exxos's Atari, Amiga & retro forum

And it's incredibly difficult to do remote support, hardware fault finding is another order of magnitude.

stephen_usher wrote: 27 Oct 2024 09:44 And it's incredibly difficult to do remote support, hardware fault finding is another order of magnitude.

Indeed it is.

ijor wrote: 27 Oct 2024 04:54 Question, you said you tested the MMU chip in another computer and it worked fine? But you tested the other way around, the other MMU chip in this Mega?

Yes I did, same thing :(

stephen_usher wrote: 27 Oct 2024 09:44 And it's incredibly difficult to do remote support, hardware fault finding is another order of magnitude.

Sure it is and I very thankful for all the knowledge shared here.
You guys are amazing.

I know understand the ST a lot better!

rubber_jonnie wrote: 27 Oct 2024 09:17 Just going back through this thread and I think it would be useful for you to post what you're currently seeing on screen when you boot as the last image was all the way back at the beginning.

Are you still getting a black screen with white border? Or is it something else?

When threads get long a summary of what is happening now and what you've done to this point are helpful for new people joining the thread.

Also, you mentioned at the beginning that the previous owner could get it running by slapping it, which suggests a mechanical issue. Do you know where it was slapped? on the side, on the top? a location might allow you to home in on a specific IC, or at least area of the mainboard.

One of my Mega's had a super fragile PCB, so don't assume that something you've already looked at has not since failed if you've been handling the board a lot.

Obviously the other folks are providing some great technical help, so I'm sure it will live again :)

Yes, the screen is (was) every time the same, black with white border.

I exchanged some messages with the former owner and he had this computer for about 15 years now and it was previously owned by another brazilian guy which is supposed to be know in the "retro scene" but I never knew about him. And this bad contact problem was a issue since then.
In my opinion, each slap, somehow put the loose IC in contact again and for a while it was a solution.

But not know as I have tested all continuity and ICs.

Guys, first of all, thank you very much for sharing your knowledge and your time helping me out to find the problem of my Mega ST4.

I really, really appreciate all I learned and am very thankful.

As just said in this topic, remote support is very difficult.
You don't know my skills, what tools I do have handy, my knowledge, a lot of things but you never let me down here.

Yesterday morning I, following the agenda I created to this repair, was to check the RAM.
But I almost didn't do it as the logic analyzer was here and I understood it would help you guys pointing me in the right direction as the options now are fewer then before.

But, forgive me if I quote it wrong, stephan_usher said something like this, after one post of the logic analyzer data: now we can see the processor is doing its thing for a while....
This hit me hard and I remembered the first input I had about this problem, black screen with white border: usually it is RAM related.

I know the diagnostic cartridge would have helped a lot here (I already ordered the PCB - tools are never enough!!) and also, maybe, the DiagROMs would have helped. I received the 27C1001 EEPROMS, I didn't burn the DiagRoms into them (yet) :D .

The only thing which as pushing me back was: I probed every DRAM IC with my scope and couldn't see any difference from the signals.

I know there wasn't CAS, but all other signals were there and with no difference.
I don't know if instead of using the scope, using the logic probe would be better.

Well I decided to follow my instincts and removed the first lower RAM (1-2 MB) 16 DRAMs ICs. And voilá, two bad ICs.

At this point I asked here if the computer would boot with 2MB, so 16 ICs were needed. Without spares, I removed the other 16. Two more bad in these.

Cleaned the board, and after resoldering the first 16 DRAM ICs, BLAM! The magic of TOS in the screen!

I am really happy even a bit upset not finding the bad RAM probing with the scope.
I would have saved a lot of time as having the diagnostics cart handy.

And for learning purposes, after receiving the replacement RAM and sockets I ordered, I will put the computer in the same state it was before, with these bad RAM in their places to probe the signals @ijor and @stephen_usher asked for, as it is important to this topic to have them not only to increase my and the vistor's knowledge how to diagnose a problem in their ST/Mega STs but also to come to a "closure" to this saga.

A big thank you to you all who helped me posting here or just passing by to read and learn a bit.

I enjoyed it a lot!

I will give some love to the enclosure to make it shine again and disk drive to see if it is working.
I came with an external SCSI drive which I think will need some recap.

