What makes a good *non linear* switch?

[codemonkeymike]

I have been thinking about this question a lot as I have been getting deep into investigating different clicky and tactile switches and how they work/feel. I have gathered Buckling springs, Beamsprings, Marquadt butterflys, Alps Blues (SKCM), Zealios and others that I cannot think of off the top of my head. The one thing that seems to make a good tactile switch stand out is a sharp drop in force which lets you know that a switch has been actuated as well as keeping a steady rhythm to typing. What do you all think makes a good *non linear* switch.

Non linear as in clicky, tactile, parabolic and so on and so forth were force is not constant before or after actuation.

[Ratfink]

I think the sharp drop is important, but even more important is that it be closely correlated to the actuation. And if there's a click, it had better be exactly at the actuation, or it just comes across as odd and unhelpful.

[Daniel Beardsmore]

I've wondered about this for ages. My feeling is that distinctly tactile switches (tactile or clicky) present an impediment to typing. They offer a much steeper initial force gradient, sometimes almost a force brick wall, which means that keys have to be struck harder. In effect, they're hardly any better than the stiffened rubber domes that they replaced. Linear switches give you much longer to build up the actuation force, but then the force keeps climbing and climbing.

Linear keys that are too heavy tend to feel spongy, and tactility helps to ease this feeling, but too much tactility and you just make the keys unnecessarily heavy.

I feel that it's a series of trade-offs where the drive towards high tactility is actually taking things too far. It's something that has to be carefully balanced. That's why I'm curious about Romer-G — since it's only weakly tactile, it seems like it's a more reasonable compromise, giving keys a bit of extra resistance against accidental presses, and a bit more oomph without being heavy-handed like white Alps or just all around heavy like membrane buckling spring.

The funny thing is that, having bought an MX Red keyboard, I've decided that I like linear after all! I'm curious about MX Black now, too. A whole keyboard of something like Omron B2R or RAFI RS 76 C would be excellent. Ultra smooth linear.

I was using my G84-4400 again earlier. That's got excellent tactility — very sharply defined yet very soft at the same time. You get clear feeling in your fingers without hitting that solid wall as you get with Alps-type switches. Part of the issue is that we just don't have every combination that we ought to — everything comes with a caveat. G84 keyboards have tiny keycaps, cramped spacing, horribly cramped layouts and the switches are prone to binding (though my G84-4400 works really well), but they're also up at the top of best feeling switches for me. There's no reason why you can't have a better version, and maybe the "robust" version (ML1B-11JW) with its taller slider shaft was better at off-angle keystrokes. Trouble is, they're no longer made!

[paecific.jr]

I'm a capacitive buckling spring guy myself so I for one love tactility and clickiness. I believe that understanding where the bump/actuation is going to be is really helpful. After spending time with my keyboard for a number of months I found that I don't bottom out and I don't ever go to the "wall" as Daniel was saying. Even when I'm only pushing one key in a game I know where that actuation point is and I stage my keypress so that I have a shorter travel. So that drop is very important to me. If it could be sharper I would be happier! Maybe if the key almost fell away from you for a second, that would be cool!

[Daniel Beardsmore]

No, the "wall" is the lead-in ramp. Basically, how far can you press the key before you're pushing against strong resistance. With Cherry MX, you can press it some distance, while with Alps and Alps-like switches the key hardly moves before you need to apply a lot of force.

I'm not sure I want to explain this all over again, as it just goes over everyone's heads.

[zslane]

The best non-linear switch is one that is felt, but not heard. A silenced Topre is a good example. Anything clicky is just a royal nuisance that should never be allowed out in public.

[seebart]

I can tell you what makes any *non linear* switch good or even great: the sum of it's characteristics. If all these work well together that can make the switch great for the user. The tactile feeback and the audio feedback foremost. Of course the keyboard as a "housing" for the switches also plays a significant role. You could have the greatest switch in the world but if the rest of the keyboard is just horrible... Now I understand you'd like specific signle characteristics that make more of a difference than the rest? I don't think it really works that way. Great switches are liked by many because they work well in the sum of their "events". Then of course there are variations of the same switch where it get's really wild since the manufacturers got smart as to tweak them just a bit. There are numerous examples but Alps SKCM is a good one. Same basic switch in various tweaked variations.

