• Please review our updated Terms and Rules here

Testing/Repairing ST's

Difficult question, I suppose what you really need to know is who made that memory expansion so you can find the installation instructions for it so that you can reverse them. Are there any identifying brands, marks, dates, model references on the memory PCB?

The folks over on atari-forum.com (which is working again now, no security warnings) would be the most likely to recognise that memory expansion and maybe point you to some information about it - I confess I have not seen that particular type before. They should also be able to tell you whether it is likely to cause any problems if you just unplug it.

If this was a 4MB expansion in its own right then obviously the onboard RAM would have had to be disabled somehow and so removing the expansion without reversing that change would leave the machine with no enabled RAM at all, but I think your reasoning is good and it sounds like the memory is the original 1MB plus an additional 3MB. In those days memory was really too expensive to waste, so it would not have made sense to use 4MB where 3MB would do.
 
Can you run YAART and show the results showing the faulty addresses?

By Pass 1, Pass2, etc, it will usually mean that it is running a particular phase of the test, each time trying a different specific test or writing a different value to the RAM locations. For example if you have a RAM location with a bit stuck low and the first pass writes 00 to that location it will successfully read the same value back, so to test that location more thoroughly pass 2 may write FF to all the locations, pass 3 may write 0xAA (10101010) to all of the locations, pass 4 may write 0x55 (01010101) to all of the locations, and so on. The tests can get increasingly sophisticated so the software may make many passes before the test is complete and only some of them may report errors.
 
The kind of errors I got where like 8080 != 8000, and as far as I remember often in the third bit. So that does smell like one or more faulty memory chips or the logic steering them.

I couldn't get a report out of YAART and after 5 passes the amount of errors was near 2000, so impossible to screenshot or manually write down. :) I can try to run it again and see where the first error(s) occur.
 
Yes, if you can maybe just identify the lowest address at which errors begin to be reported, then we may be able to see whether only addresses in the upper 3MB are affected, but it is also possible that a fault on the expansion could corrupt the test results for the onboard RAM as well, since everything sits on the same buses.
 
I just reran YAART RAM test.

YAART said it would scan from 0x000008 to 0x3ff1c0 which looks a lot like 4MB to me.
YAART said the "low memory" from 0x000008 to 0x011c48 is "OK!".

The errors came fast after each other in a group of 85, so I didn't get the very first one but here's some:

Pass 1: error 56: 0x3c5dba (8000 !=8080)
Pass 1: error 85: 0x3fef48 (8000 !=8080)

Then shortly after that, I assume towards the end of pass 1 but I only saw it when pass 2 was already going on, I had this one:
Pass 2(might have been during 1): error 87: 0x335e96 (5e14 != 5e94)

I didn't take notes of subsequent passes (takes a LOT of time) but at least here they appear to be in the same range (and very high up in the total range) and all have an 8 ofset on the third bit. So far for my knowledge; I don't know how to determine a cause, fix or anything else from that. The range doesn't tell me whether that is in the motherboard RAM or expansion RAM, as I don't know the first thing about it.

I have not yet tried the trivial such as reseating those expansion connectors.
 
Last edited:
Yes, all those errors do seem to be up at the high end, well into what would be added on memory. The 'low' memory end address is not very high up the range, only around 72,000+ (Decimal)

Does YAART allow you to specify a FROM and TO range to test? It would be interesting to see what would happen if you limited the test range to only the first 1MB of RAM.
 
For information, there is a modern version of what you have there available from Exxos, who makes various interesting things for STs:-


I note his is a true 4MB upgrade which means that the original onboard RAM has to be at least disabled by a method explained in the fitting instructions when this one is fitted. I remain convinced that you could revert yours to its original and probably reliable 1MB configuration if you can find the fitting instructions for that upgrade and 'un-fit' it.
 
Yes, the YAARTTT variant apparently allows you to set a range. The YAARTTT variant can't test memory that is already in use (by TOS) though, but that will be in the first 1MB anyway.

Which range should I enter, 0x000008 to 0xFFFFF? (1024*1024-1)

I'll think about undoing the extra 3MB once I've seen the first MB do splendid. :) But as you said, it's still possible that the extra MB's mess with the first one at the moment but we'll see how it goes.
 
OK I've just YAARTTT'ed the living daylight out of the STFM. I couldn't start at 0x000008 as you need to make sure your range consists only of unused RAM. I don't know exactly how much that is in my case, but TOS occupies somewhere between 75kb and 100kb. In the hex range for 100.000 and 1024*1024 I let YAARTTT run 2 passes and it returned zero errors. In the lowest 100.000 we're mainly talking about stuff used by TOS and TOS works, so I assume the first meg is just fine.

