Factual updates and other interesting changes are reported here. Adjustments and tweaks are not listed.
Updates for 2021
Sunday, 1st August
Because Magsat “flying magnet” switches are so unusual, and so difficult to explain, I felt that such an interesting design justified a set of illustrations showing how the switches operate. A set of diagrams to this effect is now posted to the Magsat page.
As with all my other diagrams, they are in SVG format. In Waterfox, Waterfox Classic mostly manages to render them OK, but all the arrowheads are missing (which is a new bug). Microsoft Edge draws the arrowheads, but it renders numerous lines in the wrong place or in a “low resolution” look, leaving a fuzzy and indistinct image, and Chrome will likely do the same. The inability to get a simple diagram rendered properly is the icing on the cake of an afternoon and evening of Inkscape 1.1 constantly crashing, as it does on most work now. The 21st century is a long way away yet.
Wednesday, 28th July
Freshly added to the list of unseen switches is the Magsat “flying magnet” system. Although these switches can be seen in advertisements, they join a growing array of switch types that have yet to be encountered by anyone in a keyboard. With no avid vintage keyboard collectors presently active, it may be quite some time before such a keyboard ever shows up. Magsat Corporation itself has all but vanished into oblivion.
Monday, 26th July
Lots of rummaging around on the Internet Archive in the last few days has turned up a number of interesting finds:
- Oak Series 475 is older than expected, having been found advertised as far back as 1973, only a few years after Oak Series 400 is likely to have been introduced
- The first known advertisement for Micro Switch SD Series with a verifiable date, being printed in June 1974; the advertisement does not state the readiness of SD Series however
- Licon Series 555 was advertised briefly in Computerworld in December 1972, months before any of the patents for it were filed; discovered examples are not presently known from before 1975, as early ITW keyboards are rare to the point of nonexistent
- Maxi-Switch 3100 Series is now conclusively identified via a 1972 advertisment in Electronic Design, where it is clearly depicted; this and a clearer advertisement also in Electronic Design show the whole of Maxi-Switch’s 2700 Series photograph, which can be seen to precisely match the Amphenol/Bunker-Ramo patent
- An advertisement in Electronics Australia from January 1973 depicts Mechanical Enterprises SS-10 and SS-11 switches from their Mercutronic family; these are the switches from the “intermediate” mercury contact patent that were still awaiting discovery of their identity
- An advertisement the Electronic Engineer from September 1971 for General Instrument’s AY-5-2376 single-chip keyboard encoder gives us a new low water mark for the introduction of off-the-shelf scanning encoders
Wednesday, 21st July
Strange fact for the day: according to a former Key Tronic employee, Key Tronic’s Butterfly™ switch never entered sale. He is not at liberty to disclose its characteristics, only that it turns out not to be a name for any switch type that we have ever seen. (There is still no answer to the question of how it came to be touted as a linear complement to an existing tactile design, as that does not seem to be in line with reality.)
Tuesday, 20th July
Revisiting the Electronic Engineers Master catalogues to look for anything that got missed on initial inspection, turned up Amphenol’s 601 Series of keyboard switches. Details were vague, but the specifications indicated that reed switches were most likely. More curiously, the advertisement suggested that Amphenol was part of Bunker Ramo, and thus 601 Series could relate to their reed switch patent. This turned out to be true: all the keyboard switches attributed to Bunker Ramo were in fact Amphenol products.
In fact, 601 Series happens to be the switch that Maxi-Switch sold as Series 2700: the Series 2700 photo matches the patent exactly, and the descriptions of the switch also match the patent and how Amphenol advertised it themselves.
Further investigation has found a single advertisement for a lower-cost Amphenol switch, with only one model: 601-M11A. This will be the double break bifurcated-contact type that was sold as Maxi Switch Series 3100. These have been attributed until now to Bunker Ramo, and more properly they should be attributed to Amphenol.
Thus, another Maxi-Switch type that one could have argued was their own innovation, turns out to be someone else’s product again!
Sunday, 11th July
Another day, another coincidence. Having only just documented the Harris HD-0165 16-to-4–line binary encoder and the intention that these would be used in pairs for keyboards, a Zbrojovka Brno keyboard appears in eBay’s suggestions box that works in pretty much the same manner. The Consul 259.13 keyboards uses a pair of TESLA MH1KK1 16-to-4–line binary encoders in conjunction with Hall effect switches. Instead of defining the output in the switch wiring, a single Fujitsu 1 kB PROM appears to contain the output codes. (There appears to be only one, but not all the chips can be clearly seen.)
The starting point for this was an eBay listing for an IBM 5576-002 keyboard, in its original box. The Alps model number is printed on the box: KFCREA087A. This model number reveals the existence of another Alps series: SKCR switches and KFCR keyboards. So what is SKCR? With the IBM 5576-002 keyboards only known to use Alps “plate spring” switches, one could take a reasonable bet that, since we know that Alps KCP/SKCP is the plate-mount “plate spring” type, then Alps SKCR would be the square PCB-mount “plate spring” type. (There is a second PCB-mount type to find an identity for.) Frustratingly there is insufficient data at present to indicate whether SKCR began life as KCR; there does not appear to be a separate utility model for either of the PCB-mount types. For now, “SKCR” is being used in accordance with the only available data.
