I purchased this PET 2001 from Yahoo auction (Japan). From the auction listing I could tell it was one of the earliest versions of the 2001 with the blue monitor surround and the blue gold faceplate sticker. I could also see that it had a Japanese chiclet keyboard, and I have never seen or heard of any other like it. It’s a rare computer indeed.

When the computer arrived it was fairly grubby and suffered from a very light coating of surface rust, but nothing major.

I pulled the power leads off the motherboard and did a quick voltage check, all seemed good so I plugged it all back together and fire her up.

Nothing. Dead. I found that power wasn’t getting to the  motherboard. With help from Ethan Dicks we tracked the problem down to some dead diodes on the power circuit. Thanks Ethan

FInally I could power he up. The screen took some time to warm up and come on, but when it dis I saw a screen full of random characters which is a typical symptom of problems with these machines. Unfortunately this screen lasted about 5 minutes before everything suddenly changed. Gone were the characters, to be replaced by blocks and vertical lines. After some testing It turned out that the character ROM had failed which is unfortunate because I did see some kana characters displayed on the screen so I knew that it was the Japanese version. The Japanese character ROM has not been backed up by anyone, and the binaries can not be found anywhere on the net...and mine, the only known ROM still in existence had just died!!!

Well, there is no use crying over spilt milk, the ROM was gone and there was nothing I could do about it.

I needed to burn a new character ROM as a replacement, so I ordered a 6540 ROM adaptor boards from Nicola Welte. While I was at it, I also ordered two 6550 RAM adaptor, and a couple of RAM/ROM boards (I have another PET 2001 board that need fixing and I thought having two of these RAM/ROM boards was a good idea). After a short wait the boards arrived, and I quickly built up the ROM adaptor and burnt an EPROM containing the english character set. The was an instant improvement, now I had random characters again where I once had blocks. However on closer inspection I could see that there there vertical lines running vertically through all the characters and a lot pixel noise. In fact that is what you can see in the image above. I turned to the friendly folks at VCF (The Vintage Computer Forum). These guys are very knowledgable and always excited to help out.

neoncluster.com

After some discussion on the forum, Ruud Baltissen emailed me a very simple test kernal ROM replacement program that displays the character set on the screen. I removed all the ROMs from the board and burnt the Ruud’s .bin file to a 2534 and inserted it into the socket at H4. The simple test program worked as it was supposed to, which confirmed the CPU was working and able to run code. There was still lots of noise on the screen, and I could also see now that the left most column 1 and 2 were repeating!!!

There was also a suggestion to replace the 74165 at B3 to see if that clears up the video noise. as this chip was socketed it was a no brainer. A new 74165 fixed the noise problem and the characters were now displaying clearly.

At this point  it was a fairly safe bet to assume that either some of the original ROMs were bad, or maybe an issue with RAM. It was time to build up Nicolas Weltes marvelous RAM/ROM board. With this the 6502 CPU is removed from the 2001’s motherboard and is installed in the RAM/ROM board. this board is then inserted into the empty 6502 socket on the motherboard. The RAM/ROM board replaces all the RAM and ROM on the main board, and allows you to switch between different versions of Commodore BASIC. Very handy indeed for fixing these old machines. Unfortunately when building my board up I had accidentally put the resistor pack in the wrong way, and it took me days to figure out my mistake,

Finally with the RAM/ROM board installed I was able to get the Commodore BASIC splash screen, although the repeating column issue became very apparent.

I also realized that I couldn’t get the machine to boot into BASIC4. This problem turned out to be a corrupt flash ROM which came with RAM/ROM board. I programed a new one and that issue was solved. I also built one of Nicolas's 6550 video RAM replacement boards which was installed, although that did not solve the screen problem.

Solving this last issue became a long drawn out process with me posting scope reading to the VCF list on a daily basis for about 2 weeks. One suggestion from the group was to wire in a 33pf capacitor connecting pin 7 (GND) and pin 13 of the flip flop at C8...surprisingly this resolved the screen problem quite nicely, but it still didn't fix the underlying problem, it just masked it. Finally dave_m figured out that the problem lay with the 74LS00 at D8 which had some slight timing issues. Nothing very noticeable, but enough to cause problems. Replacing this chip fixed the issue.

The last thing to do was to recap the monitor. It suffered from serious screen wobble. Working with monitors is always tricky because of the high voltages involved, so great care should be taken. Once recapped the monitor was nice and stable and the machine was finally fixed.

Conclusion

That was the most complex fix I have ever done. I learnt so much about how to use my oscilloscope in the process, and without this knowledge I wouldn't have been able to fix anything. It was a great learning experience. I also wouldn’t have been able to fix the machine if it wasn’t for all the people who I mentioned in this post, and to the countless others that helped on the forum.

It was quite upsetting that the character ROM died. I keep thinking that I should have backed it up right away, but at the time there was no knowing that it would die within 5 mins. I have attempted to make new binaries that I believe are the same as what Commodore used and you can read about that here.

After the death of the Japanese character ROM.

A very rare 1977 Japanese PET 2001-8B. Featuring a Japanese chicklet keyboard, blue screen surround, and blue gold sticker. That means it’s a super early production model.

The game ‘Android Nim’ like you’ve never seen it before. Note the repeating column 1 and 2, and the character offset that cascades down the screen.

Timing issues showing up in C8. Green line shows the broken board. Red line shows a known good board.

After replacing the dead character ROM, and instaling a small test kernal ROM at H4. Note the issue with the first and second column repeating.

Replacing the 74LS00 at B3 fixed the repeating column issue.

A kludge fix for the repeating columns.

The motherboard I have is an unusual hybrid featuring both
6540 ROMs + 6550 RAM.