Some teasers of what's to come...

User avatar
HaaTa
Master Kiibohd Hunter

06 Feb 2014, 22:16

This is the reason I haven't posted any pictures as of late...

Image
Image

Image

Still some refining left to do. There is a bit of deflection on the support beam when doing the reverse direction so the blue line needs to be shifted to the right. As well, the sample rate is a bit low atm due to the 2400 baud on OO2s force gauge; mine does 19200 though).

Graphs should get quite a bit better as I find more ways to remove error in the measurement process :mrgreen:

User avatar
Jmneuv

06 Feb 2014, 23:05

slime.png
slime.png (111.65 KiB) Viewed 1827 times
Last edited by Jmneuv on 07 Feb 2014, 07:55, edited 1 time in total.

User avatar
Muirium
µ

06 Feb 2014, 23:23

It's exponentially better when you know what the motive is. But my lips are sealed!

User avatar
Daniel Beardsmore

07 Feb 2014, 00:21

Kishsaver is a Model F, right? That graph is over 9000 80 gf and virtually linear! The actuation seems too far along, too.

User avatar
HaaTa
Master Kiibohd Hunter

07 Feb 2014, 00:47

Yep, Kishsaver is a Model F and there's nothing odd about it in terms of the switch.

The actuation for Buckling Spring is very late in the press. I wouldn't use this curve as it has some issues with it I know how to resolve, but it's reasonably representative of Model F switches.

User avatar
webwit
Wild Duck

07 Feb 2014, 00:55

For point of reference, I measured the activation point only, with accurate weights and calibrated precision scale of a number of keys on a Model F AT, and it averaged on 64g, ranging from 60 to 66g. The results were pretty consistent (unlike Topre and Model M. Cherry MX Blue is incredibly consistent).
http://webwit.nl/input/witometer/
In feel, capacitive Model F is a bit different, but not heavier. If not the same, lighter than AT. And not as heavy as BS. This confirms my activation point measurements in some way (it's not a full force graph).

User avatar
clickykeyboards

07 Feb 2014, 01:17

I tried similar (force) * (distance) precision measurements and found that to get accurate graphs, you really need a digital micrometer and a way to incrementally and consistently apply force to get the distance measurements right.

Also the mounting of the force gauge must be on a stable, solid frame.

http://www.clickeykeyboards.com/model-m ... d-and-ibm/

User avatar
webwit
Wild Duck

07 Feb 2014, 01:22

Nice pictures, but give us the results! :ugeek:

User avatar
clickykeyboards

07 Feb 2014, 01:30

My colleague recommended the following gauge stand.

http://www.ebay.com/itm/Test-Stand-for- ... 4897.l4276

I had access to the force rig that I used in the pictures in the summer of 2011 from the mechanical and aerospace engineering department at work and at the time I used it to measure a bunch of model M keyboards. Very time consuming to do it by hand and had to return the rig before I could finish a proper study. Was hoping to find an undergraduate intern to become interested in the project for them to writeup for their senior thesis work...

Never got back to it.. but if I were to do it again, I think the more interesting results would be to see the quantitative results and graphs for different keyboard manufacturers and key switches.

User avatar
Daniel Beardsmore

07 Feb 2014, 01:32

HaaTa wrote:I wouldn't use this curve as it has some issues with it I know how to resolve, but it's reasonably representative of Model F switches.
That's odd, as that graph implies that it's even less tactile than membrane BS, and it's supposed to be more tactile. This is the curve I was expecting to see:

http://deskthority.net/wiki/File:Force_ ... spring.svg

Much lower pretravel and a fairly standard force reduction on actuation. Also, 80 gf is a lot of force for actuation, and I thought the whole problem with membrane BS is that the membrane hammers forced the switch to be heavier (though that graph has actuation at 70 cN, which isn't far off membrane BS).

I presume you've spoken with jacobolus about measuring all his Alps switches — I'm really looking forwards to seeing measurements for lots of switches we're so far only able to guess at.

User avatar
HaaTa
Master Kiibohd Hunter

07 Feb 2014, 03:31

One of the problems about getting a pre-made scale (with calipers and force gauge) is that I'd still have to tear one open and tap the protocols because the data needs to be correlated.
Because I do this, I can move the tester/up down as slowly as I want and not spam me with data (works using distance sensor increments which are around the um range).

clickykeyboards, the stand you linked looks interesting (does about the same thing mine does), but I don't think it would work that well for keyboards. Not enough clearance to place a keyboard under it (more for testing out springs and single components). Might be ok with a really large bracket though.