Tuesday, 6th July
Recently, I updated my description of two-of-N encoding and in doing so, noted the three different switch arrangements that are suitable for it: dual isolated outputs (e.g. Hall effect), double-pole switches (also isolated outputs) and single-pole switches with the output split in two isolated lines via a pair of diodes. For a while I have been vaguely aware of the Harris HD-0165 16-key encoder—that has been sat in my pending pile of collected advertisements—but not how it worked. Having obtained the specifications and application note, I can now see that it is a 1-of-16 encoder. Essentially it is 16-to-4 binary encoder with input validation and strobe. For larger keyboards, two such encoders must be used together, in a 2-of-32 manner. The Harris documentation confirms that two encoders require double-pole switches or split diode-protected outputs. Unlike discovered TTL two-of-N keyboards that favour 8-input encoders, this is a 16-input variation. Of course, not all 2-of-32 codes are valid: of each input code, only one bit per encoder is permitted, as with any multiple-encoder arrangement.
This is another encoder IC yet to be found in use in a keypad; its only known use to date is in Jameco keypads.
Sunday, 4th July
Tentatively, the second production two-of-N encoded keyboard outside of Micro Switch has been identified: Bendix Flight Systems Division’s NASA control panel. With four binary encoders, double-pole switches with separately-wired poles and no matrix co-ordinate processing, it seems most likely that this is a two-of-N arrangement. This is also the first ever confirmed production mechanical two-of-N keyboard, although Maxi-Switch certainly designed such a keyboard.
Wednesday, 16th June
In February I posted a page on TESLA keyboard encoders: specifically models MH103 and MH113 single-chip contactless encoders. Few details on these parts are known to be documented. An eBay listing for a Consul 262.3 keyboard shows the complete keyboard schematics, including a clear indication of the encoder functionality: they are two-of-N encoders. Micro Switch schematics don’t show (in all examples observed to date) the encoding arrangement, while the 262.3 schematics do depict this. Confirmed two-of-N arrangements outside of Micro Switch are extremely rare, so this example is an important find. The encoding is shown in the schematics as a regular grid, but as with later Micro Switch keyboards, the encoding integrated into the switch wiring instead of using a dedicated connection grid.
The rate of discovery has slowed lately, so this was a most welcome encounter.
Monday, 31st May
zrrion sent me photographs of an Alps Low-Profile Keyboard Switches sample pack that had been archived from eBay over a year ago in April 2020. With over a year having passed, the new owner of this extraordinary find (has never knowingly come forward, and the archived images have languished all this time. I did not feel particularly inclined to act on this information, frustrated by how poorly such discoveries continue to be treated: the keyboard community as a whole has to realise the need to take research and documentation seriously. zrrion himself has posted the archived photos to a forum topic, Mechanical Keyboard Switch assortment/samples [misc Alps], and unsurprisingly the significance of the discovery is lost on everyone. (What’s more sad is that this sample pack appears to be the same one that alps.tw photographed SKCMAF and SKCMAG from: he seems to have ignored everything else in it at the time.)
Left as it stands, the information in the photographs in the forum topic would remain largely inaccessible. While it’s true that search engines could conceivably archive the text within the images, that data still needs to be extracted into textual form and presented in usable manner so that it can be indexed using search engines and readily identified in search results. In general, information gathered during research needs to be presented in a variety of ways, to make it as accessible as possible. Photographs should be captioned, with a mention of anything in particular that the reader should be looking out for; without captions, it can be difficult to determine the purpose of some photographs, as the reader will not know which details to observe. Data should be tabulated where applicable; if nothing else, this helps people who cannot understand the written material to pick out useful data. I cannot read Japanese, but I can recognise and understand various switch characteristics when placed in table form. Always bear in mind that the people perusing the material may not speak the language, and may need every clue possible to help interpret the data. Anything that helps make the information easier to search for, or pick out in search results, goes a long way to help others track down the information when they need it.
As the community appears intent to nod, smile and then forget about this discovery, I was left with little choice but to go through and document it myself, which I have now done. In summary, we have:
- A newly-discovered KKC Series capacitive keyboards, seemingly rubber dome based in some form (possibly covered by one of the Japanese patents I turned up, but difficult to say at this stage, with no access to the sample switch to inspect its structure)
- KCP Series confirmed for the plate mount types, with four model numbers; curiously, KCP alternate action is just KCL alternate action (SKCLJB) with a replacement model number (SKCPJB)
- Three more switch model numbers for KCL Series
- KEW Series confirmed, with the model number just for the white plunger type
- KEU Series confirmed, with the model number just for the white plunger type
- KNB Series is one of the spring-over-membrane types; due to the terribly low level of documentation of these keyboards and the poor quality of the photographs of the sample pack, it’s not possible to be sure what KNB is (in relation to three different styles of these switches known thus far)
The force curve is given for each type, which at some point I may decide to recreate.
Now if only I could get my hands on more of the pre-1986 model numbers!
Wednesday, 26th May
One of the most infamous computer keyboards is the chiclet keyboard supplied with the IBM PCjr in 1984. This keyboard was produced by Advanced Input Devices, now Advanced Input Systems; it appears to be an EKI (ErgoKey International) model. The replacement PCjr keyboard was another Advanced Input Devices keyboard, a virtually identical EKT (ErgoKey Truncated) model, differing only in the use of conventional keycaps instead of miniature keycaps intended for use with overlays.
