Keyboard Builders' Digest / Advent Calendar
The Data Driven Future of Switches
The always verbose and never short on switches ThereminGoat gives his take on the data he has collected with switches in the previous year as well as what the future may hold for such.
Published December 5, 2022
I must say at the start of this that I feel both incredibly honored and out of place in a series of articles as such. When I was initially reached out to by dovenyi, who is the awesome individual who put all of this together, I was more than welcome to sign up for the prospect of getting to write about switches more than I already do. However, looking at the significant slant toward the DIY side of the hobby in the articles before and after me, I have to admit that it feels a bit out of place. The people who have all contributed to this project are people with some pretty serious hands-on technical knowledge or produce real, historical driven content and I more or less just rant my opinions about switches over on my website. In fact it is for this reason that I am surrounded by so many talented people in this hobby that I felt I should write about something that I am much less known for than the switch reviews – data.
While there is no doubt in my mind that I am known (and probably somewhat hated) for my overly verbose writing and switch reviews, a lesser-known series of projects that I’ve taken on in the past year or so has been around collecting as much tangible data as possible about switches. On one side of this project I’ve got my caliper-based switch measurement sheet, which takes a look at 13 different dimensions present within switches including things such as the overall stem length, dimensions of the top housing hole for the stem, and quite a few more. On the other side of this project, I was gifted a rather professional force curve meter by Drop earlier this year which has kicked off a downhill pace of collecting force curves for as many switches as I can. While these have produced infinitely more files and consumed a lot more time than I think many people recognize, I’m sure there’s one question that comes to everyone’s mind when they see the extent of data that has been collected thus far: “Why?”
Measurements and data about switches on every single sales page are among the most considered and also least explicitly recognized pieces of information that inform switch purchases.
Sure, things such as price, color, manufacturer, and materials all ring out in our minds as the first things we consider when purchasing switches. But what about bottom out spring weight? What about the type of switch? Has anyone ever looked at the travel distance of a switch to find out if it’s ‘long pole’ before? With the nuance in switches being increased seemingly daily with all the recent releases from every manufacturer you can think of, these types of small details have begun to become the separation points to help people decide between ‘Switch A’ and ‘Switch B’ for their endgame board. And for the record, while that might not mean much for someone’s Keychron board who just joined the hobby, these are the types of decisions that more experienced community members have to weigh when building a $5,000 TGR board, or a $3,000 Keycult – and that’s an expensive decision to make with so little data. While I am but just one goat, it’s because of these sorts of considerations people are having to make, as well as purely selfish curiosity, that I’ve picked up this data collecting project at mine.
Aside the explicit points of note that I brought up above, with force curves for over 400 switches and measurements for over 200 different ones as of the time of writing this article, there’s a surprising amount of trends that I and other switch nerds have been able to pick out of all of this. On first glance, looking at all of the switch measurements on the caliper-based side, of which there are more than 24,000 different ones, it may seem like a bit of a mess to pick anything out. However, if you plot all of the stem length measurements (marked as ‘G’ in the raw data sheet) together, what we actually find about the average stem length is quite contrary to popular opinion. Whereas many people may initially think of Tecsee as the long stem pole champion, it turns out that on average TTC has the longest stem poles, beating out Tecsee by over 0.200 mm in total stem length.
Looking next at the force curve measurement data set, we’ve actually been able to identify a rather interesting trend that is noticed in switches that come unlubed from the factory. Very specifically, if you look at the post bump linear travel region in unlubed tactile switches, it’s not all that uncommon to see a sort of serrated edge type look. While the exact mechanism of this has yet to be fully fleshed out, and there are various hypotheses being tossed around as to what may cause this, the appearance of this in a good handful of these switches definitely points to some sort of mechanistic nuance that completely escapes the average user experience.
To be entirely honest, these two examples just touch the tip of the iceberg in trends and interesting data points that have been able to be pulled from the woodwork over the past year of collecting data. In fact, these projects also both have incredible potential to expand beyond their current frameworks and likely will be doing so in the upcoming years’ time. With respect to the caliper measurement sheet, I have currently been working over the last few months to expand these measurements to the keycap mounting stem and would like to release it once I’ve passed something like 150 switches measured in order to provide a really neat data set that doesn’t need to grow as slowly as this first one. (Plus, it may be tied up in other content projects as well!) As for the force curve measurement sheet, I can assure you that there’s even more potential out there that hasn’t been considered nor attempted yet as far as I can tell.
Take an average linear switch, for example. Given that these articles are written in majority by DIY people, I can imagine that the average audience member would likely recognize a linear switch’s force curve rather easily – it’s flat. The entire downstroke region, save for the initial start and bottoming out, are as the name implies in the type of switch mostly linear. However, upon testing over 400 of these switches I’ve begun to notice that not all linear switches are created equally – some have very steep curves and others have flat ones where the initial downstroke weight is within only a few grams at most of the bottoming out weight of the spring. From this, I’ve begun to rationalize and flesh out a way (that I would like to share soon-ish) to quantify and describe the sort of ‘strength’ of linear switches – in which more steep switches with a big change in force between the start and end of the downstroke are “strong” as opposed to the “weak” switches that hardly have a difference between the two ends. While quantifying tactility has been attempted in the community quite a few times prior, to the best of my knowledge this would be the first time that anyone has attempted to ever squeeze more information out of solely linear force curves. Don’t worry though, I definitely do want to take on the big task of quantifying tactility in due time.
At the end of this year’s worth of data collecting, as frenzied and fast paced as it has been, I definitely have appreciated the chance to get to sit down and flesh out my thoughts for an audience like this. While I will almost certainly continue to expand and extract more data from the numbers I have and will keep continuing to grow, it’s only with the recognition by the community at large of the value that these types of measurements have to offer will they begin to be utilized as I’ve initially set out to provide. Though, I don’t want you to have left this article thinking “Damn, this Goat guy really needs to get a life, that’s way too much work”, rather I want you recognizing that there is so much switch data out there to be collected that you also could be a part of. You don’t need a fancy set of calipers or even a force curve machine of your own – something as simple as tracking mold markings, switch colors, spring lengths, etc. are all things that can be done with a bare minimum set of documentation skills that could unlock trends that none of us are immediately aware of yet. I implore any and all of you out there interested in switches of any type, shape, or size to go out there and collect information about your favorite switches and to help drive the future of switches to a more data driven one.
Resources
Published on Mon 5th Dec 2022. Featured in KBD #106.
Related
SP Star Sacramento switch
ThereminGoat reviews the SP Star Sacramento switches.
1000+ switch collection
ThereminGoat celebrates passing the 1000 switch mark writing about the collection, the background behind some of the switches and his collecting history.
Microswitch SD to MX adapter
Jpl3k designed an adapter to run a Microswitch SD keycap on an MX mechanical keyboard.Some switch sounds
Sound test of some switches by TheHapaLens.
Magnetic levitation Void switches
Chyrosran22 reviewing Riskable's Hall-effect switches.