The wiki thread

[Daniel Beardsmore]

At first sight, the only difference appears to be the code on top of the casing and mounting direction on the pcb: the whites have the code on the north side, the yellows south. They appear to be of same activation force and clickyness.

I can't tell if the writing on the white ones is north, or whether it's just soldered the other way around (or the top shell is just on backwards). This is the problem with using numbering for identification — it's so hard to capture and make out in photos.

[webwit]

I haven't checked yet if it's really soldered the other way round, will open it up this evening or tomorrow for a closer look.

[Daniel Beardsmore]

You should post some photos of that board showing the switches, both white and yellow.

I'll add a note to the page tonight that the F key switches are yellow, not white. I'd probably seen that before, but I'd forgotten when I rewrote the page :(

[webwit]

I took a closer look, casing is the same except the code on top, but of the six switches I exposed (4 white, 2 yellow), they all had different codes on top of the casing. Yellow switches are indeed mounted the other way round. I measured the function row (yellow switches) and numbers row (white switches) using my wit-o-meter, and the yellows activate at 66-68g while the whites activate at 70-72g. Little difference, but it's consistent.

[Daniel Beardsmore]

I just meant posting a photo so that people can see what's what. I found a photo that suggests all ivory (60±15 gf AK-CN2) switches in the MiniTouch, so it looks like there were at least two clone versions:

http://geekhack.org/index.php?topic=11589.0

Digging around for the SMK-85 turned up various other brandings of the Monterey board that got badged as the SIIG MiniTouch. Still no evidence from anyone as to why the FCC ID is from more than a decade before SIIG sold it — rebadged new old stock?

As for the switches ...

Does anyone know for certain what purpose the numbers serve? Mould number I guess.

Funny thing is, I took a photo of a Xiang Min KSB-C next to an APC BSW 065WHN (or 075WHN) and both just happened to be "A12"!

Co-incidental numbering.jpg (155.09 KiB) Viewed 7411 times

How strange is that.

Numbering is another clue as to origin: APC is always a narrow-font code at the top-left. Xiang Min is normally top-centre, though the font and size changes a lot. Unless it's a KSB-LE, which has a corner number. And the KSB-LE has Hua-Jie internals. (I'll give him a few days, then I'll chase up Channing at Hua-Jie to see if he'll agree to sell me some samples. I really want to know if their switches are now identical to Xiang Min, or still different.)

Bottom-centre numbering and diagonal corners are two of the more confusing schemes that suggest that visibly different designs may have come from the same factory. Hard to be sure until everyone here posts photos of their switches so that we can piece everything together. And that will never happen.

[webwit]

Sorry, my camera is fucking up at the moment. It started when it showed no image when powering on, and it took a few power off/power on steps to kick the live into it. That number of steps is increasing... I'll post some pics when I solved that or got a new camera.

[Daniel Beardsmore]

I should replace mine actually — there's an uncleanable smudge on the lens (or something), and the LCD is slowly disintegrating :) But that's still better than the soul destroying experience of trying to replace something without ending up with a worse product ;-) (And I would end up with something worse than my 7-year-old Pentax.)

[webwit]

It's alive!

Case codes are 4 characters, most center aligned.

001.jpg (131.33 KiB) Viewed 7386 times

002.jpg (117.69 KiB) Viewed 7386 times

003.jpg (137.69 KiB) Viewed 7386 times

004.jpg (137.42 KiB) Viewed 7386 times

[Daniel Beardsmore]

Fascinating! See your 3A71 that's off-set to one side? That's in Alps.tw's gallery as type T4. T4 is basically a later Type IV simplified with a copper leaf. Also, Type IV (Alps.tw code OA1) shows a switch with code 1B25, and you have, there, 1A25. I'm now convinced that Type IV moved from steel leaf to copper leaf.

Any chance of a photo of the internals?

I'd like to verify whether yours are Type IV or Hua-Jie. If they have Hua-Jie internals, then the Type IV will be an older Hua-Jie, and Hua-Jie would have kept the same shell moulds. If they're Type IV internals, I shall go back to Monterey and ask if they bought from someone other than Hua-Jie.

We're getting close to cracking Type IV, my Number 1 Mystery Switch.

[Daniel Beardsmore]

I'm also curious about my Nan Tan board — the switch internals are almost Type IV (the click leaf has angled corners, which is strange), but the numbering is plain numeric.

