Kailh Box Switch Stem Measurements and Possible Problems


I just switched my Kira preorder from box royals to speed coppers, waiting on the fixed stems at novelkeys.

The metal keycap method takes a long time per stem, and is very rough on the fingers. I have reproduced the results with 10 switches, but I don’t know if I will attempt more. I do have another method to test once supplies come in this week.


So today i received my 1.5mm chisel, and set about cutting some jade stems!

It works, is fairly fast to do, but you need to be VERY careful. This thing is sharp, and it is very easy to take off too much material. The amount that needs to be removed is .03 mm or so, so think half the thickness of hair.


Instead of a chisel, could a very small jeweler’s file be used to remove just a hair of the plastic on the stem? I would think it would be easier to control how much of the plastic is being taken off compared to a chisel. if it’s just .03mm that needs to be taken off, a pull or two with a good file would do that. It’d be quicker than using a metal keycap by far, and easier/more accurate than pushing a chisel into the plastic. I don’t have any box switches yet, otherwise I’d try myself.


but how do you get a file down to the base of the stem inside the box? At least with the chisel you could make sure you go to the very bottom of the stem.

IDK, i’ve never used a jeweler’s file before


Yesterday I shaved the stems on 100 Jade Switches, and was fairly consistantly at 1.25mm +/- .03. It took under a minute for each, but I cannot stress enough how careful you need to be, I suggest having a firm base to hold the switches. After removing the material I tested each stem with a variety of keycaps (GMK, SA, DSA) to confirm that they were not too thin now, finding only one sent caps flying - it measured 1.18mm.


Thanks for testing, good sir (or madam). I’ll order a file and get to work. 1000 or so switches to shave.

Are you just removing material form the horizontal stem line?


I used the chisel I linked above. Just the horizontal x-axis, you want to look to remove the ‘nubs’ detailed in previous images.


Which one do you think worked better? Metal keycap or the chisel? I want to try out with my Box Navies, Jades, and Hako Trues.


My nantucket set does have prob with box orange. Removed immediately


Chisel was easier by far


Thanks! Shaved 60 Hako Trues last night using a home made chisel. On to 90 Box Jades and 140 Box Navies.


My caliper just came in, now I’m waiting on the Kira switch tester. I’m hesitant to buy a tool I have no further use for after this, so I’ll test the needle files/x-acto set I have already before buying the wee chisel.

While I’m waiting, My readings on the Cherry MX switch tester on my desk shows a range of 1.25-1.27 mm, and a sampling of unused Gateron Blacks show a range of 1.27-1.30 mm


I got a chisel and spent 5 min on 6 switches. 1 feels like it may be still too tight, the rest are now too loose. I think it would be safer to just desolder and replace these things.


Yeah. Some of them became too loose but not enough that the caps pop out.


Have you tried the plastic wrap/Scotch tape fix?


Oh I could make them work with plastic wrap but we’re talking about 900 switches that would be going into high dollar custom boards. It’s pretty much just a loss


I’m still measuring my GMK keycaps with my gauge pins, but I should have results posted this weekend. I’ll probably make another thread for that post in case other people want to post their data or discuss that sort of data specifically


This is a basic engineering task of understanding the tolerance on an interface between two parts.

To create a Cherry compatible switch, one would go and measure ~100-200 Cherry stems (preferably with a micrometer which is higher resolution and more repeatable than just calipers), calculate the min, max, median, and standard deviation, write a specification and work with the clone switch vendor to make sure that their switches fall within that specification.

I’m seeing so many people making different excuses for why this isn’t their fault, or they had no idea the stems could be a problem, or whatever… and the one ‘fix’ I have seen proposed is to change the spec to 1.30 +/- .020 mm, which is still not matching up with the range that Cherry switches fall in

All that you have to do is read any tutorial on sourcing electronics from China and the name of the game is verify, verify, verify, verify. Don’t assume things will be okay or take anyone’s word for anything, or you’ll get your genuine parts substituted with “compatible” parts that don’t actually work, etc.

These dimensions are something that should have been checked and verified by the people who have been selling thousands and thousands of dollars of switches to the community, that the switches they were importing were truly compatible switches. Not “oh woe is me there is no way I could have seen that”.

Don’t even get me started on the fact that Input Club is going to still sell bad switches with the Kira, and refuses to delay the project to make this situation right for all of the switch options.

Best thing to do here - open trash can, toss box switches in trash can, refuse to do business with any vendor that won’t hold their stems to as-measured tolerances of Cherry stems.


naa, don’t trash, just mod them; not the ideal solution but it works!


So this is really interesting. Seth at Keychatter made this post - https://www.keychatter.com/2018/08/16/unpacking-the-kailh-box-switch-debacle/.

But in this post he has the official Cherry spec sheet. Cherry stems are rated for 1.31 ± .02mm. So Kailh’s updated BOX stems will fall well within this spec.

I think its great that we can finally get to see official spec sheets from Cherry, and not just from Kailh. I would really like to see more spec sheets on keycap stems, as well.