- General terms
The following definitions are explained in terms of major manufacturer literature and advertisements.
Electronic Engineer magazine guide, 1971
Electronic Engineer magazine published “this is your keyboard reference” in the September 1971 issue, compiled in conjunction with Micro Switch as an advertisement for their product range. (Bitsavers scanned it and incorrectly list it as being from 1972.) This document divides all keyboard technologies into two types: “mechanical” and “solid state”. Membrane keyboards are listed, but not in a form where the membranes have any kind of surface-printed traces.
The term “mechanical” in this instance was used to refer to anything where the switch contacts physically touched, no matter whether they were solid or liquid, sealed or exposed. Unfortunately, the lack of elastomeric rubber and silver-carbon membrane keyboards this far back means that we cannot know for certain how these would have been classified.
The full list of terms is as follows:
TV Typewriter Cookbook, 1976
Don Lancaster’s 1976 book TV Typewriter Cookbook covers many details of keyboard design and operation. On pages 136–138, he divides keyboard switches into three groups: mechanical, elastomeric, and exotic:
|Mechanical||Mechanical switches are said to give “metal on metal contact”. He also notes that if such switches are properly designed, they “should combine both a wiping action and a cross-point contacting [action]”. The only series specifically named is Mechanical Enterprises T-5, while a switch presumed to be Mechanical Enterprises LFW is depicted as the example type.|
|Elastomeric||Elastomeric switches are described as having “a piece of flexible conductive plastic or foam.” Elastomeric switches are divided into types where the elastomeric material is pressed against a stationary contact (as is the case with Datanetics elastic diaphragm and Mitsumi’s two-membrane “hybrid” types), and pressure-sensitive materials that lower their resistance under pressure. Curiously he notes that “light-duty, short-term keyboards have even been built out of the protective foam shipped with many MOS integrated circuits.” No product examples are cited.|
“Exotic” is used to cover a number of other types:
PC Magazine, 1989
In PC Magazine, Vol. 8, No. 21 from 12th December 1989, an article entitled Wanted: More than just a replacement (by Bruce Brown and Kellyn Betts; pages 225–260) covers a range of keyboards on the market. This article also provides definitions of the different switch types found in commercially-available keyboards (pages 240 and 241). These are illustrated with rather poorly-drawn diagrams. The description of capacitive switches indicates that the authors do not fully understand the technology involved, and thus their descriptions cannot be relied on as being completely accurate.
|Mechanical||Described as “a simple switch relying on contact between two conductive materials”. The conductive materials are not stipulated to be metal, but the article notes that “the cost can be high, depending on the contact material—often gold or gold alloy—it uses.” The article also notes that mechanical switches “are reliable, have a relatively long life expectancy, generate an audible click, and (depending on the spring tension used to return the key) can have a very positive feel.” The term “positive feel” is found extensively in literature and patents, but is seldom if ever defined. The example manufacturers given are Chicony, DataDesk, NMB, Northgate and Zeos.|
|Capacitive||Capacitive switches are said to “detect a change in capacitance as a circuit is opened or closed; they don’t make a mechanical contact between conductive elements.” The description is somewhat confused, as it notes that “a dielectric … cushion is pushed down, forcing the conductive elements farther apart (or closer together, depending on the design) to create a closed circuit.” In reality, the conductive element is always kept separated from the sense pads, typically by the solder mask on the PCB.|
|Conductive rubber dome||This covers typical conductive dome keyboards, such as those by BTC. The manufacturer that they cite is Maxi-Switch, and their review model is the ME 101, a 2186 type with conductive domes.|
|Membrane||The term “membrane” here is applied solely to flat keyboards with short travel: “Keyboards that use [membrane switches] are quiet, have little or no feel and are hard to repair, yet they are considered a good choice for harsh environments because the continuous rubber sheet that overlays the full key-switch set helps to keep out harmful dust, crumbs, and liquids.” The article does not review any membrane types. Even though full-travel membrane did exist by 1989, it seems that from the perspective of the journalists, it had not become recognised.|
The term “mechanical” is not precisely defined. So far, there seem to be two separate trends for how the industry defined it. In early 1970s usage, “mechanical” seems to have been used for any switch type that involves electrical conduction. This was used in contrast to “solid state” types to cover any type that relied on electromagnetic detection, as suggested in the table above.
