Raven060 (luciodra build).

luciodra
Site sponsor
Site sponsor
Posts: 96
Joined: Fri Jun 28, 2024 1:59 pm
Location: Rome

Re: Raven060 (luciodra build).

Post by luciodra »

agranlund wrote: Wed Nov 20, 2024 4:49 pm
luciodra wrote: Wed Nov 20, 2024 8:50 am with quartz 48.00 x1 the system is unstable (presents numerous artefacts on the screen).
1x is not yet completely implemented as you noticed.

I noticed...
luciodra wrote: Wed Nov 20, 2024 8:50 am With quartz at 48.00 x2 the system is very unstable
with quartz at 24.00 x2 the system is quite stable
Since my 060 is a 75 MHz, if I used a 35.00 x2 quartz what results should I expect?
The answer would have to be "don't know" :)
It's probably fine but I'm mostly running 96/48 myself.
I used to test 50/25 quite a bit as well because I like the idea of having some kind recommended / tested configuration for if one has something that isn't happy with the overclock.
Perhaps I should treat 64/32 as that instead of 50/25.. 32mhz is still going to be within spec of the slower 33mhz version 68150, and most 68060's will probably do 64mhz?
(64/32 got me to the desktop last I tried it but I don't think I did any exhaustive tests)
I think I will try to buy a 35.00 to try, but now I have to clarify the role of the MC68150FN33 or 40?
User avatar
agranlund
Site sponsor
Site sponsor
Posts: 1121
Joined: Sun Aug 18, 2019 10:43 pm
Location: Sweden
Contact:

Re: Raven060 (luciodra build).

Post by agranlund »

luciodra wrote: Wed Nov 20, 2024 4:10 pm My 68150FN40 was also purchased on Aliexpress. Could it be that?
Unfortunately I only have one and I should try to buy another... but are the 68150FN33 also good?
68150FN33 is rated for 33mhz so it's probably even less likely to accept being overclocked to 48mhz.
I would personally stay away from those unless you cannot find a 40mhz version (68150FN40)

I'm generalizing that a 40mhz might be more likely to accept 48mhz. In reality, who knows, perhaps manufacture date/rev plays part too.
I had one FN33 running happily at 48mhz and another that didn't accept any overclocking at all.
The FN40 I normally use is happy with 48mhz but that can't guarantee all of them are.

In terms of speed you are looking at both the 68060 and 68150 in terms of how much overclock (if any) is going to be stable.

68150 is clocked by your main oscillator = bus speed
68060 is clocked by bus speed x2
luciodra
Site sponsor
Site sponsor
Posts: 96
Joined: Fri Jun 28, 2024 1:59 pm
Location: Rome

Re: Raven060 (luciodra build).

Post by luciodra »

I have some doubts:

- I can't install Freemint :(
- the system is always unstable, more at 96MHz than at 48MHz, but always unstable. Can it depend only on the 68150 or can I see other things ?

Thanks.
User avatar
agranlund
Site sponsor
Site sponsor
Posts: 1121
Joined: Sun Aug 18, 2019 10:43 pm
Location: Sweden
Contact:

Re: Raven060 (luciodra build).

Post by agranlund »

I was going to suggest checking the ram perhaps but since you mention it being more unstable at 96 compared to 48 it makes me doubt that being the problem. Still, it's easy to do a few tests so I would test with only a single ram inserted in slot 0. And then again switching that stick out for one of the other ones.

Never hurts giving YARTT a try too:
https://www.atari-forum.com/viewtopic.php?t=30110


When it's unstable, what does that mean specifically?
Just a completely silent hang or do you get an Emutos panic message? If you press the NMI button, do you get the usual printout of the registers (if so, what do they say)? No detail is too minor really.

Can any kind of pattern be noticed for when the symptom is more often triggered? Ie; during disk activity, running certain programs etc.
Is the disk activity LED always lit when and during whatever the symptom is?
If it's just sat at the desktop, is it completely stable then or will it eventually crash(?) anyway?
Can it depend only on the 68150 or can I see other things ?
With the information at hand it could be anything really. It could be the 68150 but it could also be anything else.
Anything from aliexpress/ebay usually automatically ends up at least somewhere on my suspect list.

The 24mhz would make the 68150 go at 24mhz which is within spec, and the cpu would go at 48mhz which is also within spec.
That said, I never tried 48/24 so I cannot guarantee there aren't timing bugs related to that particular speed that I don't know about. I don't think so, but I have no proof since I haven't verified myself. However, I was running mine just fine at 50/25 for some time before though so I'd file that lead under;
"if you have a 25 then I'd use that and get rid of one variable, but if you don't I wouldn't go out of my way to get one of those just yet"

Removing things is usually an easy way to quickly dismiss suspects, but there's not much you can remove since I assume you need a running desktop to trigger it. Replacing socketed stuff regardless of evidence is also quick and easy if you happen to have spares lying around.

Ideally some information can be had from when the symptom happens and clues can be gathered from that.
luciodra
Site sponsor
Site sponsor
Posts: 96
Joined: Fri Jun 28, 2024 1:59 pm
Location: Rome

Re: Raven060 (luciodra build).

Post by luciodra »

agranlund wrote: Tue Nov 26, 2024 2:45 pm With the information at hand it could be anything really. It could be the 68150 but it could also be anything else.
Anything from aliexpress/ebay usually automatically ends up at least somewhere on my suspect list.

