT60e to monitor behaviour. But if the joy demo works then all should be good anyway..

[Swissfalcon]

If you move the PSU out of the way, you can probe the fpu clock pins from above as shown in my investigation.The ct60e might be in the way though..

You mean here? Second pin from the top ?

IMG_0047.JPG

The left clock pad on the V4 IIRC goes to the expansion port clock. It looks ok. But it's what's on the destination which is important.

hat's what I'm beginning to think. In mode 030 everything's stable, in mode 060 it's unstable. Can I measure the signal on the CPU connector (where the CT60e goes)?My feeling is that the clock signal is good on the SDMA but not so good on the expansion port.

As to why falcamp crashes I don't know. I assumed it ran in TTram on the CT60 . I assume you're running nvdi ? Are you over clocking the CPU ? But if you can zip me files to test I can try my CT60e to monitor behaviour. But if the joy demo works then all should be good anyway..

I borrowed a friend's Falcon this morning for a test. I installed everything that's in my Falcon (CT60e, RAM extension, CF card, CF adapter, keyboard, mouse, Netusbee), only the motherboard changes in the setup. My friend's Falcon has an Atari V1.3 clock patch.

PHOTO-2024-07-12-22-33-44.jpg

By changing only the motherboard In my setup, Falcamp has been streaming MP3 radio for 6 hours now.

IMG_3510.jpeg

So, the conclusion is that my motherboard is the cause of the instability problem. To be clear, with my motherboard, without the V4 clock patch, Falcamp works for a little while, between 10 minutes and an hour. With the V4 patch, it crashes almost immediately, sometimes holding out for a few minutes. When it crashes, the machine is completely frozen. And it's not only Falcamp, the whole system is now instable and will crash / freeze after some minutes, where it was very stable before I applied the V4 patch (minus Falcamp, but as it use the DSP at a high rate, it seems logical to me).

I'm uploading the whole image of my system, so you can test it, I will PM you the link.


EDIT download link submitted in PM. If someone else want it just PM me, it's a whole complete system image with MINT, Magic an TOS extended.

[Steve]

For now I think you should try the V3 patch. Bare in mind on the V3 you have various SDMA delays you can experiment with too. So I'd say, give that a test for stability.

[exxos]

You mean here? Second pin from the top ?

Yes.

In mode 030 everything's stable, in mode 060 it's unstable.

The clock patch does not really have much to do with the 060 other than the clock feed to the expansion port, which you already documented was reasonably good.

Can I measure the signal on the CPU connector (where the CT60e goes)?My feeling is that the clock signal is good on the SDMA but not so good on the expansion port.

If you look at the Falcon port pinout you should be able to see where the clock signal goes.

I borrowed a friend's Falcon this morning for a test. I installed everything that's in my Falcon (CT60e, RAM extension, CF card, CF adapter, keyboard, mouse, Netusbee), only the motherboard changes in the setup. My friend's Falcon has an Atari V1.3 clock patch.

Unless you put the V4 into that same machine, the test doesn't really prove anything unfortunately.

By changing only the motherboard In my setup, Falcamp has been streaming MP3 radio for 6 hours now.

Changing the motherboard is completely changing the entire system.

So, the conclusion is that my motherboard is the cause of the instability problem. To be clear, with my motherboard, without the V4 clock patch, Falcamp works for a little while, between 10 minutes and an hour.

where it was very stable before I applied the V4 patch (minus Falcamp, but as it use the DSP at a high rate, it seems logical to me).

You also said

In mode 030 everything's stable

sorry I don't follow, those statements seem completely contradictory to me ? It's either very stable, or it is with or without the V4, which is it ?

If your machine is unstable in any way without the V4 clock patch, and adding the V4 clock patch your machine is still unstable ..
You probably have a problem with your CT60 or something else.

With the V4 patch, it crashes almost immediately, sometimes holding out for a few minutes. When it crashes, the machine is completely frozen. And it's not only Falcamp, the whole system is now instable and will crash / freeze after some minutes,

I don't know how you could end up in that situation unless your CPU is overheating etc. I've been running the V4 in my falcon with the DFB1X for months without any problems. So I don't see how it can be totally unstable in your machine. Your machine completely stable before fitting the V4 ?

It sounds more like you are having overheating problems with your CT60 at this point to me. Have you simply tried removing the CT60 and running falc amp to see if its stable ?

IIRC , the DSP I it's on its own dedicated clock. Also IIRC, it goes via the SDMA.

I suggest you try removing the 47pF capacitor and see if that helps. Also measure the clock on the expansion port. There is only 20R in series with that to help with the ringing. You could try bridging over the left side resistor for more sharper rise and fall times to the expansion port. But you haven't said if you are overclocking the CT60 yet. Pusing things to the limits , even 1ns would be enough to cause problems.

I'm uploading the whole image of my system, so you can test it, I will PM you the link.

Thanks, but I'm afraid I cannot download a 32GB file, I have a 40GB download limit per month. But me trying to restore your image would likely end up not working anyway. I need a minimal set up to try which ideally should fit onto floppy as I cannot easily transfer files to my Falcon currently. Anything outside generic TOS404 gets into to much complexity in resolving problems because of the amount of software layers involved.

