Which component is the most important part of a keyboard?

[Menuhin]

Many people 'swear by' something, e.g. IBM capacitive buckling springs, Alps SKCM blues, Nixies, etc.

I don't have enough experience to 'swear by' anything about a keyboard, but I can tell when I definitely cannot stand a keyboard for a certain property:

• I won't use a keyboard in the first place when its layout is for sure not sufficient for my usage, e.g. some important keys are lacking.
• I may not want to touch a keyboard when the caps look oily with food grease, hair, dead skin, etc otherwise, all caps and all profiles are good to use for a while.
• I will stop using a keyboard when its switches irritate me to an unbearable degree, e.g. "@#%&* so gritty, some sands are inside?!" or "So loud... Sshh... it's 3am".
• I actually care about the case least, until a keyboard flexes or when the clicks echo like there is a valley inside, or when I pick it up (rarely) to take a good look at it as an object instead of a device.

I can imagine I'm happy as a compromise to use an IBM ThinkPad USB/Wireless keyboard for its functionality, does it mean that I put the 'PCB' first when other things are kind of acceptable for me?

I also see some component can be more of a 'constant' and some keep on changing. A set of nice caps 30 years ago are still a set of nice caps, while some may not care about those non-reusable vintage switches or toss away the PCB (and therefore the case) of vintage boards. It is rarer to reuse vintage cases and plates, and I have yet to see someone reuse (but not revive) a vintage PCB.

Buy the classic caps and then a sandblaster (plus an ultrasound cleaner and an ultraviolet setup, or just collect PBT and POM caps and I'll only need the cleaner), then in 30 years, I will still be happy about the look of my keyboard.

[mike52787]

I believe all parts are important. Switches are important, but only combined with the right caps. A board with amazing switches and caps can be let down by a shittily built case, or a bad layout, but many will choose a modern layout with inferior switches over a odd layout with superior switches (see model M vs F). It has to be a combination of all factors to do it for me. There really is no "most important" thing without the others.

[Menuhin]

What make us give up or harvest from vintage keyboards is all because of the 'PCB' part, if I can combine layout/missing keys and protocol compatibility in one 'PCB' factor.

Let's say in a parallel universe where keyboards 40+ years ago use some magical 'modular' PCB structure capable of being rearranged to any other layout, and that they used a universal protocol and all backward compatible. And if there's no need to have planned obsolescence kind of consumerism to run the world, people perhaps will be using elegantly carefully built keyboards from their grandparents' era until each keyboard is not repairable.

It's just looking from sort of a keyboard practicality point of view.

[zslane]

If I were to take the factors that matter to me and put them into priority order, the top item would be switch mount type. If a keyboard doesn't support MX-stemmed keycaps, it is absolutely useless to me, no matter what its other virtues are.

The complete list, in priority order would be:

1. Stem type (must be MX)
2. Layout (must be ANSI-104/108 or ANSI-61 with Filco-style bottom row)
3. Switch type (must be silenced Topre or MX silent red...but willing to try Hall Effect)
4. Case style (conventional plastic or aluminum top-cover-over-bottom-shell preferred)
5. Case color (white or black depending on keycap set to be installed)

Everything else is negotiable, but any keyboard that does not hit all five of the above points is of no use to me.

[scottc]

As a direct response to your 4 requirements, OP:

#1: If I can remap the keys using an external converter, flashing firmware, etc then a sub-optimal layout can befine. I do this all the time (eg. the IBM Model F AT is perfectly usable with heavy remapping and changing the keys to have a more ANSI-like layout).
#2: Dirty caps can be cleaned! I've seen keyboards as disgusting as you can imagine. With a lot of love and affection, even an ugly keyboard can be beautiful again.
#3: Sometimes grittiness can be fixed by lube or cleaning. If not, I agree with you -- keyfeel is very important.
#4: I agree about the case -- it usually doesn't bother me. Sometimes a case flexing and not being rigid might even help keyfeel - for example, I love the PCB mounted MX clears in my G80-1800 with a flimsy PCB and case and no plate, but I don't think I'd like them as much in a plate-mounted keyboard.