Advanced Input Devices have produced more keyboards than I realised; looking back at old forum topics, it was a brand that appeared from time to time but never registered in my consciousness. What makes them interesting now is the realisation of just how far back they were producing rubber sheet (elastomer mat) keyboards, something they told me they were first manufacturer of in the States. Further details should be forthcoming from them in time, and hopefully sufficient to verify whether or not they were indeed the first. The oldest elastomer mat advertisements I have are both from 1984: one for Advanced Input Devices (rubber sheet over PCB) and several for Cherry’s Next Generation Keyboard (rubber sheet over membrane).
Details are being added to the Advanced Input Devices page as it comes to light. This should at some point include historic product literature, that they still have in their office. Curiously, all Advanced Input Devices keyboards to date have been found to be dome-over-PCB, with not a single membrane example. Elastomer mat over PCB is of course better known in the guise of BTC’s keyboards.
Sunday, 23rd May
As far as mysteries go, a five-year resolution is far from the 11-year record for Datanetics DC-60. This mystery relates to the switch type in an unidentifiable Univac-like keyboard, photos that I can see that I archived locally in 2015. The part number cannot be read, and the plungers did not match any other known keyboards, until a Xerox 860 IPS example showed up. This latter find indicated that, whatever this switch is, it was not Univac-specific.
A revisit of ITW’s patents showed a remarkable resemblance to US patent 4352144 “Capacitive keyswitch with overtravel plunger mechanism”, filed in January 1981. The patent was the right age for the Xerox 860 keyboard, and indeed these switches look capacitive. (The patent was also left out of the patents table on my ITW page.)
Finally, further digging turned up details of a Xerox 820 keyboard. This page tied the plunger to Cortron CP-4550, which was their capacitive system. That keyboard is the same model as the Xerox 820-II keyboard shown at q7.neurotica.com, whose gallery of photos lacks a critical detail: the plunger design. Putting both sets of images together, all the implementation details can be seen.
Thus, these mystery plungers are Cortron CP-4550 metal leaf capacitive, a design very closely related to Digitran Golden Touch.
There are still lots of Licon and Cortron discoveries remaining to be made, in particular:
- The mechanical types corresponding to Series 555 and 35-xxxxxx keyboards, both patented
- Confirmed examples of both FC-2550 and FC2500 such that their part numbers can be observed, along with the differences between these two series
- What Series III (ss3) switches look like
- What Series 54 (Superwitch II) switches look like
- What series the 35-xxxxxx keyboards belong to
- The original wired inductive ferrite core keyboards (possibly 55-0xxxxx)
- Another example of the old 55-1xxxxx type, that can be seen in more detail (all the photos of the one known example are all lost)
Some of these details are waiting for me in an envelope on a desk at Cortron, and one day they will be revealed … One day.
Friday, 21st May
Something that I have never given sufficient consideration to, is the origin of rubber dome keyboards. Unsurprisingly, like many other designs, the details remain obscure. However, a variety of information has come to light over time, and the history section on the rubber domes page now contains a number of examples of when various companies introduced rubber dome keyboards, or at least filed patents on them.
The examples listed so far only represent a fraction of the prominent manufacturers across the industry. Brands where no introductory advertisement has yet been collected include Advanced Input Devices, IBM, BTC, Maxi-Switch, Hi-Tek and Key Tronic amongst others.
The information to date suggests that discrete rubber dome switches came about around 1980, and rubber dome over membrane around 1984, but these dates could change hugely as more data comes to light in the months and years to come.
Monday, 3rd May
One of the most frustrating companies to track down is Pye Electro-Devices, or PED. Various details exist on their keyboard products, but not a single confirmed photograph or example has shown up. This stands in sharp contrast to fellow British manufacturer Alphameric, from which we have discovered numerous examples to date. Like Alphameric, PED is now gone; some Alphameric staff still work for Devlin (who bought out their keyboard business), but are forbidden to speak, while with PED it’s not clear what became of the keyboard line of the business.
Although PED switches were used in some BBC Micro keyboards, the actual manufacturer of those keyboard assemblies is not confirmed. We do know that PED made the console panel for the Monarch 120 digital PABX, which is the first visually confirmed example of capacitance-controlled transistor sensing. The design here is different to that of Control Devices. This Monarch example may be PED’s Series 90, but this is not confirmed.
Those details that are still discoverable are now added to the PED page. The “PED switch” write-up remains on the same page for now, as none of the information available at present offers any suggestion of their identity, only that they appear to have been available for sale separately.
Friday, 23rd April
A Sharp ZX-510 typewriter in Shallot’s collection has now demonstrated that Alps KFF series survived long enough to be split into SKFF and KFFF series. Thus, although I was incorrect to use the name “SKFF” on the Deskthority wiki based on the evidence available at the time (which I have since learnt indicated KFF series, not KFFF), this error has proven irrelevant as the series became SKFF anyway. The question now is, how long will it be before the first switch part number shows up? After all, electronic typewriter service manuals are especially elusive.
Friday, 16th April
My understanding of electronics is not great, to say the least. In the past, I specifically avoided documenting electromagnetic sensing techniques (e.g. capacitive, Hall effect) as I did not understand any of them; consequently I also avoided certain brands, in particular Micro Switch. As time passed, it became clear that all the brands that I was leaving to others to document, either because they seemed beyond my understanding or because they seemed to have a sufficiently loyal fan base to take care of that, have mostly fallen to me to research and document, as the only person willing to do so.
