How to best mimic Gateron Yellow weight in a JWK housing?

I found my end game switch for now, the JWK EV-01 switch.
Smooth and marbly-clacky on the KBD67 Lite it’s also comes un-lubed.

But it’s too light, the supposed 63.5 spring is noticeably lighter than the Gateron Yellows I have, swapping the springs from the Gats makes it significantly heavier (probably due to a different housing shape).

Pylon’s measure of Durock switches suggest it’s actually 55-60g compared to the 60-65g Gateron yellow.
I would have just gotten the 65g version but I’ve found the (heavy) factory lube doesn’t pair well with Tribosys 3204, and the EV-01 version is the only un-lubed version I can get currently.

I presume you’re referring to this graph from Pylon’s 63.5g measurements?

If I recall correctly, Pylon noted that the spring in question from that Penguin is actually an incorrect 55g spring that was installed into the Penguins. So you should really be looking for actual 62-65g springs that are roughly the same length. Based on his data, the Sprit Slows at 63.5g should be a pretty close matach.

Jimmy Nguyen measured a Gateron Yellow spring at 15.3mm long, which is longer than the standard Durock spring length of 14mm that I’ve measured from a pack of 63.5g springs. Longer springs have more preload because they get compressed more at resting when installed in a switch, which increases the actuation weight. This is why the default springs from your EV-01 feel lighter than the equivalent spring from the Gateron Yellows you have.

1 Like

Interesting, I had wondered why 63.5g penguins required spring swaps for stabilized keys when most switches with that weight have no problem.

Thank for the info, seems like the issue revolves around Durock’s incorrectly releasing data all over the place.

Durock claims the EV01 and POM switches to be rated at 63.5g and Lupine at 62g yet Pylon’s measures the POMs [and other 63.5 JWK switches], and Lupine to bottom out at 55g.


This also happens on the Alpacas which are rated 62g but also measured at 55g bottom out.
Yet the rest of the Durock switches (mainly the tactiles) seems to bottom out at the correct force.

Though I’m still unsure about what springs to choose, if I get the Sprit Slow that acts basically the same as the Gat spring (in what I assume is a Gat housing) then wouldn’t it end up the same and feel heavy?

On the left (Durock’s 63.5g spring, on the right KS8 Gat yellow spring).
If the length difference is that small yet it becomes as heavy as Ink Black what spring should I get?

I wager its down to spring variance which is causing the lighter bottom out, which means that JWK should tighten up their spring production.

Your post title and initial post implied that you wanted a Gateron Yellow spring weight, but if you want something lighter then stick to the 60-62g range and find a spring that’s around the 15mm mark since that’s longer than the standard 13.5-14mm Durock but shorter than a Gateron Yellow and slightly lighter. TX 62g medium should do the trick.

I am looking for a Gat Yellow spring weight, I was under the impression they’re around the 60-65g? Pylon’s Gat Yellow measures are in that range.

Anyway I appreciate the help, I’ll look for what you suggested.

Nick7790 did send me some springs (most of which I still need to get to measuring) - one of the ones I did measure (but forgot to post to GitHub) was a Durock 63.5g spring. This was measured in a Gateron Yellow housing with the leaf removed, with a JWK Queen stem (since Gateron Yellow stems have oval spring holes that tend to cause springs to catch/bind)

Every spring I’ve measured from a 63.5g Penguin ended up coming around 55g. Spring bottom out ratings are usually rated at 4mm travel, so a switch with reduced travel will bottom out at a lighter weight than what the spring bottom out nominally is, but even accounting for that they are still consistently coming in under. I bought some NK Silk Yellows recently (to give linears another shot), and they’re also coming in under:

I did find it interesting that a Gateron Yellow spring inside a JWK linear might feel heavier than inside a Gateron Yellow. The leaf/stem interaction does affect the force needed to press down, even on a linear switch - the leaf ends up pushing the stem downwards, so the leaf’s presence makes the switch lighter than if it weren’t there. Perhaps differences in the stems and leaves of the Gateron Yellow and the JWK linears account for the force difference. (see the next post - this explanation is incorrect)

Out of curiosity I did a quick measurement of a Gateron Yellow, and of a JWK x Moyu.studio Queen switch with the spring from that same Gateron Yellow swapped in, and it seems to be the case that the JWK is heavier with the same spring:

This seems to be an odd finding, and it’s also possible my load cell (used to measure force) drifted between the two measurements, so I want to take a couple more measurements to verify that finding and to rule that out. However I need to go to sleep (it’s 1:31AM), so that second measurement to verify this will have to wait until tomorrow. If this is the case - that different springs feel different in different linear switches - that would throw a pretty big wrench into spring ratings.

