exxos's Atari, Amiga & retro forum

I guess one way to check the address lines at least at the CPU end would be to remove the CPU and pull up (or down) each address line and see if any other the others follow it.

stephen_usher wrote: 31 Jan 2025 18:06 I guess one way to check the address lines at least at the CPU end would be to remove the CPU and pull up (or down) each address line and see if any other the others follow it.

As I said earlier though, I made a mistake and A2 & 5 are actually ok, it was data in and data out that looked odd, and having just looked It looks like they are tied together.

All the address and data lines plus RAS/CAS all look fine.

I'm done in for the night TBH, non retro stuff so I'm just gonna try and chill.

rubber_jonnie wrote: 31 Jan 2025 16:48 I have been wondering if the card is not compatible with this particular issue Amiga, but that would be rather odd.

That isn't going to happen, these should be rock solid for all revisions of the board.
You have a good revision board and it should not be an issue.

Did you clean the Agnus and Gary?, it isn't invasive like soldering and it often gives result.
Is there a break in the Agnus socket maybe?

Maximilian wrote: 01 Feb 2025 00:07
That isn't going to happen, these should be rock solid for all revisions of the board.
You have a good revision board and it should not be an issue.

I didn't think it would be, but you can't help but wonder.

Maximilian wrote: 01 Feb 2025 00:07 Did you clean the Agnus and Gary?, it isn't invasive like soldering and it often gives result.
Is there a break in the Agnus socket maybe?

I did. In fact I did this for not just Agnus and Gary, but also all the other socketed IC's.

By way of a status update, this is where I'm at:

1. Removed the socketed ICs for clean/reinstall/contact cleaner spray.
2. Removed the pin header for the trapdoor expansion and cleaned ultrasonically. Cleaned up the corrosion on the PCB, checked for damage and reinstalled the pin header.
3. Used Amiga Test Kit to test the trapdoor RAM memory space - it fails/crashes the machine.
4. Tried piggybacking RAM on the trapdoor board - no change.
5. Tested the RTC, that at least appears to work. It's visible/settable in Amiga Test Kit and if I remove the card, the clock is no longer visible.
6. Checked the RAM with my oscilloscope. All the address pins, CAS & RAS look normal and power is ok. What looks odd is the Din & Dout pins which look very low voltage wise and have a weird looking noisy signal. The trapdoor RAM is 16x 41256 BTW.

I have reviewed with a thermal camera and i don't see anything off, but I have ordered some 74F244's (U34 & U35) as they seem pretty instrumental in what 's going on with the trapdoor expansion. In the meantime I may socket them and test them in my TL866.

I don't really want to try buying another RAM card as it's money thrown away if it doesn't work when I have the parts to upgrade internally. I understand that you don't agree with installing the missing RAM, but I don't se it as a problem if it gets me to 1MB total.

If it were to do with the RAM card the machine wouldn’t crash when you did a RAM test with Amiga Test Kit as the OS isn’t using that memory. This suggests an issue on the motherboard.

What happens if you try the memory test without the card fitted?

stephen_usher wrote: 01 Feb 2025 07:41 If it were to do with the RAM card the machine wouldn’t crash when you did a RAM test with Amiga Test Kit as the OS isn’t using that memory. This suggests an issue on the motherboard.

What happens if you try the memory test without the card fitted?

I'll give it a try and see what happens.

stephen_usher wrote: 01 Feb 2025 07:41 If it were to do with the RAM card the machine wouldn’t crash when you did a RAM test with Amiga Test Kit as the OS isn’t using that memory. This suggests an issue on the motherboard.

What happens if you try the memory test without the card fitted?

I get pretty much the same result. I get a flashing repeating pattern on screen that can't be escaped from. They vary slightly depending on wether the card is plugged in or not.

I've ordered a couple of 74F244's just in case as they are fundamentally involved in the RAM. I was looking at JP3 but that can't be a problem because it swaps internal/trapdoor RAM, however it may be useful as a troubleshooting mechanism.

I'm very tempted to just fit the missing 4 RAM ICs at this point. Either way I'll probably socket the 74F244's at U34 & U35 just so I can test them.

I was going to have a close look at the soldering on the bottom of the board, but my microscope is broken.

I would imagine that a full "Chip RAM memory test" might corrupt the screen (at least for a short while) as it might have to overwrite screen memory?

If you remember the patterns that used to be on the screen as a packer decompressed, that is a program using screen memory.

I'm not 100% familiar with Amiga architecture, perhaps it is possible a well crafted test could test part of ChipRAM then move the screen base address and test the other part of ChipRAM?

alexh wrote: 01 Feb 2025 12:28 I would imagine that a full "Chip RAM memory test" might corrupt the screen (at least for a short while) as it might have to overwrite screen memory?

If you remember the patterns that used to be on the screen as a packer decompressed, that is a program using screen memory.

I'm not 100% familiar with Amiga architecture, perhaps it is possible a well crafted test could test part of ChipRAM then move the screen base address and test the other part of ChipRAM?

Well I think I'll divert to the Acorn A5000 as I now have the CMOS ICs for that and I'm waiting on parts for the Amiga now.

EDIT: I may set a RAM test off in the trapdoor memory area and walk away, see if it finishes if I walk away, then do it again with the trapdoor ram installed.

I'm fairly sure the test should be quit-able by hitting ESC and it isn't, but we'll see.

I did remove and socket both 74F244's however both tested ok, though they might still be marginal, you never know.

If there's no RAM there and you're only testing the SLOW RAM addresses then the machine should not crash but the memory test should fail. You are manually testing only 0xC00000 to 0xD7FFFF aren't you?