As the popularity of solid state keyboards declined, and new cheaper technologies emerged—conductive rubber and silver-carbon traces on polyester membranes in particular—the term seems to have been repurposed to indicate metal contact switches, also known as “hard contact”. Whether or not reed switches were classified as “mechanical” also varied.
Many switch types are defined as mechanical by the manufacturer, including the following:
- Forward Electronics SKBL/SKBM series was described as “High-reliability mechanical contact” on their website in 2004.
- Fujitsu FES-360 Series discrete leaf spring switches are also decribed as “メカニカルスイッチ” (“mechanical switch”) in the datasheet.
- Omron B3G Series was described as “メカキースイッチ” (“Mechanical key switch”) in the 1984 catalogue (curiously, the later B3G-S series is not described as mechanical).
- The SMK J-M9031 keyboard specification datasheet gives the switches as “SMK J-M0404 series, mechanical contacts”.
- Tokai MM9 Series was described on their website as “接点が回転するボールコンタクトスイッチ！” (“Ball contact switch with rotating contacts!”) as well as being classified as “メカニカルスイッチ” (“mechanical switch”).
Monterey International also used “mechanical tactile click keyswitches” in their 1992 catalogue under the K110 keyboard and KP110 keypad listings.
The definitions above however are not found in context against other offerings from the same manufacturer and era. The following are examples where “mechanical” is contrasted directly by the manufacturer against other contact types:
- RAFI’s RS 74 and 76 series keyswitches—which date back to the 1970s—are subdivided into M and C types. In the 2001 Electromechanical Components catalogue, M denotes “mechanical”, and comprises the metal contact types, while C denotes “contactless” and comprises the Hall effect versions. The Electromechanical Components catalogue from 2015 uses the same terminology.
- Clare-Pendar Series S950 was described as a “mechanical keyswitch” in their 1986 catalogue, under the Reed and Hard Contact Switches section.
- ITW’s second generation switch design, as shown in US patent 4227163 (filed in 1979), depicts and describes that this shell style can be “either of the mechanical contact type or of the analog contactless type.”
- Cherry gold crosspoint, in the KB79-2 and KB79-2R catalogues; they are referred to primarily as “hard contact” (compared to capacitive) but the “keyboard designers’ work sheet” from KB79-2 offers a choice between “Capacitive” and “Mechanical”. KB79-2R revised this to “Mechanical (Hard Contact)” and “Solid State (Capacitive)”.
- Alps advertisement from 1983: this is not as clear as it could be, but it indicates that Alps offer a choice of “mechanical or conductive rubber contacts” (KFL and Mem-Tact are the only product names given and the switch technology used in KFL is not stated).
- Similarly, Alps advertised also in 1983 that they offered a choice of “mechanical, conductive rubber or tactile feedback”: this is before SKCM was introduced, and the advertisement only depicts KCC and only mentions KFL, so “tactile” here may mean TACT miniature switches.
- Sejin America’s desktop keyboards, 1997, offering mechanical (which would have been Futaba MA series, based on the models depicted, such as the EAT-1010MB) and membrane options. (The Sejin models in question use -MB, SKM- and SWM- for mechanical (not membrane) and -RB, -SKR and -SWR for rubber dome over membrane.)
- In 1999, NMB described their Series 725-based Right Touch Model 8200W keyboards as “mechanical”, compared to their “membrane” offerings. Their datasheets also depict “Membrane Keyboard Keyswitch” and “Mechanical Keyswitch” types.
