Buckling spring keyboards. Yes. Rubber domes? Hell no!
So, recently[ish] I bought a buckling spring keyboard from Unicomp. These guys make brand new buckling spring keyboards with USB. If you want a no-fuss B/S USB keyboard, they are the place to go. I did and so that's where I went. I decided to go for the PC 122. The decision was based mostly on how cool it looked. I don't mind the slight impracticalities of it if it looks cool. It has at least 22 buttons that never get used, that's cool.
Anyway... Shortly after buying it, I found that some keys were a little dodgy. I contacted Unicomp about my worries, I said that I had a faulty keyboard. It had to be shipped in to the UK from the USA so I didn't want to have to send it back. They replied first by suggesting removing the faulty keys from the keyboard, wiggling the spring and pivot plate around a bit and then putting the keys back in. I did this and nothing happened, still intermittent trouble with some keys. They were letters too, I sent all my messages to them with these letters missing to prove the point. So their next reply said something along the lines of "Do you have a socket that will fit the hex screws on the back of the keyboard?" I knew what was coming next. I replied 'yes' naturally and then they said words to the effect of "Open your keyboard, tighten the screw near the main circuit board until the keys work again and then reassemble the keyboard."... Wow. This should have happened in the factory!?
This solved the problem for a while, I was happy with the service, albeit unusual. A little while later, I found the same issue. They had come back. I told Unicomp and they said they would send me a new one. Great! Until the new one does the same. I have kept the new one going for a while, It types just fine, the keys don't go dodgy any more. The downside is that it only works with a very specific amount of pressure holding the PCB in place. Having the case screwed shut applies too much pressure and so I have to suffer with a loose case. This is fine. It sits on my desk and so a loose case is not a big problem.
So now I had a working but not ideal loose keyboard and a faulty keyboard. Unicomp didn't want the old one back. I don't blame them. As a hacker, naturally the next step for me was to try and fix the bad one to get two good keyboards for the price of one. The warranty on the first one was no longer valid and so the keyboard was officially mine. Anything that is mine and is broken will be fixed (or at least I'll try to fix it). :D
What's wrong with their keyboards. Why are they not reliable?
The first step was some more thorough disassembly, let's look at the construction from top to bottom...
The construction of the keyboard is, naturally, unusual. (That is calling rubber domes usual. They are common, I suppose this counts.) At the centre of the keyboard is a huge moulded plastic panel with "chimneys" (as Unicomp calls them) where the keys are attached. The springs lie within the shafts of the keys inside these chimneys. On the key side of the spring (on the bottom of the keys) is a sort of cone shape that keeps the spring in one place. On the bottom of the spring is the pivot plate. This is the small flap of plastic that makes the sudden movement when the spring buckles. This hits the back of the membrane, this changes the capacitance between the front and back membranes. The pivot plates are held in by grooves in the large plastic panel. There is actually a thin rubber sheet between the pivot plates and the key matrix membrane. I suppose this is simply to provide better mechanical contact between the bottom of the pivot plate and the membrane while dampening the sound slightly. The sound is of course part of the appeal (for me anyway). And now we get to the bottom of the keyboard. This is the place where I started when pulling the keyboard apart. This is also the place where the factory stopped when making it. (You guessed.) The bottom is one large metal sheet that is curved to match the profile of the large plastic panel.
Having pulled it apart, I found out exactly what was wrong. The metal plate on the back of the keyboard has lots of holes in it. The plastic panel had posts on it that went through these holes. In the factory, to join these two together, the metal plate is placed over the posts and then the posts are melted into pads that sit on the back of the metal panel. I HATE THESE THINGS! I don't know what this method is technically called but I'd call it the "Hacker's Bane" method. To tear the thing down and get inside, I had to remove these melted posts. That was easy. It was easy because most of them fell off. When I first opened the keyboard, several of them had already fallen off. This explained the faint rattle as the keyboard moved around. I rubbed my hand against the back of the metal panel and most of these melted posts fell. a few of them held on long enough to need to get the stanley knife out. The machine that melted them went too far! The layers of the keyboard were coming apart and so the keys couldn't properly reach the membrane matrix. It didn't help that the electronics was dodgy too. The angle of the PCB in its flimsy holder was awkward to get right. This is part of the reason Unicomp asked me to tighten that screw.
Here are the melted posts on the unmodified keyboard; not as bad...
So, poor QC after the melting part of the manufacturing process. Considering that a keyboard is mostly molded plastic, you'd think that any company making keyboards would be able to get this right. Perhaps even have a plastics expert in charge of things like this. what made it worse was the fact that the large plastic panel had a slightly shallower curve to the metal plate. One of them must have been made incorrectly. The plastic has to bow significantly to reach the metal plate.
