deskthority

I recently discovered this site and the world of vintage keyboards, and I have a question:

Looking over all these old keyboards, it's apparent that many of them use rather novel and complicated ways of registering a keypress. Unlike the seemingly obvious use of a 'normal' switch with conductive contacts that just closes a circuit (like a 'standard' Cherry/Alps switch, ordinary buckling spring, or even cheap membrane switch), you have these designs that do crazy stuff like measure capacitance, or the Hall effect with magnets, and I even think I saw one that used lights and optical sensors under each key...

It seems like massive engineering overkill. What was the motivation behind all these designs? It's not like early computer keyboard vendors were obsessed keyswitch-feel aficionados, so I doubt it was in pursuit of "a feeling of oneness", and basic switch design must have been quite mature by the 60's and onwards...

Can someone enlighten me?

You have much to learn, young padawan. Let's put it that way.

Rather than write a lengthy screed right now (which I may do later) I'll just point out that discrete switches are the niche nowadays. Cost cutting killed the mainstream keyboard. Time was, every computer company put some talent and pride in their keyboards, but generations of bean counters have worked on destroying that legacy. There's still plenty of engineers today who'd love to make their own key switch (some are members here, working on private experiments) but the market didn't evolve past them, it just degraded. Why over serve people when they don't even notice their keyboards? Cheapskates get what they deserve…

Capacitative sensing and Hall effect are both about more than key feel. The idea behind them was to isolate the switching mechanism from the wear and tear of movement. That way they could work reliably for years. And indeed the few of them to stick around still do.

Cool, thanks. I look forward to a lengthier screed should you find the strength to write one

Reliability was something that occurred to me but I assumed that plain old contact switches could be durable enough. I guess computers were not the cheap disposable items they are now and so things were built to last.

As for assuming that keyboard feel wasn't a big priority back then, this may have been colored by the (many?) examples of lackluster feeling mechanicals from the 80's (like Apple's early compact Mac keyboards, which used mechanical switches but had a pretty crummy feel). This made me think that keyboards used mechanical switches just because that's what was available, not because vendors generally cared too much about the subtleties of the feel. Also, the existence of things like this: http://en.wikipedia.org/wiki/File:ZX80.jpg

Things were already going pear shaped in the 80s. 60s (!) and especially 70s keyboards are generally the great ones. And although I'm into Apple hardware otherwise, keyboards were never something they mastered. The Apple Extended II is nice (and I'd love a smaller, modern version of its layout) but everything since then, urgh…

robo wrote:Also, the existence of things like this: http://en.wikipedia.org/wiki/File:ZX80.jpg

Ooh, that illustrates Murium's point perfectly:

Wikipedia: ZX80 page wrote:It is notable for being the first computer (unless one counts the MK14) available in the United Kingdom for less than a hundred pounds.

To repeat:

Muirium wrote:Cheapskates get what they deserve…

Also, a lot of the complex switching mechanisms back in the day were due to the nature of conductive switches bouncing. The complex mechanisms were far less susceptible to bounce.

Once ICs got cheap enough in the 1970s, the debouncing could be practically done in the matrix scanning hardware, and in the late 1970s, early 1980s, microcontrollers got cheap enough, and debouncing was done in software. Then, key bounce became a non-issue, and keyswitches could be made much more cheaply.

Great point.

Even our own controllers handle debounce these days. (Debounce is when you clean the dirty signal you get from switches, where they'll shake between on/off states right around the moment of activation. Get it wrong and you'll miss key presses or get reppppeating keys!) In fact, if you check that link to Soarer's Controller, he's written some good documentation about what he's up to. Not only switches have dirty states, but so does the entire diode matrix. When you probe it, you have to wait a tiny length of time while it stabilises, before you read the switches' state from it.

A lot of what's changed over the years has been plain simple cost cutting. The caps are a great place to look for that. You can tell a classic keyboard by those alone! Doubleshots were once the standard, but have since become niché and indeed élite. But not everything was penny pinching alone. Some things, especially the move to laptops, are simply progress. As soon as mobility became feasible, keyboards have become ever lighter and ever thinner, neither of which really suits key switches. Once most people are on touch screens for life, perhaps we'll wind up in the ironic situation that the only keyboard buyers best are high end, and the mechanicals will triumph! Certainly, the traditional desktop's demise means doom and gloom for crappy desktop keyboards.

Cool! That's great info. Bounce had never occurred to me. Out of curiosity (and rudimentary electronics knowledge) could this not have been solved with small capacitors to smooth out the voltage change? Or is that too slow or cumbersome?

Capacitance can certainly be used, although I think it'd act a bit odd on a matrix, so you might end up needing a cap on each key?

The way the AY-5-3600 did it was a hardware delay that was timed by an RC circuit, and when it detected a key press, it started the delay before checking the key.

A matrix isn't a steady state. Switches are arranged that way to reduce connections to the controller. The way keys are read is by polling: you light a strobe (which can be a row of keys for instance) and read the senses (columns in this case). Then you're off to the next strobe, reusing those senses. You pulse waves of voltage across the matrix, even when no keys are pressed.

In the '70s and '80s there was a lot of talk that computers would replace typewriters. There were even dedicated word-processing machines that ran nothing else than the built-in program. The people who used typewriters as well as the new machines then were often professional typists and designers wanted to make the transition as smooth as possible.
Therefore, many key switches of that time were designed to mimic how typewriters felt, worked .. and even sounded. Many of these switches were also used in electronic typewriters.
If you look up the history of IBM's Buckling Spring, and its predecessor the IBM Beam Spring, you can find that mimicking IBM's best-selling typewriter was stated as an explicit goal.

I believe that the mimicry of typewriter keys is also the main reason why all major mechanical key switches actuate in or near the middle of the key travel. Even the very cheapest most simple key switches did.
Some rubber dome keys even have an inner dome which buckles after the key has been pressed, but it is often very small so it tends to just make the landing mushier...