In ITW’s case, they describe mechanical contacts as follows in the patent:
The mechanical type of keyswitch has the advantage of being relatively low in cost, and for many applications this factor makes it desirable to employ such a mechanical keyswitch. However, mechanical keyswitches have a number of disadvantages that make them undesirable for use in applications where high reliability is required and the added cost of a analog switch is, therefore, considered to be warranted. These disadvantages include contact bounce, the possibility of arcing, lower life times due to pitting and corrosion and possible deformation of the contact members.
Thus far, mechanical has been contrasted against solid state, membrane and conductive rubber. The above description also strongly implies that the switch contacts are metal.
For my purposes, “mechanical” will be taken to mean directly-operated metal contact switches, per 80s and 90s convention. This also includes ball contact switches, on the basis that Tokai classified MM9 series as mechanical, but excludes reed switches.
An awkward case is Datanetics DC-50 series. Meryl Miller of Datanetics affirms that these are “diaphragm switches”, putting them into a classification of their own. He notes that the Mylar membrane facing the actuator has the responsibility of separating the metal switch contacts, which are glued to the membranes. Although these switches are metal contact internally, they also use a three-layer membrane. Although it might seem reasonable to class them as mechanical due to being metal contact (the switch contacts are the same pieces of metal as the terminals), from manufacturer standpoint they were never classified as mechanical.
Alps SKCL/SKCM and related types use metal foil contacts. Alps specifically refer to their foil contact system as “mechanical” in their catalogues.
In Modern Data, April 1970, Mechanical Enterprises Mercutronic mercury tube switches were described as using a “mechanical switching approach based on the movement of mercury in a sealed flexible tube.” 1970 was around the time that Licon Series 550 ferrite core switches came out, and less than two years after the introduction of Micro Switch SW Series Hall effect switches. As such, having a moving tube filled with mercury that is pinched by an actuator to separate the mecury when the switch is opened, was classed by MEI as “mechanical”, in comparison to solid-state designs. Just as with DC-50, these switches are hard to classify, as they used an extremely unusual design.
The normal definition of “solid state” is the use of semiconductor circuitry. Thermionic valves (vacuum tubes) are excluded from this: even though electronic computers using valves replaced electromechanical computers, computers did not become solid state until the introduction of transistors. Solid state storage uses semiconductor memory to hold the data. A broader use of the term “solid state” covers anything without moving parts, which for keyboards excludes everything except capacitive touch screens, as even resistive touch screens have a flexible plastic cover as part of the sensor mechanism. Even so, a number of keyboard manufacturers marketed their full travel switch technology as “solid state”, from as far back as the 1960s.
For keyboards, “solid state” refers to switch types where there are no switch contacts of any kind, be that metal (as in mechanical switches), conductive rubber (as in some Mechanical Enterprises T-15 variants) or silver–carbon ink (as in most membrane keyboards). Keystroke sensing is achieved using capacitance or by electronically detecting the presence of a magnet. Such keyboards may still require one or more additional parts as part of the sensor.
Photoelectric sensing (also called opto-electric) is an old technology. Photoelectric keyboards can be self-encoding, or they can function equivalently to simple pushbutton switches with a separate encoding means (e.g. matrix scanning or diode matrix). Photoelectric keyboards fell out of favour decades ago, but optical sensing has seen a revival in gaming keyboards due to its extremely low latency: photoelectric sensing is instantaneous, being entirely free of contact bounce.
The exact origin and the early utilisation of photoelectric keyboards is unclear. A number of old patents describe photoelectric encoding means, including US patents 2234832 “Photoelectric transmitter” (filed 1938 by Teletype Corporation), 2168886 “Actuating means for typewriters and other mechanisms” (filed 1938), 2228780 “System of light control for selenium cells” (filed 1941) and 2432527 “Keyboard control system” (filed 1945).
Photoelectric encoder keyboards use beams of light running across the width of the keyboard. A shutter attached to each key selectively blocks these beams, with the undisturbed beams providing the binary code for the key that was pressed. The oldest known full keyboard type is that of Invac; Monroe filed US patent 2641753 “Photoelectric keyboard” in 1951 for a photoelectric encoding keyboard for calculators, nine years before Invac’s patent for a full alphanumeric keyboard.
