What I have learnt about my mechanical keyboard preferences

2024-11-21

Anyone who has put some time in learning about mechanical keyboards understands how deep the rabbit hole is. While all the relevant magnitudes are measurable, the complexity of their interactions is such that one person’s optimal setup is another’s bane. Against this backdrop, I think it is futile to search for the one keyboard to rule them all. This quest only guarantees a considerable financial investment in endeavours which, at best, constitute lateral steps on the path towards optimality.

My approach is thus that of the pragmatist: if I can get something that is “good enough”, I can forget about the keyboard and focus on other tasks. I have been doing as much for a long time, having recovered almost completely from Repetitive Strain Injury (RSI). Though I have had my fair share of relatively small experiments along the way.

I am sharing what I have learnt to (i) keep a record of my progress hitherto and (ii) provide a resource for others who are interested in another user’s perspective on the matter.

One-shot keys are my favourite feature

Perhaps the biggest improvement to my setup is the firmware level (QMK) feature known as “one-shot keys”. I use this for the modifiers and some layer activation keys and, as such, call them “one-shot modifiers”, else OSM. With OSM configured, I type a modifier and then within a configurable time window any key press is registered as Mod+Key. I no longer put any pressure on thumbs or pinkies by forcing them to hold down keys for too long over the course of an extended typing session.

Another neat feature is the “modifier lock”. I double tap a modifier (the exact number of taps is configurable) to make it sticky. This means that a double tap on Shift followed by 1, 2, and 3 will produce !@#. A single tap on the locked modifier undoes the effect. This makes it extra nice to input all capital letters or those long key combinations in Emacs like C-x C-k C-p (bonus points if you know what this does without looking it up).

I tried the popular “home-row modifiers” option of QMK but could not make it work in a manner that is consistent with my typing. Whether my technique is not compatible with it or the setup is inherently limited is beside the point. What matters is that I cannot consistently type prose when some key combinations mistrigger a modifier. And I do not like to write prose by going back and forth in editing it—I prefer the continuous writing approach, followed by a final spell check (also why I activate the spell checker on demand).

The keyboard I am using is the Iris by Keebio (thanks to “Andreas” for the gift! (not affiliated with that company, by the way)). It has a thumb cluster where I place the modifiers I use the most. Check the implementation details here: https://github.com/protesilaos/qmk/tree/master/keyboards/keebio/iris/keymaps/prot. I think that with OSM I would be content to even use a traditional keyboard for a while without fear of suffering a relapse in my RSI. This is because I would not need to twist my wrists to press and hold any modifier. Make no mistake though: an ergonomic keyboard is the superior option, ceteris paribus.

Linear switches are my favourites

Taken in a vacuum, I find tactile switches very fun to type on. The feedback they provide is unmistakable: it does not fail to capture my attention. In a real-world scenario though, this quality is a major weakness. When I am typing out a wall of text such as this one, I do not need any extra indication that characters are printed on the screen. I see them and can still hear the keyboard (no mechanical keyboard is completely silent). Furthermore, the tactile bump is considerably higher than the actuation point of the switch, meaning that over the course of a prolonged typing session I will be exerting more effort than necessary for a feature that has no added value to me.

The Iris keyboard I got shipped with the Durock Shrimp switches (Durock Shrimp Silent T1). I picked them because I thought I liked this type of switch more than their linear or clicky counterparts. It turns out that I do not appreciate the added weight to each key press. Consider this: if you are comfortable typing on keys that actuate at 50gf, with tactiles you will be applying force that is closer to 60g. You need to look at the peak tactile point to get a better sense of how heavy the switches are: the actuation point will mislead you. (And if the tactile point is too close to the actuation point you might as well go directly for linears as the bump will be too small.)

When I started out with linear switches, I felt like typing against a pillow. I could not tell by touch alone if my key presses were being registered. There was no feedback other than an abrupt bottom out event. It was disorienting. I did give them a fair chance though, which helped me improve my technique as well become more familiar with the linear feedback in general. It also made me realise how the in vitro perception of a switch is distinct from its in vivo performance. When I was trying the linear feedback without actually doing any work, I was paying too close attention to where exactly the actuation point is. In practice, this is not how it unfolds: I am writing something, the fingers move around fairly quickly, and I am not focused on the feel of each key press.