As time passes, I have also ventured into the electronics side of keyboards. This includes capacitive sensing, in part because some very interesting techniques have turned up, as past updates have revealed. In the last few days, I have made some improvements to coverage and clarity of the capacitive sensing page. There are still no clear details on how any matrix scan keyboards actually work; this may be addressed later as and when I encounter or seek out patents that detail the approaches taken.
The illustrations on the IKOR page are also improved, to better show how the sensing technique works and to bring them more in line with the patent. The actual implementation seems to have rapidly deviated from the patented design, but for the purposes of illustration, the page shows the approach as patented. The modifications taken in production keyboards are not explained in the detail found in the patent.
PS Wave if you are following these updates!
Sunday, 11th April
Another day, another largely forgotten vintage keyboard manufacturer. This time, it’s Control Devices, Inc. Control Devices invented possibly the most convoluted capacitive system, one that required (in addition to the variable capacitor) a transistor, a fixed capacitor and two resistors for every key. These small circuits then drove a diode matrix, which was then fed through a sense amplifier to yield adequate signals strength and a synchronisation circuit to wait for all the sense lines to settle before the output code was delivered. The small circuit within each key station provided each key with its own capacitive sensor. One of the inventors, Donald Gove, was the inventor of the IKOR self-encoding capacitive switches mentioned recently.
A couple of people have come across Control Devices keyboards, but nobody was ever forthcoming with details or illustrations. Some instances were described as “foam and foil”, which raises an interesting question of whether they moved on to more “conventional” capacitive keyboards (using a matrix scanner to centralise the capacitive sensing circuitry), or whether they merely simplified (or cheapened) the variable capacitor portion but retained the per-key capacitive sensors.
Saturday, 10th April
After Marylou Bassi-Dolmans got in contact with me regarding Clare and Pendar, I started investigating these brands anew. During my searching, I found a number of pages with details on Pendar’s capacitive keyboards, but did not get around to reading them, instead simply making a mental note of the summarised contents in the search results while I looked for more details on the history of the US Pendar. Having gone back to read up on this interesting new line of research, all those search results had vanished without trace. Even the browser history entries for prior searches found nothing. Sadly I have no recollection of what sites had those details, only that possibly it was all duplicates of the same material. It was a description of how Pendar in France—the buyout company—had set up a line of capacitive keyboards in France, something for which there is unfortunately no patent to examine.
Someone called Furieux Furet posted a photo of a single switch from a Pendar keyboard at Deskthority, and then—as so many people do—vanished without ever completing the investigation. There is not even a single photo showing the complete keyboard, and no attempt to investigate the switch workings. The switch with its external springs had a strong suggestion of being capacitive. Finally, a second keyboard has shown up with this switch, an unknown Goupil keyboard, this time depicted in considerable detail, but (from the photos to date) not identifiable as Pendar. Ricardo Guzzetti did indicate that Goupil was a Pendar customer, so it seems that this keyboard is one of theirs.
Hopefully more details will come to light one day. After sending messages to a number of ex-Pendar staff via a French social media site, not a single one has ever responded. Not one.
Thursday, 8th April
A brief page is now up for Applied Dynamics International. Applied Dynamics is the primary name for the business that also traded as Collimation Keyboards. The well-known Collimation D40.592 photoelectric encoder keyboard can now be seen marketed as the ADI Series 5000 in 1977. The advertisement also gives some brief specifications.
Saturday, 3rd April
When I first encountered the idea of encoding switches—specifically Mechanical Enterprises Mercutronic and Micro Switch KB, it seemed like a pointless and short-lived notion. This was not helped by a misunderstanding of when KB encoding switches were introduced, now suggested to be 1964.
In reality, self-encoding was a fairly common technique, found thus far in conductive, capacitive, photoelectric and inductive forms. What makes them seem uncommon is their rarity: self-encoding keyboards are scarcely encountered, with known examples seemingly limited at this stage to photoelectric keyboards from Invac and Collimation.
Joining these ranks is another manufacturer whose products also remain undiscovered is Synergistics. Synergistics keyboards, obtained from the acquisition of Peripheral Equipment Corporation in 1968, use a semi-reconfigurable conductive implementation. Keys can be rearranged at will (albeit seemingly only by removing all keys to its right in the same row) but changing the encoding requires sourcing a replacement coding mat, the small insulating sheet with break-off tabs that determine which pairs of switch contacts are active. This approach, including seemingly a complete lack of any on-board circuitry besides the output buses (including the absence of an electrical monitor or its accompanying resistors) led to a keyboard deemed at the time to be very cost effective. Indeed, a 1969 selection of products on the market gave a $50 quantity price, the cheapest product listed by a wide margin. Mercutronics came in second place at $75, and all other offerings were priced at $100 or more.
Sunday, 28th March
There are certain keyboard designs that, for the moment, seem specific to a single company. One of these is IKOR’s capacitive encoding system, which may be their only keyboard design. No such keyboard is yet know to have been observed, and it seems that they did not remain the market for many years. IKOR keyboards used a self-encoding technique where each output bit is detected separately via capacitance. An electrostatic shield is fitted to each key, with a hole punched in it for each output bit to be set as a 1. Details are given in the patents, and some rough illustrations are also posted to help explain this rather strange sensing and encoding method.
