deskthority

Not in any particular order, we have:

Micro Switch

The most common switch is now confirmed to be [wiki]Micro Switch SD Series[/wiki]. SD series covers both the keyboards and the switches, and the switches finished up as 1001SD Series, but were also referred to as SD16 Keyswitch Modules. For now, "SD Series" will have to suffice as not enough documentation exists to resolve this confusion. The reason why "SD" never appears on the switches is that they are only labelled with the unique part of the part number. So switch 1001SD4A2K is only labelled "4A2K".

Documentation found by Mark Fisher and passed to Al Kossow of Bitsavers also contains a breakdown of most part number codes. 4A2K for example is: momentary action (4), sloped keyboard (A), 0.556 N/2.0 oz (2), timed repeat (K).

It's also clear that the dual-magnet type is [wiki]Micro Switch SW Series[/wiki] (per both the switch and keyboard part numbers), and SW seemed to be followed by [wiki]Micro Switch SN Series[/wiki], which has yet to be opened for examination to see if it still has has the two magnets. SD and SN were the two types still around at the end.

The reed switches are likely RW Series as that's what's written on them.

SMK

Thanks to a previous Bitsavers find, we have the brochure to snuci's SMK keyboard. This also contains the series name ([wiki]SMK J-M0404 series[/wiki]) and basic specifications for what was previously called "SMK vintage linear". Internet searches indicate that part numbers look like J-M0404#nn, where the range …#01 to …#10 is confirmed. This range is large enough to hold most of the known switches (at least 13 types confirmed so far, 11 of which are depicted on the wiki) and while it's not proven that every type fits into this range (short/tall, upright/angled, standard weight/low weight, normal/low-friction, live/dummy, sprung/unsprung, keystem/no keystem, momentary/alternate action) certainly the upright stem switches are included.

One part number exists for the second generation (KKM0611-0001), but SMK are not able to locate a shred of data on any switches and they appear unable to derive a series name from the part number. KKM series is possible, but SMK part and series names are a little odd.

ITT

This is snuci's find: discovery of [wiki]ITT ETL18[/wiki]. This is likely to be a whole series, as Oxygen list ITT part ETL18H1X1GC, but it's extremely rare, with one French computer being the only known instance. It's a curious design, being a cross between buckling spring and snap-action, using a single lateral spring instead of a stacked-spring arrangement. For a tactile switch it's very smooth and pushes back with strength. At this point, experience with it naturally very limited!

Futaba

We now have some identifier codes for a variety of Futaba switches; these codes may be part numbers, models, series or sub-series:

• MD-4PCS — standard momentary complicated linear (this one we've known for years)
• MD-4P — depicted as Sandy's black shell bar mount version of the above
• MD-7 — illuminated complicated linear
• MR-6C — simplified linear, both cruciform and bar mount
• ML-3-CM — basic low-profile linear
• ML88 — also looks to be basic low-profile linear
• MA41/MA42 — clicky switch, non-illuminated (MA42 is the heavier weight, reported as non-tactile)
• MA41JF — part number listed at Oxygen Electronics
• MA72/MA72 — clicky switch, illuminated (MA72 is the heavier weight, reported as non-tactile)

Although no documentation has been found, it is clear that the switch series are centred around two-letter codes: MD for complicated linear, MR for simplified linear, ML for low-profile linear, and MA for clicky.

MEI

A phone conversation with Robert Twyford Jr resulted in the correct designations for MEI switches: [wiki]MEI T-5 series[/wiki] ("WEAB") and [wiki]MEI T-15 series[/wiki] ("mini-elastic"). T-5 (and its precise spelling) was later confirmed with a catalogue excerpt obtained from RS, while no documentation on T-15 has yet to be recovered. The catalogue covers part of T-5 series: tall upright keystem (T-5C) and flat-top illuminated (T-5L). The box-top illuminated may still exist, and no explanation so far exists to cover the single-gold-plated and dual-gold-plated terminals or the meaning of "WEAB".

Sasse

Part number "200MN" remains unverified, but the Wayback Machine yielded up the series name, of series 25/Baureihe 25. What's most remarkable about this is the photograph on the old Sasse website of a series 25 switch with a four-legged black box inside. Although no mention of this is made anywhere (and Sasse themselves are hopeless) it looks very much like the Hall sensor ICs that RAFI use, suggesting that Sasse also made Hall effect switches.