WTF switch 3.jpg (245.14 KiB) Viewed 7374 times

I also managed to destroy the contact mechanism and tear a hole right through the shell trying to get it apart!

WTF switch 1.jpg (87.89 KiB) Viewed 7374 times

WTF switch 2.jpg (330.98 KiB) Viewed 7374 times

I also severed the end off my old plastic letter opener in the process, which is a nuisance, as it was an ideal tool for opening switches, especially Cherry MX, KPT and Omron.

I suspect I have another clone from someone else, and one that's incredibly hard to get back together (the support grooves in the shell are too tight).

(I did manage to open and reassemble one before, twice IIRC, but this switch wouldn't play fair. I suspect they're all clones of APC, and APC switches are a doddle to work on.)

[webwit]

I dare you to make something like this for Alps.

Spoiler:

[002]

The human eye can only discriminate up to 10 million colours though
Seriously though If I had to guess I'd say such an Alps timeline would be similar in size to that image - a bit longer though obviously.

[Daniel Beardsmore]

That's a long way off. And it's more complicated. For example, KPT has a click leaf copied from Alps, but a contact mechanism based on … well, the switches that are branded SMK, NEC and Maxi Switch and we don't know who designed them. All companies who are believed to have designed their own switches, so any one of them could have been responsible for that design. A design with a case that's very similar to vintage tee mount Alps ... (which may have introduced the switchplate design also used by Omron and Mitsumi switches as well as Alps CM)

Any dates that I do know, go here: [wiki]Keyswitch timeline[/wiki]

[Daniel Beardsmore]

Hm.

Alps.tw type OA4 has an "83" shell, exactly the same numbering scheme as my 82 above.
T9 (a weirdo switch) has the same numbering as T1, and a Xiang Min rear contact plate.
T8 is a yellow switch with the same numbering as T4 and OA4.

There's a huge amount of crossover between a number of these designs. It almost seems as though Himake (Hua-Jie's switch factory) was burning through design ideas.

I'll see if Channing knows.

[Daniel Beardsmore]

Here's my badly-drawn diagram. Going to send this to Channing along with Alps.tw's photos:

Hua-Jie and Xiang Min switches.jpg (345.88 KiB) Viewed 7333 times

Nothing is conclusive, but every one of those switches has a very visible tie to at least one other.

[002]

Shit Daniel I think I lead you astray this morning on IRC. I believe that I said the Monterey I was typing on had the same number format as webwit's. What I meant to say is the *minitouch* I was typing on had that number format...sorry

In any case, here is a picture with a few keycaps popped off. This morning I randomly chose the Num Lock and the hyphen keys which happened to have the 4 digit numbering. Seems the neighbouring keys aren't quite the same...

[Daniel Beardsmore]

Pretty sure I knew you meant MiniTouch (which is of course a Monterey board), as you confirmed to me that yours, unlike webwit's, doesn't have the yellow top row switches.

Your photo is really interesting — you have the #A## numbering, and the ## numbering side by side. Again, this means that either Hua-Jie did both types, or that Monterey used two vendors (which, if we have reason to suspect this, I'll just ask them).

Do the two types have the same internals?

[002]

Yep, they seem to be the same.

[Daniel Beardsmore]

Now I might have to send webwit his keyboard back :-) Or I'll keep it, minus one spring :)

I forgot to put the numbering on OA4 on my diagram above (not the only mistakes!) but OA4 is shown with your numbering (and a different style), your contact mechanism, and a steel leaf. Yours is the copper leaf version. Several designs appeared to go from steel to copper leaf, including APC, Type IV, and the MiniTouch one.

Some months back, I speculated that all these four-tab clones were "XM" switches, whoever XM were. Strictly speaking, this is false, but I think my hunch may have been right from the start: Hua-Jie looks to have been the big cheese in clicky switches in the 90s, since all the common clones appear to be theirs. (I have no expectation of cracking all the rare ones, like ATW and D-TRNS, though; with that said, ATW internals are a lot like KPT … including that unique click leaf design)

Hua-Jie samples look like they're on, too.

Great photo, most grateful!

[webwit]

In other news, theoretical physicists have at last bound together the Higgs Boson, the eventual destruction of the Universe at light speed and the Alps vortex. Daniel, unaware of this (and CERN unaware of Daniel), is just traveling an alternative path to proof the same thing.

