The machine also came with some extra S100 boards:

1. •CDC64k memory card
   

2. •CompuPro EconoROM 2708
   

3. •DRC 16K ROM
   

4. •Processor Technology 3pps
   

I removed all these boards to simplify the repair process. The first problem I encountered was that the video out connector was a non-standard type. I jumped on eBay again and purchases an adaptor. It would be a while before the adaptor got to Japan, so I temporarily hooked up some test clips to the video out and fed the signal into the back of my composite monitor.

After an initial inspection to see if anything was broken I then powered her up. What I saw was very promising. I could get a nice clear steady image on the monitor, however I was getting a screen full of strange characters, and not the  ‘>’ prompt and single flashing cursor that I was hoping for. Again, the internet was no help here and doing some searches produced no results. I did however download a slew of scanned manuals from sol20.org. These manuals turned out to be very helpful.

As always I checked to see if the CPU was running and I was getting good pulsing in various places around the motherboard. I also tested the voltages from the power supply and all seemed good. I then tried playing with the DIP switches that effect some basic operations. Although it did change some random characters on the screen and how the screen looked (inverse screen on/off etc), the outcome was generally negative. One important setting was when switch 1-1 was in the on position the screen would fill with a square graphic followed by a ‘9’ repeating, and looking something like ‘090909’...This setting turned out to be important later.

I first turned my attention the Personality Module. This is a small card that plugs into the motherboard and holds the monitor ROM chips. I wondered if the ROM was suffering from bit rot. I soon realized it would be difficult to dump the contents of the 9216 ROM with it’s non-standard pin configuration. At the same time I also happened across an eBay auction selling bare board remakes of the PM2708 Personality Modules. It seems the the seller had too many boards and was selling a few off very cheap. I decided it would be simpler to get one of these boards and build it up than it would be to build some adaptor to try to read the 9216 ROMs. In the end I purchased two boards. Once the boards arrived I build one up and programed some 2708 EPROMS from the SOLOS binaries I found online. Once plugged in the results were exactly as before. Although this didn’t eliminate the ROM’s from being the cause of the problem, it certainly puts a lot of doubt on the possibility.

I decided to then move on to testing the RAM. As I mentioned before, when DIP switch 1-1 was in the on position the screen would fill with square graphics followed by and a ‘9’ repeating... After reading the assembly manual, this seemed to indicate that the CPU was running and that the video RAM was good. I also discovered that the onboard RAM came in two banks, One bank was the video RAM, the other is the user RAM. I realized this was good news as it would allow me remove the user RAM and then, one by one, swap that RAM into the video RAM bank. If the screen changed with the DIP switch 1-1 set for off, then that would be a good indicator that the RAM that was swapped in was probably bad. After doing this I found 3 dead chips. Unfortunately now I didn’t have enough user RAM to test the machine. I did however have the S-100 CDC64k RAM card that came with the machine. I removed all user RAM from the motherboard and using the DIP switches on the RAM card, I remapped 8k of memory to the 0000 address space. Restarting with the card installe produced a flashing cursor on the screen. There was still no ‘>’ symbol. Maybe there was some sort of bus conflict going on that didn’t allow full initialization of the boot code. Anyway, this was all good news. I ordered some new 2102 RAM chips to replace the dead ones and they arrived about 2 weeks later. I removed the RAM card and refilled the user RAM banks with known good chips. On power up the machine sprung to live with the ‘>’ symbol and the flashing cursor. Working, hooray!

...Wait, not so fast Philip. Seems that the keyboard was totally dead. Not a single key worked.

(Just as a side note, I now believe may be possible to test the RAM with the DIP switch 1-1 in the on position. If the pattern in the screen looks like 090909, then the RAM is good. If the pattern is something like 070707, or 080808 or anything else, then there is a problem. This would have probably been a much simpler test for bad RAM that just looking for changes in screen corruption with switch 1-1 off)

Actually I had anticipated the keyboard to be non functional. I had done some research before the Sol-20 arrived, and non functioning keyboards are a very common problem with old computers that use foam pads in the key switches (Apple Lisa and some Sun Microsystem machines). Over time the foam degrades and turns into a sticky or flakey mush. The only way to fix this is to either make some new foam pads (difficult process) or steal some from another keyboard that uses the same basic design. I was lucky enough to find someone online who was selling a set at a very reasonable price. Although I was expecting the keyboard to be an issue, I did think at least some keys would work. Having no keys work was a little bit of a concern as it may be an indicator of another issue unrelated to the foam pads. I went ahead and started the foam pad replacement. It quickly became obvious that the foam pads were very dead indeed. A gooey sticky mess. It took quite some time to clean all the gunk and finally install the new pads...shock!!!! I was one foam pad short of a full set. Oh no, what to do. I contemplated making one, then I remember that my New Zealand friend and fellow retro computer freak, Terry Stewart, had recently fixed an Apple Lisa keyboard using the same type of foam pads, and may, if I was lucky, have some spare. I was lucky, he had a few left over and we arranged a swap. A couple of 2716 EPROMs for a foam pad...seemed like a fair deal to me. Finally I was able to fully fix the keyboard an the machine.

I presently only have the CDC64k memory card installed. I found this card to be fully functional with no problems.

Conclusion

A very satisfying fix as I didn’t have to rely on other peoples knowledge. The Sol-20 is now at the top of my favorites list along with my Rev 0 Apple II, Commodore PET 2001j and Ohio Scientific 4P.

I have purchased a Versaflopy II cards for the Sol-20, and sometime in the future I will try attempt to get CPM up and running. Stay tuned...

Update - As it turned out, the Sol-20 still had one last surprise for me.

I’ve been wanting a Sol-20 for quite some time. It’s a classic, and considered by many to be the first true personal computer that came with a consumer friendly case, and predating the original Apple II by a year. I love the sturdy sheet metal case and the walnut wood side panels. The CPU is an Intel 8080 and is the precursor to the famous Z80. I was very surprised at how heavy the machine was, probably weighing as much as my Commodore PET 2001, which is quite an accomplishment considering the Sol-20 didn’t have a build in CRT screen like the PET does.

I got my Sol-20 from eBay in non working condition. In retrospect, and considering its age (35 years old), the fix was relatively simple. The machine was in very good condition, both physically and electronically.

Unfortunately there wasn’t a lot of repair information available on the web, and no dedicated forum to ask questions. This was one of the few fixes where I felt I was totally on my own.

neoncluster.com

Funky screen with DIP switch 1-1 in the off position.

Funky screen with DIP switch 1-1 in the off position.

With the DIP switch 1-1 in the on position I get a screen full of repeating squares and 9’s. This tells me, according to the manual, that the CPU is working and the video RAM is probably good.

After fixing the keyboard and replacing some bad RAM the machine lives. Here we can see it playing a game of chess.