exxos's Atari, Amiga & retro forum

I recently upgraded three 520STFMs to 1MB, and have hit some interesting issues.

I have 2x C070789-001 REV.D main-boards that aren't happy with more than 512K fitted. Both will boot, but both exhibit issues which prevent use with more than 512k RAM fitted.

One gives on screen banding which constantly changes/flickers. It's as though the screen memory is corrupted or is changing within the bands, and you're seeing the changes happen live.

The other is weird. With the extra RAM fitted, the floppy drive doesn't work. With just 512k, it works just fine.

All have new 41256-12 RAM chips, 100nF decoupling capacitors, and 68R resistors.

For the first one, I'm thinking maybe upset timing, as the existing bank isn't 41256, but 6256's. I'm thinking about putting in RAM of all one type.

The other is really odd, with the extra RAM it's like the FDD is disconnected, and the light stays on. Gotek doesn't work either, though not really surprised by that.

Both by the way are TOS 1.02. I will be upgrading to 1.04 at some point, though annoyingly one is 6 chip, the other 2. I'll have to get the missing U68 so I can fit the 2 chip ROM to the 6 chip board.

So I'm wondering if any of you hardware Guru's might have any advice on the above.

What are the chances that both of the C070789-001 REV.D boards have a fault?

Or is it more likely that there may be some incompatibility between the old ram and the new?

Also, I used sockets (Also all new) for the new RAM, is there a possibility that the sockets are an issue? Maybe high resistance or something in the chip to socket connection?

I also have a C070789-001 REV .F which was upgraded in the same manner, with the same parts, with no problems at all, and happily ran the memory tests with 1MB RAM.

Any thoughts gratefully received

Was this the epic where I suggested you try each RAM chip in the known working board ? Basically one chip at a time ?

If you can upload the datasheet's both the chips I will have another look..

It could also be possible both chips you have or not 120ns even though it is printed on them like that.. Random colours and such are normally timing issues, mostly I see this when using 60ns RAM for example. So I think it's a good idea to test each RAM chip out individually, and probably leave the machine running on the YAART program for maybe a hour to make sure it gets tested thoroughly..

exxos wrote: ↑Thu May 03, 2018 3:15 pm Was this the epic where I suggested you try each RAM chip in the known working board ? Basically one chip at a time ?

If you can upload the datasheet's both the chips I will have another look..

It could also be possible both chips you have or not 120ns even though it is printed on them like that.. Random colours and such are normally timing issues, mostly I see this when using 60ns RAM for example. So I think it's a good idea to test each RAM chip out individually, and probably leave the machine running on the YAART program for maybe a hour to make sure it gets tested thoroughly..

It was, and resulted in my RAM testing rig. I have tested a number and all came out ok, but I'm hoping to get sometime over the BH to look into it. Also hoping to have a slightly better 'scope so I can do some deeper testing. I will get there!

rubber_jonnie wrote: ↑Thu May 03, 2018 3:22 pm It was, and resulted in my RAM testing rig. I have tested a number and all came out ok, but I'm hoping to get sometime over the BH to look into it. Also hoping to have a slightly better 'scope so I can do some deeper testing. I will get there!

Did you test every single one you used in the upper 512K bank ? Of course it will only take one chip to cause chaos

exxos wrote: ↑Thu May 03, 2018 4:13 pm Did you test every single one you used in the upper 512K bank ? Of course it will only take one chip to cause chaos

I soldered a socket into the first chip of bank 1, bank 1 and bank 2 being the original 512k RAM. I used my test rig and tested 32 chips (I'm going to test more), and the machine booted up with no issues to a desktop. 16 are running fine in my REV.F board.

I didn't do any deeper testing than that (Such as YARRT), and I'm planning soldering in a socket at the end of bank 2 to test that way as well.

Hopefully time over the bank holiday....

So I finaly did the epic buzz out you'd suggested, and found that on the first board tested, the A7 connection was missing on the top 2 chips in the 3rd bank of RAM (1st new bank of RAM)

I added a 'Bodge wire' to make the circuit from the last good A7 pin, up to the two that were missing A7 :
After re doing the buzz tests to make sure the connections were now there, the RAM was then refitted:
On power up, the memory corruption that was previously present was completely gone, so I fired up SYSINFO to see how much RAM was reported:
So, the screen corruption was gone, and the RAM was correctly being reported as 1MB.

I then ran a quick test with YAART, because previously it would error as soon as testing of anything in the new bank of RAM started. There were no errors, so it looks like it's fixed, though I will give it a more thorough YAARTing to make sure. Blitter is next for this board.

Great troubleshooting technique, thanks to exxos forthe advice

There is one final thing to come from this, and that is that both boards were of the same type, C070789-001 REV .D.

I have another, and have buzzed that out too, and it's much worse, so it does make me wonder, was there a problem with C070789-001 REV .D mainboards, and were they only ever sold with 512K RAM. I guess we may never know.

I guess the next job will be to map the myriad of faults with the 2nd board, as a quick buzz through showed a lot more connection problems than the other board.

Although it's going to be the board that is a donor for my STFM remake, I'd like to get it working before it gives up any body parts, as I'd really like to have it as a spare PCB.

Well done!

The old bodge wire. Classic!

Smonson wrote: ↑Mon Aug 06, 2018 7:22 am The old bodge wire. Classic!

You can't beat it