The 24mhz would make the 68150 go at 24mhz which is within spec, and the cpu would go at 48mhz which is also within spec.
That said, I never tried 48/24 so I cannot guarantee there aren't timing bugs related to that particular speed that I don't know about. I don't think so, but I have no proof since I haven't verified myself. However, I was running mine just fine at 50/25 for some time before though so I'd file that lead under;
"if you have a 25 then I'd use that and get rid of one variable, but if you don't I wouldn't go out of my way to get one of those just yet"

Removing things is usually an easy way to quickly dismiss suspects, but there's not much you can remove since I assume you need a running desktop to trigger it. Replacing socketed stuff regardless of evidence is also quick and easy if you happen to have spares lying around.

Ideally some information can be had from when the symptom happens and clues can be gathered from that.
IMG_1662.jpg
IMG_1662.jpg (560.51 KiB) Viewed 128 times
First of all, thank you as always for the support. This is the case I found from Cooler Master and it is really beautiful, all metal and extremely modular.
I spend a lot of my free time doing test after test trying not to bore many. Some problems depend on the fact that I had my last Atari Mega 4 with ET4000 a few years ago and therefore I had to make many attempts both for NVDI 5 and others. Now the configuration is satisfactory but still at 48MHz. I did the memory tests several times, connecting the serial cable to have the output also on the Mac. The test passes 30-40 times without an error and then everything freezes, even when pressing the NMI key. It gets stuck doing everything, sometimes not right away, doing everything I see done by you. Castway, on the other hand, never seems to crash.
I tried two PSUs, another 68060 rev 5 in my possession (on Amiga 4000TX + BFG9060 they are perfect).
I understood why when I buy two chips on Ali Express they send me 5-6.
They only sent me one of the MC68150...
Last thing really: the front panel cable of the case has 4 wires and two are for the power switch and another two for the hard disk LED and they work, but I think that the power LED must work somehow...
User avatar
agranlund
Site sponsor
Site sponsor
Posts: 1121
Joined: Sun Aug 18, 2019 10:43 pm
Location: Sweden
Contact:

Re: Raven060 (luciodra build).

Post by agranlund »

luciodra wrote: Tue Nov 26, 2024 4:57 pm Last thing really: the front panel cable of the case has 4 wires and two are for the power switch and another two for the hard disk LED and they work, but I think that the power LED must work somehow...
Assuming the pinout of the 4-wire connector matches you'll plug it into where the blue outline is.
(I think the pinout matches, it looked like it on a video about that case)

Then you'll want to put a jumper where the red outline is.

This selects the "single LED mode" were it functions like this:
LED will be ON when power is on, except for when:
a) reset button is held down
b) there is disk activity
Thus the one and only LED flashes off when there is disk activity but is normally on otherwise.

If you jumper it for "dual LEDs" then the power LED will be on when power is on except for when reset it held down.
The disk LED (connected to -DL+ pins) will be on when there is disk activity and off otherwise.
Screenshot 2024-11-26 at 20.41.png
Screenshot 2024-11-26 at 20.41.png (90.9 KiB) Viewed 105 times
luciodra
Site sponsor
Site sponsor
Posts: 96
Joined: Fri Jun 28, 2024 1:59 pm
Location: Rome

Re: Raven060 (luciodra build).

Post by luciodra »

Ok that's how it works. Too easy. I tried but it didn't work, but now...
Rather, I'm trying YARTT with one module at a time but after a few cycles it crashes, both ST ram and TT ram and pressing NMI doesn't go into debug.
luciodra
Site sponsor
Site sponsor
Posts: 96
Joined: Fri Jun 28, 2024 1:59 pm
Location: Rome

Re: Raven060 (luciodra build).

Post by luciodra »

IMG_1663.jpg
IMG_1663.jpg (810.37 KiB) Viewed 95 times
Are these values ​​correct?
User avatar
agranlund
Site sponsor
Site sponsor
Posts: 1121
Joined: Sun Aug 18, 2019 10:43 pm
Location: Sweden
Contact:

Re: Raven060 (luciodra build).

Post by agranlund »

luciodra wrote: Tue Nov 26, 2024 8:32 pm Are these values ​​correct?
That looks normal yep
User avatar
agranlund
Site sponsor
Site sponsor
Posts: 1121
Joined: Sun Aug 18, 2019 10:43 pm
Location: Sweden
Contact:

Re: Raven060 (luciodra build).

Post by agranlund »

I'm not sure yet how to split this information into action points but these are very valuable informations;
- NMI button does not work when it hangs.
- An identified scenario which is different, where the issue is not triggered (Castaway)

My best guess based on this is that the CPU is waiting on something to respond. We don't know any details but at least there is an initial path to focus on - as opposed to if it was a crash, then the troubleshooting would have taken a different direction.

Going with that theory, the most likely target is on the 68150 side (rom/ram has fixed waitstates and we don't wait on replies).

The fact that it works fine with Castaway is interesting. There could be others thing too but one obvious difference is that while Castaway hammers the ISA bus quite hard it only ever does 16bit writes to videocard ram. no reads, and no I/O access.
Could perhaps hint at getting stuck on one of the other access types either never finish or the message does not correctly propagate to the cpu.
It could be something else too of course. It could also be failing at ATF1508<->MC68150 communication.

A logic analyzer would be ideal to see the state of things during the lock-up, but even just measuring with a multimeter when the lock-up has happened could help shed light on things.

Some signals will be impossible to get at with normal probes but there are still quite a few interesting ones that can be accessed very easily on large pins - and if it turns out necessary after that, could always run the machine upside down to have easy access to ATF1508 and MC68150 pins on the underside of the board.
I can highlight some points for some initial measuring.
Post Reply

Return to “RAVEN 060 - USER BUILDS”