Possible improvement: would this work?

[FoxWolf1]

So I was thinking a little bit earlier today about mechanical switches, their advantages, and their shortcomings...

Now, as we all know, typical mechanical switches differ from their membrane counterparts in that there is a significant amount of travel between the activation point and the maximum travel of the switch. In general, this gap is A Good Thing, improving both comfort and responsiveness compared to domes of the same travel. It does, however, lead to certain weaknesses in gaming situations, where keys are often held down at full travel:

Firstly, because the switch must travel upwards a distance before passing the reset point, there is a loss of precision for actions that are done by ceasing to hold a key. The extent of this weakness varies considerably by switch type; on MX Blacks, the heavy spring and linear action reduce the severity of the issue, while green XM switches have the problem to quite a pronounced degree. Even with MX Black switches, however, the distance between full travel and reset exists, and therefore there is theoretical room for improvement-- and even an improvement that is hard for lower users to notice is better than no improvement at all. On some, but not all switches, it is theoretically possible for the problem to be reduced if users refrain from holding the key all the way down, but requiring such behavior in high-energy gaming settings would be to place more burden than necessary on the user.

Secondly, because the switch must reset before it can be re-pressed, very fast multiple presses will be read as holding the key down, which may not be desired. The speed at which the key can lift from full travel to reset imposes a limit on keypress rate in situations where concerns of speed, excitement, or imperfect user coordination make it impractical for the user to refrain from applying force in excess of that required to activate the switch; furthermore, with some switches, a difference between activation and reset point might result in a limitation on repeat rate even where no excessive force is applied.

It seems to me, though, that these issues are avoidable, and without even having to design a new switch to solve them. Basically, all you'd have to do is detect when a key is bottomed out; then, when it stops being bottomed out, even if the switch itself is still active, have it send that the key has been raised. If bottoming out again is detected while the switch itself has not been released, have it send that the key has been pressed again. As for how, exactly, you detect bottoming out, it seems that there are plenty of possibilities even without resorting to modifying the switches. For example, conductive O-rings plus conductive tape running up either side of the switch housing (and ultimately connected back to the PCB) would have bottoming out close a circuit. I'm sure it'd be possible to come up with other methods, too.

Hopefully, the potential of such a system is obvious...for example, you could have a keyboard using a light, tactile, and clicky switch (MX Blue, for instance) but still have precision and repeat-press-rate similar or superior to heavy linear switches.

Thoughts?

[Binge]

The actuation distance of a mech switch is 2mm, the return is 2mm, and for the rubber domes a similar issue exists.  The mech has lower travel times overall, and a switch can travel that 2mm distance faster than the polling rate of the keyboard controller..  The user can just compensate when double tapping to make sure they are hovering around the actuation zone.  With blue switches there is no issue with the key not working for double taps, but there is no tactile feedback when you hover over the actuation point and so it is harder to notice than say MX browns and it is also hard because the user will be used to tactile feedback.

Even the most intense and time-sensitive e-sports like fighting games and music games use mechanical switches and the #1 in the world swear by these input methods.  If they can do sub-2ms actions then obviously there is no issue with the use of mech switches.

[fohat.digs]

Get some really fat O-rings and your travel will be cut way down.

[FoxWolf1]

Even the most intense and time-sensitive e-sports like fighting games and music games use mechanical switches and the #1 in the world swear by these input methods.  If they can do sub-2ms actions then obviously there is no issue with the use of mech switches.

Just because something is the best method that exists currently doesn't mean it's the best that it could possibly be; and as I said, even a small improvement is better than no improvement. Or, to put it a bit differently, "good enough" is a far cry from "no theoretical room for improvement remains."

For sure, overall, mechanical switches are far better than domes, but that doesn't mean that a hybrid couldn't be even better still. And remember that the area of improvement that I am aiming for has to do with performance from bottomed out, not when oh-so-carefully or -skillfully hovering around the actuation point. If you want to see what I mean, just do a simple test. Pick a key with a LED, so that you have a visual indication of how many times a key registers distinct presses, and hold it down-- really grip it tightly against the plate, as if you're in the heat of battle. Then, as quickly as you (rather than your keyboard) can, re-press the key. There is a degree of speed at which, because you are no longer lifting the key enough for it to pass over its reset point, even an MX Black will fail to register the re-press, despite the user feeling like he has un-pressed and re-pressed the key (and, on plate-mounted boards, there will be a second audible event of the key re-bottoming out). The system I have described would make it possible to rapidly repeat a key from bottomed out and without requiring particularly careful force control, because it reduces the minimum distance that you have to allow the key to lift before you can press it again.

Get some really fat O-rings and your travel will be cut way down.

But then you lose some of the typing advantages of having mechanical switches. I want a system that can take something like an MX Blue switch and give it the gaming advantages of an MX Black, without sacrificing any of its own advantages as an MX Blue.

[tp4tissue]

Since almost zero games/applications depend on Keyboard precision, there's little incentive for anyone to retool and build better switches.

[TheProfosist]

Someone should make a keybaoard using omeron switchs

[pitashen]

One thing you have to ask yourself is whether the current mechanical keyboard design is encountering any sort of problem in performing certain tasks in practice (not some theoretical imaginations in your head). IF there is, then you have to ask yourself whether the so called improvement for overcoming the "problem" would make any economical sense.