[Techno Trousers]

My feeling is that distinctly tactile switches (tactile or clicky) present an impediment to typing.

This is not at all true in my experience. To me, as a 90-100 wpm touch typist, the biggest impediment to quick and accurate typing is the "did I, or didn't I catch that key?" feeling and subsequent slow down that comes with rubber domes that require varying force levels, or linear switches not pressed quite far enough down to register.

With tactile feedback from a great switch (my favorite is capacitive buckling spring, second are ergo clears), I could literally be blindfolded and still catch 9 out of 10 of my typing errors from tactile feedback alone. That gives me the confidence to go ahead at full speed and not have to read back what I just typed, searching for missed keystrokes.

So to reply on topic, to me the most important aspect of a great tactile switch is a crisp, consistent feel of key activation. The activation nose of cap BS is a bonus, but not required to have a satisfying typing experience.

[Daniel Beardsmore]

I can't work out if a key registered no matter what tactility I felt or what click I heard. My brain just can't form that association.

I don't mean that tactility itself is an impediment, but rather keys that have to be struck so hard that you feel like typing is a battle against the keyboard. Even good white Alps feels like the keys are your enemy, and Acer switches are evil. However, if the keys really are too devoid of feeling then I do get completely confused and have no idea what I typed — that's why I despise Apple scissor switches. For me, key travel is a more important determinant for key registration — with light weight short travel it feels like you never pressed anything. Apple managed to spec a keyboard where you have hardly any force and hardly any travel, and that's insane. Cherry MX Red doesn't have much feeling but it's still full travel, but I do think that it needs making a bit heavier and able to offer some sort of clue — that's where you end up in the ML direction, but normal ML (ML1A-11JW) was always a problematic design. If you mixed ML's internals with a more stable slider setup, it should be better, but we don't yet know why it's so terribly scratchy.

It's all a matter of balance though -- too many designs fall outside of the optimal balance. Early Matias quiet click isn't too bad, but it could do with being smoother and more stable; I've not typed on the newer grey slider version, so I don't know how this issue stands.

[cumwagondeluxe]

IMO a rubberdome type force curve is the best type of tactile switch, as least as far as the ones I've tried. I don't know what it is, but having all the resistance at the top and a sharp drop after works best for getting my fingers to react and release the switch as soon as possible. Since I bottom out on literally any switch with non-insane weighting (e.g. under 120g), I don't find that the smaller and lower bump on Cherry brown/clear type switches does that much for me. I type a solid ~20 or so WPM faster on my HHKB compared to my linear (Gateron Yellow) Let's Split and I've seen similar results for others as well with tactile vs. non-tactile switches so there does seem to be some advantage in having a physical sensation to let your fingers know they need to come back up.

As far as clicky goes, I haven't really typed on any decent ones long enough to say. MX Blues are very weakly tactile, and I find the high pitched click annoying to type on for too long. Interestingly enough, I also tried buckling springs (and beam springs) which I enjoyed the sound of, but also found to be distinctly non-tactile. Maybe it was because I tried them while standing or something, but they basically felt like linear clicky switches to me if that makes any sense. I'd imagine sound (as long as it lines up exactly with the actuation) isn't a bad way at all to subconsciously signal the fingers to come back up as well, though.


Realistically there's no good consensus though - WPM is only really a useful measurement if you're typing less than e.g. 60wpm and can't afford to lose 10-20wpm or some crazy switch bumps you from 100wpm to 30wpm. Higher WPM is fine for chatting and all, but pretty much anything worth doing on the computer productivity wise with your keyboard (coding, writing, etc) is going to be limited by how fast you can think. People have wildly different preferences as far as weight and tactility go, but I imagine the three most important universal traits you'd want in a switch would be sound, smoothness, and ruggedness (resistance to dirt/grime over time, does it matter if you hit the keys off-center, reliability, etc)

[Daniel Beardsmore]

Buckling spring and beam spring are what I think of as "path divert": linear but with an instantaneous drop in force at actuation.

[Chyros]

A properly designed tactile element that produces a strong, sharp tactile event that is nonlinear in nature. I disagree with Daniel Beardsmore's assessment of "path divert" linearity, as I don't consider those linear or linear-based; I would consider non-divert ones linear. If clicky, using a metal clicker, and for weighting 60-70 gf (less than that doesn't allow for all that much tactility to be implemented unless a seriously weird keyfeel is adopted).