[seebart]

Very good question Menuhin! Problem is all of those parts are almost equally important, with the PCB (layout and LED support, NKRO, firmware etc) possibly playing a slightly lesser role in the direct experience but that's relative. So I cannot answer with one option.

[vvp]

The overall keyboard shape must be ergonomic enough. Preferably split for easier angle/distance adjustment. Everything else does not really matter provided it works at all (e.g. the keys do not stay struck after hitting them etc).
Yeah, and firmware should support remapping.

[emdude]

My answer is basically the same as some of the previous ones, that all parts are important. I do place great value in cases, at least as far as vintage keyboards are concerned. A keyboard's appearance matters a lot to me and so yellowing is a big blegh. I also apply this to ABS key caps.

[Chyros]

If I had to pick one part it'd be the switches.

[ideus]

The most important part of a keyboard is its usability. It implies more than one of the items in your list. It may include most of the items, because usability is the effect of the interaction of most of the keyboard components. Instead of choose one, only; the deal is to prioritize them:

1. Layout.
2. Switches.
3. Key caps.
4. Case.
5. Plate.
6. PCB - However this rules the layout.
7. LEDs.

[Matt_]

I can adjust to another layout, to different switch types, to PCB feature set (programmability, backlight), however if I have to type on crappy keycaps I know I'm going to feel miserable.

So I'd put keycaps in 1. Thick PBT or ABS in Cherry profile or bust. The rest I can accommodate, but I'd put case construction & layout in 2, switch type in 3.

[Mr.Nobody]

I'd better keey my own councel on this topic. If you ask me which part is lest important my answer might be "rubber feets".

[Phenix]

I can adjust to another layout, to different switch types, to PCB feature set (programmability, backlight), however if I have to type on crappy keycaps I know I'm going to feel miserable.

So I'd put keycaps in 1. Thick PBT or ABS in Cherry profile or bust. The rest I can accommodate, but I'd put case construction & layout in 2, switch type in 3.

I would agree to Chyros: switches are the most important, followed by layout+programmability

I mainly mind if the keycaps are good readable, as I like both ABS and PBT (tough Im touch typing)

After case/plate/feet/cable/LEDs

[Chyros]

Interestingly, the results so far aren't completely in line with my experience of other people's priorities. Based on what I read on DT, GH and in the comments to my videos, the most important are, in this order:

1) layout
2) switches
3) keycaps

[seebart]

Interestingly, the results so far aren't completely in line with my experience of other people's priorities. Based on what I read on DT, GH and in the comments to my videos, the most important are, in this order:

1) layout
2) switches
3) keycaps

This question and hence the poll is really good since all these factors are so "interwoven". In fact I think I many even say that my answer may depend on the keyboard in question.

[matt3o]

I believe you grouped too much into "PCB". I don't give a damn about LEDs, but layout it's #1 priority.

[Laser]

It will *have* to be the switches, for me: no other component, when it's BAD, can have a worse effect for me than a keyboard with bad switches. The layout can be worked "around" by changing what the keys do, even in software, if not possible otherwise.

[Daniel Beardsmore]

I believe you grouped too much into "PCB". I don't give a damn about LEDs, but layout it's #1 priority.

I like LEDs, but otherwise that was my own thoughts. Layout matters the most, as you can't touch type if people move the keys around! (Looking at you, evildoers who swap Fn and Ctrl, or the stupid keyboard my boss has, with "F" (Fn) in place of the left Windows key, ISO backslash and ANSI enter … crackheads …)

[Harshmallow]

The 'Any Key'. If my board doesn't have an Any key, then it's no keyboard for me!

[THATGUY69]

Switches are very important to me but they come close second to the layout of the keyboard. I use a model M at the moment and its a great board, but it doesn't feel nearly as good as my model F XT but I cant use the XT on the daily since that damn layout just drives me insane. The smaller shift, back and enter keys drive me insane and lead to many typing mistakes. I'm currently looking for either a model F AT or model F 122 to maybe fix the problem of layout since they both have an almost normal modern layout compared to XT .

