geekhack forums
Go Back   geekhack forums > wiki

Exclamation Default:IBM Wiki
WORK IN PROGRESS



Post suggestions and corrections in the discussion section.

Note: In order to see the pictures in this article, you must be registered and logged in.

Unless specified otherwise, part numbers are for US English keyboards.

Text property of ch_123.



Thanks to Brandon Ermita of ClickyKeyboards.com for his pictures and his informative site.

Thanks to Sandy55 for his pictures and information on IBM's Japanese keyboards, and his great site.

IBM Datamaster picture property of Flickr user Marcin Wichary. IBM Model B picture property of Flickr user Salvajada. Flickr pictures released under Creative Commons licence.

Other pictures sourced from within Geekhack.org, all rights reserved by appropriate owners.


Contents


[top]IBM's Keyboards

[top]In the Beginning...

[top]Typewriters

IBM's keyboards date back to 1933, when IBM bought the Electromatic Typewriter Co. - who made a popular electric typewriter also called the Electromatic. IBM sold this as their Model 01 Electric Typewriter.



This was succeeded with the Model A in 1949.



The Model B in 1954.



The Model C in 1959.



And the Model D in 1967. this latter being the last member of the IBM Electric family.



Models A - C are relevant to us as they formed the basis of some of the original computer consoles and terminals; not only for IBM computers, but ones made by DEC and other manufacturers as well. When the original IBM Selectric typewriter came out in 1961 it replaced the earlier models for terminal use, as it was much easier to interface with computers.



Even still, adapting Selectric typewriters to computer use was not a trivial task . From Wikipedia -

Quote:
Despite appearances, these machines were not simply Selectric typewriters with an RS-232 connector added. A Selectric is a marvel of mechanical and production—but not electronic—engineering. As with other electric typewriters, and electric adding machines of the era, Selectrics are best thought of as electromechanical devices: The only electric components are the power cord, power switch, and electric motor. The electric motor runs continuously. The keys are not electrical pushbuttons, as they are on a computer keyboard. Pressing a key does not produce an electrical signal, but rather engages a series of clutches which couple the motor power to the mechanism to turn and tilt the element. A Selectric would work equally well if hand-cranked at sufficient speed.

Adapting this mechanism to the needs of computer input/output was nontrivial. The keyboard and printing mechanism were mechanically separated (so that keystrokes do not necessarily result in immediate printing), microswitches were added to the keyboard, solenoids were added to allow the computer to trigger the typing mechanism, and interface electronics were needed. Several mechanical components, in particular the motor and the main clutch, had to be upgraded from the typewriter versions to reliably support continuous operation. Additional microswitches had to be added to sense the state of various parts of the mechanism, such as case (upper vs. lower).
Even after adding all those solenoids and switches, getting a Selectric to talk to a computer was a large project. The Selectric mechanism, as documented in its service manual, had many peculiar requirements. If commanded to shift to upper case when it was already in upper-case, the mechanism locked up and never signaled "done". Same thing for shifting the ribbon direction or initiating a carriage-return. These commands could only be issued at particular times, with the Selectric in a particular state, and then not again until the terminal signaled the operation was complete.

In addition the Selectric spoke neither ASCII nor EBCDIC, but a unique code based on the tilt/rotate commands to the golf ball. That and the bit-parallel interface and peculiar timing requirements meant the Selectric could not be directly hooked up to a modem. Indeed it needed a relatively large amount of logic to reconcile the two devices.

Particularly vexing was the Selectric's lack of a full ASCII character set. The late Bob Bemer wrote that while working for IBM he lobbied unsuccessfully to expand the typing element to 64 characters from 44. The Selectric actually provided 44 characters per case, but the point remains that with 88 printable characters it could not quite produce the full printable ASCII character set.

Nevertheless, between 1968 and about 1980, a Selectric-based printer was a relatively inexpensive and fairly popular way to get high-quality output from a computer.
A popular Selectic-based terminal was the IBM 2741 released in 1965.



Some units, such as the IBM 029 card punch machine of 1965 used a Selectric mechanism under what looked like a modern keyboard.



Other contemporaries such as the IBM 2250 and 2260 terminal may have used a similar arrangement.

[top]Beam Spring Keyboards

Eventually typewriter-based terminals gave away to modern keyboards which used individual electric switches to register keys. These were cheaper, more reliable due to lack of moving parts, and were much easier to interface with digital circuitry. Most companies used some sort of a linear switch - these essentially consisted of a spring loaded plunger which actuated an electrical contact. However, these did not provide any tactile or auditory feedback. IBM, recognizing that typists used this feedback to aid their typing designed a keyboard switch that used the snap action Beam Spring to not only raise and lower a capacitive contact plate, but to provide tactility and a click when the user's keypress caused the beam to invert. Most IBM keyboards made between 1972 and 1981 used these Beam Spring switches.

Keyboards that have been confirmed to use the Beam Spring switch include the 17427xx keyboards used with the IBM 3278/3279 terminals -



The 736xxxx keyboards of the IBM 5251/5252 terminals -



The 5641316 keyboard of the IBM 3101 terminal -



The 2682678 keyboard of the IBM Displaywriter -



The keyboard of the IBM 5100 Portable Computer.



While these are just the ones where the switch type has been explicitly identified, in reality, just about every IBM terminal, console and computer designed during the 1970s had a beam spring keyboard. Some of the more noteworthy examples include the 3101 terminal, the System/38 console, and the IBM 5110 and 5120 portable computers.

Unfortunately, none of these keyboards are compatible with modern computers as they used 15-25 pin parallel connections that were specific to the units they were designed to be used with. In theory an adapter/converter could be developed, but this has never been done with these keyboards.

All known Beam Spring keyboards were manufactured by IBM Canada at Don Mils, Toronto between c. 1972 - 1986. Although it has never been explicitly confirmed, it is possible that these keyboards that came between the Model D Typewriter and the Model F Keyboard might have been designated the Model E.

