deskthority

Hey guys, mechanical switches are great... but understanding how they work is giving me a headache!

This is my first post so hello everyone I'm enthusiastic about old keyboards and is currently working on a keyboard just like this :


My switches are SMK second generation, "circled cross" Cherry MX mount variant.

On my way toward new knowledge I opened one of those mystical black boxes and spent hours putting it back together ; And I still don't know how switch contacts work.

So there's a circuit that gets closed when you press the key, right, but I don't get how it's done.

---

In this case for example:

• How does the right copper wire interacts with the flat metal leaf to its left ?
• What about the long slightly u-curved prong to the very left ?

Those two pieces are soldered to the PCB and interact with each other to close the circuit but I don't really understand how.



---

I haven't found a video or gif or schematic that explains it.

I mean this schematic looks like the right stuff to answer my questions, likewise, this page comes close but still not quite there

I don't get what's happening dynamically:

• Where do they touch ?
• Does the metal contact leaf springs toward the wire and is prevented/allowed by the plunger movement ?
• Or does the plunger going down pushes the two pieces in contact ?

Any help appreciated!

From what I recall the metal leaf is actually behind the prong and is allowed to move forward when it clears the slider, but my memory is a bithazy. I also haven't been able to reassemble the switch I took apart xD . I do plan on doind a teardiwn video on these sometime in the future though.

Oh right..! In that picture it would be on the wrong side.

The bump would allow it to move toward us or toward the back of the screen depending on the position of the plunger..!

I sort of successfully got it back together but I damaged the leaf and I'm not sure it will contact the wire correctly anymore. I will try to reshape it with some brucelles pliers.

I don't get what's the use of the left-most prong though... Why does it go up like this ? It already extends to the PCB below so why rise ?

Possibly for stability reasons?

That image on the wiki (as used above) bears the description "This switch arrived jammed, and you can see that the movable contact is on the wrong side of the crossbar; this was fairly easy to repair. The switch is from an Apple Keyboard II and the mini PCB is still attached."

This design is patented, and you can read the patent here:

http://www.google.com/patents/US4866228

It's not one of the more descriptive patents nor one of the best illustrated, but it does briefly mention how to automate the assembly of these wretched things. There's even some suggestion that it's damped.

Yeah, that's what I thought, it's on the other side of the prong. Funky switches!

I do have a series of unpublished photos of the earlier series (J-M0404) including one or two of the contacts — one day I'll clean those up and upload them. The principle is the same. I have some more photos of second generation too, but I'm not quite sure on the details as those were taken a while back — I was a lot less happy with those, while the J-M0404 photos really aren't too bad.

Thanks a million @Daniel Beardsmore! I love nice descriptive texts and schematics it's going to help me a great deal!!

You're right about the description, I should have attached it. At that time I didn't understand half of the words.

I'm now wondering about the notches denoted "34" on the schematic

leaf-spring contact including at least one notch formed for obtaining plural contact points with the stationary wire contact

They're not disclosing a reason for their existence, apart from saying it's plural contact points. Wouldn't that just add some short circuit potential?

@Chyros I'm sure you're right about the part of "29" that goes up, it'd be just for stability/solidity.
I broke 29 at the base of the switch, pulling it madly with pliers; I thought that the switch would open but it desoldered from the PCB! WOULDN'T ADVISE DOING THIS.

I may cut the top of 29 to resolder it below and then again to the PCB.

I'm still unsure about the possibility of opening switches without desoldering them, I'm not seeing anything preventing it and it's straightforward once you've got the piece in hand.

I'll be providing pictures of all of this, yours are also welcome!

During all my searches I haven't come close to anything more bleeding edge than where we are now. May this thread serve as a knowledge pool for SMK 2G fiddlers

I think the plural contacts are basically their equivalent of a crosspoint contact.