FR Electronics

[wiki]FR Electronics RSM 82-1A[/wiki] was terrycherry's discovery: a British reed switch very similar to the Polish DOLAM switch. Examination concluded that it supports DPST operation and illumination. Ownership of the product range transferred to Reed Relays and Electronics India Limited, and indeed RRE still have a very small stock of the illuminated version: [wiki]RRE RKS 80-1AL[/wiki]. Sadly they have no documentation of any kind, not even the series name.

General Instrument

mr_a500 picked up on the unusual SKCC switches in his TI-99/4A but never desoldered one, so nothing could be concluded. Via STRONIC in France, these were found to be [wiki]General Instrument Series S950[/wiki], an adaptation of SKCC series that accepts Clare/Pendar keycaps. They are all "GI" branded on the base, suggesting that the Clare-Pendar brand was being phased out by this point. The series mirrors SKCC in having both the latching switch and the eye-stalk illuminated design, as well as angled keystems never seen in an Alps keyboard. A number of changes were made to the switch over its lifetime.

Hi-Tek

What do you do when you break a Hi-Tek switch grid? You can't simply swap out a switch, as there aren't any discrete modules.

Instead, according to [wiki]computer service manual findings[/wiki], you use special Hi-Tek tools to repair the keyboard. Instructions were provided for desoldering and removing the contacts and using the special tools to insert a new pair of contacts into the switch and then replace the spring and slider. This process is clearly illustrated, and Hi-Tek part numbers are given for virtually every component of a momentary switch.

Datanetics

2016 marked the end of an eleven year period between Sandy publishing details on the Fluke Y1700 keyboard, and getting the switches identified as [wiki]Datanetics DC-60 series[/wiki]. At the same time, the Datanetics diaphragm switch was identified as [wiki]Datanetics DC-50 series[/wiki] (the operation of which is now finally understood: it uses a sliding leaf spring to introduce hysteresis via lost motion). Meryl Miller didn't think DC-70 ever made it into production, but apparently it did; it may be the same switch as Mike Muller's [wiki]ITT snap-action array[/wiki] (known from some Apple II computers) or it may be something wholly different again. At least [wiki]Apple hairpin spring[/wiki] can't be blamed on Mike Muller! (The improbably-named Gary C Butts was responsible for that joke.)

Others

• The AWC/WEC mystery appears to be solved. AWC seems to be Atari Wong.
• Alps vertical plate spring appears to be [wiki]Alps SKFF series[/wiki], based on a label in a typewriter; there's still just enough time to solve Alps plate spring similarly before the year is out.
• The [wiki]Clare-Pendar[/wiki] story is now a lot better understood.
• The metal contact Clare-Pendar switches appear to be [wiki]Clare-Pendar Series S840[/wiki]; this is the one type for which Electro-Mech have not retained the documentation.
• SKCC has been found with a second keycap mount, a tall side-angled bar mount keystem, in Sweden ([wiki]Alps SKCC Side-Angled Cream[/wiki]).
• Stackpole single-key modules are now known to exist, in the [wiki]Lear Siegler ADM 11[/wiki] — these interlock just like Hi-Tek dovetail modules but using a different attachment design (to work around the dovetail patent presumably).
• The [wiki]Stackpole discrete switch[/wiki] has finally been sighted in a keyboard.
• The meaning of codes like "A12" in mould numbering has been explained by Tai-Hao as the mould cavity identity; A12 means the 12th cavity of mould A. The encircled numbers in Alps switches are likely batch/shift codes set using drop-in rods, leaving the top-right codes (e.g. "12D") to indicate the cavity.

While that is a lot, it is certainly not everything that we have uncovered and learnt in this past year. What else have we discovered this year that I've missed?

E&OE.

Wow, that's excellent work, thanks for this update - it's easy to miss how much progress has been booked.

Excellent summary as usual. Were you going to mention Cherry and the ADDS terminal manual or are you still working on that?

Very good work and good idea to post this since many DT users don't really follow the wiki. The Micro Switch findings are the most imporant to me for personal reasons but it's really all equally valuable knowledge!

snuci wrote: ↑Excellent summary as usual. Were you going to mention Cherry and the ADDS terminal manual or are you still working on that?

Cherry is still ongoing confusion, but it's true that there is now reasonable evidence to suggest that M4/5/6/7 are simply different subsets of the same series (now called [wiki]Cherry gold crosspoint[/wiki], a compromise name based on a variety of terms used in old catalogues). Unfortunately it's still not clear what the distinction was between them as that itself seemed to be forever changing.