[top]IBM Model F

The Beam Spring keyboards were very well designed keyboards that were particularly pleasant to type on, however, there were some issues with them -

* The complicated mechanism would have been slow and expensive to produce.
* The keyboards were quite heavy, weighing anything from 5-7 Kg depending on the variant.
* The Beam Spring mechanism itself was unusually large for a keyboard switch, this resulted in rather ungainly tall keyboards with a high-angled slope.

None of these would have been serious issues when used with terminal keyboards - the terminal would have been stationary, thus negating the size and weight problems, and as computing equipment of the time was usually expensive but well built, cost-cutting was not a huge priority. However, with the move towards smaller and cheaper computers like the IBM 5100, IBM needed something more affordable and compact.

The successor to the IBM 5120; the System/23 Datamaster, used a 83-key keyboard based on the 'capacitive' Buckling Spring mechanism patented in 1978. This was the first Model F keyboard, and the earliest confirmed use of the Buckling Spring mechanism in a commercially available keyboard. It used a layout derived from the 5251 terminal keyboard.



Many of the engineers who worked on the Datamaster went on to work on the original IBM PC and managed to convince their superiors to use the Datamaster's keyboard with the PC. The keyboard kept the Datamaster's physical layout, but some of the key functions were changed. (particularly the left-most row - the Datamaster's keyboard had application-specific functions, the PC replaced them with the generic F1-F10 keys) As the PC's keyboard was an external unit with a long cable, the cumbersome parallel link of the older IBM keyboards was replaced with a 4-pin Serial connection using a 180-degree DIN-5 connector. This keyboard had the part number 1501100, and was bundled with IBM's PC and later PC/XT, and is often referred to as the XT Model F.



The keyboard met mixed reviews. On one hand, it was far ahead of any other personal computer keyboard available on the market, and it was praised for it's high quality and pleasant tactile action, with one comparative review of the IBM PC and a competing Compaq model noting that the keyboard was "half the reason to go for the IBM". On the other hand, the layout of the keyboard came under much criticism - IBM's engineers had attempted to come up with a keyboard that mixed the ANSI layout popular with Americans with the ISO layout popular in Europe. The result was a layout that American users in particular disliked a lot. The keyboard also squashed in a lot of keys together, and failed to separate the numeric pad from the main body of the keyboard.

These concerns were taken into account, and when IBM released the PC AT in 1984, it came with a re-designed Model F keyboard - part number 6450200 known as the AT Model F. In addition to a 84-key layout more similar to modern keyboards, the serial link between the computer and the keyboard was redesigned to be bi-directional. This allowed, amongst other things, lights to indicate whether or not the various Lock keys were enabled. This protocol has remained constant since then, and thus the PC AT keyboard is the earliest design of keyboard that is readily compatible with modern PCs.

People who have used both the PC AT and PC/XT designs claim that the PC/XT was more pleasant to type on and was better built. Irrespective of the validity of these claims, the PC AT is still an excellent keyboard in terms of both feel and build quality, and always a favorite of those who have used one. The superior layout and compatibility with modern computers make it a much better keyboard overall than the XT model.



In addition to the aforementioned variants, other Model F keyboards include -

* The keyboard of the IBM 5531 Industrial PC, released in 1984. This was identical to the PC/XT keyboard except for the colour, was IBM's 'Industrial Grey'. The grey colour is quite unusual, and a lot of pictures make it look either brown or green. In person, it is predominantly grey, with a pasty green twinge.



* The keyboards of the IBM 3178 terminals, sometimes erroneously called the "Mod C" because of the designation on the back. This actually refers to the model of the terminal it was designed to be used with. Part number 56409xx depending on version. It used a 15-pin parallel connector. The Mod C1 keyboard had 75-keys, the C2, C3 and C4 had 87. A similar looking keyboard was used by the IBM 3104 terminal.



* The 122-key keyboards used originally for the PC 3270 (an XT with terminal emulation software and hardware) and later the 3179 terminal. Part number 611034x, depending on version, layout, and whether it was for use with an emulator or actual terminal. It used a 240-degree DIN-5 connector. The layout was arguably inspired by the DEC LK-201, and seems to have been the basis for the layout of the later Model M. These can be adapted for PC use, as will be discussed in the Model M terminal keyboard section.



* The keyboard of the IBM 3290 terminal, part number 1387033 (1st picture). Similar to the 122-key models but without the numpad. Unfortunately, there is a similar looking rubber dome version of this keyboard for the IBM 3190 terminal that is far more common, part number 73X3832 (2nd picture).



* The keyboard of the IBM 5291 terminal, part number 4176191 or 1397950. It used a 15-pin connector. It's layout and shape was very similar to that of the original Datamaster keyboard.



* A number of unusual and rare keyboards. For example - a keyboard for IBM's Electronic Medical Records equipment, part number 100A536 was similar enough to the PC/XT keyboard with a slightly different layout. It used a DB9 connector with only four pins enabled.



With the exception of the PC AT keyboard, Model Fs are incompatible with modern PCs. However, adapters have been developed to allow the XT keyboard to work with modern PCs, and 122-key Terminal Emulation keyboards can be adapted to work with PCs by changing the cable for an AT or PS/2 compatible one. However, drivers will be needed to make use of the extra keys.

Unlike the later Model M, the Model F name was not particularly well known about, mainly due to the fact that information labels were usually hidden inside the keyboard (except for the 122-key Terminal models). This also meant that the part numbers are not all that well known about. When attempting to find one of these keyboards, you are better off searching for the name of the computer they were designed to be used with.

[top]IBM Model M (IBM Enhanced Keyboard)

By the mid-80s, the computer market had become quite different to the one IBM envisaged. A large number of companies were making clones of the IBM PC that were 100% compatible for the most part. These clones were much cheaper than IBM's offerings, and IBM doubtlessly needed to save costs wherever possible. The Model F, despite being the result of cost-cutting excercise itself, was too expensive compared with some of the cheap Keytronic and Asian-made keyboards used by the competition. One must bear in mind that the PC/XT Model F sold by itself for between $350-400 in early 1980s dollars. Even though IBM were/are well known for their high prices, one can imagine that the Model F design was still very expensive for IBM to manufacture.