A generic depiction of this arrangement is given in the diagram below:
Photoelectric encoder keyboards were an early innovation, offering bounce-free sensing and a means of encoding keys without requiring logic circuitry or large diode matrices. Conversely, they were bulky, complex and formed from a great number of mechanical parts. Further, they lacked the simplicity and flexibility of discrete switches to be arranged anywhere on a keyboard of any size and shape.
In spite of these drawbacks, photoelectric encoder keyboards were produced by several manufacturers, including Invac and Collimation.
The advantages of optical sensing can also be attained by placing the light source and light detector within a switch. The plunger interrupts the light beam as the key is pressed. As with the encoding types, this sensing method is bounce free, which means that the keyboard encoder does not need a debounce delay. Consequency, optical switches have no inherent latency.
Such an arrangement was rare in past decades, with Burroughs Opto-Electric being the only discovered example. Tokai produced the SPT-0101 opto-electric type, but this has never been seen, and it was only discovered following Tokai’s bankruptcy, so no details could be obtained from them. Tokai patented their design in 1994 as “Semiconductor photocoupling switch”, suggesting when it was introduced. More recently, new types including Adomax Flaretech and Gateron KS-15 series have been introduced. These switches are targeted towards gamers due to their responsiveness. Additionally, the Flaretech design is fully analogue, allowing the keyboard to detect how far the key has been pressed. Unlike with Hall effect, analogue sensing is not a feature of all optical switches, and the majority provide simple on–off switching.
Reed switches have magnetically-operated contacts sealed within a small glass tube. The field around the magnet in the plunger causes the metal reeds to attract each other, make contact, and close the circuit. Reed keyboards date at least as far back as the 1960s, with Navcor being an early adopter, along with Micro Switch (with KB), George Risk Industries and many others. Reed switches have potentially a long lifetime (although lifetimes of only one million cycles could be found in decades past), are not susceptible to dust and moisture, and have a low bounce time of 2 milliseconds or less.
See How reed switches work (magnetically operated switches) for a clear explanation of reed switch operation, including normally-closed contact types.
Reed switches are still metal contact, but the switch contacts are closed magnetically, allowing the contacts to be sealed against moisture and debris ingress. Consequently, some manufacturers classify reed switches separately from mechanical switches. As early as 1968, Raytheon distinguished their KBSR-1 and KBSR-2 reed types from their KBSM-1 mechanical type. (Previously in 1967, Raytheon had described the non-sealed type as “wipe-action”.) Conversely, US patent US4370533 for Fujitsu FES-360 switches (filed in December 1980) notes the following:
Switches are divided into two types, that is, switches having a mechanical contact element, such as reed switches, and switches having a non-contact switch element such as hall IC. The present invention is directed to a keyboard comprising the former type, i.e., switches having a mechanical contact structure.
Clare-Pendar advertised Series S950 as a “mechanical keyswitch” type and S820 and S880 as “reed keyswitch” types in their 1986 Switchlight and Pushbutton Switches catalogue. All three types fall under the general classification of “Reed and Hard Contact Switches”. In their Semiconductor Memory Data Book for Design Engineers from 1975, Texas Instruments indicated that the TMS 5001 keyboard encoder is “Compatible with Reed and Mechanical Switches”.
Fujitsu would later separate the types out in the May 1985 edition of their magazine, with Fig. 5 “Development of keyboard switches” (page 429) giving the following descriptions of their switch product lines:
- “リードスイッチ” (“reed switch”), covering FES-5, FES-9 and FES-4 reed switches
- “メカニカルスイッチ” (“mechanical switch”), covering FES-300 and FES-301 leaf spring types
- “メンブレンスイッチ” (“membrane switch”), covering membrane leaf spring
Broadly, the consensus is that reed switches (being magnetically operated) are classified separately from mechanical switches (whose contacts are physically operated).