And now, the fixing. Or perhaps the reinventing...
The posts came away easily with a small drill and a rotary sanding bit. This left smooth flat sections where the posts were.
[Knee in shot]
Near the centre of the image is an example of a filed post. The pivot plates sit between pairs of 'T' shaped sections. The springs go through the large holes.
The first thing I though of doing was drilling fairly large holes for bolts to go through. The trouble with this was space. The board has few places to put bolts that won't restrict the motion of a key being pressed. There is also little space under the board where the nuts won't get in the way.
So, on to plan B. Drill some small holes where the posts were and use some really tiny self tapping screws to hold the back on. I didn't have any really small screws lying around in ample quantities and so it was off to google... a short time later, I came across this fantastic website that sells good value small fixings for hobby purposes: http://www.modelfixings.co.uk/
Modelfixings.co.uk looks like it hasn't changed since the 90s. The website is perfectly functional however, knowing my luck they'll update the site the moment I publish this. So I went for some super small 'Panel' screws and some washer to suit.
When they arrived I found I didn't have a small enough screwdriver. These things take 0.9mm (AF) hex drivers. 0.9mm!!! Crazy small!!! I went to maplin to get a set of small hex keys that went down to 0.7mm. These were useless. After about two uses, the hexagons had become circles. Not great. To be fair, the guy that sold me them told me I'd have to bring them back. When I did bring them back the same guy was perfectly happy to refund them but couldn't resist an "I told you so" (jokingly). I bought a wera driver on ebay the same day for much less and it worked so much better. Most importantly, it is still a hexagon. Anyway... The washers were so small that they almost fell through the holes in the metal sheet for the back of the keyboard. I used some larger washers I had to add to the size. Each screw therefore needed two washers.
On the right in the image above is one of the screws in one of the M1.6 washers I bought with them. The left shows the larger ones I had to add. I think it is M2.
To get these in to the plastic, I needed to drill some holes, introducing the trusted mini drill...
The drill bit used for the final holes was 1.2mm. The screws have a 1.5mm thread diameter. This didn't leave much for grip especially when the drill tilted slightly and widened the hole. Other than a few minor problems, the drilling went well. It was time to reassemble.
The first thing was to put the pivot plate and spring assemblies back in. This is a simple job but a little awkward because the springs are slightly taller than the chimneys and so you have to have the whole thing off the desk slightly before the plates rest in their proper place.
Here's the box of springs and pivot plates from the disassembly;
This is them resting in place;
One is missing. Just for the photo.
After this was the non-photo worthy process of replacing the rubber padding and the membrane. With this done, I put some screwdrivers under the keyboard as I pushed the metal sheet on to the back. The screwdrivers pushed the centre of the plastic towards the metal so that they met perfectly flush. This had to be pushed down fairly hard as I put the screws in one by one. Most of the screws went in feeling fairly satisfying, like they were contributing to holding the keyboard together. Others simple stripped the plastic out right away. By the end I had enough of them in and this was the result:
I put a few keys in, connected the electronics, plugged it into a computer and gave it a go.
First results looked promising. The keys all worked. I put the rest in and gave it a proper test.
This is where it started to unravel. Some of the keys didn't slide up and down properly. Some of them didn't work despite a satisfying clicking sound. Some of them would only work when you pushed them really hard after their springs have already buckled. Hmm. After a little fiddling I got it down to just a single key that was dodgy. It was the letter 'P'. you had to press really hard but it did work. If you have ever typed on a keyboard where a key doesn't work then you'll know that it is one of the most frustrating things you can do. As this was the only key that wasn't working, I thought it reasonable to try a quick fix instead of pulling it apart to find the problem. I don't know why it was 'P' that drew the short straw but it was fairly straightforward hammeing the metal plate just behind the P until the plate had pushed the membrane close enough to the pivot plate for P to work fine again. HOORAY!
As you can see, this left some ugly marks. This isn't seen so that's fine. So then I tried all the other keys near it to make sure they weren't affected and found that... well... the minus sign or underscore key no longer worked. Also the 'O' key wasn't great. One step forward, two steps back. The rebuild was mostly okay until this point. I still have the keyboard I am using now and so thought I'd just leave the project there for now. Something to come back to perhaps.
I'd like to try cutting the rubber backing into lots of small squares so that it is more free to move for each individual key. This may help the bad keys come good. This requires removing the metal back again however. I need much more energy and enthusiasm before I feel like going through putting all those screws in again. They are so fiddly and awkward, it's unbelievable.
Overall. Not a success. Oh well.