In the mid-1980s, IBM moved over to the membrane-based Buckling Spring mechanism. With the new keyswitch mechanism and user feedback, they introduced a new and improved design of keyboard. The keyboard cable was detachable from the keyboard as well as the computer using an SDL connector on the keyboard side. The keycaps were split into two pieces, a removable top which had the markings on them, and the stem which was attached to the Buckling Spring mechanism. This allowed custom keycaps to be easily provided to the customer. IBM also sold clear plastic keycaps that allowed the user to create their own keycaps with paper inserts. The keyboards had a revised layout, similar in many ways to that of the DEC LK-201 and the 122-key Terminal Emulator boards (which in turn were likely influenced by the LK-201. The F keys were moved to along the top to correspond better with their functions displayed on the monitor, and arrow keys and home/end/pg up/pg down/insert/delete keys separate to the number pad were placed between the main body. Other tweaks to the layout included the duplication of Control and Alt and space for a speaker. Keyboards for terminals and RS/6000 workstations utilized this speaker, but most Model Ms did not have a speaker despite having a speaker grille on the underside of the case.

These newer keyboards were known as the IBM Enhanced Keyboard or 'Model M'. The Model M's 101-key layout became industry standard and forms the basis of the keyboards we use to this day. By means of comparison with the Model F, the original Model Ms would have sold for about $250 by themselves. The first Model M appeared to be the part number 1386303 keyboards used by the IBM 3161 terminal, which entered production in mid-1985. These used a 102-key layout, which differed from the standard 101-key layout by having an extra key in the numpad.



Soon after came part number 1388032, an "Industrial Grey" keyboard designed for the PC AT-based IBM 7531 Industrial PC. This was released in the fourth quarter of 1985, and was the first Model M compatible with modern PCs. As with the Model F, these Industrial boards were identical to the later regular Model Ms except for the colour.



Beginning April 1986, IBM introduced versions of the PC XT and PC AT that used Model Ms - part number 1390120 was for the XT, and part number 1390131 was for the PC AT and the AT-compatible XT/286. The only difference between the two keyboards was the absence of LED indicator lights on the '0120 - as XTs did not support these. The 1390120 and 1390131 used the same controller, so the 1390120 was forwards compatible with ATs, and the 1390131 was backwards compatible with XTs. The 1390120 and 1390131 were manufactured from Jan 1986 to as late as 1991, and a batch of 1390131s made by Lexmark as late as 1996 have been seen. Lower end models of the XT and AT, and the original PC still used the appropriate Model F keyboards until they were discontinued in 1987.




In 1987, IBM released the PS/2 line of computers.. The PS/2's keyboard was part number 1391401 - often considered the quintessential Model M due to their ubiquity. It came with an SDL cable that used the 6-pin Mini-DIN PS/2 connector, although the keyboard could use the earlier AT SDL cable, and vice-versa. It adopted the ovular label most commonly associated with Model Ms, but was otherwise identical to the earlier 1390131.



The 1388032 was replaced in 1987 by the 1390653. Some examples of the 1390653 have been seen with the black logo of the older 1388032 which can make identification more confusing. Aside from the label there does not appear to be any significant difference.



Model Ms made from about mid-1989 onwards did not have any stabilizer bars for large keys (except the space bar). There were also some internal differences - the back plate was made of thinner steel, and whereas older models had the status LEDs connected to the controller board with a yellow wire cable, they now had a thin ribbon cable instead.

Around the same time, the 1390653 began to be superseded by the 1394946.



[top]Lexmark Model Ms

In March 1991, IBM divested its printer, keyboard and manufacturing arm in Lexington, Kentucky, doubtlessly due to competition and the need to concentrate on more profitable areas of their business. The resulting company, Lexmark, took up the job of manufacturing Model M keyboards for IBM. The labels on the underside of the Model Ms made for companies other than IBM began to indicate "Manufactured by Lexmark", although IBM ones were still marked as being made by IBM until 1993. In 1992, Model Ms had a slight logo change - the black writing on the IBM logo was replaced with blue writing. Keyboards made since that point had four drainage holes on the bottom of the underside of the case to make the keyboards more spill resistant.



Lexmark continued the manufacture of 1391401s and related Model Ms, with the only difference being the addition of drainage channels. However, these became rare after about 1994 (which was also the year IBM discontinued the PS/2) and were replaced by some simplified models.

Lexmark's first new variation on the Model M design were part numbers 52G9658 and 52G7000 . These were largely similar to the 1391401, except the SDL port was removed, and the cable was permanently attached. For the most part, the former had double piece keycaps and the latter had single piece ones although there are exceptions to both.



There are a large number of Model Ms with other part numbers that were made by Lexmark (Most of which followed the form xxGxxxx. These part numbers seemed to depend on which system it was intended to be used with as opposed to it's features. Most of these miscellaneous Lexmark keyboards were based on the 52G design, although some were based on the 1391401 or the later 42H.

One particularly common part number was the 82G2383 which appeared, depending on example, to be based on either the 52G9658 or the 52G9700. It came as part of a bundle with an IBM PS/2 mouse that IBM may have sold separately from their PCs. Another one was the 92G7453, which depending on which year it was made in was either based on the 1391401 or the 42H1292. Part number 60G3571 was given to 52G9658s that were sold as retail units seperate to IBM computers.

One part number that potential buyers should be aware of is the 71G4644. This keyboard is similar to the 52G7000 except that it uses Rubber Dome switches instead of Buckling Springs. Unfortunately they are identical to other blue label Model Ms except for the keyswitch, so besides taking a look under the keycap the part number is the only way to identify them.



The 42H1292 is a rather noteworthy variant. It entered production in 1995 and is the most common variant of the Model Ms made from 1996 onwards. It appears to be a modernization or cost-cutting redesign of the Model M. The case was made of a different plastic and was slightly redesigned (42H based keyboards do no have a speaker grille on the underside unlike earlier Model Ms). The internals were redesigned - there was a lighter backplate and a different controller board. The keyboard was lighter than the 1391401 by about 0.2Kg. They had either single or double piece keycaps depending on the example. They can be easily identified by their permanently-attached straight PS/2 cable (as opposed to the coiled cables that had been standard on IBM keyboards since the PC/XT days) and different design of border around the lock lights.



