I went ahead and worked on a TRS-80 Color Computer 1, the same model I used as a kid, and this one needed a few things right away. It powered up and displayed video, but the keyboard had dead keys, the case had sticky residue, one foot was missing, and I wanted to see whether a composite video mod would make it more usable on an LCD. On the bench I compared the RF output on both a CRT and an LCD, and the CRT already looked better, with sharper text and a more solid image.
Before opening it up, I checked the basics and confirmed the machine was running. This was a 26-3003, the 32K model, and the warranty seal screw in the middle of the case had never been broken. Inside, the board was clean and the machine still had the original shielding arrangement and internal power supply layout. I also found the usual handwritten markings inside the case and on the shield.
The motherboard layout was a nice reminder of how serviceable these early machines were. The chips were socketed, and I walked through the major parts including the Motorola MC6809E CPU, the MC6883 SAM, the ROMs, the PIAs, the DAC, the VDG, and the RF section. The RAM setup was one of the more unusual details: the board physically had 64K worth of chips installed even though the machine was sold as a 32K system. The explanation was the common Coco 1 arrangement where only one 32K bank was used.
The keyboard turned out to be a rubber dome design with separate little metal contacts under each key. Once I got the keyboard apart, the main issue was dirt and corrosion on the contact board rather than broken parts. I cleaned the board with IPA and was careful with the loose metal contacts since they came out of place very easily. Later, a little DeoxIT on the contacts helped with extra key presses I had seen during testing.
For the video mod, I built a composite circuit on stripboard based on an older Coco 1 mod design and a later revision of that design. I laid it out in KiCad first so I could map the strip cuts and jumper wires cleanly, then wired it into the machine using audio and video points at the RF modulator. The first test failed because I had the RF modulator pins reversed, but once I corrected that I got a working picture out of the composite board.
The problem was that the result still wasn’t good enough to keep. I tested the new composite output against the original RF output on both an LCD and a CRT, and the RF image stayed sharper and cleaner. I also tried a simpler direct tap into the signals feeding the RF modulator, and that produced a picture too, but it still didn’t beat the stock RF output.
At that point I went ahead and recapped most of the motherboard to see if dried capacitors were affecting the video quality. Getting the board out was more work than expected because of the foil-covered cardboard shield and the way everything was fastened together. I replaced the electrolytics I had on hand, left a few larger or special-value caps for later, and retested the machine. The recap made no visible difference to the composite quality.
The keyboard worked much better after cleaning, but the composite mod came back out and the machine stayed on RF output. While cleaning the case, IPA reacted with the finish and marked the top shell, so the Coco still needed a replacement foot, a few remaining capacitors, and possibly a repaint later.
— Aaron
Links
- “W.A.V.E.” by Emil Rottmayer
- “Rise And Shine” and “Fakebit World” by Malmen
- Dragon 32 image – Rept0n1x / CC BY-SA
Timestamps
00:00 – Intro
01:24 – Initial Testing
04:48 – Brief History
06:54 – Inside The Case
09:45 – Layout Of The Motherboard
15:17 – 64K or 32K
17:57 – Repairing The Keyboard
21:20 – Composite Mod Design
26:05 – Installing And Testing
33:33 – Re-capping The Motherboard
39:00 – Cleaning
42:25 – Final Thoughts And Wrap Up