Commodore PC40-40 repair, mainboard is a PC/900-2, PB85065-(I)-2F

[Wobblestone]

Hi,
this is not a request for help; I just want to document my repair of a Commodore PC40-40. It seems that the mainboard is quite rare and maybe this thread helps somebody else to repair his machine.
I did not find much information about that machine and board. There are no videos on Youtube. Just a couple of threads here and there showing pictures. But no repair threads.

As I purchased this machine it came complete with a graphics card (VGA with a Trident RVGA 8800CS and 512kB Video RAM), a controller card WD 1003-WA2, a random serial/parallel card and a 5.25" floppy drive and a harddrive Hitachi K521-5 with 40MB. The machine has 1 MB or RAM == 4 banks with 41256.

It was not working as I received it, and it was the dirtiest machine I've ever bought.
After disassembling it, I washed the mainboard and then everything else. The PSU was dead. The fuse was blown. I replaced it, but it instantly got destroyed again. No wonder, since the main switching transistor - a BUV48A was completely shorted. I ordered one and I'm still waiting for it.

I continued with the mainboard itself, which I powered with a modern ATX PSU with an adapter cable. The mainboard was completely dead. It has also 2 ROMs like the NCR, which I repaired recently.
In order to diagnose the mainboard, since it's one of the clones of the IBM 5170 with 80286 and 2 ROMS, I used again the Landmark/Supersoft ROMs:
Put Landmark U27 into U82 of PC/900-2
and
Landmark U47 into U80 of PC/900-2

But no change. The machine remained dead....
....until I poked around with my oscilloscope. Suddenly I got a picture, but the machine stopped at the memory check.
Ok, time to check the RAMs. The first two banks were socketed. Checking the chips revealed 3 dead ICs. I did not check the second and third bank.
The machine still started up only very rarely.
I began to remove ICs from their sockets, putting Deoxit in the socket and so on. That did not help at all....
.... until I removed the CPU itself from its socket.
I cleaned the socket and the CPU and finally the machine booted up and worked perfectly fine.

I can't explain the behavior completely. But maybe poking with the oscilloscope bent the board a little, which closed some contacts within the CPU-socket and that brought the machine back to life. Maybe...

The graphics card was tested in a known good mainboard and it worked out of the box, as well as the controller card.

Surprisingly even the harddisk worked. I just formatted it using Spinrite and Speedstor and installed MS-DOS 6.22.
Only the 5.25" floppy was and is still dead. Most cables were completely rotten. As I tried to repair them, I completely destroyed the cables for the stepper motor of the read/write head.

 

[Ruud]

.... a Commodore PC40-40. It seems that the mainboard is quite rare ....

No, not really. You have various types like the PC30-III, PC35-III, etc. Have a look at Bo Zimmers' site. They all have the same board. But some boards have onboard VGA and others have not. A good example is this PC30-III board but clearly shows the text PC40-III at the bottom-right between the ISA slots. Notice the unpopulated part at the board at the left: this is the VGA part.
Having a VGA card I assume that is also the case with your board.
But... the PC40-40 could be different. Would you mind sharing a photo of the board, please?

 

[Wobblestone]

Unfortunately I forgot to make photos. It is now already inside of the case and I do not want to remove it again, since I hate these tabs, which hold it in place.
The photos which I just made are already in the case. But you should see, that this is a completely different board than the one of the PC30-III.
So far I have 4 Commodore PCs (not Amigas...) - two are in these small cases of PC30 and so on. Two are in the big cases, which look like an Amiga 2000.
This case is different to the other one. I only found one photo of this machine on Bo Zimmers Site: https://www.zimmers.net/cbmpics/cpcs2.html
There is a photo of the PC40-40 (but with 2 floppy drives).
It seems that the Commodore PCs of that era were a complete mess
Fun-fact: On the front of the computer the badge is Commodore PC40-40 like on the photo on Bo Zimmers page. On the backside the badge shows Model PC 40-01

 

[Wobblestone]

I forgot to mention, that I replaced the original clock chip HD146818P on position U120 by a replacement from Necroware. It's a nwX287, which I bought from eBay.
But you can't use that out of the box. As Necroware mentions on his Github page, you either need to put a pullup or pulldown on pin 1 of the chip.
I got this also from Adrians Digital Basement, where he replaces the same chip on a BBC Master.
But on the Master he had to put a wire from Pin 1 to VCC, while here - on an Intel machine - I had to put a pulldown from Pin 1 to Pin 12.
Now it works great and should last a couple of years. And no bloody external battery thing.

 

[Ruud]

That's indeed a complete different board and I have never seen it. Regarding the badge differences: I almost consider it normal

 

[Wobblestone]

That's indeed a complete different board and I have never seen it

That confirms, what my many google searches revealed: There is almost nothing to find about this machine. Hopefully this thread will be used in future to collect information about it.

 

[ronaldvd]

That's indeed a complete different board and I have never seen it.