It should be noted that IBM's factory in Greenock made keyboards in the style of Lexmark's, so it appeared to be an IBM decision to make the cheaper models as opposed to Lexmark's. Some have said that keyboards such as the 42H1292 were not as stiff as the IBM-made keyboards, which may be an advantage depending on individual preference.

Lexmark continued to make keyboards until they sold their keyboard making division to Unicomp in 1996. From that point, IBM manufactured most of their Model Ms in their Greenock plant. Unicomp manufactured designs that has been specific to Lexmark such as the Space-Saver, the M4s, M5s and M13s (Unicomp had also bought out the patent for the Trackpoint). Maxi-Switch continued to manufacture Buckling Spring keyboards for IBM, in particular the black M13. This arrangement continued until 1999 when IBM-branded Model Ms appeared to have gone out of production.

[top]International Model Ms - Greenock and Mexico

After the 'all-size-fits-all' approach taken to physical layouts with the Model F, IBM decided to implement region specific layouts on the Model M. The 102-key ISO-style keyboards favored by most of the world outside of the US and the Far East were produced in IBM's plant in Greenock, Scotland (which is also where they made their PCs and laptops). All Model Ms had the necessary membrane contacts to support both ANSI and ISO layouts, so the only physical difference aside from the different keycaps was the location of the spring/hammer assemblies to accomodate the appropriate keys.

Early Model Ms were just simple ISO versions of the appropriate US Model M. The 1390136, for example, was a UK layout version of the 1390130. Similarly, the 1391406 was a UK layout version of the 1391401. Lists of country codes can be found on Wikipedia or the Keyboard Reference in the Geekhack Wiki.

Unlike in the US, the Greenock plant remained in control of IBM after the Lexmark divestiture. IBM continued to manufacture ISO layout Model Ms, but they followed the same general trend of Lexmark's keyboards and adopted the 'blue' logo and drainage holes. Unlike the American keyboards, they changed the designs without changing the part number, meaning that for each ISO part number there are three distinct variations - Around 1994 they switched from a 1391401-based design to a 52G9xxx-based one. By 1995 or 1996, they had changed to a 42H1292-style design (see the Lexmark section for details).

Greenock-made Model Ms have some distinguishable features - they have electrically-grounded space bars, and the models with drainages holes in the bottom do not have drainage channels inside.

In addition to the usual 13914xx part numbers, IBM also manufactured the keyboards under a wide variety of almost random part numbers that didn't follow any particular pattern. They also made some rubber dome equipped variants, these can be avoided by ensuring to buy a keyboards with a 13914xx part number. Most of them however had part numbers of the form 71Gxxxx.

There were also US layout keyboards made in Greenock. All 42H1292s made after 1996 were produced here. Earlier variants were also produced in Greenock, sometimes with the same part number, but there were some Greenock-specific US keyboards - the 1396790 was a standard 101-key keyboard that was either based on the 1391401 or 52G9658 depending on the year of manufacture. The different part number may have arisen out of the fact that it might have been intended for non-US markets that used the US layout such as the Netherlands or South Africa.

The 1394950 was an Industrial model made from 1996 onwards. It was largely identical to the earlier 1394946 with one distinguishing characteristic - instead of the hard plastic label with raised silved letters used by it's predecessor, the 1394950 used a simple flat sticker.

In addition to Greenock, IBM made keyboards from it's Mexican PC manufacturing center in Guadalajara, Mexico. These keyboards were largely intended for the Latin American market, although there were also some French Canadian ones, and ones for the US market. Not much is known about when production of Model Ms began or ceased in the Mexican plant, although it is known that IBM started manufacturing PCs in the plant in 1986, and some examples have been seen with dates as late as 1994 (which was also the year the PS/2 ceased manufacture, and thus the year when the Model M ceased to be standard equipment for IBM PCs). Unlike the Greenock ones, they appear to have no defining characteristics or differences from the IBM US/Lexmark Model Ms. IBM Mexico Model Ms should not be confused with the Maxiswitch made IBM Model M derivatives which were also made in Mexico, albeit in a different plant in a different part of the country.

As with US Model Ms, they went out of production in 1999. For most Model M sub-variants (with the main exception of the M15) there were ISO versions. These however are generally much more rare. They also had different part numbers which will not be listed here (check the Keyboard Reference for more details).

[top]Variations on the the Model M


In addition to the mainstream 101-key layout Model Ms described below, IBM and Lexmark made some specialized variants. These are described below.

[top]The Model M Mini (IBM Space-Saving Keyboard)

One early variation upon the Model M which appeared in 1987 was the "IBM Space-Saving Keyboard" which was largely similar to the 1391401, except the numpad was removed from the keyboard, leaving a more compact 84-key design for people who did not have room for a full size board. In order to facilitate rapid entry of numbers, Shift+Scroll Lock would activate a 'numpad emulator' which turned the right hand side of the main block into a sort of numpad.



ISO variants are very rare, but French, German and Spanish ones have been confirmed to exist, and others were probably available. Unlike other international variants, these were all made in the US by either IBM, Lexmark or Unicomp. The most common variant is the UNIC046 made by Unicomp. Often described as a UK layout keyboard, it's more like a US layout keyboard that uses the ISO key arrangement.



There were also some Industrial Grey examples, part number 41G3600, but these are also very rare.



IBM had a bewilderingly large amount of part numbers for the regular keyboard, none of which appear to have distinguishing characteristics. The keyboard reference is the best place to find a list of these different numbers. Lexmark didn't introduce any new designs, and aside from the blue logo change, they didn't modify the design - although some (but not all) Lexmark-made ones had drainage channels. Some have single piece keycaps, but these do not seem to be associated with any particular part number. As with other Model M designs, they went out of regular production in 1999, with later examples being produced by Unicomp.