That is excellent news :)

:girldance: :girldance: :girldance:

Damn that RAM, but good to know that the white border/black screen was telling the truth :)

I think that whatever you do, finishing that diag cart will be a really good idea and it'll let you practice with running diags to your PC with a serial cable. It will show things that you simply can't see on screen. Having had a machine in a similar state, being able to see the bad RAM lines over serial was a boon and helped zero things in.

Great to see that little green desktop :)

As for finding bad RAM that is installed, I found recently when repairing my Colecovision that bad RAM isn't always just hot, it can be cold: Colecovision cold RAM

The problem is that if you have no way to see it, such as a thermal camera, then distinguishing a hot or cold one with your fingers is a bit tricky. However, testing with your fingers isn't always the best way as I recently burned a fingertip on a faulty 74LS IC in an Acorn Electron that was 106C IIRC.

An alternative to find hot ICs is to spray a little IPA onto them, this often shows them up as the IPA evaporates very quickly.

All in all a good job, bet you're very happy :)

rubber_jonnie wrote: 27 Oct 2024 12:17 That is excellent news :)

:girldance: :girldance: :girldance:

Damn that RAM, but good to know that the white border/black screen was telling the truth :)

I think that whatever you do, finishing that diag cart will be a really good idea and it'll let you practice with running diags to your PC with a serial cable. It will show things that you simply can't see on screen. Having had a machine in a similar state, being able to see the bad RAM lines over serial was a boon and helped zero things in.

Great to see that little green desktop :)

As for finding bad RAM that is installed, I found recently when repairing my Colecovision that bad RAM isn't always just hot, it can be cold: Colecovision cold RAM

The problem is that if you have no way to see it, such as a thermal camera, then distinguishing a hot or cold one with your fingers is a bit tricky. However, testing with your fingers isn't always the best way as I recently burned a fingertip on a faulty 74LS IC in an Acorn Electron that was 106C IIRC.

An alternative to find hot ICs is to spray a little IPA onto them, this often shows them up as the IPA evaporates very quickly.

All in all a good job, bet you're very happy :)

Sure! I am very happy.

Despite the RAM weren't hot, in a lot of cases, probing them with the scope can determine which was bad or not.
But this time, I couldn't.
But I suspect, being a 16bit machine, things are different, as it was for example, messing up the CAS signal.

Still have a lot to learn!

asapreta wrote: 27 Oct 2024 12:32

rubber_jonnie wrote: 27 Oct 2024 12:17 That is excellent news :)

:girldance: :girldance: :girldance:

Damn that RAM, but good to know that the white border/black screen was telling the truth :)

I think that whatever you do, finishing that diag cart will be a really good idea and it'll let you practice with running diags to your PC with a serial cable. It will show things that you simply can't see on screen. Having had a machine in a similar state, being able to see the bad RAM lines over serial was a boon and helped zero things in.

Great to see that little green desktop :)

As for finding bad RAM that is installed, I found recently when repairing my Colecovision that bad RAM isn't always just hot, it can be cold: Colecovision cold RAM

The problem is that if you have no way to see it, such as a thermal camera, then distinguishing a hot or cold one with your fingers is a bit tricky. However, testing with your fingers isn't always the best way as I recently burned a fingertip on a faulty 74LS IC in an Acorn Electron that was 106C IIRC.

An alternative to find hot ICs is to spray a little IPA onto them, this often shows them up as the IPA evaporates very quickly.

All in all a good job, bet you're very happy :)

Sure! I am very happy.

Despite the RAM weren't hot, in a lot of cases, probing them with the scope can determine which was bad or not.
But this time, I couldn't.
But I suspect, being a 16bit machine, things are different, as it was for example, messing up the CAS signal.

Still have a lot to learn!

You're definitely in the right place to learn :)

Yay!!!!

I knew that you'd get there in the end.

If you don't have a digital oscilloscope the logic analyser is an excellent instrument to see what the system is doing and if it's trying to run code.

If the LA had just shown only a few reads and then stopped then this would have meant that the CPU wouldn't have been able to read the ROM. Running for a whole 100ms means that it was trying to run stuff for quite a while until it stopped, so the system must be working to some degree.

Without being able to instrument the whole of the data bus plus /AS and /DTACK and knowing what the data is supposed to be (such as the ROM values) it would be very difficult to diagnose a RAM fault if the chip isn't totally dead or pulling the bus hard high or low. So, at this point it is a fact that you either use a spare set of chips and try piggybacking as a start or just remove them all, install sockets and test the chips in a tester.

You went through everything which could be done non-destructively, so desoldering was the only way to go.