An example of a switch that does all of these well IMO is (of course) blue Alps, which has a nice, sharp tactile event, very nonlinear travel, great weighting, and a very nice-sounding metal clicker. The opposite end is MX brown, which has almost no tactile event, it's not sharp, it's linear, too light to promote tactility, and the noise is pretty crap.

[Menuhin]

I am wondering how the force curve of such an "ideal non linear switch" would look like.
Imaginary force curve can definitely be drawn out.

[Orpheo]

The answer lies in the question in itself. You say "good" or some say "ideal", and those are highly subjective terms defined to tastes, habits and your mood/the weather...

Good quest nonetheless!

I like very much vintage MX clears with GMK QMX clips! I received them from GMK themselves after the recent mass drop.
They are a real pleasure in the tactile department. The course is reduced, as is the noise. It makes for a super mechanical and highly tactile experience me thinks. YMMV

[seebart]

The opposite end is MX brown, which has almost no tactile event, it's not sharp, it's linear, too light to promote tactility, and the noise is pretty crap.

It's the opposite to you. I gurantee you there are people out threre that prefer Cherry MX brown to Alps SKCM blue! I can't say I'm one of them.

[Daniel Beardsmore]

A properly designed tactile element that produces a strong, sharp tactile event that is nonlinear in nature. I disagree with Daniel Beardsmore's assessment of "path divert" linearity, as I don't consider those linear or linear-based; I would consider non-divert ones linear.

A linear switch is one where the force graph is a straight line. This is in effect the case for beam spring and buckling spring too. The lead-in period from zero travel to actuation is linear. At actuation, which is around 2.5–3 mm, the force drops off but the gradient remains the same: the overtravel region is still linear. MBS according to Jacob has the actuation point at 2 mm, but my 1996 Model M feels like actuation occurs around 3 mm.

IBM force graphs: https://plot.ly/~haata/68

"Path divert" refers to a linear graph where the only thing that changes in the y = mx + c plot, is c. At actuation, c is reduced. The gradient m is unchanged and the switch remains linear. It's two back-to-back linear plots with a vertical downwards step in the middle.

I always did feel that my Model M felt essentially linear, and the graph shows this to be the case.

It's not an either-or situation: Alps is not path divert. Alps switches first present a peak, then a trough. This sharp build-up of force risks being so steep that the key fights back against being pressed, which is a very common issue. Blue Alps had the angle and scale of this peak such that it didn't excessively hinder the keystroke.

Alps graphs: https://plot.ly/~haata/66

Cherry MX Blue is different again. The underlying line is true linear in that the overtravel region is the same line as the lead-in line. However, it uses a tactile peak. IBM switches don't have a resistance point where the force increase causes the key to stop. They're linear switches that just "give up". Cherry switches will suddenly stop, requiring a harder push to clear the tactile peak. This results in what I perceive as a less linear feel than my Model M, as you get that sharp rise in force first, and when pressed slowly, a point where the switch will stop until you push harder.

Cherry graphs: https://plot.ly/~haata/65

Tactility comes in a lot of fundamentally different forms, and as time passes we'll come to understand more of them, as they get plotted. The most tactile for me click leaf designs, and Cherry ML — there's no graph for ML yet.

[Chyros]

The opposite end is MX brown, which has almost no tactile event, it's not sharp, it's linear, too light to promote tactility, and the noise is pretty crap.

It's the opposite to you. I gurantee you there are people out threre that prefer Cherry MX brown to Alps SKCM blue! I can't say I'm one of them.

Oh yeah, of course, it's all preference, all of these "what's the best" topics are .

[paecific.jr]

Here's a question for some of you with larger collections, what about switches that are non-linear, non-clicky, and non-tactile?

[elecplus]

Here's a question for some of you with larger collections, what about switches that are non-linear, non-clicky, and non-tactile?

Wet newspaper

[codemonkeymike]

@paecific.jr Parabolic switches or maybe worn out rubber domes probably any switch made with a maglev type action (this is theoretical). But elecplus has a point, it wouldn't feel good at all.

[Chyros]

Here's a question for some of you with larger collections, what about switches that are non-linear, non-clicky, and non-tactile?

