Over the last couple of days I've been working on DEC LK201 terminal keyboard.
When I last tried the VT220, about 7 years ago, I found that its keyboard had stopped working sometime between the late 90s and then. The symptom was that the four status LEDs would stay lit and the terminal would say "Keyboard error - 4" and that was that.
Seeing as I've not a lot of other machines which need fixing at the moment I brought it and the terminal out of storage to see if I could fix it.
After removing it from its case I set about checking for life around the microcontroller... well there was some and it was communicating with the terminal as I could see serial data going both ways. This was a good sign.
IMG_4913.jpeg
Doing some Googling suggested that the problem may be a stuck key and one blog post suggested that removing the membrane ribbons from their sockets would allow the self-test to pass. So, I did just that and.... four lit LEDs.... and around 8 volts on a load of the connections!
Looking more closely, the membrane ribbons are held against the pcb by steel hoops which go over the top and have individual leaf springs pushing on the ribbon. On one of these connectors there are two extra pads not used by the membrane and one of those was connected to all four LEDs and was powered indirectly from the 12V part of the board! The steel hoop was, of course shorting all the PCB pads together.
Hoping that I'd not fried something I launched into bending the lugs one the three hoops to disconnect them from the PCB. With those gone I could try the board without the keyboard connected... Diagnostics passed. OK, so the PCB is fine... must be the membrane.
I downloaded the DEC maintenance manual for the LK201 keyboard, which thankfully had the matrix layout.
A quick check of the lines on the matrix showed very many shorted connections. Oh...
Anyway, with a great deal of struggling and swearing I managed to clip the membrane ribbons onto the PCB. They had to be mounted by putting the ribbon in the correct location, with the PCB fully fitted to the keyboard back sheet, and then getting the steel hoop over the top, sliding it back until the lugs on the side can go into the tight holes in the PCB. Then flip the whole keyboard over, holding the hoops so that it doesn't move, and then try to force the thick steel lugs over to hold the hoop in place... and do this with all three ribbons, without them twisting to the side or coming out from under the hoops. Nightmare!
With it back together I could get the oscilloscope on the go. That showed that at least 13 keys were stuck as I could see the individual pulses from the scanning on practically every sense line.
Wondering if a bit of hot air may help the membranes relax and the contacts be pulled apart I put the rework station onto 100C and heated the front and rear of the keyboard, which took a while given the large steel plate. I then continuously pounded and mashed the keys. Eventually one by one the number of stuck lines decreased, until just four were left which stubbornly wouldn't free.
That was where I left it last night, thinking it a lost cause.
This morning I decided that as it was a lost cause anyway I may as well do radical and risky things and just see what I could do.
The way the keyboard works is that there are nylon frames into which the keycaps clip. On the bottom of these are leaf springs pressed upon by the ends of the key cap stalks. The springs have a little point which presses through a rubber sheet onto the three layer membrane underneath. This membrane is very similar in construction to the ST's. The frames are held in place by moulded pegs which go through two holes in the springs and then holes in the rubber sheet, membrane and lastly the steel back plate. The pegs are then melted on the back to hold the sandwich together. The only way to get access to the membrane is to cut these melted bits off to free the frame. However, this means that the original method of holding the keyboard together is no-longer available. Completely unmaintainable.
Seeing as this was at the point of "what the heck!" anyway, I decided to cut away the bits of the keyboard with stuck connections so I could take a look at the membrane underneath.
IMG_4911.jpeg
Lifting the rubber sheet I could see what the problem was. At some point the keyboard must have been laying with something reasonably heavy on top for many years. The conductive paint on the two membrane pads which act as the switch had glued themselves together. The only way they separated was when the paint on one side or the other was pulled away from the plastic on the other pad. Thankfully, the areas involve were small so the key switch still operated.
Having freed the last of the stuck pads the keyboard passed the diagnostics and the terminal showed the welcoming "VT220" display.
Now all I had to do was work out how to reattach the frames.
IMG_4910.jpeg
I decided to snip off a fre of the pegs and drill out holes and then use M2 self tapping screws instead of the melted ends. The other pegs would be used for location and to hold the rest of the springs. M1 screws would have been better but M2 were the smallest I had.
IMG_4914.jpeg
After a few false starts I finally got the keyboard back together and working.
IMG_4912.jpeg
It just left firstly putting it into its case and then 3D printing some clips for keycaps where the clips had broken away when removing them.
IMG_4916.jpeg
Anyway, it's back together now and all fixed.
IMG_4915.jpeg
Phew! I dind't think I'd be able to bring this one back from the brink.