The Space-Saver keyboards were optional extras that one would have to pay extra for. As such, they are nowhere near as common as fullsize Model M variants and command a hefty premium on the second hand market, with used ones in good condition routinely going for over $100 on eBay.

[top]The Model M1/M2 (IBM Selectric Touch Keyboard)

The Model M2, part number 1395300, was an attempt at making a cheap and lightweight Buckling Spring keyboard for IBM's PS/1 - a home computer released in 1990. The keyboard was also used with the later PS Value Point. Compared with the contemporary Model M 1391401 it -

* Doesn't have the inbuilt slope of the Model M - it's flat.
* Has shorter, flatter single-piece keycaps.
* Doesn't have a backplate.
* Has a straight, non-removable cable.



On paper, the compact and lightweight M2 looks like an excellent design. Unfortunately it is plagued with electronics problems. Specifically, the capacitors in the logic board are prone to failure. It is believed that the capacitors used immature surface mount technology and where therefore inherently prone to failure. The usual method of diagnosing capacitor failure is that the Caps and Scroll Lock stay on whilst the keyboard doesn't respond to keypresses. However, there is plenty of detailed advice on Geekhack on how to replace the capacitors, and this seems to take care of the problem. It really should be stressed that this is a prevalent problem - the owner of clickykeyboards.com claims that 4/5 of all M2s he deals with are duds. Potential buyers should not expect long term reliability from an M2 unless they are prepared to replace the capacitors.

Another unusual thing about the M2 is that instead of metal stabilizer bars, the large keys have plastic stabilizers permanently attached on to them. When disassembling, care should be taken to ensure that they don't snap. There are other signs of lower grade construction quality - instead of the membrane connecting into the controller PCB with a ribbon cable, the underside of the PCB has electrical contacts, and the contacts on the PCB 'sit' on top of contacts on the membrane, and are held in place with flimsy plastic tabs.

The Model M1, part number 60G3570, appears to be identical to the 1395300, except that it is sold under the "Easy Options by IBM" branding as a retail unit, as opposed to the M2 which was bundled with IBM computers.

One should also be on the watch for Rubber Dome versions of this keyboard. Fortunately, the most common rubber dome variant, part number 60G0817, can be spotted by it's large plastic border. However, there are some further rubber dome variants that may not have this border, such as part number 73G4614.



There are also POS equipment variants that connect via a RJ-11 connector. Aside from the connector, the easiest way to spot these is by their keycaps which have non-standard markings and shapes.

In general, one should make sure to stick to ones with the 1395300 part number if they want to buy one for their PC.

Guide to replacing the capacitors.

[top]Pointing-device Keyboards M5-1/M5-2/M13

Beginning 1993, Lexmark manufactured a number of keyboards with in-built pointing devices that were designed to either supplement or replace the traditional mouse. These keyboards were -

* The Model M5-1, part number 1370478. This was a 1391401-based design which had a small trackball and mouse buttons placed around the arrow keys. In addition to the two standard buttons, there are two more which are used for drag and drop operations. They didn't seem to be produced for very long, and aren't particularly common.



* The Model M5-2, part number 59G9757 or 92G7455. Similar to the M5-1 except the trackball and buttons were placed on the top right corner of the keyboard. The buttons are duplicated both on the surface and back-end of the trackball unit. While still rare, these are more common than the M5-1.



* The Model M13, part number 82G3281. It had an integrated Trackpoint II pointing stick and two mouse buttons. There was also a Lexmark branded version with part number 1403380 which is somewhat common. When Lexmark ceased manufacturing keyboards, Maxiswitch started making them. These keyboards have the part number 92G7461.



* Part number 13H6705 was an all black M13, the only black IBM Model M, and one of the most popular Model M variants. The lettering on the keys is pad printed as opposed to the dye sublimation technique used for other Model Ms, meaning that the letters are more prone to wear than regular Model Ms are. Earlier examples were made by Lexmark from 1995 to 1996, whereas later examples were made by Maxiswitch from 1996 to as late as 1998. There does not appear to be any difference between examples made by the two manufacturers.



* Part number 06H4173 was an Industrial M13 made by either Unicomp or Maxiswitch. Unicomp made ones use the Force Resisting Gauge design, and therefore isn't as desirable as the Maxiswitch variant which used the regular Trackpoint II. These two variants can be distinguished by their label - the Unicomp one used the same flat blue writing on white background as the regular beige keyboards, whereas the Maxiswitch one used the raised silver letters on black background used by the 1394946 and related Industrial keyboards. The Unicomp is more common than the Maxiswitch one. There may have been a Lexmark one with a different part number, but this has not been confirmed.



The M5s had removable SDL cables which either had a PS/2 y-cable (a plug for the keyboard, and one for the mouse) or a AT-DIN/9-pin Serial y-cable. The M13 had the same connector options as the M5, but with an attached cable only. The PS/2 M13s also had a PS/2 mouse pass-through port to allow the user to use a conventional PS/2 mouse in addition to the trackpoint in the keyboard.

One thing that potential buyers should note is that these trackballs/trackpoints were designed in the days when 640x480 or 800x600 was the standard resolution. Users of these devices with modern hi-res monitors often find them quite sluggish. This can often be resolved by turning up sensitivity and, in the case of Windows, disabling the "Enhance Pointer Stability" option below the sensitivity slider.

[top]The Model M15 (Lexmark Select-Ease Keyboard)

The M15, part number 13H6689, was an ergonomic Buckling Spring keyboard made by Lexmark between 1994 and late 1995, when Lexmark transfered the associated patents to Maxiswitch. Unlike the pseudo-ergonomic 'curved' keyboards made by the likes of Microsoft or Logitech, the angle at which the two halves of the keyboard faced eachother was fully adjustable. It was even possible to detach them from eachother if required. The arrow keys were duplicated on both sides of the board, and the two-part spacebar could be configured so that one of the sides could double as a backspace.



The M15 did not have a numpad, but Lexmark did make an optional seperate numpad that was designed to be used with the M15. These are very rare however.