Foam and foil switches :p .

[sneaux]

Who can resist a gathering of force curve nerds someday there might be enough of us for a dedicated thread! IMO the combination of light, linear, quiet switches is probably a more sophisticated preference than any other I can think of, in the same sense that a vegetarian (or vegan, etc.) diet is probably a more sophisticated preference than many others. That said, I'm not a vegetarian. I think wanting overly tactile or heavy switches has been most connected to my touch typing skill level at the time. And while my tastes are more moderate as I've improved, tactility of various kinds is just something I'm used to and makes things less boring.

[zslane]

Yeah, I'm a bottom-out typist. I don't use the tactile bump to tell me I've actuated a switch. For me, tactility is simply a nice feeling under my fingers and nothing more.

[seebart]

Here's a question for some of you with larger collections, what about switches that are non-linear, non-clicky, and non-tactile?

Foam and foil switches :p .

Hmm...the BTC foam and foil in my BTC 5100 are so very tactile that they make Topre look linear. To be fair the dome is tactile part in that switch. In my old Key Tronic with the green thick foam the switches are pretty linear but the keytravel is looong. So no I'd have to agree to disagree chyroasan22.

[Daniel Beardsmore]

Here's a question for some of you with larger collections, what about switches that are non-linear, non-clicky, and non-tactile?

I only have one keyboard that fits that description, [wiki]Oriental Tech OK-100 series[/wiki] (OK-100M).

I don't really understand how the spring forces work, but it's progressive rate. It feels interesting (maybe lighter than MX Red), but it's absolute rubbish and totally unusable (stupid key weights, ghosting or transposition, and a key that gets stuck).

Other switches with a non-standard feel are [wiki]Tokai MM9 series[/wiki] (stepped progressive rate) and illuminated RAFI RS 76 M (progressive rate with knee). I have these switch types, but not in a keyboard.

[Chyros]

Here's a question for some of you with larger collections, what about switches that are non-linear, non-clicky, and non-tactile?

Foam and foil switches :p .

Hmm...the BTC foam and foil in my BTC 5100 are so very tactile that they make Topre look linear. To be fair the dome is tactile part in that switch. In my old Key Tronic with the green thick foam the switches are pretty linear but the keytravel is looong. So no I'd have to agree to disagree chyroasan22.

No, I mean "linear" ones obviously, the tactile ones are just tactile. But the linear ones are nonlinearly linear. If you get what I mean xD .

[Daniel Beardsmore]

Progressive rate, I presume. Should be progressive rate with knee as there is a specific point where the foam starts contributing to the force.

[seebart]

No, I mean "linear" ones obviously, the tactile ones are just tactile. But the linear ones are nonlinearly linear. If you get what I mean xD .

Right, the "linear" ones in my old heavy Key Tronic aren't "linear" in the sense I would call them linear but I don't know what to call them. One of those switches where a video might not do much good in conveying it either if you ask me, HaaTa's force gauge might help.

Progressive rate, I presume. Should be progressive rate with knee as there is a specific point where the foam starts contributing to the force.

Right. I have not typed on those for ages but if I recall the progressive rate is "delicate".

IMG_20151018_220103.jpg (958.4 KiB) Viewed 6947 times

[czarek]

And what makes a good non sweet food?
Very personal and everyone will have different view on that.
One will be happy with a tiny bit of tactility provided by Cherry MX Brown, others will like huge bump forcing you to bottom out present in Rubber Domes (and of course Topre), and some will hate that both saying that the best non linear switch is capacitive Buckling Spring with lowish force and super sharp tactility
Myself, I like all those options, but I also like nice linear switches too!

[Daniel Beardsmore]

Food doesn't have quite the same manufacturing limitations.

Statistically speaking, switch popularity is not going to be uniform. The probability of person P liking switch A is not equal to the probability of person P liking switch B. A manufacturer in a small market would want to aim for a design that has a reasonable probability of being appreciated, otherwise sales won't be good.

As you know, there are designs that draw heavy criticism, such as Mitsumi miniature mechanical and Cherry ML.

Even though you can't please everybody all of the time, it is still possible to improve your market share by making a design that people are more likely to appreciate.

Also, imagine you wanted to populate a new office or computer lab with decent keyboards, and you need to order a batch of a single model: what keyboard would you choose?