No doubt there is more that I've forgotten or never became aware of. For example, Jacob #1 has measured the force curve of the fat RAFI switches and they are exactly what I expected to see from RS 76 C. I passed that information to RAFI to see if they could ID the switches, but they never responded. (They also have the same Berlin wall–era HFO Hall sensor chips in them.)

That's it? Nobody else remembers any discoveries throughout the entire of 2016? The forum is mostly asleep now but I didn't think everyone slept through 2016!

I found the "daylight robbery" Omron switch in 2016, and better pictures have now emerged now that I've placed an order for some. The family is B2R, which arcade aficionados may be familiar with. It turns out that in addition to the "G1" panel switch, there is also an "M1" version (simply marked "M") suitable for keyboards (in the same vein as the different versions of Tokai MM9).

G1 is a reed switch (there are some disassembly photos of one), so I assume M1 is too; I had a feeling it would be magnetically actuated from the shape of the slider. Although I only just learned of the family identity a little earlier, the switch itself is a 2016 discovery.

I found a huge number of small things that hadn't been documented yet during last year's reviews, but the one that stood out for me most was when I found those tactile magnetic reeds which show Fujitsu invented tactile leaf springs rather than Alps.

Thanks for the round-up about keyboard related discoveries in 2016, definitely need workers to organize these knowledge.

In summary, with mostly just MX-boards and rubber-dome boards in the stores, and not a single line of the 2016 round-up is about those, I can tell the golden era of key switches is already bygone.

Chyros wrote: ↑I found a huge number of small things that hadn't been documented yet during last year's reviews, but the one that stood out for me most was when I found those tactile magnetic reeds which show Fujitsu invented tactile leaf springs rather than Alps.

Poor Old Mr. Alps

The idea of a near-zero-motion modular actuator assembly was not their idea either. In fact, the membrane keyboard came first! The oldest known near-zero-motion actuator is that of Datanetics DC-50, and it was simply a way to put their even older (1969²) membrane system ([wiki]Datanetics elastic diaphragm array[/wiki])¹ into a discrete switch. DC-50 is not a true membrane system: it uses true metal contact, in a sandwich:

1. Stainless steel backing plate
2. Gold-over-nickel-plated cadmium copper alloy switch contact/terminal
3. Mylar spacer
4. Second switch contact/terminal
5. Mylar cover

At the top left, you can see a production version. I got four of these, and one was in bits, so I finished off the disassembly. Some sort of blue adhesive holds the steel plate on, but the Mylar parts are seemingly affixed using invisible adhesive. (The brochure simply refers to "thermo-activated adhesive".) Below this is a pink actuator, which will look very familiar.

Top centre is a chunky cardboard-backed experiment, and to its right are two beige versions, one in bare form and one with the lead-outs stamped to form strengthened switch terminals (except one has snapped off at some point in the last 44 years). Both variants are fully working, but the beige one was quite stiff. Below these is a golden-brown version that doesn't work. Below all of these are alternate colour actuators.

In 1972, while this was being developed, the technology was licensed to Futaba. Meryl's notes don't say precisely what was licensed, but this may be where [wiki]Futaba MD series[/wiki] came from, since MD series is the most similar design, albeit without the hysteresis present in DC-50.

Alps SKCC seems to be from around 1979 onwards, so unless their switchplate goes back further, they appear to have come into this game fairly late. Their design is of course very different to both that of Datanetics and Futaba, but the overall idea is exactly the same: sealed or partially sealed contacts with near-zero movement that are operated by a perpendiculator. (A new word: something that converts motion in one axis to motion in a perpendicular axis. Also functions as a motion reducer, which I wanted to call a motion condenser but I settled on [wiki]motion reducer[/wiki].)

¹ if anyone actually understands their membrane system I'll be well impressed. I don't think even Meryl still remembers. It's ludicrously complicated, with about seven layers, and it looks like it was simplified over time to something far more akin to what we know of as a membrane keyboard today
² the wiki says 1968 but Meryl's notes say 1969 — I need to double-check where I got 1968 from

Daniel Beardsmore wrote: ↑That's it? Nobody else remembers any discoveries throughout the entire of 2016? The forum is mostly asleep now but I didn't think everyone slept through 2016!

Why, is this a quiz? I assume HaaTa or someone else looked at these magnetic Nixdorfs long before I did:

wiki/Nixdorf_SM_8890

Daniel Beardsmore wrote: ↑

Chyros wrote: ↑I found a huge number of small things that hadn't been documented yet during last year's reviews, but the one that stood out for me most was when I found those tactile magnetic reeds which show Fujitsu invented tactile leaf springs rather than Alps.