Not all linears are the same

There is the obvious metric of weight as well as other qualities like sound profile and smoothness, but what I realised recently is that the real deal for me is the delta between the actuation point and the bottom out force. It has a noticeable effect on typing feel while writing. A fairly “horizontal” force curve like that of the Keychron K Pro Silver switches slows me down while writing at speed because the keys do not have anything to prevent me from bottoming out hard and losing momentum. By contrast, a more “diagonal” force curve provides extra resistance at the bottom of the spring. It makes it easier to notice you are past the actuation point and it also results in both a cushioned bottoming out as well as a stronger rebound event.

As such, when buying mechanical key switches, always pay close attention to the data. You want to get an accurate force curve or, at least, to understand how flat or steep that would look like. For my use-case, the flatter profile of the Keychron Silvers was appropriate when my RSI was at its worst state. Now I am all for a more pronounced slope to the force curve, like that of the Gateron G Pro 3.0 Yellow switches. There is no one-size-fits-all. What matters is to understand what your needs are.

Cherry keycaps are peerless

I tried low profile keys, the MT3 keycap set, OEM, and the MOA profile (check my Final thoughts on MOA keycaps and spherical tops in general). What practice has shown me is that Cherry are the best for me:

• The cylindrical tops are forgiving to off-centre key presses.
• The surface area of the cap tops is not too large, which balances out the above point when it comes to getting a feel of where you are on the keyboard.
• The cylindrical tops do not cradle the fingers and thus let them slide around easier.
• The sculptured or stepped heights between the rows further improve the sense of orientation through touch. For my Iris keyboard in particular, they also allow me to place taller keys at the furthest and closest point in the thumb cluster, thus reducing the amount of stretching or curling involved.

I admit that I was sceptical about the popularity of the Cherry profile. I was thinking that maybe the reason they are ubiquitous is because of market hype or some trendy fashion statement in picking the premium designs manufactured by GMK (the company that has the original Cherry company moulds). While I still have no interest in paying a small fortune for keycaps, I do recognise that the Cherry profile as such (which you can find for cheap everywhere), is the real deal and a testament to engineering excellence.

Looking at Keychron’s wide range of options to get a sense of some market dynamics, they seem to favour spherical tops (whenever they are not selling a board with Cherry caps). Their OSA and KSA sets are proof of that, with the latter looking very similar to MT3. If I was buying a keyboard from them right now, I would be opting for the barebones option and then picking up one of their Cherry sets.

For another comparison from a well-known manufacturer, Tai-Hao offers two types of sculptured profiles: OEM, which is basically Cherry with minor tweaks (I guess historically done to avoid patent issues?), and Cubic which have the OEM height but a bigger surface area at the top. If I had to make a choice between those two, it would be in favour of OEM: the smaller surface area must be understood holistically with the way cylindrical tops do not cradle the fingers, as I noted above.

Between Cherry and OEM, I cannot really tell the difference while typing. It all depends on the build quality of the specific set. What I am using right now is the Keychron Cherry Dolch Red set that a keyboard enthusiast who prefers to remain anonymous gave me as a gift. It is produced from PBT and is done using the double-shot technique. On this note, I do not care about PBT versus ABS materials. The MT3 set I have is made out of ABS and is very nice to work with, my dislike for the profile itself notwithstanding.

Determine what you are working towards

All my experiments with the keyboard are done to bring me closer to a comfortable typing experience. I write a lot and thus need a setup that empowers me accordingly. Understanding the “why” we are doing something is a precondition for refining it with intent. Otherwise, it is easy to get lost in the sea of options and end up paying exorbitant fees on gimmicks.

Market prices are a good indicator in this regard: a 30 EUR Cherry profile keycap set will deliver almost the same experience as those extra costly GMK designs. (More than 100 EUR for a base set!? Come on mate!) While you are still learning about your peculiar requirements, refrain from all those products that promise to deliver top-class quality: you might not like it and it is questionable how much you need it, anyway.

Unless you are specifically keen on tinkering with mechanical keyboards for the sake of exploration, consider this a trap: some experimentation is necessary to figure out what you need, but when it becomes a distraction from your duties you know you are overdoing it. Same principle if you are hoarding keyboards with the rationalisation that you are keeping them in your daily rotation. As such, try to maintain a pragmatic outlook in settling for a build that is “good enough” and, perhaps, committing to make incremental refinements to it over the longer-term.