The M15 was a special order item, and was quite expensive - costing $179, or $199 with the numpad. Between this, and it's short production life, they are quite rare. Consequently, M15s are one of the most valuble keyboards on the market.

The patent for the M15 ultimately ended up in the hands of Goldtouch, who make a variety of split keyboards with rubber dome switches, so it's unlikely that Unicomp or anyone else would be able to make an M15-like Buckling Spring keyboard for quite some time. It also should be noted that depending on the example, M15s would either have an IBM logo, a Lexmark logo, or an "Options by IBM" logo - the latter appearing to be the most common.

[top]Terminal Variants

When IBM switched over to the membrane buckling spring from the capacitive one, many of their dedicated terminal Model F keyboards had the internals switched over to the membrane buckling spring design, and were designated Model M, but otherwise looked and worked the same. These designs went out of production between the late 80s and early 90s as the terminals they were intended to be used with ceased to be supported.

As terminals became less important due to the popularity of PCs, IBM streamlined the design of their terminal keyboards, and as such, a limited number of designs were adapted to all situations.

[top]Unusual Variants

IBM and Lexmark manufactured a large number of custom Model M variants for their customers. Often these came in the form of standard Model Ms with special keycaps to facilitate the keyboard being used with the customer's software of choice. Compiling an exhaustive list of these variants would be largely pointless, so it should be sufficient to note that these keyboards are generally PC compatible, and usually have their own unique model number. Pictured here is a keyboard with special keycaps for IBM's APL2 programming language, which uses a large number of special mathematical symbols.



There were also some variants with more esoteric modifications. One such board was the Soft Touch Keyboard, part number 8184692, made by Lexmark from about 1994-1996, was a 1391401 variant that had each keyswitch spring coated in a lubricant grease that would lower the noise without having a serious impact on key feel. The keyboard also included an internal speaker, and was it was therefore likely that these were originally intended for IBM's RS/6000 workstations. These keyboards are quite rare, and should not be confused with Rubber Dome versions of the Model M which are sometimes referred to as 'Quiet Touch'. Unicomp offers a greasing service for their keyboards. They look identical to other Lexmark-made 1391401s, the primary recognition feature being the special 'Soft Touch Keyboard' logo.



IBM, and (in particular) Lexmark made re-branded Model M keyboards for other companies. Again, compiling a list of companies that got their logo on a Model M would serve no purpose, however, one example that always surprises people is Dell. It seems that from about 1991 until at least 1994, Dell shipped PCs with re-branded Model Ms. Earlier examples have the original Dell logo, whereas newer ones had the modern logo with the slanted E.

In the early 90s, IBM had a (ultimately unsuccessful) subsidiary named Ambra which manufactured low end PCs. In most places, Ambra had standard Model Ms with an Ambra logo. However, in Germany, Ambra keyboards had unique keycaps - they were all white, and used a font and letter positioning similar to that of Apple keyboards -



Other OEM branding includes Alaris, Wang Laboratories, Allen-Bradley and Lexmark.



[top]The Model M4 and M6

Despite what the name may suggest, these keyboards had little to do with other IBM Model M variants, and were instead a family of compact keyboards based around the Buckling Rubber Sleeve mechanism that appeared around 1991. Most common amongst these was the Model M4-1, an 84-key (or 85-key for ISO versions) keyboard with an integrated Trackpoint II pointing stick. These were shipped with the IBM PS/2E, an early 'Green' computer which used Thinkpad parts such a 10" LCD screen, although they would have been used for other applications where a compact keyboard was needed, such as with servers or POS equipment.

Part numbers were 84G2525 (black version) or 1379590 (beige version). These keyboards had a PS/2 pass-through, an SDL socket for an optional numpad and a standard SDL connector for a PS/2 Y-cable.



Other members of the family included the Model M4, part number 1397451 (white), which was similar to above except without the Trackpoint and PS/2 pass-through, and with a standard detachable PS/2 cable. The keypad of the M4 without the housing was used as the keyboard of the IBM L40SX laptop of 1991, part number 1396181.



The M6-1, part number 42H3979, was a version of the M4-1 used in IBM's early Thinkpad laptops. This had a slightly different layout, and had only a ribbon connector for internal connection with the laptop. Later M6-1s had Scissor-Switch mechanisms instead of the Rubber Sleeves. The keyboards of later Thinkpads did not have Model M designations.



These rubber sleeve keyboards were made by Lexmark and Keytronic (particularly for the M6-1). After 1996 these were made by Unicomp for IBM until 1999. Unicomp sold these under their own brand as the 'Mighty Mouse', although as of about 2008, the Trackpoint variants are out of production, and they only make the M4 style unit.

[top]Unicomp

By 1996, Rubber Dome keyboards were the market favorite, and keyboards with mechanical or other such high quality switches were becoming more and more of a niche market item. Lexmark doubtlessly saw the way the wind was blowing and decided to phase out it's keyboard production in favor of concentrating on it's far more profitable printer brand. One of the engineers who had been working with on keyboard production since the IBM days; Neil Muyskens, and a number of his colleagues, bought out the keyboard making machinery and set up their own company in order to continue production of Buckling Spring keyboards. This company was called Unicomp and was found in 1996. After IBM discontinued the Model M in 1999, Unicomp continued to sell Buckling Spring keyboards to the market directly.

Unicomp offers customization services for it's keyboards, allowing customers to have keyboards with custom keycaps, logos and key layouts. (albeit within the confines of the standard ISO or ANSI physical arrangements) They can repair old IBM keyboards - generally this involves them replacing the internal keyboard assembly with one of their own new ones. As discussed above in relation to the Soft Touch Keyboard they offer a greasing service which quietens their keyboards. This modification costs $40 (presumably due to the labor involved in greasing 100+ individual springs) and will void your warrant (due to concerns about the grease leaking downwards and damaging the membrane). A home-made version of this mod will be discussed below in the appropriate section.