Poor Old Mr. Alps

The idea of a near-zero-motion modular actuator assembly was not their idea either. In fact, the membrane keyboard came first! The oldest known near-zero-motion actuator is that of Datanetics DC-50, and it was simply a way to put their even older (1969²) membrane system ([wiki]Datanetics elastic diaphragm array[/wiki])¹ into a discrete switch. DC-50 is not a true membrane system: it uses true metal contact, in a sandwich:

1. Stainless steel backing plate
2. Gold-over-nickel-plated cadmium copper alloy switch contact/terminal
3. Mylar spacer
4. Second switch contact/terminal
5. Mylar cover

IMGP3196.jpg

At the top left, you can see a production version. I got four of these, and one was in bits, so I finished off the disassembly. Some sort of blue adhesive holds the steel plate on, but the Mylar parts are seemingly affixed using invisible adhesive. (The brochure simply refers to "thermo-activated adhesive".) Below this is a pink actuator, which will look very familiar.

Top centre is a chunky cardboard-backed experiment, and to its right are two beige versions, one in bare form and one with the lead-outs stamped to form strengthened switch terminals (except one has snapped off at some point in the last 44 years). Both variants are fully working, but the beige one was quite stiff. Below these is a golden-brown version that doesn't work. Below all of these are alternate colour actuators.

In 1972, while this was being developed, the technology was licensed to Futaba. Meryl's notes don't say precisely what was licensed, but this may be where [wiki]Futaba MD series[/wiki] came from, since MD series is the most similar design, albeit without the hysteresis present in DC-50.

Alps SKCC seems to be from around 1979 onwards, so unless their switchplate goes back further, they appear to have come into this game fairly late. Their design is of course very different to both that of Datanetics and Futaba, but the overall idea is exactly the same: sealed or partially sealed contacts with near-zero movement that are operated by a perpendiculator. (A new word: something that converts motion in one axis to motion in a perpendicular axis. Also functions as a motion reducer, which I wanted to call a motion condenser but I settled on [wiki]motion reducer[/wiki].)

¹ if anyone actually understands their membrane system I'll be well impressed. I don't think even Meryl still remembers. It's ludicrously complicated, with about seven layers, and it looks like it was simplified over time to something far more akin to what we know of as a membrane keyboard today
² the wiki says 1968 but Meryl's notes say 1969 — I need to double-check where I got 1968 from

Wait, how is any of this a tactile leaf spring though? Oo

seebart wrote: ↑Why, is this a quiz?

No. I just don't remember everything that happened last year.

Chyros wrote: ↑Wait, how is any of this a tactile leaf spring though? Oo

Who said it was?

Daniel Beardsmore wrote: ↑

seebart wrote: ↑Why, is this a quiz?

No. I just don't remember everything that happened last year.

Obviously! Who would? Good thing we can always go back to the respective threads, although in some of the very old DT threads I found the image links to be gone.

That's what the wiki is for.

Daniel Beardsmore wrote: ↑

Chyros wrote: ↑Wait, how is any of this a tactile leaf spring though? Oo

Who said it was?

I was talking about how Fujitsu used tactile leaves before Alps did (as everyone assumed Alps invented the practice, don't know if it's still in the wiki), and then you replied to that with:

The idea of a near-zero-motion modular actuator assembly was not their idea either. In fact, the membrane keyboard came first!

But I don't understand what the connection is with tacile leaves oO .

OK, what Alps switch first used the tactile leaf?

Can I make a wish for keyboard research 2017?

What did Matias do to the [wiki]Alps_SKBL/SKBM_series[/wiki] design to 2.5x it's lifetime rating from 20m to 50m?

Where do these switch ratings come from anyway and who verifies them?

Wodan wrote: ↑Can I make a wish for keyboard research 2017?

What did Matias do to the [wiki]Alps_SKBL/SKBM_series[/wiki] design to 2.5x it's lifetime rating from 20m to 50m?

Where do these switch ratings come from anyway and who verifies them?

Good question, I am generally quite skeptical regarding these ratings...

seebart wrote: ↑ Good question, I am generally quite skeptical regarding these ratings...

Especially when they all SOMEHOW exactly match the rating of Cherry switches, who gold-plate their contacts for a 50m rating ...

Wodan wrote: ↑Can I make a wish for keyboard research 2017?