“Solid state” is an awkward choice of term in the context of full-travel keyboards. All such keyboards require moving parts, with at minimum a plunger and some kind of spring. However, the oldest switch type so far discovered that was marketed as “solid state”—Micro Switch SW Series—did indeed have actual semiconductor sensors. Introduced in 1968, the Micro Switch SSK—the original brand name of SW Series—was advertised to be the “first of its kind”, using an integrated circuit within each switch to detect a pair of magnets within the plunger via Hall effect.
Hall effect provides a means to to detect the proximity of a magnet entirely electronically. As the magnet is moved near the Hall sensor chip, current flowing through a special conductor area within the chip is diverted by the magnetic field from the permanent magnet, creating a minuscule potential difference across the conductor. Following amplification, this potential difference can be read in an analogue manner to indicate proximity, or the Hall sensor can set its own trigger and release thresholds and provide digital switch output.
RAFI’s Hall effect switches and keyboards are described as “contactless” and “solid state” depending on the document: “contactless” in the Electromechanical Components catalogues, and “solid state” in the Standard Keyboards catalogues.
Magnetoresistive elements increase their electrical resistance in the presence of a magnetic field. The only confirmed implementation of magnetoresistive keyboards is RAFI’s magnetoresistive range introduced in 1970, for which there is a complete product catalogue.
Other manufacturers filed patents for magnetoresistive switching in keyboards, but confirmed products are not presently known. IBM’s US patent 3848252 “Magnetic keyboard” filed in February 1972 with a priority date of March 1970, describes both Hall effect and magnetoresistive sensing; the patent depicts an IBM Selectric as the equipment that will be using the switches described. Sperry Rand filed US patent 3768095 “Magnetoresistive keyboard” in September 1972, with a self-encoding design. In the Sperry arrangement, under each key there is a separate PCB track per output bit, and only the bits to be set (or cleared) have a magnetoresistive element. Pressing a key moves a magnet in place that changes the resistance of those lines that have a magnetoresistive element, generating an output code. Whether this arrangement is intended for calculators or alphanumeric keyboards is not indicated, and the term “keyboard” at the time was ambiguous, encompassing not just data entry keyboards but also any kind of keypad.
The only other known mention at present of magnetoresistive keyboards is that of Nucleonic Products in December 1970, and there is presently no evidence to indicate whether these keyboards ever entered production.
Ferrite core sensing is a family of contactless techniques that use very simple 1:1 transformers as the switching element. Broadly, ferrite core switches can be divided into moving magnet and moving ferrite types, but moving shield was also patented. The two approaches can be observed in the diagram below:
Unlike “proper” transformers, the wires are not wound around the core; instead, they are simply passed through or around the core. Licon Series 550 was described in the advertisement as “All solid state”, although in the strictest sense this is not true as it does not use a silicon chip as the sensor.
The best-known manufacturer of such switches was ITW, through its ITW Licon and ITW Cortron subsidiaries. Licon Series 550 was advertised as far back as 1969, and Cortron ferrite core keyboards were still being manufactured into the 1980s. The concept pre-dates ITW; for example, Clary Corporation filed patent “Keyboard controlled circuitry” in 1956 (granted in 1961 as US patent 2997703) for magnetic core inductive switches. The patent notes:
As is well known, input controls for electronic computers, etc. generally comprise one or more contact closure per key, which contacts are either opened or closed upon depression of a respective key. Although such switch contacts are, in general, satisfactory, they tend to arc and to cause contact chatter under other than ideal conditions. Furthermore, such keyboard contacts are subject to wear, pitting, and to dust or dirt particles being lodged between the contacts which may increase contact resistance or even cause open circuits. Also, undesirable transients and other objectional phenomena may be caused by such key controlled contacts.
Clary’s design places a fixed permanent magnet beside each set of magnetic cores, which saturates the cores and prevents the inductive coupling. The plunger of each key is “preferably formed of soft iron or similar material capable of shielding magnetic flux” and as the key is pressed, the plunger blocks the path between the permanent magnet and the magnetic cores, and allows the key to be sensed. Clary’s design uses wired inductive encoding, the approach that would be later adopted briefly by ITW. In this implementation, each output bit has its own core, with four separate cores per switch. Each core corresponds to one output bit; cores corresponding to bits that should be a zero are left unconnected. ITW chose to route a single wire per bit selectively through the switches, requiring only a single core per switch.
