A500 Rev6 1MB Chip RAM DiagROM 1.2.1 AddressError Detected

[BSzili]

I have a Rev6 A500 with the JP7A trace cut so I can use the trapdoor memory expansion (Gigatron A500 SE) as Chip RAM. Since the system is a bit flaky I started DiagROM to look for some possible faults. When doing the extended Chip RAM test the first 512KB was fine, but after that at one point I got an address error. When I unplug the expansion it goes away, but it still tries to test the area above 512KB. I guess this happens because of the JP7A hack.
Anyway, what does this error mean? Is the RAM expansion faulty, or maybe there's some other problem on the board itself? Here's a slightly blurry image of the actual error screen:
https://photos.app.goo.gl/7JCgezw26A8JQUpE8

[brett71]

I think there was also a jumper pad that needed to be shifted to enable the trapdoor expansion to be used as Chip RAM. Did you also make that change as well? Can't remember if it was JP1 or JP2 that needed to be cut and then the other pads connected.

[hooverphonique]

as brett hinted at, you need to set/cut another jumper (exram) similar to "trapdoor ram not present" when using it as chipram..

[BSzili]

I checked, and the JP2 jumpers is in the correct position as well, I verified with a multimeter. Sorry, I did this mod about 10 years ago, and I completely forgot about the other jumper. Does Address Error mean the RAM expansion is faulty?

[jasonsbeer]

Might be a red herring. Diagrom tests up to 2MB of chip ram. If it hits the end of the 1MB, it will throw an error because it can't test the next block, since it does not exist.

When doing the chip ram test, you can watch the memory range count up as it tests. If it hits 1024kb and then throws an error, your memory is not bad, it just ran out of address space to test. As I recall, it counts up in 32k increments. Check this and let us know what you see.

[LIV2]

Address error means that the memory diagrom uses for the stack is likely getting corrupted, when it's trying to return from a subroutine it's trying to return to an odd address triggering that error.

This could mean that the lower 512k is being mirrored at the higher addresses for some reason and that the memory test is overwriting the stack.
Off the top of my head I can't think of what would cause that but I'll have a look and get back to you

Can you check JP3? If it is set incorrectly it could cause the ram to be selected for both banks, it should look like the one here: http://amiga.resource.cx/photos/phot...res=hi&lang=en

Also with the expansion installed check if there is any continuity between one of the top and one of the bottom pads of JP3, there shouldn't be.

I tested my own rev 6 with diagrom 1.2.1 and found that it throws an address error whenever it goes past the end of the installed memory, this is a bug and is fixed in more recent versions where it shows the bit errors and lists bad blocks.
If you can't burn a more recent version you can try the memory tests in AmigaTestKit to see what it says

[BSzili]

That makes sense, it probably just goes past the 1MB boundary. I was a bit worried because the trapdoor expansion had some battery damage before the NiCad cell was removed, but it mostly affected the RTC part. I don't have an EEPROM burner handy right now, but I'll try 1.3 later to see if the issue is still there.
I'd check with AmigaTestKit too, but this machine has some issues reading floppies both with the internal an external drive. This is the original problem I was trying to track down, as otherwise it's working fine, it even booted from a SCSI drive just fine in the past.

[jasonsbeer]

Originally Posted by BSzili

This machine has some issues reading floppies both with the internal an external drive. This is the original problem I was trying to track down, as otherwise it's working fine, it even booted from a SCSI drive just fine in the past.

Try reseating Gary and the CIAs. Try switching the CIAs if the problem persists. There is some logic in there as well, but start by reseating those chips and let us know what happens.

[brett71]

Originally Posted by jasonsbeer

Try reseating Gary and the CIAs. Try switching the CIAs if the problem persists. There is some logic in there as well, but start by reseating those chips and let us know what happens.

I think Paula also has some part to play in floppy I/O.

[Jope]

You can roughly split it so that the read/write bitstream is handled by Paula and the control signals are CIA/GARY.

[BSzili]

I reseated Paula, Gary, and the two CIAs, I also swapped the odd and even CIA. Now I can boot from floppies but it's still not perfect. There's some issue with the disk detection. For example I can put in a disk, and hear the drive click a few times before it notices that there's a disk inside. The opposite is also true, when a disk is in, and it tries to click it someties behaves as if I ejected and reinserted the disk. This doesn't happen during active disk operations, only when the drive is idle.

[jasonsbeer]

Likely a bad disk detect switch in the drive. Squirt some contact cleaner into it and exercise it.

[BSzili]

I'm finally back to my machine, so I was able to check this. At first I thought the drive mechanism didn't push the disk far down enough for the switch, but I checked with a multimeter and you are right it's the switch. I'll drench it in IPA and give it a good workout, and if it doesn't work out I'll desolder it to see if I can find a similar one from a PC drive.

[BSzili]

It looks the little exercise fixed the switch, the disk detect problems are gone.