Kailh Box Switch Stem Measurements and Possible Problems


#64

I made a pretty lengthy post on reddit. You can see that here.

Basically the short story is this.

DAREU was the original larger purchaser of BOX switches when they first came out. They requested that Kailh increase the axis to 1.32mm. Thats the reason why they are 1.32mm.


#65

Thanks for the update @mgsickler :slight_smile:

Follow-up question,

Do you know GMK’s current specs for their stem inserts?

If we go by Cherry’s 1987 specs they claim that the thickness is 1.17 ± 0.02mm (If I’m reading it correctly) and that’s definitely on the thin side compared to future Box Switch’s 1.30mm ± 0.02mm. But at the same time we haven’t had issues with any other mx style keycaps cracking which mostly sit around the 1.27mm ± 0.02mm range according to @MehAdviceGuy’s Datasheet. Only a few of his data points for non-box style switches reach the thickness of 1.3mm, but we don’t have enough information to determine if those are outliers or closer to the mean/median for those switches.

While it is nice that we will have consistently thinner x-axis switches, will that be thin enough? I’d like to know if GMK has any updates on that 3 decades old spec sheet? At any rate I think we can consistently agree that 1.27 ± 0.02mm seems to be in the safe range, I just hope that 1.3mm ± .02mm will still be in the safe range.

Thanks for the update, hopefully we won’t have any issues with the future box switches


#66

I do not know GMK’s or even other manu’s specs on their stems.

1.30 is the same size as the other switches in Kailh’s line up, with actually tighter tolerances. The updated BOX switches will be 1.30 ± .02mm while the Pro and Speed switches are 1.30 ± .03mm. There have been 0 reports about those stems causing issues. So I would think its on the safe side.


#67

I just measured some speed and pro switches and the x axis thickness was between 1.26 and 1.28 mm. If their spec says they should be 1.3, they consistently run small. I can collect more measurements if needed, most of my builds have pros or speeds.



#68

Hm, another interesting observation, all the Novel x Kailh non box stems seem to be thicker than the non collaborations, from the small sample I have I am seeing they range form 1.27 to 1.30mm


#69

Today I sent an email to PMK/SP to see if they’ve had any reports about stem inserts cracking due to box switch stems. I’ve also asked if they have any specifications they could share with us about their stem inserts. Hopefully they get back to me soon. I’ll share their response as soon as it comes in.

I’ve also linked this post to them so hopefully they can find time to look it over.


#70

Cherry is laughing over a plate of Sauerbraten right now.


#71

I have got a workaround for the problem, if one has some loved box switches.
Just put a metal keycap a few times on the box switch, and the stem will be a little tighter.

I know it is not the ideal solution, but no need to throw away some good switches.

I used a cheap metal keycap which I got from ebay.


#72

For those who haven’t seen, Mike posted an update for this issue. Any pending purchases of box switches will be delayed but will be corrected at no cost. This includes Novelias and any Massdrop orders.

@lekashman do you know what will happen with the group buys that include switches from Input Club?


#73

I noticed the exact thing. However, couldn’t take proper pictures due to poor lighting in my room. If you put the BOX Jades next to the Royal Trues, the fine stripe-layer is quite apparent. Also, I used SA Carbon keycap to try fitting and see if switches affected it.

To my surprise, Cherry MX Blue and BOX Jades required some force to push in the unused, new SA Carbon cap. After that, Royal True required a bit more force. But when I reversed the order, the same SA Carbon keycap slipped easily on MX Blue and BOX Jade switches. So yeah, even a couple of seconds on Hako Royal True widened the SA Carbon.


#74

It would be great if you can post some before and after measurements of the Hako Royal True switches.


#75

What do the letters and at the back of the switches signify?
Royal Trues have 1 at the bottom.
BOX Navy and Jade have 3 at the bottom.


#76

Hey, tha’s a good idea. Thanks for sharing.

I wonder if any of you with calipers could try a mod like this to measure the results.

I would love to make use of all of these box switches I currently own :heart_eyes_cat:


#77

It’s something we are looking into now - we have had no issues with any of our products like the one shown so far - but I will sit down with Kaihua and ask about this topic.


#78

I understand the viewpoint of @lekashman and @kailh , b/c it “seems” that box switches have been in the wild for a year with no issues. However, there is zero chance that the switches are not problematic.

The only reason you’ve had “no issues” with any products like the one shown is b/c you failed to notice it. There are many reasons for that. I didn’t notice that 20 or so of my GMK hyperfuse keycaps were cracked for nearly 1/2 a year. It wasn’t until I moved them to another switch that I found out.

The switches crack keycaps. It’s plain and simple. PBT keycaps seem to be mostly resistant to the cracking, which may be another reason why no one has noticed up until now.

At this point in the game, selling these switches as-is without a clear warning that using them on ABS keycaps (and specifically GMK keycaps) will possibly crack your keycaps would be unethical.


#79

I don’t know how difficult it would be with calipers, but has anyone tried measuring the part of the box switch that isn’t raised? Might give a better idea for what the thickness would be after reverting to the “1.3mm” tolerances (assuming the ridge is the result of them adding thickness per Dareu’s request)


#80

Melissa from SP/PMK responded to me that their measurements are as follows:

MX Mount is: 0.217 x .159 x .047 ROUND CRUCIFORM, RECESSED .005 (measurements are in inches)

aka 5.512mm x 4.039mm x 11.938mm RECESSED 0.127mm.

Followup email:

“The cruciform itself is 0.217 inches in diameter with the cross bars measuring 0.159 inches long and 0.047 inches thick - the 0.047 inches is where the issue would be with the Kailh switches.”


#81

I had an Iris with gateron yellows, Commodore64 caps with 3d printed adapters.
The caps had wobble, but they stayed on just fine.

Build a new one with box blacks because they have less wobble, so with the adapters that’s better.
Really used it for the first time yesterday after building it, keycaps were literally popping off during use.

Now obviously, these are 3d printed adapters so they aren’t as strong as a regular keycap. But, knowing that there was 0 problem on the gat yellows and after just an hour of use on the box blacks they immediatly pop off, I don’t need anymore proof. I have 2 boards with box blacks and gmk keycaps > cracks. Iris with the adapters thus not cracking but popping off. I’m done with the switches sadly. For now my solution with the iris is some tape and see how that works. Luckily the adapters are replacable.


#82

So we’re looking at 1.19mm or so then? That’s interesting considering most switches (even from different brands) are at least 1.25mm. Presumably that’s because you need the stem to be a little thicker so that they friction fit. Hard to say then what the ideal size actually is.


#83

Agree with @pixelpusher. Saying box switches have been out for year without any issues is not very convincing because it took a while for the switches to reach a certain level of market saturation to where a good amount of people were using them. Additionally, as it’s been pointed out, the problem can be hard to notice. The cracking is not something you would be in a position to discover on a daily basis. Ultimately, a few anecdotes of people reporting issues prompted EVERYONE to check their keycaps at the same time which is why this is all happening now. There are multiple factors and the measurements speak for themselves.

But I think at this point it’s more important about how this issue is dealt with going forward rather than trying to play the blame game. Let’s make sure new switches are manufactured so that this doesn’t happen anymore. Let’s also make sure there is full disclosure to those buying the existing switches.