Saturday, 27th March
It seems that Micro Switch SD Series is a little older than I thought. I have been reviewing the collection of advertisements and articles that Marcin Wichary sent me. I tend to disregard much of it, as the scan quality is generally terrible, and much of is completely unattributed: the only trace of the origin of the advertisements is when the scanned pages bear the publication details, and full-page advertisements in magazines often lack even a page number. With the publication details recorded, I could at least request a fresh scan from the Linda Hall library, as I did with the Maxi-Switch Series 6000 and 8000 advertisement where it was critical to have the illustration in a usable form. I have been forced to delete some of the advertisements as the lack of even the year of publication means that I cannot use them as a historical record; perhaps they are older than the copy I found myself elsewhere, but I have no way of knowing.
One advertisement that did offer some insight is one for Micro Switch SD Series, which indicated that production was expected in late 1974, and that SD Series was designed to compete with mechanical keyboard pricing (based on, it seems, projected 1975 pricing). The SD Series drawings date from 1975 and the oldest known SD Series keyboard was made in 1976. The advertisement suggests that the first keyboards would be delivered in 1975, which would fit.
This advertisement is now added to the SD Series page.
Thursday, 18th March
Joining the list of seemingly short-lived companies with no confirmed successful sales, is Fort Electronic Products, later known as FEP Associates. Their keyboard speciality was the Fero-Snap™ ferrite-based switch. With a company lifetime of under two years, the chances of finding one of their keyboards is slim to none, but somewhere out there, one of them may exist.
Wednesday, 17th March
Amongst the assortment of archived PDFs that Marcin Wichary sent me, that I had never noticed until today, was an announcement in Design News magazine from December 1983 for Maxi-Switch Series 6000 and 8000. This article was accompanied by an very poorly-scanned illustration of at least one switch type; comparing the text with what little could be made out of the illustration strongly suggested what 8000 Series was …
The relevant issue of Design News is carried by the Linda Hall Library, and after ordering a scan of the page in question from them (which they turned around promptly), the illustration turned out to depict exactly what I hoped it would:
- Maxi-Switch Series 6000 is now confirmed to be “Maxi-Switch vintage linear”: the illustration shows the two switch contacts in front of an assembled switch
- Maxi-Switch 8000 Series is the type that until now has been called “Maxi-Switch integrated dome”: the dome and the base plate with the contact pads can be clearly seen
I knew that Maxi-Switch were never sure whether “Series” should be a prefix or a suffix, but in this announcement, “Series 6000” and “8000 Series” are used verbatim.
The codes found on miniature elastic contact keyboards suggest either 8500 or 8700 series for those, which now fits in nicely with 8000 Series being the full-size elastic contact switches.
The encoding and output page is also updated with an intriguing discovery: an article from 1978 described a keyboard implementation where the keyboard matrix was treated as memory-mapped I/O, using a regular 8-bit microprocessor instead of a microcontroller.
Saturday, 13th March
For several years, the brands list has only covered manufacturers of switches and keyboards. More recently, it has been extended to cover the semiconductor manufacturers responsible for single-chip encoders. The component manufacturers list is now split into a column of its own, and is extended once again to cover a number of other manufacturers responsible for components, materials and equipment used in the keyboard industry, all backed by contemporary literature.
Covered thus far are companies responsible for keycap engraving machines (New Hermes Engraving Machine Corporation), PBT raw material (Celanese Corporation), tickers resembling the “solenoid” in IBM beam spring keyboards (Endicott Coil Company, Inc.), printed circuit boards (Photocircuits Corporation) and keyboard testing machines (Osawa & Company).
Monday, 8th March
Some time ago I wondered why Licon Series 550 was advertised in 1970 with so many wires passing through the switch. The answer is surprisingly simple: wired inductive encoding, in which the encoding of each switch is defined by which bit lines are passed through the ferrite cores. This seemingly short-lived idea may well have been unique to Licon; it seems that PCB-based encoding soon replaced it.
Sunday, 7th March
On the subject of forgotten products, here is another: Hathaway Drireed low-profile reed switches using a rotating magnet assembly. The rotating magnet design was first described by Hathaway in a patent for a tall reed switch filed in 1970, but their low-profile keyboard switch design (with a vertically diagonal reed to reduce height) filed two months later in January 1971. These switches used an interlocking system to avoid the need for a mounting plate, an idea that would become better known with the later Hi-Tek Dovetail Series design.
Also, another Futaba part number is now known: “MD4P-R”: this is for the blade stem version of MD-4. The “C” is now confirmed to mean a cross stem, implying that there is also an MD-4PC. An excerpt of the Futaba diagram along with the switch specifications can be seen in a Eagle Electric Co advertisement.
Friday, 5th March
There is a surprising number of keyboard manufacturers that have become completely forgotten with time, and whose products never seem to be encountered. One such company is the now-rediscovered Colorado Instruments, who in 1970 introduced clicky capacitive keyboards that used a metal dome as the movable capacitor plate.
Sunday, 28th February
Omron B3K has suddenly become a lot more complicated. Omron themselves–who for big name YouTube channels were willing to provide factory tours in both 2015 and 2019—refused to say a thing (about anything, not just B3K), such is my irrelevance. This has earned them a coveted place in the Keyboard Hall of Shame.