Also glad people are finding my force curve measurements useful :slight_smile:

3 Likes

I’m not sure if this is going to make any sense, but after talking with HaaTa - the underlying reason is likely because the Gateron and JWK housings compress (preload) the spring to different lengths. The JWK housing likely compresses the spring a bit more than the Gateron housing, which results in a higher force and therefore a heavier switch even if using the same spring.

A typical spring has something like over 10mm of travel if it’s not mounted inside of a switch.

A spring’s full natural force curve (sitting outside of a switch housing) would look something like this:

When you install the spring into a switch, however, it will only use around 4mm of that 10mm+ of natural travel. However, which 4mm of travel out of the 10+mm of travel available the switch uses may vary from housing to housing, depending on how much the spring is compressed when it sits inside the switch (in fancier engineering terms: how much the spring is preloaded by). Some housings compress the spring more, and some housings compress the spring less (e.g. one housing may let the spring sit at 11mm long when installed inside it, while another housing may compress the spring to 9mm when it rests inside that housing). The housing that lets the spring sit longer at its resting state would feel lighter than a housing that compresses the spring more.

Visually - the Gateron Yellow housing perhaps uses this portion of the spring’s full natural force curve:

While the JWK housing, which compresses the spring more, may use a heavier part of the spring’s full natural force curve:

If this is true - the implication is that the same spring would feel lighter or heavier in different housings. Which throws a wrench into spring force ratings.

5 Likes

That’s a great explanation, really appreciate it.
From my personal [non accurate] testing using a Gateron Yellow spring in a JWK houding makes it as heavy as an Ink Black when pressing them into each other.
I guess Gateron conforms to Cherry’s dimensions while JWK doesn’t, wouldn’t have been that bad if their released numbers weren’t completely wrong.

Weird it hasn’t been brought up by anyone, people mark the hyper tactile 67g tactiles as the gold spot but happily use 55g linears unknowingly they’re significantly lighter than what they’re thinking.

BTW your measurements are always useful, probably the most accurate form of comparison I’ve seen in this hobby yet.

1 Like

This is fully the case, but… I don’t know what spring manufacturers or vendors can do about it?

There’s not a component in an MX switch that doesn’t affect top-out, bottom-out, or actuation weights, and the leaf is (maybe?) the only component that affects only one of those things (actuation weight, specifically, and even then it’s usually also going to matter in terms of determining peak force for the bump on tactiles or clickies). Every other component of the switch is going to affect how much preloading is happening (even the stem!) and how much travel is experienced (and, thus, what part of a spring’s curve is relevant).

I think it’s still extremely worth looking into these things, but perhaps there’s more framing information that needs to be provided in talking about springs and force diagrams?

1 Like

I think the natural progression would be for spring manufacturers to specify the actual spring constant of each spring next, and choose one switch to measure the bottom out weights by which would probably be Cherry for simplicity since that’s original spec. Bottom out weight doesn’t account for variables like what Pylon described in terms of preload, or the reduced travel switches.

Actually I think it would be the opposite, TX springs had an old chart of the equivalent springs being lighter in a Gateron switch than the Cherry switch which I never really thought of it being from preload - I always assumed that the switch manus used the same internal measurements and compressed the spring equal amounts. Based on what you said with the Gateron Yellow spring being as heavy as an Ink Black spring in a JWK switch, I would err more on the side of the JWK switches fitting tolerances correctly.

2 Likes

Well I don’t know either but I would appreciate if they could provide accurate numbers on their own switches though haha

Oh for sure. Switch manus could definitely stand to improve a lot in terms of what info they’re putting out about things.

From some testing it seems like Gazzew’s 62g springs from the Boba U4Ts in the JWK produces the same weight from comparing them against each other.