[FoxWolf1]

One thing you have to ask yourself is whether the current mechanical keyboard design is encountering any sort of problem in performing certain tasks in practice (not some theoretical imaginations in your head). IF there is, then you have to ask yourself whether the so called improvement for overcoming the "problem" would make any economical sense.

I don't really see the issue as one of solving a problem, but rather as a case of wanting to try and grasp at every last bit of performance in a struggle to maximize a user's advantage over his competition. And as for the cost issue...how expensive would it really be, if it can be done using existing switches? Plus, enthusiasts are already willing to spend lots of money on very high-end products, and on aesthetic customizations without even a slight performance advantage.

To be honest, I'm surprised at the sort of reactions that people are having. Whatever happened to the idea that being an enthusiast meant enjoying the pursuit of ever-greater performance for its own sake?

[pitashen]

To be honest, I'm surprised at the sort of reactions that people are having. Whatever happened to the idea that being an enthusiast meant enjoying the pursuit of ever-greater performance for its own sake?

It is great to have a spirit like yours, but... welcome to the real world.

Unless you can show/demonstrate people the actual practicalities and the need of maximizing the "performance," it is hard to motivate them to move away from what they already feel comfortable with.  And since we can't see the "obvious," it is up to you to really show us the light (i.e. build one and prove your point) in a way that we would say "that's what I need/want!"

[FoxWolf1]

*shrug* I never said that other people should move away from what they have now, or that there's any "need" for performance increases. I was just wondering if a system like the one that I described would be doable-- and if so, what would be the best way to go about doing it. And surely I can't be the only one who considers the "goal" of having an interest in keyboards to be to obtain the best possible performance?

For me, my interest in keyboards is part of a larger hobby of trying to optimize my peripherals setup in terms of gaming advantage, typing and general use comfort, and durability. I guess that makes me a bit different from many people here-- for instance, I would consider any custom keycap with a non-function-oriented shape to be worth less than nothing on a keyboard, since it is further from functional perfection than one with a normal shape (also, as with any form of "bling" or "swag", such things are inherently distasteful, but that's quite unrelated to the point that I'm trying to make). But surely there's enough diversity that I can't be the only one with this sort of attitude?

[alaricljs]

Doable, sure... almost anything is possible.  I recall some musical project where they used Cherry MX switches and then some funky sensor underneath the switch assembly for pressure sensing.  Could probably use the exact same design with different software to get the effect you're looking for.

[urbanus]

It seems to me, though, that these issues are avoidable, and without even having to design a new switch to solve them. Basically, all you'd have to do is detect when a key is bottomed out; then, when it stops being bottomed out, even if the switch itself is still active, have it send that the key has been raised. If bottoming out again is detected while the switch itself has not been released, have it send that the key has been pressed again. As for how, exactly, you detect bottoming out, it seems that there are plenty of possibilities even without resorting to modifying the switches. For example, conductive O-rings plus conductive tape running up either side of the switch housing (and ultimately connected back to the PCB) would have bottoming out close a circuit. I'm sure it'd be possible to come up with other methods, too.

You're proposing two switches per key -- the existing one at the halfway point, plus another at the bottom?  I think this also suggests changes to the controller chip.  You wouldn't want the keyboard to be sending multiple key down/up events per stroke.

I think what you're proposing is not ideal, but a step towards it (arguably): to be able to detect rapid changes in direction at any point in the key travel.  Rather than having one actuation point, and moving the key down and up past that point, you detect when the user starts moving the key down and that becomes the down event; then when the user starts to release the key that becomes the up event.  That would give the user the freedom to toggle the key any way they please.  I wonder if there's a switch that works this way.

[alaricljs]

Check out the back space switch... I'm told the space bar is the same.  http://deskthority.net/photos-videos-f8/canon-typewriter-keyboards-t4242.html

[FoxWolf1]

Check out the back space switch... I'm told the space bar is the same.  http://deskthority.net/photos-videos-f8/canon-typewriter-keyboards-t4242.html

I like that thread! Guess I'm not the first to think that a double-switch system would be a good thing.

I wonder if that method would be possible for regular-size keys, though...even with a purpose-built housing, it seems quite large. It might also be a bit excessive to have a whole secondary mechanical switch, since the user already gets tactile and audible feedback when he bottoms out.

You're proposing two switches per key -- the existing one at the halfway point, plus another at the bottom?  I think this also suggests changes to the controller chip.  You wouldn't want the keyboard to be sending multiple key down/up events per stroke.

I think what you're proposing is not ideal, but a step towards it (arguably): to be able to detect rapid changes in direction at any point in the key travel.  Rather than having one actuation point, and moving the key down and up past that point, you detect when the user starts moving the key down and that becomes the down event; then when the user starts to release the key that becomes the up event.  That would give the user the freedom to toggle the key any way they please.  I wonder if there's a switch that works this way.

Yeah, pretty much...it's a "double switch" system, and it's not as awesome as a "fully variable" switch-- but might be easier for a custom or low-volume manufacturer to implement, since the primary switch could be something that already exists, while the secondary "switch" could be a fairly minimal affair that makes as much use of existing components/surfaces as possible (like an "o-ring" of some conductive material on the keycap stem being used to complete a circuit upon bottoming out).

[urbanus]

Yeah, pretty much...it's a "double switch" system, and it's not as awesome as a "fully variable" switch

I wonder if a fully variable switch could be implemented with a capacitive switch, and an onboard controller that registers keystrokes by detecting a relative change in capacitance.  Would that be possible?