Андрей Студенцов (chromov113) has provided two interesting sets of photos, however. B3K-T13LN-L is a model used in the backlit Logitech G413, but lacks the light pipe found in all other models. B3K-T135 is a Creative PRES model. Note the lack of suffix on the model number: the reason for this is unclear, but the example he found is branded only by Omron. There is a separate B3K-T135C, listed in Omron Product Code Change AMP-Z-2007029, but whether that is a separate Creative-branded model or whether the “C” suffix here has a different meaning, is not known.
Omron Product Code Change AMP-Z-2007029—found on two websites of vendors who previously dealt with B3K (and now added on the B3K page)—lists four models: B3K-L135, B3K-T135 (used by Creative), B3K-T135-L and B3K-T135C. B3K-L135 could be linear (from the model number) and is shown with the same grey plunger as the linear Romer-G model B3K-L13L. A second type is depicted, possibly B3K-T135-L; like B3K-L135 this has only Omron branding. The “L” suffix here does not appear to denote Logitech, possibly similar to how B3K-T13LN-L has both “L” and “-L” present in the model number. (The three vendors who list B3K parts—TTI, Anglia and Arrow—were of no use. TTI did not even respond.)
The document also implies that lot numbers containing a “C” in the fifth position indicate Chinese manufacture, while those with “R” in the fifth position indicate Japanese manufacture. (Vintage Omron B2x switches uses “R”, as one might expect.) Inspection of the lot numbers on the Romer-G switches sold in China suggests that there may be production lines in both China and Japan, with a shift towards China as of the 1st of October 2020 as a reward for the pandemic.
Unfortunately, the more we discover about B3K, the more we realise that we do not know. It is a remarkably complex series considering that the mechanical keyboard market had collapsed, and is largely avoided by major switch manufacturers (with far too much reliance on an exponentially-growing list of secretive, muddled and often anonymous Chinese firms). One has to wonder whether any of these B3K models are used outside of the gaming market, such as for industrial control panels. This may never be known. Sadly we are likely to learn more about products from decades past than we are about a product currently on the market.
Monday, 15th February
A new listings page is available that shows tables of available keyboards as presented in historical publications. None of these tables are exhaustive; each one contains a selection of representative or interesting products on the market at the time. These tables help illustrate not just the brands and models that existed, but the types of sensing and encoding techniques in use at different times in the industry history. In some cases, the publication from which the table was obtained also illustrates and describes the working of these products.
The tables show a number of brands that are yet to be written up here: IKOR, Synergistics, Soroban, Teletype, Ultronic, Digitronics, Killian and Transducer Systems. Details of many of these product ranges are in the relevant publications or are in the many magazine back-issues scanned by Bitsavers and WorldRadioHistory.com. Further details on some of these will be forthcoming.
Saturday, 13th February
This time we stretch back a lot further, to the early days of computer keyboards, to the era when electromechanical keyboards were still being produced. A lot of interesting details on these keyboards were written up in the article Manual Input Devices in Computer Design, Volume 4 No. 12, December 1965, which is now added to the references page.
The Navigation Computer Corporation, also known as NAVCOR (later KDI Navcor) was an American electronics manufacturer whose product range included, over time, three different designs of reed keyboard switch. Their keyboards used diode matrices for encoding. Other keyboard types may have also existed: as with so many such brands, few examples are known. NAVCOR curiously took out a patent on the idea of switches with a sloped base, an idea that would be later used by a number of manufacturers including Raytheon, Veetronix, George Risk and Fujitsu.
Thursday, 11th February
Another early adopter of membrane keyboard technology was Chomerics. Alphanumeric products general manager Len Halio claimed in 1981 that Chomerics invented membrane switch technology in the late 1960s, although they did not apply it to full-size (alphanumeric or “QWERTY”) keyboards until long after Datanetics did. Until this point, their market was keypads, such as telephones, robotics and toys. There is no indication that any Chomerics keyboards have been found by anyone, but some must surely exist somewhere.
Sunday, 7th February
Having just uploaded the specifications of Oak Full Travel Membrane (kindly scanned in by TG3 Electronics), this seems a perfect time to upload an article from around the same time that describes the workings of Micro Switch’s own offering: SC Series capacitive membrane keyboards. It is not clear which of Oak and Micro Switch introduced membrane keyboards first, but both types seem to date from the start of the move to full travel membrane keyboards. (Datanetics batch-fabricated array is much older, but they found that discrete switches were more practical.)
When I first encountered suggestions of capacitive membrane keyboards from Alphameric, it seemed like some kind of a mistake, but the article from Computer Design from 1981 depicts and describes the operation.
Friday, 5th February
Photographs have turned up depicting the reed variant of SMK JM-0400 series and SMK’s elastic contact type (covered for the moment under JM-0400 in order to have somewhere to list it). The elastic contact switch is unsurprisingly identical to Maxi-Switch’s elastic contact type (as found in a few TRS-80 Model 4 keyboards and seen exactly once in a single photograph), right down to the green pigmentation of the rubber dome. This strongly suggests that SMK manufactured those switches for Maxi-Switch. With that said, no SMK branding is visible in either instance, nor any indication of where either switch came from. The images were uploaded without a single word of description and no permission to comment on them.