[Menuhin]

I believe you grouped too much into "PCB". I don't give a damn about LEDs, but layout it's #1 priority.

That is true - I actually wanted to separate 'Layout', 'Firmware/NKRO', and 'Protocol compatibility/USB/PS2/AT/XT' into 3 items. However, for the modern keyboards and it seems also in the future development direction, the component on a keyboard that determines all these aspects is really just the PCB.

PCB determines the plate arrangement, and the case too. That is the case and the plate are pretty much built for the PCB, or is it the reverse that the PCB is tailor made for a certain layout, case and plate design? Does the later happen more in Korean keyboards? I believe the ErgoDox followed the later development process, which is more holistic and reasonable approach.

We might have "grouped too much into the 'PCB'". Can alternative design separate these three into modules, so that changing or updating one of the three does not affect the others?

[vvp]

... or is it the reverse that the PCB is tailor made for a certain layout, case and plate design?

Definitely this. Imagine e.g. Kinesis Advantage. I doubt one can claim that the case was made for PCB. Even for standard keyboards. First thing decided is the case shape, design, aesthetics and PCB is made so that it fits in. As I look at it, PCB is irrelevant. Layout, overall keyboard shape (split / non-split), and firmware features (mouse keys, remap, macro, layers) are important. But these things have hardly anything to do with PCB. PCB is just that shit inside so that you do not need to hand-wire it.

[davkol]

PCB is a poor expression indeed, because the circuit can be printed on a membrane, or just wired together. And if you have, say, a touch screen, then there's the screen with a digitizer and a controller. Or consider a projected keyboard...

I still voted that way (as one of the first voters), though, because I care about the controller and layout (which is related to matrix geometry).
Keys (switches&caps) don't really matter, as long as they work. For example, today I've used only a high-quality but worn-out rubber dome with completely glossy keycaps (due to wear) and a generic gaming keyboard with modern MX Black switches; it could have been better—silent, smoother, and non-sticky with sweaty fingertips—but it worked okay. OTOH, I have many keyboards with brilliant keys, but never use them, due layout/keymap annoyances.
A case doesn't matter at all. Some of my keyboards don't even have one; I just use the PCB/plate on a mat.

[vvp]

I guess people care about controller only in the sense what firmware they can load on it.

[davkol]

Even if it's a blackbox, its interfaces, memory, architecture support or flashing mechanism matter.

[y11971alex]

Beamsprings are switches of epic proportions and phenomenal design calibre, but the quality of the case is also part of their charm. For me, switches come first, and case second, and buttons third. This is mostly because the case can influence the performance of the switch; if beamsprings were PCB-mounted in a flimsy plastic case instead of being clamped on a metal plates, they might perform differently, in a less satisfactory manner.

[matt3o]

not sure I follow, you guys would prefer to type on the most uncomfortable keyboard on the planet if it has the switches you like?

[Laser]

Do you prefer to use a 5$ membrane keyboard, only because it's in the layout you want? I can always configure the layout in software, if I don't like it. I think it's rather we all start with different assumptions about what a 'worst switch' is, and what a 'worst layout' is, and then the apple-pear comparison is skewed in one direction or another. In any case, I aim to solve all of them (switch type, layout, firmware, PCB and case quality etc.).

[matt3o]

Do you prefer to use a 5$ dollar membrane keyboard, only because it's in the layout you want?

by layout I mean keyboard shape not just QWERTY vs QWERTZ or ANSI vs ISO. An ergodox lover would always pick that over a standard keyboard even if the switches are cherry brown instead of clear.

[Laser]

That's what I'm saying, if keeping the switch type in one area (brown vs clear), but varying the layout greatly (ergodox vs qwerty), one is tempted to highlight the importance of the layout. By keeping the layout type in one area (qwerty vs qwertz) and varying the switch type greatly (cheap membrane vs. beamspring), others would favor the switch type.