TI99/4a

So far the only computer in my collection that meets the '2 bytes' criterion. The TI99/4a is a 16 bit home-computer made by Texas Instruments. It took a while before I found one at a decent price (€25,-) but I finally got one from another collector who was clearing out his collection.
It came complete with power supply and RF-modulator. The RF-Modulator was a bit of a surprise. So far I had not encountered a computer that used an external one, and certainly not such a huge, heavy black box as this one. Fortunately the output leading to the modulator carries standard YUV (or Component Video) signal, which is supported by my GBS8200 Video Converter Board.

The TMS9900, one of the first 16-bit processors.

After powering up the screen indeed shows the TI99 intro screen. But it is unstable. Graphic characters just randomly seem to change shape. Pressing the '1' key does bring up the basic prompt, but that's about it. It appears a mere coincidence that the '1' key actually works, since almost none of the other keys do.

So it has to come apart..

Using the illustrated instructions on Mainbyte, this is easy. And I soon find out that tapping the 'power bar' that runs past all the video RAM chips causes flicker and the changing of the characters. After touching all solder joints with my iron and some fresh solder the video problems are greatly reduced.

Now for the keyboard. Initially I suspect the keyboard scanning chips and resistors. But after checking the signals with an oscilloscope it soon shows that the keyboard contacts are causing the problem.
The TI99/4a has been produced with several different types of keyboards, and mine appears to contain the worst of all: the 'Mitsumi' membrane based keyboard.

When searching for possible fixes I soon found out that most of them indeed don't work any-more and also that everybody more or less gave up on these keyboards. So there's a challenge: find a way to fix or replace it.
After posting this question to the Retrocomputing StackExchage, I followed the advice and  cleaned both sides with 'Chemtronic Pow-R-Wash', a contact cleaner the leaves no residue, and I gently rubbed the contact carbon using an eraser. The keyboard now seems to work great again, not sure how long this will last though.

Meanwhile the video failed again. And after a lot of tinkering the system does not work at all. It looks like the 5V is down, but when I disconnect the PSU the 5 V is correct. So it looks like there's a short circuit somewhere.
What puzzles me are the white bars, which I expected to be full metal bars carrying either power or ground. But that does not match the connections to the RAM chips. It looks like its bot conected to the 5V and the GND pins.
I found the following comment on the 'AtariAge' forum:


 "that the 'weird rails' are actually two metal strips with a paper separator, and actually carry both +5V and GND. By trimming it, you've possibly created a short circuit and/or you've only jumpered one side of it. I'd give that a careful inspection."

Right. I did not expect that. And this means I probably created a short somewhere. Looking further I found it's even worse. The bars may also carry +12V and 5V. The top one shown in this photo is the +5V / 12V bar. It supplies the video RAM chips, 12V to pin 8, 5V to pin 9.

1.With bar.  2.Bar removed.  3.New connections.

 
It the end I just removed the complete bar. Which indeed solved the short circuit, suggesting I was on the right track. So I just wired all 5V and 12V points manually using some solid wire. Which worked. Video is back again and the computer runs as normal.
Unfortunately it failed again after a week, same problem. So the short must be somewhere else. I suspect a capacitor, but finding that will be a lot of work.

FlashRom99

A bare TI99 is nice, but limited. After typing a few simple BASIC lines, but not having any means of saving and loading programs the fun is more or less over. But I discovered the FlashRom99.
The FlashRom99 Module is created by Ralph Benzinger and uses a Secure Digital (SD) Card for storing ROM images. After power up the content of the SD Card is displayed in a menu. When a selection is made the ROM image is read from he SD card by the ATMEGA8515 CPU on the module and copied to the 32KB RAM chip. It's available as fully built cartridge, a complete kit, or just the PCB with a programmed CPU. I chose the last one. After all, building is 90% of the fun.
Assembly was really easy, and it worked immediately.
The .zip file that can be downloaded from the FlashRom site contains a bunch of sample games and programs so you can get started right away.

A set of useful utilities like an editor/assembler, ready for installation on the FlashRom can be found on the TI99 site of Fred Kaal 

 

Useful links for the TI99/4A

The MainByte site has got a lot of really good information.
Stephen Shaw's 'Getting Started' book online.
Ninerpedia. The Wiki for TI-99/4A, TI-99/8, Geneve 9640, and all related hardware and software.
The TI99/4A Forum on AtariAge
A recent article on The Register about the increasing popularity of the TI-99.