Bad reads after bus resistor replacements with ACSI2STM

[Cwiiis]

I was experiencing some instability (Bad Apple crashes the machine a second or so in) and thought it'd be a good idea to make my way through the relevant mandatory fixes - I've just done the bus resistor replacement and now the ACSI2STM seems to be getting junk reads (picks up all drives on boot, but on TOS loading, I only see the C drive and it has `.'/-,((*/` bytes free and the directory listing is partially nonsense). Does anyone know what the cause might be of this, or how to go about debugging it? I've already done the 1772 pull-ups and floppy disk access appears to be fine.

I've tested the bus resistor values after fitting - I did the badwolf method of soldering an extra network in parallel on the bottom of the board - my values are closer to 2.4k than 2.2k (and the one that should be 1.2k is closer to 1.3k), I figure these small discrepancies are likely irrelevant.

I haven't put a pull-up resistor network on the DMA chip data lines, but thought it'd be better to consult here first rather than start taking shots in the dark...

Edit: This is also with an HC 68000, fwiw. I've not tried with non-HC.

[exxos]

Other than the DMA pull ups I have no idea. I did half read somewhere a few weeks ago that something like ACSI2STM had some issues, but think a firmware update solved the problems.

Problem is, these machines have that many faults it's impossible to keep up with them. I know there is some new DMA issue which has surfaced recently which I haven't had time yet to look into either.

But wasn't you the guy doing something with that DMA fix board thing a few weeks ago ? If that's in the mix as well its a whole bunch of unknowns to me as well.

[Cwiiis]

Other than the DMA pull ups I have no idea. I did half read somewhere a few weeks ago that something like ACSI2STM had some issues, but think a firmware update solved the problems.

Problem is, these machines have that many faults it's impossible to keep up with them. I know there is some new DMA issue which has surfaced recently which I haven't had time yet to look into either.

But wasn't you the guy doing something with that DMA fix board thing a few weeks ago ? If that's in the mix as well its a whole bunch of unknowns to me as well.

I have the DMA fix board in, but it isn't even connected to the data bus pins of the DMA chip, so I doubt that comes into the equation. The data bus was previously 10k and switching it to 2.4k does seem like a big jump relative to the jump for the address bus(?) Maybe worth going to some higher value to match that, say 5k instead?

[exxos]

I have the DMA fix board in, but it isn't even connected to the data bus pins of the DMA chip, so I doubt that comes into the equation. The data bus was previously 10k and switching it to 2.4k does seem like a big jump relative to the jump for the address bus(?) Maybe worth going to some higher value to match that, say 5k instead?

AFAIK that fix board changes the RDY signal timings which is one of the main causes for corruption. It has nothing to do with the databus etc. The bus resistors also affect the same timings and using the HC CPU also changes the same timings. So doing all 3.. who knows what the outcomes will be. Likely back to corruption again.

AFAIK nobody has done in depth testing of that fix board ( I don't follow progress of projects outside of this forum) . Though I bet nobody has done all my fixes and combined them with other fixes as well. You are in new territory here. I would suggest not mixing fixes by different people until someone has verified it all properly first.

[Cwiiis]

I have the DMA fix board in, but it isn't even connected to the data bus pins of the DMA chip, so I doubt that comes into the equation. The data bus was previously 10k and switching it to 2.4k does seem like a big jump relative to the jump for the address bus(?) Maybe worth going to some higher value to match that, say 5k instead?

AFAIK that fix board changes the RDY signal timings which is one of the main causes for corruption. It has nothing to do with the databus etc. The bus resistors also affect the same timings and using the HC CPU also changes the same timings. So doing all 3.. who knows what the outcomes will be. Likely back to corruption again.

AFAIK nobody has done in depth testing of that fix board ( I don't follow progress of projects outside of this forum) . Though I bet nobody has done all my fixes and combined them with other fixes as well. You are in new territory here. I would suggest not mixing fixes by different people until someone has verified it all properly first.

