In case the CMOS data were corrupted I tried to clear the DS12887A by wiring the GND to RCLR bar for a few seconds. But nothing have change in the boot sequence. I still got the error code 00 08. I have redone the soldering of the socket and tested some traces, all seems fine. Originally the job was well done. I begin to think the problem is elsewhere. Each time the thing occurs after changing the configuration and having done something with the motherboard disks ports. Maybe there are bad soldering around the HDD and FDD ports. I'll check that tomorrow. I'll remove all socketed chips and clean the contacts too. Who knows.
I will try everything I can to save this pesky Commodore.

Which Dallas chip did you use, can you clear the CMOS? I'm assuming you've tried unplugging all the drives again? Also, I'm not clear from a quick look through the service manual whether the floppy interface is IBM style (twist in cable to select drive) or Shugart style (drive select jumper on the drive). I think it might be Shugart, in which case you'll need a floppy drive that can be jumpered to fit.

snufkin wrote on 2022-02-07, 00:34:

Which Dallas chip did you use, can you clear the CMOS? I'm assuming you've tried unplugging all the drives again? Also, I'm not clear from a quick look through the service manual whether the floppy interface is IBM style (twist in cable to select drive) or Shugart style (drive select jumper on the drive). I think it might be Shugart, in which case you'll need a floppy drive that can be jumpered to fit.

The DALLAS is a DS12887A that I clear the RAM with no change unfortunately.
The floppy connector is a Shugart so it needs drives with a jumper setting which I have in the machine (5.25" and 3.5").
The floppy connector is a normal IBM PC compatible connector. I checked the Chinon 5.25" drive and indeed it has no jumper. So it means 1) I can use classic standard 3.5" FDD, even modern ones so I put a more reliable Mitsumi in place instead of the Epson. 2) The original ribbon cable is not original as it is not twisted... 3) I remembered jumpers on the 5.25" but that was surely for another machine.

Correction on the part 3) of my explanation as it was not very clear, here are the detailed steps of what I have done.

• Tried a few start/restart with, cold and hot with no floppy drive connected with success.

• Then started with floppy drive connected with the original cable. Entering the BIOS configuration setup, that is done after the boot sequence at any time, by pressing on the keyboard Ctrl + Alt + Esc. I configured the drive, setting no HDD (as I would like to use a XT-IDE Deluxe), enabled COM and LPT ports and adjusted the time as it was little in advance. Saved and restarted (hot reset done by the Setup). Everything was detected and the drive was seeking at boot and a message to insert a floppy to boot was displayed.

• Then, I switched off the machine and replaced the floppy cable. And when switching it on again, it hasn't started. That is why the two suspects where the CMOS and the handling of the floppy cable.

I tested the cable since wire by wire and it is good. So the problem was not with it. Maybe with the electronic or solder near the connectors. To insert or remove HD or FDD cables bring some constraints on the board. The chassis of this machine is quite badly engineered. Amongst many things, there is little space, only a hole, for accessing the disks ports. And when the PSU block is installed (the PSU is attached to a support where the HDD and FFDs are attached too) and as there is no holder under the MB where the ports are, the nearest ones are on on each side on the keyboard connector (located on the right side of the machine) it could be a problem with damaged traces, pads or solders in the area.

A post error code 08 is at the very beginning of the POST sequence. I may be wrong... but on such machine, does stopping at 08 means it passed 08 and stopped at 09, or does it means it engaged 08 and never passed it?
That is why I also wrote the 09 post code for both Phoenix and the manual.
The MB is very big with a lot of ICs. I must be sure about the error code. Logically, if the card display a 00 08 error code and the manual say the error correspond to xxx so that should be that...
Meaning if it engaged 08 and not passed it, that is a defective RAM or RAM logic. This test is just writing a 128 KB sequence to check the RAM, no to test it, as detecting/testing the RAM is step 00 10.
If it is on 09, then it is a problem with the video card.

As I said the machine was stored as not working because of this problem. When reinstalled it 10 years latter and it worked miraculously as if nothing where wrong with it.
Something happens to make it work again, but what?
And what is the culprits of it not working again?

