Greetings,
I have an IBM 5160 64-256k malfunctioning motherboard here that is giving me quite the headache.
First of all, I don't know the history of this mobo, as I got it in a box of junk at a flea market years ago. Turns out there was a pretty good reason for it to be there.
Conditions when I found it:
- No expansion cards with it
- 128k of ram fitted (board half full)
- No NPU
- No evident bending or scratches on top or back
- C56 tantalum capacitor missing. Remains of the leg showed it was cut away. Blown up in the past, probably
As I suspected an issue with tantal caps in the past, I desoldered all of them and replaced with MLCC caps I had around. 1uF/25v. Also added a cap where C56 should have been.
I then checked the power connector for shorts. None were found, so I powered it up.
As I side note: I use my own adapter together with a known working PicoPSU to power boards I'm testing. The adapter takes care of providing -5V.
As a first test I tried the "Minimum diagnostic configuration" described on minuszerodegrees. Got the beeps notifying me of the missing video card. Good.
I fit my clone CGA card, adjusted the switches, powered up and... nothing: No beeps, no image.
Ok then, time for a run with the Landmark diagnostic ROM to see what happens... and this is where the real odyssey began.
- Burned an AM27C256 EPROM, 200ns access time. Fit it into U18.
- No extra cards installed except CGA adapter
- No keyboard connected
The diagnostics started. Most of the time I got them brutally locking up at the "Interrupt level 0" test (often with the speaker emitting the hi/lo beep from the landmark tests, continuously), but not always. I also got a few runs going through all the tests (obviously failing keyboard, ram above 128k and floppy controller/read).
Ok, as the docs mention that the interrupt level 0 test involve the 8259, the 8253, the 8288, and the CPU, I tried replacing all of them, one at a time, in that order (obviously desoldering and fitting sockets for soldered in ICs. No tracks or vias were harmed!). No dice, same issue...
While at it, as I had an 82C37 around, I tried that too. Nothing.
As the 8284 was already socketed, and I had a few CMOS spares around, I wondered why not try that too... was getting desperate.
I replaced the 8284 with a CP82C84, also had to remove R1 and R2 and fit 33pF caps in their place (required by CMOS 8284, see here http://www.malinov.com/Home/sergeys-projects/xi-8088), and this is where things got interesting:
As soon as the 8284 was replaced, I never got the lockup at interrupt 0 again, and now most of the time the first round of checks complete, locking up the computer (or resetting it) around the 2nd and 3rd loop of tests.
Now the errors I usually get are the following:
- Lockup at the floppy read phase, often including screen corruption
- Resets or screen losing signal at floppy controller test/read phase
- Very rare, errors on the first 16k, often different bits
I tried shuffling the RAM chips, removing the second bank, removing U9 (the buffer on ram data lines) and fitting a new one. Nothing: same issue.
Then I decided to remove the video card and just go blind with the acustic error codes: the tests ran longer, around 30 minutes before locking up (and they restarted abnormally a few times in these 30 minutes).
So, starting to get a little out of ideas here. I hate intermittent issues.
The only thing left to do is performing the 1mb mod, so I get to replace the RAM decoding PROM and fit new chips.
Anyone got ideas on what else it could be tested/investigated?
I have an IBM 5160 64-256k malfunctioning motherboard here that is giving me quite the headache.
First of all, I don't know the history of this mobo, as I got it in a box of junk at a flea market years ago. Turns out there was a pretty good reason for it to be there.
Conditions when I found it:
- No expansion cards with it
- 128k of ram fitted (board half full)
- No NPU
- No evident bending or scratches on top or back
- C56 tantalum capacitor missing. Remains of the leg showed it was cut away. Blown up in the past, probably
As I suspected an issue with tantal caps in the past, I desoldered all of them and replaced with MLCC caps I had around. 1uF/25v. Also added a cap where C56 should have been.
I then checked the power connector for shorts. None were found, so I powered it up.
As I side note: I use my own adapter together with a known working PicoPSU to power boards I'm testing. The adapter takes care of providing -5V.
As a first test I tried the "Minimum diagnostic configuration" described on minuszerodegrees. Got the beeps notifying me of the missing video card. Good.
I fit my clone CGA card, adjusted the switches, powered up and... nothing: No beeps, no image.
Ok then, time for a run with the Landmark diagnostic ROM to see what happens... and this is where the real odyssey began.
- Burned an AM27C256 EPROM, 200ns access time. Fit it into U18.
- No extra cards installed except CGA adapter
- No keyboard connected
The diagnostics started. Most of the time I got them brutally locking up at the "Interrupt level 0" test (often with the speaker emitting the hi/lo beep from the landmark tests, continuously), but not always. I also got a few runs going through all the tests (obviously failing keyboard, ram above 128k and floppy controller/read).
Ok, as the docs mention that the interrupt level 0 test involve the 8259, the 8253, the 8288, and the CPU, I tried replacing all of them, one at a time, in that order (obviously desoldering and fitting sockets for soldered in ICs. No tracks or vias were harmed!). No dice, same issue...
While at it, as I had an 82C37 around, I tried that too. Nothing.
As the 8284 was already socketed, and I had a few CMOS spares around, I wondered why not try that too... was getting desperate.
I replaced the 8284 with a CP82C84, also had to remove R1 and R2 and fit 33pF caps in their place (required by CMOS 8284, see here http://www.malinov.com/Home/sergeys-projects/xi-8088), and this is where things got interesting:
As soon as the 8284 was replaced, I never got the lockup at interrupt 0 again, and now most of the time the first round of checks complete, locking up the computer (or resetting it) around the 2nd and 3rd loop of tests.
Now the errors I usually get are the following:
- Lockup at the floppy read phase, often including screen corruption
- Resets or screen losing signal at floppy controller test/read phase
- Very rare, errors on the first 16k, often different bits
I tried shuffling the RAM chips, removing the second bank, removing U9 (the buffer on ram data lines) and fitting a new one. Nothing: same issue.
Then I decided to remove the video card and just go blind with the acustic error codes: the tests ran longer, around 30 minutes before locking up (and they restarted abnormally a few times in these 30 minutes).
So, starting to get a little out of ideas here. I hate intermittent issues.
The only thing left to do is performing the 1mb mod, so I get to replace the RAM decoding PROM and fit new chips.
Anyone got ideas on what else it could be tested/investigated?
