PS/2 Model 25 troubleshooting finds a bad BIOS EPROM!

A week ago I received the planar of an IBM PS/2 Model 25 that wasn't working - no video came up on the screen, and no beeps from the system, despite being able to hear the system fan and hard drive running. I benched the planar and connected my parallel port POST code reader - the POST code displayed was '01', the planar wasn't getting very far at all.

I first swapped the 8086-2 (8MHz - Intel had an odd way of marking speeds back in the day) CPU for another with no change. On the first ('U17' - of a remarkable coincidence today with YouTube channel "CuriousMarc") ROM BIOS chip I swapped, the planar immediately started POSTing and had video! A different fully functional planar will be sent back for his system (I'm sending quite a few working PS/2 planars to a few people to fix their systems!) - but here the troubleshooting process and parts I used is documented on video.

[hughnisbet]

I loved this video, it showed the steps that you went through to home in on the correct problem, a lot of other sites seem to presume that you already know the steps that they took. Thank you once more for an interesting video.

[homelate]

I stumbled upon this video through a Google search. I happen to have a non-functioning Model 70 with a bad rom. I saved the rom to an image and compared it with the official rom. It was different so I'm now waiting on some EEPROM's to make a new rom and I hope that it will work again.

[johnstory]

Another great video! Thank you for the assist!

[GeckonCZ]

Nice job, Dave, one more PS/2 saved!

I agree that the ROM is at fault here, but things don't quite add up. Based on the diff results it may seem that bit 5 of the even ROM has a problem. That however doesn't make much sense in this context, because the system was able to reach checkpoint 01 - that means it made through some 50 (!) instructions with no problem. There would be nothing on the parallel port if one of the ROM data lines was disconnected. It would fail on the very first instruction - unconditional far jump stored at the reset vector address FFFF0h. The opcode is stored in the even ROM and normally its value is EAh. With bit 5 disconnected we would get code CAh instead - which is RETF (return from subroutine). With the stack not initialized, this will most likely crash the system almost immediately.

[douro]

Mine ended up with bad BIOS ROMs and they were the mask programmed variety. It was a good thing I had a good set of spares in storage since I couldn't find any images of them online back then.

[TravellingVikings]

Looks like pin 13 O2 doesn't deliver it's content in the readout. That explains the difference between the chips. No wonder it didn't boot up properly! It's good to know when I dig into my collection of PS/2's!

[richardsantink]

Awesome videos... I'm former IBM (14yrs) and I still have many of my machines including my PS/2 9577-ANB, and also my 8525-004. Right now, my 8525 won't give me video, but it does respond to keyboard (two-beeps, F1, beep). Any ideas of anything I can check? I've disconnected, clean and reconnected power and video connectors to be sure something didn't come loose.

[douro]

Have you ever had trouble with a floppy drive where it has no trouble writing but has trouble reading towards the end of the disk?