Though I think this is going to be too involved to any sort of "remote diagnostics". For this level of diagnostic work, I''m afraid you would have to send me your motherboard with your drive setup for me to diagnose with my own equipment here.

[exxos]

I found Falcamp here http://deunstg.free.fr/sct1/falcamp/download/index.html Last updated 2006, is it even stable on the CT60 ?!

I played 3 MP3s and it all seems fine with and without the DFB1X.

If anyone knows of a really long free MP3 track, I can video it and leave it all playing though and upload it to youtube after.

I only noticed the DSP does get rather hot..

[Swissfalcon]

ok, I try to summarise better the situation:

- before I applied the V4 clock patch, my falcon was very stable. I could run mxplayer for more than 48h with more than 10 opened programs without having the computer hanging. The only problems I had was the Joy demo that wasn't able to play correctly most of the time in 030 mode (but the 060 version that don't use the blitter always worked) and Falcamp that hanged after some playing time.

After fitting the V4 clock patch, the joy demo work now always, I tried to launch it 50 times with success. But when I activate the 060 mode, then the machine become unstable. Everything I do will lead to a total freeze of the machine after some minutes. Falcamp crash almost immediately and mxplayer don't play more. than 20 / 30 minutes before the computer freeze.

My 060 run at 100 MHz with a good cooler. The exact same CT60e run well at the same speed on the other Falcon (the one that I borrowed to my friend) right now. And it worked good before the V4 fitting.

If you look at the Falcon port pinout you should be able to see where the clock signal goes.

I will try to find that and measure it.

Unless you put the V4 into that same machine, the test doesn't really prove anything unfortunately.

the purpose what to show that the CT60e run well and that the software installation isn't the problem.

sorry I don't follow, those statements seem completely contradictory to me ? It's either very stable, or it is with or without the V4, which is it ?

If your machine is unstable in any way without the V4 clock patch, and adding the V4 clock patch your machine is still unstable ..
You probably have a problem with your CT60 or something else.

no, the machine was stable before the V4 patch, with the exception of joy and falcamp. And I'm pretty sure that if joy can crash because the clock signal isn't good enough, then falcamp can too, as it make heavy use of the DSP.

I don't know how you could end up in that situation unless your CPU is overheating etc. I've been running the V4 in my falcon with the DFB1X for months without any problems. So I don't see how it can be totally unstable in your machine. Your machine completely stable before fitting the V4 ?

Yes, was completely stable before fitting. CPU run at 42 degrees.

It sounds more like you are having overheating problems with your CT60 at this point to me. Have you simply tried removing the CT60 and running falc amp to see if it's stable ?

The falcon run stable without the CT60e, yes. And there is no overheating. 42 Degrees on the CPU, and the PCB is cool when I touch it.

I suggest you try removing the 47pF capacitor and see if that helps. Also measure the clock on the expansion port.

Totally remove 47pF? Ok, I can try.
I will try to find where to measure the clock on the expansion port. I'm pretty convinced that the problem sits there.

There is only 20R in series with that to help with the ringing. You could try bridging over the left side resistor for more sharper rise and fall times to the expansion port.

Where ? On the V4 itself?

But you haven't said if you are overclocking the CT60 yet. Pusing things to the limits , even 1ns would be enough to cause problems.

Running at 100 Mhz. Was always fine before. I tried to low the clock down to 90 MHz, it doesn't help either. Still crashing.

Thanks, but I'm afraid I cannot download a 32GB file, I have a 40GB download limit per month. But me trying to restore your image would likely end up not working anyway. I need a minimal set up to try which ideally should fit onto floppy as I cannot easily transfer files to my Falcon currently. Anything outside generic TOS404 gets into to much complexity in resolving problems because of the amount of software layers involved.

Ok. I tried Falcamp under TOS 404 (with local MP3) too, in 030 mode ok, in 060 mode crash too, so it's not OS related. I make you an archive with my version of Falcamp and 2 MP3 that I use, so you can maybe just load them both and let play then in repeat mode in 060 mode, should do the trick.

Though I think this is going to be too involved to any sort of "remote diagnostics". For this level of diagnostic work, I''m afraid you would have to send me your motherboard with your drive setup for me to diagnose with my own equipment here.

Ok. It's maybe a solution if the V3 can't help. Before I will try the V3.

[Swissfalcon]

I found Falcamp here http://deunstg.free.fr/sct1/falcamp/download/index.html Last updated 2006, is it even stable on the CT60 ?!

I played 3 MP3s and it all seems fine with and without the DFB1X.

If anyone knows of a really long free MP3 track, I can video it and leave it all playing though and upload it to youtube after.

I only noticed the DSP does get rather hot..

Yes, this is this version that I use. great. Can you try it with the C60e? With the 060 de-activated (in 030 mode), it run well. In 060 mode on my machine it crash.

But on the other falcon, on the CT60e, it run and stream a web radio now since 10 hours. So, yes, it's a good question, because I know that many say that Falcamp isn't stable, and this was the purpose of my testing on another motherboard : see if the problem come from the software of the CT60e. It's not the case :

IMG_3514.jpeg

[Swissfalcon]

For now I think you should try the V3 patch. Bare in mind on the V3 you have various SDMA delays you can experiment with too. So I'd say, give that a test for stability.

I think that it's maybe the best option yes.