Wednesday, 3rd February
A Marylou Bassi-Dolmans, who used to work for the Clare and Pendar companies, recently contacted me. Following correspondence and further investigation, including going back through all the information that R Guzzetti posted to the Deskthority forum, the Clare and Pendar history page is now updated. There is more work to be done, but even already, the story is clearer than it was before.
Sunday, 31st January
I never realised that Samsung made one of the Amiga 500 keyboard types. I never thought of Samsung as a keyboard manufacturer at all. Strangely, the switch mechanism bears a strong resemblance to one of the designs in the patent that appears to cover Alps KFNR. Sadly there are no complete sets of photographs of this keyboard, and nothing showing any branding or part numbers.
The Amiga keyboards are clearly not yet fully documented on Deskthority. WMF-4E1Q2 is another barely-documented type, this time from the Amiga 600. This model appears to be NMB, but using conductive domes instead of the standard pressure domes.
Further research has turned up a number of extra single-chip keyboard encoders. These are EA2000, EA2007 and EA2030 from Electronic Arrays; MCS 1007, MCS 1008 and MCS 1009 from MOS Technology; and TMS 5000 and TMS 5001 from Texas Instruments. The point in time that single-chip encoders were introduced is still unclear. Thus far none have been seen before 1972.
Sunday, 24th January
Correspondence with MinebeaMitsumi has turned up a rather surprising fact: at the time that they bought Hi-Tek Corporation, they were already in the keyboard business, manufacturing “keyboards for IBM's high-end personal computers”, production of which began in September 1982, following an agreement in January 1982. Until now we have just assumed that the Model F—which is presumably the type they describe—was manufactured solely in the United States, but this seems not to be the case. D’Milo Hallerberg, who was employed at Hi-Tek at the time, confirms that Minebea were manufacturing IBM keyboards in Singapore. The production moved to Thailand in 1984, where production of Series 725 was also relocated.
Additionally, all of the remaining literature that Susan Kennedy provided to Benjamin Rockwell, and that he scanned in, is now uploaded to the Hi-Tek and Series 725 pages.
Friday, 22nd January
Magazine trawling has turned up a short article on a Cherry keyboard kit that used their magnetic-separation reed switches. The article briefly details how these keyboards are designed to be wire wrapped during prototyping; this was likely to have slightly pre-dated the introduction of LSI encoders and external encoding PROMs.
A new SMK JM-0400 page now exists to collect up any fragments of information that turns up about SMK’s interchangeable low-profile switch series. There is also a “stub” page for SMK JM-0200, which is some kind of unidentified keyboard switch, which could be anything from their tall reed switches to a short-travel keypad button. An advertisement mentioning “JM 0404” lists it alongside “JJ0019”; JJ-0019 or J-J0019 is some other unidentified series of keyboard switch or switch component.
Sunday, 10th January
Based on the information within the Geekhack topic Hall Effect Keyboard Sensing and Wiring Explained from dorkvader from January 2018 (discovered accidentally), the matrix scanning section of the Hall effect keyboards page is updated, reflecting an additional method of matrix scanning, and the implications that the different approaches have on rollover.
Saturday, 9th January
An illustration of Maxi-Switch 2700 series has finally turned up, in an advertisement in Electronics magazine from 1972. They are cylindrical reed switches with fins protruding from them. The advertisement is posted to the Maxi-Switch section and may help someone identify these switches should they show up somewhere.
Friday, 1st January2020 knowledge round-up
2020 has been a strange and ultimately disappointing year for reasons unrelated to the unexpected export from the Far East of something more dangerous than cheap electrical goods.
In terms of keyboard knowledge, it seemed fairly likely that 2020 was not going to top the level of revelation of 2019, and while this seems to be true, progress has continued in spite of everything. The big breakthrough was getting catalogues from RAFI in March for their RC 72 and magnetoresistive types, the latter being to date the only known example of magnetoresistive switches. This was the first but by no means the last instance of finding details that seemed highly unlikely to be found, showing once again that so long as you keep searching, you will find.
The RAFI discoveries then led to an in-depth investigation of German Hall sensors, not least because RAFI were offering a variety of output options not covered by the HFO B 461 G sensors that we have encountered thus far. Discussion with various people in Germany and examination of various websites and catalogues has allowed me to list out both the Siemens Hall sensor ICs and the HFO Hall sensors derived from them, although the early history of the Siemens types remains poorly understood.
Several years ago, Meryl Miller offered me an assortment of switches, and I selected just the keyboard types for myself and Jacob Alexander. More accurately, I selected the ones that I believed were keyboard switches, missing the fairly obvious Micro Switch KB reed switch (which was a solder terminal type, rather than the quick connect alternative that is all I have been able to obtain to date). Less obvious was a curious contraption that turned out to be Fujitsu FES-2. The real shame was in passing up a Raytheon keyboard switch. Extensive trawling through old electronics and computing magazines at Bitsavers turned up considerably more detail on these switches than expected, giving us the identities of KBSM for the mechanical switches and KBSR and KBFR for the reed switches. (In theory, KBFM—mechanical with a flat base—should also exist, but no sign of it has been discovered to date.) Only two KBSM switches have ever been seen: the one that Meryl has since discarded, and one that had already sold on eBay, from a seller too mean to permit the now-redundant photos to be published here.