Unicomp is highly regarded amongst keyboard enthusiasts for their excellent customer service. It should be noted however that sometimes they can take several days to reply to emails, and they appear to prefer telephone inquiries instead.

[top]Unicomp's Keyboards

Unicomp started out with the following models -

* The Customizer, a re-branded 42H1292 design.



* The On-The-Stick, an M13-style design that used a Force-Sensing Resistor pointing stick instead of an IBM-style Strain Gauge. Unicomp cited cost and power requirements as advantage of the FSR design. Amongst Geekhack members, the FSR is universally considered inferior to Strain Gauge Trackpoints due to the wobble in the stick, and the low sensitivity.



* The On-The-Ball, a Customizer with an M5-2 style trackball. The On-The-Ball Plus had an FSR Trackpoint. Rubber Dome versions of the On-The-Ball are very common on eBay (part number 96U1114), so one should take care to ensure they get a Buckling Spring version. The Unicomp version is far more common than the original IBM edition. There was also another variant with a pointing stick as well as the trackball known as the On-The-Ball Plus. OTB Plus's were not produced very long and are very uncommon.



* The Mighty Mouse, a re-branded M4 and M4-1. Unlike the On-The-Stick, it used the old Strain Gauge design of Trackpoint. However, the keys themselves are considered quite wobbly compared with the Lexmark made models.



* A family of 122-key terminal emulator, dedicated terminal and POS keyboards. There are also some 102-key terminal keyboards.

* A 'Value' keyboard which used Rubber Domes (made by Silitek, not Unicomp). There was also a wireless version.

In 2000, Unicomp released a new model, the 'Endurapro'. This was in fact the Japanese 5576-C01 trackpoint keyboard converted to the US layout with Windows keys, and the FSR Trackpoint.



In 2003, Unicomp began offering all-black versions of Customizers and Endurapros.



There were problems with the white lettering on the black keycaps wearing away (see the section above on the black M13 for further description), so the black keycaps were discontinued around 2006 in favor of metallic grey ones.

In 2006, Unicomp began selling USB versions of the Customizer and Endurapro, and a version of the Customizer with Windows keys, which was only availabe in black.



In 2007, Unicomp began selling the 'Spacesaver' - an Endurapro without the Trackpoint (the marks for where the mouse buttons are meant to be can still be seen on the Spacesaver).



In 2008, Unicomp discontinued the On-The-Stick, On-The-Ball Plus, and the Trackpoint version of the Mighty Mouse. (one can only assume they ran out of Strain Gauge Trackpoints and could make a low-height version of the FSR design)

[top]IBM Japan's Keyboards


[top]Keyboards for Asia

Due to complications with oriental alphabets, IBM Japan had to devise novel means of accommodating such alphabets on their computer keyboards. One such early attempt was this monster, for use with an unknown Japanese terminal from the 1970s-



Not much is known about it, it may have been a Beam Spring keyboard, although, judging by the trend of later IBM Japan keyboards, it may have used Alps switches. There was supposedly a Chinese version that consisted of multiple keyboard 'trays' to accommodate the necessary characters. On the Japanese keyboard, there are 12 characters per button. Despite this, it has well over 200 keys.

In the 1980s, IBM's first PCs had no means of supporting non-Latin alphabets on either the hardware or software level, and IBM Japan developed their own PCs that differed greatly from the American/European equivalents. The first such design was the Multistation 5550 of 1983 - marketed as a combination PC, word processor (dedicated word processors were supposedly popular with the Japanese market at the time) and terminal, and was roughly analogous to a PC/XT with terminal emulation hardware.

The 5556, the keyboard used by this design was rather unusual by IBM standards in that it used Alps linear mechanical keyswitches with either Green or Yellow sliders. Owing to it's terminal role, it used a Japanese version of the 122-key terminal emulator layout. These extra keys were also used by a popular Japanese word processing software that was bundled with the Multistation. It shared the approximate physical shape and size of the PC/XT Model F, except was slightly wider to accommodate the larger layout. There were four versions of the board, the 5576-001 through 5576-004. The 001 and 002 used "hiragana" characters while the 003 and 004 had "katakana" ones. The 004 had a different layout to the 001, and probable that the 002 had the 004 layout with the same characters as the 001, and the 003 had the layout of the 001 with the same characters as the 004. The 001 is pictured below. These designs were only compatible with the Multistation.



In 1987, IBM Japan began manufacture of the PS/55 line of computers which were broadly similar to the PS/2. The keyboards associated with the PS/55 were the 5576 family. These used the tactile 'plate spring' mechanical switches developed by Alps for IBM Japan. Like the Model M, these used two-piece keycaps, a detachable SDL to PS/2 cable, and had space for a speaker. However, the keycaps were of a different size to the Model M, and they used a different shape of SDL connector, meaning that these parts were not interchangeable between the two boards. Unlike the Model M, all 5576s had a speaker, and they also featured a volume slider for the speaker. There was a PS/2 terminal emulator keyboard, the 3479-JA4, which had a slightly different layout - like the 5556-004.

The first member of the family was the 5576-001. It had essentially the same layout as the older 5556.



This was followed soon after by the 5576-002, this dispensed with the extra keys, and settled on the 106-key design. Despite having the same amount of keys, the layout was different to the later standard Japanese 106-key layout.



The 5576-003 appears to have been the associated keyboard of the PS/55Z of 1989. Compared with the -002, it dispensed with the numpad, and used a membrane buckling spring mechanism. IBM Japan used a unique implementation of the Buckling Spring, designed for them by Brother Telecom, and consequently known as the Brother Switch amongst Japanese keyboard enthusiasts. Unlike the Model M spacesaver which had a 'Numpad Emulator' built into it, the 003 had a PS/2 pass-through that would allow the user to attach a seperate Buckling Spring numpad. There was a terminal emulator version called the 3472-JK3.