Sound advice. Out of interest, I've taken the 10ks off the data bus and my 8gb card using ICD Pro was still corrupt but putting an older 32gb card I had around using the pp driver worked fine. I ran Bad Apple and it ran for a good long time before crashing back to the desktop (a much, much better result than it's ever achieved previously), and now the C drive has become corrupt.

So the data bus pull-ups perhaps not the culprit in this case. It appears the address bus pull-ups may well have solved my instability, but have thrown some other marginal thing out of whack... Or perhaps all of this is coincidence... I'll restore back to the previous working state, as I can't really have it randomly corrupting the hard disk, but it would appear the problem lies around here somewhere...

[exxos]

Sound advice. Out of interest, I've taken the 10ks off the data bus and my 8gb card using ICD Pro was still corrupt but putting an older 32gb card I had around using the pp driver worked fine. I ran Bad Apple and it ran for a good long time before crashing back to the desktop (a much, much better result than it's ever achieved previously), and now the C drive has become corrupt.

I think you should leave the 10K pullups on the DMA. Its long known to cause floppy / HDD errors without the pullups. It's a different "DMA issue" going on there.

The address/databus pullups solve stability issues but can also tweak DMA timings.

I really would use HD11 instead of ICD.

So the data bus pull-ups perhaps not the culprit in this case. It appears the address bus pull-ups may well have solved my instability, but have thrown some other marginal thing out of whack...

Possible yes. But it's better to clarify which databus your referring to..

Or perhaps all of this is coincidence... I'll restore back to the previous working state, as I can't really have it randomly corrupting the hard disk, but it would appear the problem lies around here somewhere...

I think you should try removing that DMA fix board and see what happens if your doing the CPU pullups and using the HC CPU. If that doesn't solve your problems, then take the machine back to stock and re-fit that DMA mod board. These machines have so many issues, a "one size fits all" fix just isn't possible. My fixes solves all the common problems I (and others) have seen over the years, but there are always more "corner case machines" where more investigation is needed.

Problem is I think, my fixes assume a stock machine, mostly early STE motherboard revisions. I also ( I don't know who designed that DMA fix board) assume they also assumed a stock machine as well. Doing both fixes , well, its not starting from a stock machine anymore. So likely other modifications could well cause new malfunctions which could turn into a mini nightmare by the sounds of it.

[Cwiiis]

More experimenting, I was having a look at the schematics and at least seeing how things correlate and so I figured P100 is the most likely resistor network to actually make a difference for this - and indeed, removing the change there immediately caused my problems with Bad Apple to reappear (almost immediate crashing) - so changing the network here seems to give me corruption issues (maybe a bad interaction with the DMA changer, perhaps some values on that board need changing to correspond to a different value here) but also fixes this particular instability.

Worth noting that my other STE that has all the fixes and a changer board didn't exhibit any of these issues (but has other problems that stop me from using it right now...), but I seem to be learning more and more that everything on the STE is extremely marginal... I may try another value on this resistor network to knock it down just a small amount to see if it helps.

I realise I said I didn't want to go shooting in the dark and now am doing exactly that, but c'est la vie!

[exxos]

P100 should be 1.2K. That is Atari's own fix. IIRC it was busgrant and other control signals having additional pullup resistance on them. Of course busgrant issues plays a huge part in DMA transfers.

[Cwiiis]

So it would appear either I've done something bad while fitting those resistor networks, or some other thing has happened at the same time - having reversed all my work, things are getting corrupted in the same way - just booting, changing the desktop config, saving and rebooting ends up with a corrupted C drive. So either my ACSI2STM has coincidentally started failing or I've caused some kind of damage somewhere... Which is a bit of a bummer...

To be honest, I think I'm about ready to give up on Atari hardware and just stick with MiSTer. Seems I just can't get a system that doesn't have some kind of issue and various problems just keep cropping up that make it impossible to work with for any length of time. Disappointing, but it is what it is, I guess I can save some desk space now!

[Steve]

I know it can get depressing when things go wrong, but don't worry things will seem better tomorrow you said you have another STe? Does that one work ok? Also are you installing sockets for bus resistors? It makes swapping / testing easier and less chance of a breakdown.