It's hell trying to reassemble these switches without desoldering them. I tried and it is REALLY hard to align the wire mech and generally ends up with bending something. The best course of action I found was to desolder, remove 100% of the solder from the legs then pull it apart, when putting it back together you put everything in its place in the top housing starting with the plunger then the 2 contacts together. Keeping the 2 contacts together in the correct place when putting them into the switch housing you can use a little masking tape, I found this to work well. I was going to make a post on how to do this, maybe next week just got back from vacation.

@codemonkeymike, if you end up doing that post please don't hesitate to ping me.

I believe for now I'll try doing it the dirty ugly way but a good documented assembly is more than welcome!

When assembling the leaf spring I must admit I sighed and finally swore loudly the 10th time it sprung in my face.

EndersKeyboard wrote: ↑They're not disclosing a reason for their existence, apart from saying it's plural contact points. Wouldn't that just add some short circuit potential?

By "short circuit" I guess you're suggesting that one of them could get bent towards the crossbar and hold the switch permanently open.

The multiple-finger approach was not new by this point in time. Older switches that worked similarly include (in chronological order):

[wiki]Hi-Tek linear[/wiki]
[wiki]Stackpole switch grid[/wiki]
[wiki]Hi-Tek Series 725[/wiki]

Patents are listed for all of those types, so you may find a clearer explanation there. Surprisingly only one such patent is cited by the SMK patent.

Lots of switch types on the wiki have patents listed.

The Hi-Tek linear patent notes, "If contamination such as dust or grit should prevent an electrical contact between one pair of the contact points 21 and 22, the remaining contact points 21 and 22 on the remaining fingers 36 and 37 will still form an electrical connection between the contact members 15 and 16."

This implies that the foreign object doesn't hold all four of the contacts apart. Flexible fingers would help there, and I don't know how that interrelates with bounce. Gold contacts are used in part to prevent bounce, as gold is soft and it absorbs impact (according to some other patent I read).

Certainly the SMK contacts didn't seem particularly flexible, so I'm not sure what the specific mechanical objective is. I'm guessing a long flat surface might be considered too easy to go out of alignment, causing virtually none of it to make contact.

Multiple contact points themselves aren't unknown – various Alps clones have two or three contact points and two or three corresponding fingers, and the new iRocks switches (Cherry MX mount Alps clones) have a three-finger arrangement. The only Alps clone patent is for a classic switch with only one finger; Tai-Hao don't recognise the patent so it could be Himake's — old Tai-Hao and Type OA2 are almost identical.)

Crosspoint also has very low current capacity, so the more contact points you have, the more current you can pass. Not an issue for keyboards, but potentially of more concern for use in machinery.

Very well written explanation Daniel. I'm glad I asked!

It all makes sense.

Old keyboard are reputed for their high current input aren't they ? So a large flexible contact surface would be good.

EndersKeyboard wrote: ↑Very well written explanation Daniel. I'm glad I asked!

That's what he's officially here for. That and looking through my lousy wiki pages.

There is also a glob of some sort of thick grease on the contact which I would think is there to prevent bounce. Only a guess though but on Monterey Blues with the Alps Mount, they have the thick grease on the contacts but none on the click leaf on the opposite side leading me to believe the grease is not for lubrication. Again only a guess.

Pretty sure it's there for lubrication.

EndersKeyboard wrote: ↑Old keyboard are reputed for their high current input aren't they ? So a large flexible contact surface would be good.

I'd have guessed that the onboard microcontroller would be responsible for the current draw. With high grade gold contacts I don't imagine that there'd be a need to raise the voltage in the switch matrix to ensure that reliable current is passed. You'd want to ask someone like hasu if you want more details on the circuitry implications for conductive switch implementations.

I'm not "officially" here at all. That's assuming I really exist — nobody here has ever seen me …

Maybe I'm seebart's absurd sockpuppet.

Whether you're the fruit of my imagination or someone else's I'm still glad about the pieces of information I've gathered.

I tested my keyboard and it works perfectly, I just need to resolder the part of the switch that I broke: No more "Inser." key for the time being. I'll get some pics.