One of the benefits of the current setup is that it's really heavy. I can let go of the lever, and it will stay (for days...). There is an adjustable tension spring, so I can tweak it just right so it's still relatively easy to turn the lever.
Last edited by HaaTa on 07 Feb 2014, 03:55, edited 1 time in total.

User avatar
HaaTa
Master Kiibohd Hunter

07 Feb 2014, 03:49

I might consider going for one like this http://www.ebay.com/itm/NEW-Hand-Press- ... 337d6985bb but it brings me back to my original problem, how to mount the force and distance gauges.

User avatar
phosphorglow

07 Feb 2014, 16:06

HaaTa... Great - now I need to bump up my level of OCD with a new toy at some point. :x

User avatar
matt3o
-[°_°]-

07 Feb 2014, 16:12

oh gosh... are we working on the same project?!

mine is cooler, though :P

Image

User avatar
Muirium
µ

07 Feb 2014, 16:24

Definitely cooler! Now let's see some data.

User avatar
matt3o
-[°_°]-

07 Feb 2014, 16:29

I need bigger copper coil. I'll post data as soon as I can properly read all the 1024 steps :)

actually, this is the new version

Image

User avatar
lowpoly

07 Feb 2014, 16:58

A mill should be good for this. With integrated caliper or digital readouts:

Image

Will not bend.
Last edited by lowpoly on 01 Jun 2018, 16:38, edited 1 time in total.

mtl

07 Feb 2014, 17:24

But will it blend?

Sorry.

User avatar
HaaTa
Master Kiibohd Hunter

07 Feb 2014, 21:27

matt3o wrote:I need bigger copper coil. I'll post data as soon as I can properly read all the 1024 steps :)

actually, this is the new version

Image
Neat!
We'll have to compare :D

Distance I understand, but how do you calibrate the force?

User avatar
matt3o
-[°_°]-

07 Feb 2014, 23:41

HaaTa wrote: Neat!
We'll have to compare :D

Distance I understand, but how do you calibrate the force?
with a high precision scale

User avatar
Daniel Beardsmore

09 Feb 2014, 14:28

Here's a question — how much control do either of you have of your environmental humidity — any at all?

Maybe one day in the future we can scientifically settle the hypothesis that Alps switches are humidity dependent.

User avatar
HaaTa
Master Kiibohd Hunter

10 Feb 2014, 23:29

Myself? Next to zero...

User avatar
Daniel Beardsmore

11 Feb 2014, 00:01

I wasn't sure whether you lived or worked somewhere where you'd have some sort of dehumidifier that you could leave off for a comparison. My main reservation with switch sampler kits is that switch feel depends a lot on how you position your hands: two-finger prodding even a Filco Zero XM doesn't actually feel a fraction as bad as if you try to type on it properly, when you realise just how much force is truly required. That said, I have definitely perceived (correctly or incorrectly) that the feel of Alps switches varies from day to day, and the only explanation I've seen so far from anyone is humidity.

I notice when I go back to work on a Monday morning that my office keyboard feels a lot lighter than it did on the Friday. It might be due to my nasty T-Rex typing posture at home, compared to a proper desk height at work, as a proper posture improves switch feel a lot. Filco Zero XM again: dreadful at home, but almost usable at work (and those are super stiffy switches).

Interestingly, when I tried my three new keyboards on my desk at home earlier in the year (crammed in in front of my Filco) — an NTC plastic plate white Alps, 1996 Model M, and possibly the Peerless (not sure what the third one was), the Model M was the only one that I could actually use comfortably. I was quite surprised at how usable it was. The near-linear force curve and long pretravel must have helped under these conditions; short pretravel is helped by having a correct typing posture where you can better exert the correct actuation force. I found Cherry ML to be the best switch of the lot for accuracy at home (it's rough, but really comfortable, with the right weighting for the travel), but the fn/ctrl arrangement drives me bonkers — one day I'll lay my hands on a full-size ML keyboard.

Having actual measurements will take care of all my situation-dependent perception problems, but it may never explain why Alps switches are so temperamental.

With luck we can also settle the specification difference between all the similar Alps models (brown/green/yellow/grey/cream linear, blue/white/yellow clicky, brown/cream/orange/salmon/black/cream damped/white damped tactile). I suspect Daemon Raccoon is too far away to measure her crazy double-action switches — Alps cite a staggering 500 gf for stage 2, above 90 gf for stage 1!

Post Reply

Return to “Gallery”