A2000 Repair - Help translating diagrom monitor output for chipram test issue

[snowpile79]

Hello all. I have a very clean A2000 board with hardly any evidence of battery damage. Only a slight bit of corrosion on the positive copper trace of the battery. Battery long since removed. Everything worked great until a week ago. With KS 1.3 I get an immediate dark green screen on power on. With KS 2.04 and up I get a yellow screen. Diagrom loads fine! It flashes black/green (possibly brown green?) about 6 times then does the normal boot up display where it seems to detect ram. Finally I get a menu. Ok test results: RAM shows errors. Almost looks like the same right 4 bits are dead (see screenshot). Is that 4 bits or all 8 bits? I see 8 characters/hex lined up under 4 bit positions on the right. If anyone can help me understand if those are hex addresses? is that 4bits bad or 8bits? Also randomly 4 bits will show bad/red in the middle column (from the right most middle position). Honestly I dont exactly know what the right, middle or left top columns represent exactly.
Down below see how it says good memory range starts at address X ends at address y but then says 0k? Same for the bad memory? Im not sure what this means exactly.

Ok all other tests run fine until you get to the CIA tests.
CIA odd tick will often say too fast. Sometimes it gets to the test and crashes the amiga causing a reboot. Sometimes all CIA's test fine. Sometimes CIA Even crashes on the tick-test saying too fast.

All other tests appear to pass. Ran them multiple times. Do not know how the heck Im getting video with the consistent chip ram faults.

I did trace out the CPU traces to the upgrade port to the left of the CPU as well as to the rom (which clearly works as diagrom works fine). I also traced them to the buster. All tone fine. Thats all I traced so far.

No ram chips get hot. I did try an alternate gary, Agnus, alternate/new CIA's, paula and denise (and buster) in case one was creating bus noise. I did try both positions for the tick signal jumper (power supply and CIA). About a couple months ago I did replace caps around the cpu, video expansion and the 12v cap on the power plug input area plus a few around the cpu planning to do them all but ran out of time.

Any info on what diagrom is telling me or any other ideas would be very appreciated. I can replace the ram as I have desoldering equipment and sockets but for some reason I feel it is not ram (unless someone here believes otherwise and can decipher the diagram output). Thanks all!

Sorry about the picture quality. Its hard taking a picture of a classic tube monitor screen

[dlfrsilver]

Originally Posted by snowpile79

Hello all. I have a very clean A2000 board with hardly any evidence of battery damage. Only a slight bit of corrosion on the positive copper trace of the battery. Battery long since removed. Everything worked great until a week ago. With KS 1.3 I get an immediate dark green screen on power on. With KS 2.04 and up I get a yellow screen. Diagrom loads fine! It flashes black/green (possibly brown green?) about 6 times then does the normal boot up display where it seems to detect ram. Finally I get a menu. Ok test results: RAM shows errors. Almost looks like the same right 4 bits are dead (see screenshot). Is that 4 bits or all 8 bits? I see 8 characters/hex lined up under 4 bit positions on the right. If anyone can help me understand if those are hex addresses? is that 4bits bad or 8bits? Also randomly 4 bits will show bad/red in the middle column (from the right most middle position). Honestly I dont exactly know what the right, middle or left top columns represent exactly.
Down below see how it says good memory range starts at address X ends at address y but then says 0k? Same for the bad memory? Im not sure what this means exactly.

Ok all other tests run fine until you get to the CIA tests.
CIA odd tick will often say too fast. Sometimes it gets to the test and crashes the amiga causing a reboot. Sometimes all CIA's test fine. Sometimes CIA Even crashes on the tick-test saying too fast.

All other tests appear to pass. Ran them multiple times. Do not know how the heck Im getting video with the consistent chip ram faults.

I did trace out the CPU traces to the upgrade port to the left of the CPU as well as to the rom (which clearly works as diagrom works fine). I also traced them to the buster. All tone fine. Thats all I traced so far.

No ram chips get hot. I did try an alternate gary, Agnus, alternate/new CIA's, paula and denise (and buster) in case one was creating bus noise. I did try both positions for the tick signal jumper (power supply and CIA). About a couple months ago I did replace caps around the cpu, video expansion and the 12v cap on the power plug input area plus a few around the cpu planning to do them all but ran out of time.

Any info on what diagrom is telling me or any other ideas would be very appreciated. I can replace the ram as I have desoldering equipment and sockets but for some reason I feel it is not ram (unless someone here believes otherwise and can decipher the diagram output). Thanks all!

Sorry about the picture quality. Its hard taking a picture of a classic tube monitor screen

the battery was removed, ok, but now you face something less fun : the battery acid went under some socket or chip, and you don't see the damages. On mine, some corrosion was hidden under the CPU socket.

I had to send it to hell, and install a new one. try to disolder carefully the socket and chips around the battery corner (the varta its acid in spray all around itself).

[snowpile79]

Sorry it looks like I forgot to mention a couple years ago I put a machined socket for the cpu. I did tone out everything from the socket to the accelerator slot and towards the rom socket. Ill continue to tone just odd behavior for diagrom trying to pinpoint any damage. If anyone can help decipher the diagrom output too that would be cool thanks.

(Edit: the machined socket was due to them providing better contact) I had plans to run a socketed accelerator. honestly have just ran a normal 68k in the socket since lol.

[snowpile79]

Some more info. Stuck my 2620 in the accelerator slot and it tests out fine including fast ram and zorro auto conf. same chip ram
Errors though. Will continue to tone traces.

[snowpile79]

Ok ran chipmem test and watched it. I think I know what is happening. Diagrom seems to run at the END of chipmem. In the chipmem scan all the bad addresses are the low byte of the beginning of ram and on towards the middle. The last chipram error is at $000522AB the rest is consistently fine. This is why diagrom works I think. So I think I have a bad ram chip and I think its the low byte bank and probably the first chip or maybe first two. Will calculate more and try replacing as all traces seems to be fine and the high end of chip is fine.

[snowpile79]

Just an update. Its fixed! It was ram chip U501 the first chip as calculated. Socketed and replaced. Was kinda stressful when the ole friend broke. It has panasonic ram chips. I have an a500 with panasonic ram chips that had blown two so far. I have probably five other a500/a500 plus’s two a1000’s two other a2000’s that have alternate brands and never blew a memory chip. Oh well old machines.

[dlfrsilver]

Cool!!!