Ferrite core sensing is an analogue approach, because the strength of the output signal is proportional to the proximity of the magnet (in saturated core switches) and presumably also proportional to the proximity of the ferrite core (in movable-ferrite designs). No keyboards have yet been discovered that knowingly take advantage of this property, although ITW did mention it in at least one patent.
Only a handful of manufacturers are known at present to have created switches of this type. After ITW, ADI (Advanced Datum Information Corp) of Taiwan is likely the best-known manufacturer. IMS International produced keyboards very similar to those of ADI, and it is possible that they sold the technology to ADI. Fort Electronic Products introduced their “Fero-Snap” switch sometime around 1971, which seems to have been fairly short-lived.
Presently known only from Clary’s patent, the plunger acts as a movable shield that selectively blocks the magnetic saturation from the permanent magnet.
“Moving magnet” switches have a fixed ferrite core and one or more magnets attached to the plunger. These switches operate on the principle that a magnet positioned adjacent to the ferrite core will saturate it and prevent AC pulses from being induced from the primary to the secondary winding. When the switch is operated, the magnet is temporarily moved clear of the ferrite core so that it can operate, allowing that one key to be detected. The Electronic Engineer magazine referred to this technology as “saturated core” in their keyboard guide produced in conjunction with Micro Switch.
Early Licon switches needed two magnets, but with improvements in magnet design this count was reduced to one.
The other distinct design of ferrite core does not use a magnet. Here, the windings take the form of PCB tracks. When the switch is operated, a cylindrical ferrite core is lowered into a hole in the PCB in order to couple together tracks on either side of the PCB. The best-known manufacturer of such keyboards is ADI (Advanced Datum Information Corp). A single model is also known from IMS International.
Capacitive keyboards are where things start to get awkward. Capacitive keyboards effectively take the form of a variable capacitor inside each key, controlled by the plunger. Various manufacturers have described capacitive keyboards as “solid state”, but this is quite a stretch. The majority of capacitive keyboards use a foam pad bearing a foil layer as the part of the switching element. These foam pads degrade with age.
In the 1979 Cherry Electrical Products keyboards catalogue, Cherry use “solid state” to describe their foam pad keyboards. This terminology is not unique to Cherry. Key Tronic also described their keyboards as having “Solid-state capacitive switches”, as in these KB 5151 advertisements. Computer Products United likewise advertised their unbranded BTC 5339, 5160 and 5151 keyboards as having “solid-state capacitance low-profile key switches”. (No brand is cited, but inspection of the keyboards indicates that they are BTC and not Key Tronic due to the LED and key placement, and 5339 is a BTC model number. This is the larger, wedge-shaped variant of the 5339, rather than the slimline version.)
Cortron used the description “Solid-state, capacitance unit” for their CP-4550 keyboards. These are metal leaf capacitive switches extremely similar to those of Digitran. Here, instead of a foil disc, there is a metal leaf spring that is pushed down onto the PCB. Overtravel is achieved instead by a special prong on the leaf spring. These somewhat fragile pieces of stamped-out metal stretch the “solid state” description even further.
Membrane keyboards use one or more sheets of thin plastic, called “membranes”, as part of the switching mechanism. In most cases, there are three sheets: two are flexible printed circuits, and one is a spacer placed in between the other sheets. This is the system used in virtually all keyboards made today. The use of “membrane” is found in multiple patents for keyboards using these switches as well as the literature from many manufacturers, and is not in dispute when it comes to full three-layer membrane assemblies. The manufacturer descriptions of two-sheet and single-sheet membrane arrangements are not widely known, but Mitsumi’s later single-membrane types (KPQ and KPR types) were documented in their catalogue as membrane also.
The majority of membrane keyboards use conductive sensing, but a few manufacturers produced capacitive membrane keyboards.