Well, if I may provide some commentary regarding this matter I think that this is dangerously lacking in evidence to be making the claims that it does. While I empathize with the late night nature of this post, there are assertions or even claims that are made here that have literally no evidence to them.

  1. You, nor Haata for that matter, appear to have done any sort of rigorous testing to actually isolate the cause of this effect. Claiming its one component out of a complex system of spring, stem, top housing, and bottom housing without having accurately accounted for those things via various tests is dubious.

  2. Even with Haata’s much more rigorous and assumedly NIST certified force curve meter, a lot of Imada force gauges that are offered have margins of error in the few gram range. While I am uncertain of the exact model he has, I can’t imagine that a home made force curve machine is more accurate than something NIST certified, nor with similar margins of error. I understand this isn’t something you can go get done, but it should definitely gate many of the claims here.

  3. Also, quite literally just approximating a spring force curve without having even measured one outside of a switch is even more dubious and even more handwavy usage of concepts than asserting to understand the mechanism of a phenomena after having done a small number of tests and dragging up something from TX.

I think its good to point out the phenomena, and I think it certainly merits more exploration based on isolated cases of “something weird happening”. But that’s about all that we have here: something strange happening. I like the work that you and Haata have done, but given this hobby’s nature and how many people seem to leap off cliffs with singular strands of dubious evidence to make monetary decisions, I can’t help but try and stop this before it gets out of hand.

This isn’t much better in its current state than burn tests, if I had to put a ‘hot take’ spin on it. I hope that we can see it more rigorously examined soon.

Ok some additional measurements between the two switches with different springs - the JWK x Moyu.studio Queen’s spring (which is a long, “slow” spring), and the spring from a Cherry MX Clear (a short, “fast”, heavy spring)


Both of these measurements show the same result as the earlier one with the Gateron Yellow spring - the JWK stem and housing produces a heavier switch than the Gateron stem and housing when using the same spring. In addition, the difference in force is more pronounced with the Cherry MX Clear spring, which is what’d we expect if the underlying cause is due to a difference in how much the stem/housing compresses the spring - the MX Clear spring has a much higher spring rate than the other two springs, so for a given difference in length it will produce a greater difference in force.

Yes my force curve meter is home-made, and yes its margin of error is probably around few grams. However, error is usually random or non-correlated (i.e. having nothing to do with whether I’m measuring a Gateron or JWK switch - e.g. error due to temperature drift) - if the difference is caused solely by random or non-correlated measurement error, then we would expect on some measurements that the JWK would be heavier than the Gateron. I’m still at a small sample size (currently n=3), but if I can take more measurements with results in the same direction, I would be more confident that this is a real phenomenon rather than a product of measurement error. If the error in force is non-random and/or correlated, and arises to the level of 5-10gf (the difference in my measurements so far), it would likely be showing up on the force curve as noticeable non-linearities when I measure springs, but I so far haven’t seen any non-linearities that large when measuring things I expect to be linear.

It would be ideal to directly measure the distance between the bottom of the bottom housing and the top of the stem spring cavity to verify the hypothesis. However, this is impossible to measure directly with readily available tools (you can’t just stick calipers inside of a closed switch). You can possibly measure it indirectly by adding and subtracting several more measurable dimensions (e.g. measuring the distance from the stem cavity to the top of the stem, the distance from the top of the stem to the top of the bottom housing, and the distance from the top of the top housing to the area in the bottom housing where the spring rests, and then adding and subtracting appropriately to get the distance between the stem cavity and bottom housing). However, these are still not easy dimensions to measure, and your measurement error goes up significantly when you have to add or subtract several measurements to get the dimension you’re looking for. The easiest way to get this would be to ask JWK and Gateron directly what that dimension is (they can measure this easily in their CAD model), but I’m not sure if they’ll disclose it.

Coil springs are well-studied objects, and force curves of them (outside of keyboards) can be readily found.
Coil spring force curve

Also, I should clarify that in my earlier post I meant to say stem and housing, not just “housing” (my sleepy 2AM brain thought “stem” was included in “housing” and that “housing” merely referred to a switch minus the spring , which is clearly not how it’s usually used)

2 Likes