exxos's Atari, Amiga & retro forum

Seems to be bit eleven. Don't know if it is bank 0 or 1 though. All address writing ends with hex400 = dec 1024 = eleventh bit that is stuck high.

EDIT: I clearly don't know how to read the fault diag rom puts out :D

Looks like total fail on lower RAM, its all coming back as "FFFF"

Addr $00000400Write:AAAAAAAARead:FFFFFFFF FAILED


This one...

Addr $00040400Write:AAAAAAAARead:00A000A0 FAILED

AAAA sent 1010101010101010
00A0 back 0000000010100000

Thats a lot of bit fails..

EDIT:
You could check my blog where I diagnose my RAM board... theres a couple of pages...

https://www.exxosforum.co.uk/forum/viewt ... 660#p37275

DIAGROM stuff starts here https://www.exxosforum.co.uk/forum/viewt ... 670#p37290 I think half the problem was a bad CAS line, it was giving me bit errors which were not really there...

@JezC

Owww, that is nasty.

Have you buzzed out the pins on the RAM at all? It'd be good quick check that everything is actually connected as it should be. I had one of these boards with a bad bank 1 and A7 wasn't connected between 3 of the chips despite there being what appeared to be perfectly good traces there.

Than maybe look at the data and address lines at the MMU. When I was getting a bad RAM check on my Mega 4 last time it was dead, it was data lines from the MMU to the CPU, so buzzing those out found that problem. The logic probe also showed the same 2 data lines were bad.

It showed up as bad bits, but wasn't that at all.

Thanks to you all for the advice - I've not got further with that board than capturing the DiagROM output but will probably resume the battle later today.

I did quickly try the other 520 STFM & that is not giving any serial output so that will probably need some more basic diagnosis first (the screen starts to do something & shows the outer border distinct from the inner screen area but that's about it - apart from a high-pitched whine (could easily be around 15kHz).

Anyway, I will put this second system aside for more basic checks after I make progress with the one providing the diagnostic output.

JezC wrote: 06 May 2020 13:19
I did quickly try the other 520 STFM & that is not giving any serial output so that will probably need some more basic diagnosis first (the screen starts to do something & shows the outer border distinct from the inner screen area but that's about it - apart from a high-pitched whine (could easily be around 15kHz).

Is the outer border white and the inner screen black? That usually means it isn't seeing any RAM. Or is the inner screen partly showing garbage?

@rubber_jonnie Yes, the second ST is displaying a white outer & a corrupted inner box (pattern changes a bit with time).

The first one was also showing a similar white outer border but a black inner box.
When putting that first one away, I noticed that there are some resistors missing - IIRC they are not fitted on a 520 but are usually fitted on a 1040.

I will check into this a bit later, but I'm wondering if the attempt by the previous owner to upgrade the RAM has been partially undone & that is itself causing part of the problem

JezC wrote: 06 May 2020 13:42 @rubber_jonnie Yes, the second ST is displaying a white outer & a corrupted inner box (pattern changes a bit with time).

The first one was also showing a similar white outer border but a black inner box.
When putting that first one away, I noticed that there are some resistors missing - IIRC they are not fitted on a 520 but are usually fitted on a 1040.

I will check into this a bit later, but I'm wondering if the attempt by the previous owner to upgrade the RAM has been partially undone & that is itself causing part of the problem

R71 = CAS1H
R72 = CAS1L
R73 = RAS1
Is that was is missing? 68ohm on each. Is the decoupling capacitor fitted beside each ram chip?

DoG wrote: 06 May 2020 14:03 R71 = CAS1H
R72 = CAS1L
R73 = RAS1
Is that was is missing? 68ohm on each. Is the decoupling capacitor fitted beside each ram chip?

Yes, those 3 are missing (but look to have been fitted & subsequently removed) - all 16 additional RAM chips do have a decoupling cap next to them.

So, what would people expect to happen if the RAM is fitted but these resistors are not?
Would the original 512k be expected still to work?

JezC wrote: 06 May 2020 14:14
Yes, those 3 are missing (but look to have been fitted & subsequently removed) - all 16 additional RAM chips do have a decoupling cap next to them.

So, what would people expect to happen if the RAM is fitted but these resistors are not?
Would the original 512k be expected still to work?

Makes me wonder if that was the plan. The extra 16 were fitted, but then the upgrade didn't work and the 3 resistors were removed. But I don't think that would be enough to disable the RAM.

Looking at exxos' RAM upgrade page, to disable the first bank, or 512KB RAM, you need to pull CAS0L, CAS0H and RAS0 high, which are R59, R60 & R61 by lifting the end farthest from the RAM and attaching a 5V line.

I'd suspect you'd need to do the same but for R71, R72 and R73 (CAS1L, CAS1H and RAS1) to disable bank 1 and just leave bank 0 in play.

Unfortunately the schematics I have are not great, so hopefully @exxos or @czietz can confirm this.

You must tie pins somewhere not left floating. If you have 1mb with resistors missing then you would get all kinds of ram faults.