So now we move to the next step: try to determine if it is safe to just unplug the 3MB addition. Safe as in "can't damage the computer". If there are no changes (ie lifted legs, cut traces...) on the mobo that might work as an ST apparently at boot just writes/reads some adresses and sees which work to determine how much RAM you have. During my first inspection of the addition I mainly saw loads of wires added to points on the mobo, but I'll have to look again in detail.
 
I've taken the liberty of posting your image of the RAM upgrade over on atari-forum to ask if anyone knows what it is, whether it can be disconnected without reversing any other steps and have also asked if anyone has the installation instructions for it. I'll let you know if anything comes up, or you can watch the thread yourself of course.

0xFFFFF sounds like a reasonable guess for the physical top of original RAM, although some part of that range may be used by TOS for system variables. Edit: Crossed with your last - I think you may have found where TOS keeps its variables. :)
 
Thank you! I appreciate your help here and overthere. As you see I don't have an awful lot of knowledge on the subject so I wouldn't get to this point by myself -certainly not that fast. I do however have enough background that you don't have to pull your hair out (too late to pull mine out...) to get me to make progress with the insights given to me here.
 
A question from Atari-Forum

If the PCB marking starts with "CC" it could have been manufactured by CatchComputer (Aachen IIRC). What IC is the one with the while "1" marking?

(By which they mean, what is the full number / marking on the IC half hidden by the left hand piece of tape in your image?)

I've also been asked, as indeed I asked earlier, if the PCB has any other writing, text or numbers on it which might help to identify the maker of the expansion PCB.
 
The expansion appears to have been identified as model CCMB4 made by Catch Computer who were in Aachen. I have a picture of the front page of the manual with a line drawing which looks exactly like your expansion board, but unfortunately not the rest of the manual. Maybe armed with this information you can track down a copy of the whole manual - you no doubt have the advantage of being able to speak and read languages other than English.
 
I apparently already had an account on the Atari forum, so I've chimed in there. Summary: I'll open up the STFM, first try to reseat the headers and take detail pics, then retest. If it's still faulty, I'll see what needs to be done to remake the original connections for the factory 1MB.
 
Summary of events: I've documented the RAM expansion. Then I've removed it, under the assumption that it had RAM errors.

Result: the ST with its original 1MB RAM doesn't work correctly. GEM doesn't show some words such as the top menu. Most programs (including YAART) crash. Stunt Car Racer works perfectly though. :)

My guess now is that the RAM expansion doesn't necessarily work beyond the first MB of RAM but can contain the "first" MB. So 3+1MB instead of 1+3MB.

Anyway, I guess this STFM won't be for me. I don't think I can repair it from its current state. So I think of selling it, as I couldn't think of much use for it anyway besides LANning Stunt Car Racer if I'd ever get a second monitor.

Next steps: try to sell this ill STFM, the untested ST520+, the RAM expansion. And then find something to fill the gap in the rack. Likely not an ST. (I still have the STF to run Cubase, mind you)
 
Strange, sorry you have such a negative result after all of that, but at least there is now a good record of how that uncommon expansion is connected. Stunt Car racer I would guess uses almost entirely self contained code and does not use any (or many) of the routines in TOS at all, especially not the display output routines, whereas GEM will make extensive use of ROM code and YARRT may possibly use the TOS text display output routines, so I am thinking there is a possibility at least that there may be ROM corruption. Remind me, are the ROMs in that machine original Atari mask-programmed PROMs or are they EPROMs which someone has fitted at a later date?

Edit: I see from an earlier post that the machine has labelled EPROMs in it, but also that you mention trying other ROMs in it. I think you are right, it may be time for this to be someone else's problem.

Another thought about YARRT, obviously with much less memory than before, there is a strong chance of YARRT scoring a direct hit on itself unless it avoids its own memory area by default, which would be a sensible thing for it to do.
 
Last edited:
They are custom ROMs, I think. They have writing on them and the center is taped off. GEM looked ok before removing the expansion.

I have tried pushing the ROMs in their sockets.

Anyway, even if it is a ROM issue, this ST also has other issues as it can't run most games/programs unless those are all TOS related issues.
 
I wonder what would happen if you refitted the expansion (I know....) and DISABLED the onboard RAM so that it can never get included in the working memory, ie, the machine would have 3MB rather than 1MB. Unfortunately it is beyond my knowledge to tell you how to do that.

Regarding ROMs, there is a great deal of variation in how much commercial software does or does not use TOS, apart from initial loading of course. Some titles, especially productivity software which makes extensive use of GEM, will be heavily dependent on the ROMs, some others are as 'bare metal' as they can be and use the system calls as little as possible.
 
I do not feel like reinstalling the expansion, especially as I see so little use for a game ST on top of my Cubase ST. Too much overlap with the other 16bit systems I have. I think I'll find more use in replacing it with another computer or console.
 
Back
Top