I'll check the soldering with the microscope near the disks ports. I'll try an ISA video card too. As step 00 09 is detecting the video card, I will see.
I still doubt there is a problem with the RAM, but if it is. I'll have to remove the chips, but I have nothing to test them... I'll put sockets for further easy replacement.
The PC have 1 MB of RAM. And that brings a question: is it possible to replace the modules by bigger one to install 4 MB (8 chip of 256Kx16) instead? If I'm correct the 80286 can address up to 16 MB but 8 MB is fine for all use case of such a machine. Or have I no choice to replace by equivalent size? I think the good answer is the last one as I don't know DIP20 chip like that with more than 256Kx4 as they have only 9 address lines.
Of course, that is only if there is a problem with the RAM. As the problem can also be with a controller (Faraday), a tantalum cap or anything else...
Thanks.

Edit 1: I cleaned up the contacts of every socket. I also checked if components became hot on the RAM area, in fact everything is on the MB is cold except the CPU that become a little warm after several minutes powered on.
Edit 2: RAM chips are 256Kx4 branded Toshiba and Hitachi on two banks.

Here are some photos:

Apologies in advance digging up an old thread, but I happen to have a PC40-III with the same issue. Did the OP manage to get theirs working I wonder.
Mine is the pre Dallas version so have repaired a lot of battery damage. WAS alive, then just randomly died after I left it on while I was trying to whip up a boot disk for it. POST card also stops at 08. Have spent many hours with a scope checking everything I can in the ram area and no obviously faulty signals.
I do note U1203 on mine has been replaced with a 74S175 instead of 74LS175 and this chip seems to get quite warm compared to any other logic on the board. So I think I will obtain the correct LS version since there's already a socket there. Short of that I'm running thin on ideas

***
Edit: I discovered a POST card shows code 08h even on a booting system so is no use in these machines!
Dead RAM in bank zero solved this problem.
A corrupt CMOS RAM also stopped the system booting as it thought it had logged an illegal shutdown condition. Replacing the clock chip with a necroware one solved this for me.

3VILC wrote on 2023-10-06, 05:22:

Apologies in advance digging up an old thread, but I happen to have a PC40-III with the same issue. Did the OP manage to get the […]
Show full quote

Apologies in advance digging up an old thread, but I happen to have a PC40-III with the same issue. Did the OP manage to get theirs working I wonder.
Mine is the pre Dallas version so have repaired a lot of battery damage. WAS alive, then just randomly died after I left it on while I was trying to whip up a boot disk for it. POST card also stops at 08. Have spent many hours with a scope checking everything I can in the ram area and no obviously faulty signals.
I do note U1203 on mine has been replaced with a 74S175 instead of 74LS175 and this chip seems to get quite warm compared to any other logic on the board. So I think I will obtain the correct LS version since there's already a socket there. Short of that I'm running thin on ideas

***
Edit: I discovered a POST card shows code 08h even on a booting system so is no use in these machines!
Dead RAM in bank zero solved this problem.
A corrupt CMOS RAM also stopped the system booting as it thought it had logged an illegal shutdown condition. Replacing the clock chip with a necroware one solved this for me.

@3VILC
Sorry to reply such a long time after your reply.
Thank you very much for your feedback. I suspected it, but knowing POST is not supported is good to know and will avoid me to lose time with it.
I still not managed to fix this machine this day as I have a little time for that, unfortunately. But from a recent close inspection what I found in mine is a tantalum cap above an ISA slot is fried. But it is not shorted, and doesn't explain why the computer does start once a while, and then does not. So I suspect a problem with the CMOS, maybe the replaced DALLAS chip is bad somehow (Necroware's nwX287 can be a solution here). I tested each RAM chip with an oscilloscope and they seems all good. In fact everything I could test seems good even on the ISA slots, despite the bad cap. ? Signals with clear voltages, nothing erratic. And without a proper technical documentation, that I'm not able to find, it will be hard to fix.