What did Matias do to the [wiki]Alps_SKBL/SKBM_series[/wiki] design to 2.5x it's lifetime rating from 20m to 50m?

Where do these switch ratings come from anyway and who verifies them?

I asked Matias this for the Tactile Pro video. They use gold-plated copper contacts (originally silver-plated steel, which I also have a few of). Most companies, even small ones, test them in-house.

Chyros wrote: ↑ I asked Matias this for the Tactile Pro video. They use gold-plated copper contacts (originally silver-plated steel, which I also have a few of). Most companies, even small ones, test them in-house.

That would make a lot of sense. Did they release a new revision recently?
The ones I got look like that:
wiki/Matias_switch#Gallery
And that is still super silver ...

One characteristic that you see associated with lifetime is contact resistance. It's not something that I ever document specifically, but it's not secret. RAFI RS 74 M and 76 M for example, has contact resistance of ≦ 100 mΩ when new, and ≦ 3 Ω after operating life. (RS 76 M illuminated starts out at 200 mΩ. And yes, the catalogue uses "≦" instead of "≤".)

Sejin cited, for MD-4P series, "Contract Resistance" [sic] of "100m or less", and at life end, "5 or less" (ohms, one assumes!) They also cite maximum bounce time of 2 ms when new, and 5 ms at end of life.

Switches wear out. The parts will physically degrade with use, and they will go out of tolerance. This will result in the switch struggling to pass current, or to bouncing too long.

Latching switches are invariably given substantially reduced lifetime, typically an order of magnitude less. For Cherry Serie M7, the life expectancy is 20 · 10⁶, while for the alternate action version, its 5 · 10⁶. For RAFI RS 76 C, it's 100 M reduced right down to 100 k for latching.

Datanetics tended to not provide the rating for alternate action switches, but the ITT Datanetics brochure lists DC-60 as "15 million actuations" for momentary action and "50,000 actuations" for alternate action. DC-50 was still cited as having the extremely long lifetime of "100,000,000 actuations", but a reduced rating for the alternate action version was given at "50,000 actuations" just as with DC-60 (which makes sense as both of them use the same design: a thin metal wire).

Meryl Miller designed the first alternate action DC-50 switch, with a fairly robust plastic arm, but it was ultimately rejected for having too short a lifetime, and the metal wire version was put into production. I have both versions here. It's curious that this 50 k operations version was the better version!

You can make metal thicker, or choose another material, but then the feel will change, and then you have to adjust something else to compensate, and then work out whether that component will fail sooner as a result. That's why I have a feeling that the change in feel of white Alps was due to the actuator: I suspect that the material was revised for some reason. The spring material was changed, as were the plastics, so the actuator may have also been changed, possibly to a cheaper alloy.

MEI Sabrecoil used cheaper silver contacts instead of gold contacts, under the basis that modern circuitry in computer keyboards could be made to handle longer bounce times. Gold is a soft metal, and presumably absorbs contact impact more cleanly, while a harder material may rebound more forcefully giving a longer bounce time and leading to chatter. The person who supplied the Sabrecoil images for the wiki noted that the keyboard suffers from excessive bounce, so it looks like MEI could have cut costs just too much. Alternatively, the parts may simply be life-expired. (For an Atom keyboard, I find that rather unlikely as it was a hobbyist computer!)

I see from comments already made while writing this that a change was made from silver to gold, which nicely mirrors the change made by MEI.

You may find this interesting:
It's the Linus Tech Tips factory tour of Cherry, and you can see keyboard test equipment. This is the sort of machinery needed to check switch lifetime.

Thanks very much for the insight Daniel, the resistance part is quite a nice clue.

I have a nice bench multimeter which allows 4-point resistance measurements, would be fun to compare the resistance of MX switches of various age and brands ...

Wodan wrote: ↑

Chyros wrote: ↑ I asked Matias this for the Tactile Pro video. They use gold-plated copper contacts (originally silver-plated steel, which I also have a few of). Most companies, even small ones, test them in-house.

That would make a lot of sense. Did they release a new revision recently?
The ones I got look like that:
wiki/Matias_switch#Gallery
And that is still super silver ...

They are the silver-plated steel ones. The new copper ones look coppery, there's really no mistaking them.

The Matias pictures on the wiki are from 2013. I've never added any photos of the newer switches.

The reflectiveness of the shell makes them something best left to someone else to photograph, but I did post this picture in my own switch collection:
Within my own collection, only the linear switch uses the new materials. The grey slider quiet click switches are still the old materials. I did wonder what the change was about.