The Tei Processor Terminal PT. 2

So, a lot has happened since my last post (ah the perks of working part time), when I left off all the cards where out of the system and I had just finished sorting out the damaged power supply. One of my cards had a dead short on it between the -16v rail and ground but aside from that everything was pretty much unknown to me. 

The first order of business, before I even thought about applying power to any of the cards (although the seller had already done that helpfully, grrr) was to backup any irreplaceable information. First port of call was the 2758 in the CRT controller. The 2758 is an interesting mongrel chip, being one of the first, if not the first single supply EPROM. It's effectively a 2708 that only needs 5v. If you want to read a little more about it, I enjoyed this article.

Now my particular one was presumably being used for character generation, or perhaps setting up the CRT5027. As my crt controller is so rare I was keen to get this backed up, as a 5v EPROM the process should be relatively simple, just read it as a 2716 and tie the AR pin to ground like it was in the actual controller. Unfortunately this did not seem to work. Every time I read it I got a different result. Here are 4 of my results (out of many more). I feel like I can see some sort of pattern here but nothing consistent enough to draw any conclusions from. I even got different results between the two halves of the 2716 I told my Minipro it was reading, which is weird as A10 was grounded. I do believe the first character on this rom is a Sigma but even that wasn't consistent.

So I assumed this ROM was damaged, which is a shame but not a total loss as it seems to be a character set. It wouldn't be fun but it wouldn’t be impossible to generate a new one. So I moved on to the ROMS in the floppy disk controller. Now these were interesting as after doing some reading on other S100 floppy controllers I realised these would contain any firmware for booting this computer. Given that the Processor card is set to power on jump to an external ROM located at the address that is Written on these ROMS I was sure this was the case. 

For reading these I used a set up that I have had complete success with backing up the roms in my Enertec Schlumberger 7600 logic analyser. Now this may make you cringe but I promise it isn't as bad as it looks. First of all, that's my mini pro that I've removed from its housing so I can attach an external 5V supply. This allows me to adjust the voltage to read tricky ROMS and more importantly, allows me to make up for the unstable and unreliable 5v my computer provides that tends to shut down when reading these old fashioned ROMS. Next is the converter on top of it that allows me to rob the missing voltages that a minipro cannot provide. Where better to get these from than a 2708 socket that already provides them? 

Despite how sketchy this looks, it was a success! Now I can tell you there was a disappointing lack of plain text in these roms, but they produced a sensible looking rom dump that, when disassembled, produces some sensible looking code that does uhhhhh.... something. I don't know enough about assembly to tell you what. I have however uploaded these rom dumps to Archive.org for your viewing pleasure. Do let me know if you have any input on them. 

  

So now at least some of the ROMs are safe and sound on the entirely intangible and definitely completely permanent internet, it's time to move onto the next thing. It's time to power the thing on! NO! Out of an abundance of caution and due to prior issues I decided to test each card individually with any potentially delicate components removed. After replacing one capacitor on the IO card that had gone short, all cards had a clean bill of health. Weirdly the cap I ended up replacing has already been replaced on some previous occasion. There were also small chunks dug out of the board on the adjacent solder points. I am getting the feeling I am not the first person to do work on this thing.

I then tested each card individually again with all components installed and again, a clean bill of health. No further shorts were developed and everything seemed healthy as far as I could tell. Finally it was time to try the IO card again, this time with the keyboard plugged in. The -16v rail immediately blew its fuse (again). After a little confusion about the board layout I found it was another shorted tantalum (using Adrians method of finding the problematic rail and then probing all the tantalums on it until you find the one with the lowest resistance). Once that was sorted and verified, it was time. Time for the first proof of life and I was presented with... GARBAGE!

 

Here it is, as viewed in a water damaged bathroom mirror (I still had the back of the computer facing the front of my desk) in a darkened room (because I had a migraine). I have never been so excited to see garbage on screen before. There were a few promising signs however. Although the garbage was different every time I powered the computer on, there was a capital letter F in the corner, with a blinking cursor underneath! So it was at least trying to do something. Another good sign is the CRT 5027 requires initialisation before it outputs video, so that part was at least receiving intelligible information from the processor, that was assuredly reading it from the ROM. All round good signs. 

 With this wind in my sails it was time to spin the computer round and try a few things. I tried pressing random keys for a while to no avail, aside from potentially momentary hesitation from the blinking cursor. However when reset and clear were held the screen was blanked before reloading with the exact same garbage. As in restarting the computer affected the garbage but resetting the computer did not. Which was strange. I did a cursory investigation of the processor board and concluded all was well on that side of things (data and address bus both were fully active, nothing stuck). So my suspicion fell on the RAM board. 

Upon probing various ram chips it seemed like I was getting data into the ram and garbage out, which was very concerning, especially with those beautiful gold ceramic chips I can't possibly afford to replace (no seriously). I was at a bit of a loss but decided to do a basic test before doing anything drastic. I worked my way through touching every chip and found one of them was scorching hot. Like seriously I wouldn't have been able to hold my finger on it for more than an instant. Really I should've realised something was up earlier as in my earlier testing i did find that installing the ram card dragged the -16v rail to nearly -16v from up near -18v. So with that chip removed I checked the rails on the ram board and discovered the -5v rail was at 0.7v!!! Very not good. I was concerned for the health of my RAM chips. 

The expandoram circuit for generating the -5v rail is alarmingly simple, just using a zener diode. 

 Now I've never liked this approach, it always seems like putting pence above sense. So I decided to bodge a 7905 -5v regulator on the board. Removing R7, CR1 and C13 I then connected the regulator to where they used to be in much the same way the other voltages are generated. It now lives on the back side of the board with the other bodge.

 Then it was time to replace the suspect chip, my parts ZX Spectrum bore fruit and donated a 4116 to the cause. It does stick out a little bit, but hey if it works it works. 0 Points for guessing which one it is.  

 

At this point I was ready to start the thing again, my hearts held great hopes for an elaborate Tei splash screen composed of exotic characters, I received Nothing.

 

[Imagine a picture of a blank screen here]

 

Resetting the computer like I had done previously yielded a cursor in the top left corner, and nothing else. Not what I had imagined. I was still suspicious of the ram so decided to have another investigate of the ram board and realised the /Write signal was stuck high! Weirdly this turned out to be a simple fix, the write protect switch was set to 'on' which according to the manual sets the write protect to off... which makes sense.  

However when I tried turning it to 'off' and upon my next reset I was greeted with this: 

 

Words cannot describe how ecstatic I was when this occured. I had to take a break to calm down again. Now this does not allow me to do much, in fact it doesn't allow me to do anything apart from spin the disk drives, but it proves some very important things. The system seems to be mostly operational. My issues reading the 2758 were clearly on my side at not the ROMs as there were no issues with the display. I seemingly have the only known operational Tei PT 208 and I've gained a new lead.  Abacus Computers LTD.

Who were Abacus? Well ill just leave this here.

 

 Check out the S100 section. I think ive found how this system ended up in the UK. Note the 60k RAM, perhaps the Expandoram is original to this system, and maybe even the Ithaca Audio Z80 card too.