Over the last few weeks~months, I have been repairing a Sinclair ZX81.
This was the first time I had a look at a ZX81. I am actually not 100% sure if there really was no video, but based on the things I did after I finally figured out what I was doing, it’s reasonably likely that I didn’t accidentally “repair” (and break and then repair again) something that wasn’t broken in the first place.
System
There was no video — well, I don’t know if that was due to the TV somehow not being able to tune into the channel it said it was tuned into. Anyway, I took out all the chips from their sockets and cleaned them thoroughly, as that is what fixed a VIC-20 I worked on earlier. Still no luck, so I checked continuity between the sockets’ pins’ solder blobs on the underside of the mainboard and the chip pins, and identified a lot of bad connections. I used a screwdriver to forcefully bend the chips’ pins to have them make contact with the sockets, and in the end I got continuity everywhere and plausible activity on the oscilloscope.
But still no video output, so we took the signal on the ULA’s pin 16 and fed that into the TV’s composite input. Nothing… Or wait, wrong, that’s just a very dark picture, i.e., black on dark-grey.
Keyboard
Anyway, turning the TV’s brightness all the way up I was able to test a few things, and found that the keyboard had some keys that wouldn’t do anything.
The keyboard’s ribbon cable had a broken trace. This was easy to find using a combination of staring and a multimeter in continuity mode — first stare at a schematic to see which lines the keys that aren’t working are connected to, then put the two leads in the same spot and then go up the trace until it doesn’t beep anymore. Then stare some more and you should be able to see that the trace is indeed very slightly broken around that spot.
Note that opening and closing your ZX81 repeatedly will probably take its toll on this ribbon cable. At first I didn’t even know that it was possible to disconnect the ribbon cable without desoldering it. It’s very possible and you should probably do it — you tug at the ribbon cable to pull it out of the connector, not the connector out of the board. Anyway, I only had one broken trace at first, and after a couple times closing and re-opening the ZX81 (don’t do that in the first place if you can avoid it) I had two.
The first (original) one was very close to the connector, so you take a pair of scissors and cut off a a small bit, and it’s like you have a brand-new cable again. It takes some effort to put the ribbon cable back into the connector, but it’s not that hard.
The second broken trace was way higher up, but fortunately for me it was the right-most trace, which is probably easier to “fix” than other traces. Do not use a soldering iron. I did that, and it just melted the cable. It’s impossible to bodge wire this with solder as far as I can tell. What worked for me was conductive foil tape.
Video
A lot of people seem to have had success just adding an emitter-follower between the TV and ULA pin 16, but that didn’t do anything at all. I also tried dropping the voltage using one or multiple diode drops right after the emitter-follower, but that didn’t seem to have much of an effect.
The video signal looked off on the oscilloscope, but to be honest I hadn’t seen a black and white composite signal on an oscilloscope before. I did some digging, and found the problem: there’s no black porch! Fortunately, that applies to many ZX81s, and there are people who have thought about the problem. For example, check out page: http://zx.zigg.net/misc-projects/ (or the accompanying video: https://www.youtube.com/watch?v=1irH3KuGyl0)
I used this person’s 555-based circuit, which wouldn’t immediately work. After some thinking and probing and staring at the oscilloscope, I found that the voltage during the horizontal blank wasn’t low enough to trigger the 555 (it has to be less than one third of the 555’s main input voltage). I added a resistor divider and suddenly had a beautiful signal!! This was the first time I ever saw a ZX81 boot up properly.
Resistor divider:
ULA pin 16 → 270 ohm resistor → 1k resistor → GND;
→ rest of circuit
I downloaded a bunch of software from https://www.zx81.nl/ and converted some to an audio file using tapeutils.jar from http://www.zx81stuff.org.uk/zx81/tapeutils/overview.html, and was able to load 1KCHESS. It loaded all right.
16KB RAM pack
Unfortunately, the 16KB RAM pack didn’t work. Inserting it into the ZX81 would produce a garbled screen. So I reverse-engineered the RAM pack and tested its memory using an Arduino, and was able to identify the faulty IC and replace it. More on that in this post: Testing a ZX81 RAM pack with an Arduino