Perhaps the most surprising discovery of all, was the names of several Stackpole series names, specifically KS-200 for the high-profile arrays, KS-200E for the low-profile arrays and the interlocking discrete switches, KS-500E for low-profile membrane keyboards, and finally KS-600E for the plate-mounted discrete switches derived from KS-200E. No mention has been found to date of a KS-100.
Keyboard enthusiasts are frequently acquainted with the method by which modern keyboards operate: the switches are wired in a grid (or matrix) and scanned by firmware stored within a microcontroller, with the key definitions stored within the firmware. Far less is known about the workings of keyboards going back to the 1960s, and the many ways of implementing a keyboard before the late 1970s when it started to become cost-effective to use a microcontroller. Many of the early techniques are now covered on the encoding and output page, itself still a work in progress as more details emerge. As part of this endeavour, Micro Switch KB encoding switches are now detailed on their own page; these pre-date the reed switches by a couple of years or so, something still in line for documenting according to literature.
Other discoveries made in 2020 include:
- The Hi-Tek switches in the UB80-01AA keyboards are licence-made by Cherry in their US factory, arranged by DEC, as explained by Peter Cherry. This ties in with DEC arranging for Tontec to produce Hi-Tek array mouldings in Hong Kong, possibly with some other company producing the completed switches for those specific VT-100 keyboards.
- We have the first non-Logitech part number within Omron B3K series, from José Soltren, which has helped with understanding the part number format for B3K. Now all that remains is to obtain the switch part number from a Creative PRES switch.
- There is a possibility that the keycaps on most BBC Microcomputer keyboards are the long-lost Signature Plastics SS series.
- The internal contact arrangement of Cherry M75–M78 is now known, although it remains to be documented; specifically, the design is almost the same, but the half-width contacts are arranged in side-by-side pairs. This allows two contact pairs in an alternate action switch (with the alternate action mechanism on the rear of the plunger), or four contact pairs in a momentary switch.
- Although Oak Series 400 was already tentatively identified, the smaller type was not; these are now known to be Oak Series 475, again identified from trawling old magazines.
- Another magazine find (from scans provided by Marcin Wichary) gives us Series 555 for the “Early Ⅱ” and “Early Ⅲ” styles of Licon/Cortron ferrite core switches, leaving “Early Ⅰ” to be the original Series 550. Confusingly, we now know that a Series III existed, but from the description, this is not the tall switches in 35-* keyboard models. Just as confusingly, the DIN-compliant switches seemed to be split between two separate series: FC-2550 and FC2500 (with only the former having a hyphen). Details on both series remains too vague to understand how they differed, but the suggestion seems to be that FC-2550 had model numbers beginning 25-5* while FC2500 had model numbers beginning 25-0*, something that cannot be proven one way or the other due to how few examples have been properly documented by their owners.
- The confusion around Micro Switch CT and its relationship with ST is now explained: the original two series were capacitive membrane SC and contact membrane CT. ST silent tactile keyboards superseded both, replacing the click spring with a rubber dome, and providing both membrane types under a single series.
- Electronic Engineers Master from 1989–1990 provides a list of features and part numbers for Marquardt Series 6184, indicating that the series included alternate action, tactile feedback and a choice of 2.5 mm and 4 mm travel. Later, a click-tactile type would show up in California, which for now is being presumed to be the tactile type (something Marquardt have neither confirmed nor denied). The click mechanism is something not seen before, and is illustrated on the Series 6184 page.
- Several more MEI T-5 Series subseries are now known, from old advertisements.
- Maxi-Switch “vintage linear” may be 6000 Series, although this remains to be formally confirmed. These switches are now almost certainly made by SMK or made using tooling licensed from SMK, which explains why the “integrated dome” type bears a very strong resemblance to another SMK series.
- It now seems likely that the Fujitsu “sheet keyboard” system is FES-3001 while the “typewriter” form (the full travel type) is FES-301. Following the introduction of DIN ergonomic standards, a lower-profile version of FES-301 was adopted, which is FES-301E (Fujitsu provided this series name and indicated that “E” denotes “ergonomic”).
- Another magazine find was a brief advertisement for SMK JM-0400 series. J-M0404 and the heretofore unobserved J-M0409 fall within this overall series. Most interestingly, JM-0400 included reed switches. We already knew that SMK produced reed switches, but not that they miniaturised them. It seems that SMK filed few patents, or they are still awaiting digital recording, as very little showed up on J-PlatPat for SMK, with no reed or integrated dome types appearing at all.
- Just as understanding of Alps was starting to improve, 2020 has dealt it some blows. Jacob Alexander has found a new Alps reed type, this time classified as SCF instead of SCB. This would suggest that even in the S-code era, each series had its own identity, but that does nothing to clear up the confusion around CH and SCH. An old advertisement for Alps switches and keypads gives a totally different series identity, in a way that does not make sense, but with the spring bridge switches considered to fall under SCH being listed as AKC.
- The origin of Cherry MY is now much better understood. The different types remain a mystery; even Günter Murmann was unable to recall anything about them.
- There is now clear evidence that HP bought “mousetrap” switches from Cherry; more details are given on the new Cherry Series S31 page.
All things considered, 2020 has fared significantly better than it could have, and with luck 2021 will be at least as bountiful.