[002]

I think I've fked something up pretty badly because although I can get these two switches back together, they don't activate any more. I think was a little rough with the two leaf plate things at the back. If I press them together, a keypress registers. I didn't realise they had such tight margins

[webwit]

^ As a collector I love switches... from the outside.

[Daniel Beardsmore]

I think I've fked something up pretty badly because although I can get these two switches back together, they don't activate any more. I think was a little rough with the two leaf plate things at the back. If I press them together, a keypress registers. I didn't realise they had such tight margins :(

Those are the worst type to reassemble: the two contact plates have a groove each, and they tend to want to go into the wrong groove. Very stiff, and very fiddly, but it's doable. Hard to say more without looking at the switches.

[Daniel Beardsmore]

MiniTouch from webwit arrived. Surprised how easy the Hua-Jie switches are to get apart, once you know how :) Remove three keycaps in a row, lever away the T retainers front and back, and then grip the switch and pull — you need the keycaps cleared to get a good grip. Reassembly is harder, though, with these; we shall see.

Findings:

All clone shells are virtually the same from the top, regardless of the manufacturer. The only difference is the mould numbering.

On the inside, comparing all the structures that support the slider and leaves, the upper shells are as follows:

• APC — no match
• Normal Xiang Min — no match
• MiniTouch Hua-Jie "Type II" and my Nan Tan board Type IVs — identical aside from a minor change to support the narrow click leaf (the latter has the purely numeric codes in the moulds as seen side by side with alphanumeric codes in 002's MiniTouch)
• Xiang Min KSB-LE and MiniTouch Hua-Jie — virtually identical (one small leaf support block is taller)

By identical, I mean every last angle, dip, chamfer, everything. 100% spot on.

Sandy believes that the plastics and the internals come from different companies and different factories; the plastics manufacturers (Hua-Jie, Xiang Min) would be subcontracting out to internals manufacturers. Doesn't explain why Xiang Min labelled their samples with Hua-Jie part numbers though.

In any case, I feel pretty confident that Type IV = Hua-Jie.

Maybe the subcontracted internals explains why they vary so much.

As for white vs yellow — the springs are different. The yellow switches have far more turns in the springs. I can't gauge the difference by feel.

Springs in the two types of switch

Hua-Jie spring comparison in the SIIG MiniTouch.jpg (217.99 KiB) Viewed 7105 times

[002]

Nice work and great pics. Your photos are getting better every time.
What's your setup?

[Daniel Beardsmore]

7-year-old Pentax Optio S5z ultracompact (5 MP) and a white plastic windowsill. How close I can get to the subject varies wildly from day to day, depends what mood the camera is in. Maybe once I have the ultimate switch collection I'll donate it to a keen photographer. I just suck too much to do this properly.

[Daniel Beardsmore]

Yellow vs ivory:

Ivory switches in the MiniTouch are around 4 mm travel.
Yellow switches are around 3 mm travel.

There are small protrusions in the base that stop the slider; in the yellow switch, these are 1 mm taller, to shorten the throw. One of the springs is also longer. I just noticed this reduced travel when I tried to reassemble two of the switches (which is really hard!) I thought the switch wasn't assembled right, but I compared it with the others and found that none of the yellow ones would fully depress, by design.

As to why — don't know. Maybe it will become clear when I finally reassemble it.

[Daniel Beardsmore]

OK, samples from Hua-Jie arrived.

Internals are the same as Xiang Min, sliders are both ivory (not green for the linear as per Xiang Min), and the numbering is very similar to Xiang Min's KSB-LE, i.e. the one that we think has Hua-Jie internals.

I've had one more go at trying to get this confirmed with Hua-Jie, but the evidence is strengthened that they're related.

[gmjhowe]

Hi all,

I have some pictures that I took, partly to show off, but also with the aim that they could be used on the wiki.

You can find them here -
iMate ADB adapter - http://deskthority.net/photos-videos-f8 ... t5300.html
AEKii internals - http://deskthority.net/photos-videos-f8 ... t5297.html
Which also includes this gif I made showing off the mechanism of the locking Alps switch -

CST KidTrac Trackball mouse (external and internal) - http://deskthority.net/mice-other-input ... t5271.html

Obviously I license these to be used on the wiki.

[Daniel Beardsmore]

OK, here we go. I noticed that my samples were labelled with "(2)" in the part number. Paydirt :–

AK-CN2/AK-DN2 — Hua-Jie as found in the MiniTouch
AK-CN2(2)/AK-DN2(2) — Xiang Min-style version