Going by the photos, this does not look like a board designed by Commodore themselves.

There is a brochure in German about the PC40 on archive.org which includes the specifications of the machine.

 

[Wobblestone]

I also think, that this board is not a development of Commodore. The graphics card, which is in my machine, is certainly also not the one, with which the machine came back in the day. It is an interesting little... ok - very big - machine.

 

[rodney]

Hi Wobblestone,

Sounds like you had a lot of fun with this PC, and it looks like you are a fan of these Commodore machines.
I was tempted to buy one a few times but at the time I was looking for a "chipset-less" machine for my project which the models on offer were not.
It will be interesting to look at Commodore PCs one day. I am curious if there are some typical Commodore methods of doing things that you can recognize in these computers.
I looked more closely at your photos and noticed this is also a early 286 technology machine, using only some PALs from the looks of it, and probably a few PROMs for decoding purposes.
These machines may be a little vulnerable, at least from my experience.
It's lucky yours seems in a running condition.
So DMA is working for accessing the floppy controller? Did you test some other known working floppy drive mechanism on this PC?
And I am curious to know if a soundcard will work on it.
This will confirm that the complex 8 to 16 bit conversion mechanism on your mainboard is functional.
This mechanism was invented by IBM to ensure full compatibility with all earlier 8 bit hard and software, and was instrumental to their success with the introduction of the AT models.

I replaced the original clock chip HD146818P on position U120 by a replacement from Necroware. It's a nwX287

Recently I have been experimenting with RTC ICs because I only have one MC146818 which is still alive.
I managed to use a DS12885 which is the same as the fat battery chip you see more often, but without an integrated coin cell, so a regular DIP case.
I discovered that this RTC is extremely precise in how it is powered by the coin cell, if there is any voltage slightly outside of the range, the chip will not power up at all and seems dead.

So thanks for testing this Necroware chip, I saw his GitHub I think. Since it can replace the 146818 directly, that is interesting to me for my projects.
I mean, I did make provisions for a DS12885 with a few optional parts and jumpers, but I still need to test this in the actual mainboard later.
Somehow I dislike these Dallas ICs because they are so picky about the battery voltage.
RTC chips are always an annoyance in my opinion.
If the machine is left without power for a while, I wonder if the Necroware one can still have a functional coincell or it has too, depleted.

It's really annoying to find such a lack of documentation from PCs of those days. Either they were lazy or fearing the competition.
But I believe any good competitor will be able to reverse engineer anything anyway.

From your photos the PCB does look similar to other Commodore PCBs in terms of the type and color of solder mask, it's comperable to the Amiga 2000 boards I have seen here, it seems to remind me of those.
Nice of them to feature double BIOS sockets which may be convenient for using some option ROM code inside those. It depends if the option ROM code is compatible with running from a 16 bit ROM construction of two data bytes. I am still not sure if the software needs to be compatible with this or not. I mean, if the jumps and other instructions would need to be 16 bit versions, for example. And of course, if those instructions are legacy 8 bit instructions, would that work in a 16 bit enabled ROM address space? This could tell us more about how this double socket construction could be used for our own purpose with these types of machines. For example in other PCs we more often see a single 8 bit ROM which of course will be enabled to run in 8 bit compatible data byte mode, requiring data byte conversion to even run on a 16 or more bit wide CPU.

 

[Wobblestone]

So DMA is working for accessing the floppy controller? Did you test some other known working floppy drive mechanism on this PC?
And I am curious to know if a soundcard will work on it.

As I had disassembled the machine, I used it to test some floppy drives - 3.5" and 5.25" (360k and 1.2M) - and they worked fine. But I only used the controller, which came with the machine. Also I did no extensive testing, just read, write and format.
I have absolutely no experience with sound cards. All my machines contain sound cards, but I have no clue which drivers to use and so on.

If the machine is left without power for a while, I wonder if the Necroware one can still have a functional coincell or it has too, depleted.

Regarding to Adrians Video, this should last several years. I could measure the current, but frankly - I normally take the battery out, when I know a computer works. Unfortunately/fortunately I have already many computers - too many for using all of them regularely.
The only problem which I have with the nwX287 is that building such devices becomes more difficult, since the main IC bq3285lcs is already very hard to get. A bq3285s can still be found, but it costs about 10 - 12€.
The three nwX287 which I bought from eBay are the last ones, which were there for sale. I paid 10€ each + some € for shipping.
So far I used 2 of them, one for the Commodore and one in a PS/2 SX55.
If possible, I rather reuse the Dallas/Odin-Chip and solder an external battery instead of the depleted internal one.

Nice of them to feature double BIOS sockets which may be convenient for using some option ROM code inside those.

You know much more about this than me. MUCH more I'm just happy to get that machine going....
The only thing I know is, that this design is very similar to the IBM 5170. I don't have such an expensive machine; I have that information from this forum and Minuszerodegrees.