In 1990, with the DOS/V OS launched by IBM and Microsoft to simplify Japanese alphabet input, they devised a new Japanese keyboard layout and launched the 5576-A01. Despite the physical layout being similar, it had a different key layout to the 002. The A01's key layout became the standard for Japanese keyboards, much in the same way the Model M set the standard for keyboards in most other parts of the world. It used the same Brother Switches as the -003. Unlike it's predecessors, the A01 did not have a speaker. It had two-piece keycaps compatible with its Alps-based predecessors. The A01 is generally considered to be the closest Japanese equivalent of the Model M. There were OEM branded versions, such as Ricoh, Mitac and Mitsubishi.



In 1991, Sega launched the "Teradrive", a unit that combined the functionality of a Sega Megadrive console, and an IBM PC/AT compatible computer running DOS/V. The unit was manufactured by IBM Japan for Sega. The keyboard used a buckling spring over membrane mechanism designed and manufactured by Alps, and used a standard PS/2 connection. The Teradrive was a flop - as with a lot of IBM products of the day it combined unsatisfactory performance with a hefty price tag. The units fetch high prices due to their demand with game console collectors and enthusiasts, so finding the associated keyboard 'in the wild' is unlikely.



In 1993, IBM Japan launched the PS/55E, the Japanese equivalent to the PS/2E. The keyboard used with this design was the 5576-C01. Unusually, this keyboard was not made in Japan, but by Lexmark in the US, and used regular Buckling Spring switches. This keyboard could be roughly considered a Model M13 with a redesigned case and Japanese layout, and thus had very little in common with the A01 or indeed any of the other 5576 boards, and had none of the distinguishing features like the speakers or two piece keycaps. One unique feature however was that a part of the back of the keyboard would twist to the side, and allow the keyboard to be balanced on it's back when not in use. This would have been a useful feature in Japanese offices where space is very much at a premium. The C01 served as the basis for the Unicomp Endurapro. Somewhat unsurprisingly, Unicomp dropped the swivel support from their design.



IBM Japan ceased production of the PS/55 line in 1995 and the successor designs were equipped with rubber dome boards. The first such design was the 5576-B01 made by Mitsumi. This used a rubber-sleeve like design and was superficially similar to the earlier 5576 designs, although it dropped some features like the two-piece keycaps, the detachable cable and probably the speaker/volume slider. Later designs, 5576-B05 and 5576-B06 were standard rubber dome keyboards made by Chicony, and were Japanese versions of the standard rubber dome keyboards found in other countries. The B05 was white, and B06 was black. Aside from the name, these had nothing in common with the earlier 5576s. Pictures of the B0x keyboards, the terminal versions of the 5576 keyboards, and more pictures of the other 5576 keyboards can be found on this site.

[top]Keyboards for the Rest of the World

IBM Japan produced some keyboards for computers that were also sold outside the Asian market, and are thus worthy of a seperate mention.

One of the first such computers was the IBM JX of 1984, a Japanese equivalent of the IBM PCjr, which was also sold in Australia and New Zealand. The JX was far more successful than it's American equivalent, and one of the reasons cited was the higher quality keyboard options. There was a 'standard' keyboard which had a similar layout to the American PCjr, and was infra-red wireless only (powered by two AA batteries), and a professional layout which had an expanded layout, and could be used either wirelessly, or with a detachable cable. Both keyboards have linear Alps switches, some having green Alps, others have older style Alps switches with a white slider (also linear). Japanese ones were available in black or white, whereas the English ones for the Australian market were available in black only.




In 1986, IBM replaced the archaic Portable PC 5155 with the PC Convertible 5140. The 5140, as with a lot of IBM's later portable designs (including the Thinkpad) had large portions of its design and manufacture done by IBM Japan. The keyboard used Alps linear mechanical switches (brown sliders for most keys, and half-height switches with green sliders for the half height keys). The 5140 was a flop due to it's low computing power, heavy weight, and high price, and is relatively uncommon.

In 1988, IBM launched the PS/2 Model P70 - a 'portable' computer that folded up into a briefcase-style arrangement. It used a keyboard with Alps plate-spring click switches - making it the only keyboard using such switches that was available outside of Japan. However, in the interest of making it suitable for its portable computer role, it was not as well built as the IBM Japan desktop keyboards using this switch type, having a flimsy case that snapped together.



[top]Other IBM keyboards


There are a number of IBM keyboards of interest to the collector that do not fit into the above categories.



[top]IBM's Keyswitches

Note: The descriptions of the switches in this section are not intended to be 100% accurate, but merely to provide the reader with an easy to follow explanation of how they work. For a more accurate description, one should read the patents.

[top]Beam Spring Switch





(Animated picture by forum member bhtooefr)

Officially described in a related patent as the "combination beam spring and fly spring" or "beam/fly spring interlock" mechanism, this mechanism was designed to emulate the experience of using a Selectric typewriter - the patent specifically discusses replicating the tactile feedback, auditory feedback and key pre-travel of a typewriter in order so that operators trained to use typewriters would be able to adjust quickly to using an electronic keyboard. It was invented by John Wolfrom and Richard Hunter Harris, both of IBM. The patent was filed in the US in 1971. No traces of the US patent can be found so it can be assumed that the application was not successful. Instead, most information comes from the patent filed in the UK, and a later US patent which described a two-stage variant of the same mechanism. These patents can be seen here and here.

At the most fundamental level - the switch consists of two parallel electrical contacts on a PCB covered over with a layer of insulating substrate. Resting directly on top of the substrate is a so-called 'Fly Plate' - a piece of conductive plastic. A capacitor is formed between the fly plate and the two contacts (of a capacitance of 4 picofarads according to the patent)

Attached to each end of the fly plate are two metal arms, each with a slot near the top (the so called 'Fly Spring'). A flat piece of metal with tabs on either end slots into the fly spring. The metal is tensed upwards into an arch between the two arms, consequently forming a Beam Spring (more commonly called a leaf spring or flat spring, these are a type of flat, arced spring often used in car suspensions). The midpoint of the beam spring is attached to a spring loaded plunger, on the opposite end of this the keycap is securely attached.

Pressing on the keycap depresses the plunger which in turn applies pressure onto the mid point of the beam spring. Eventually the beam spring inverts (from a dome shape to a 'u' shape) - this causes the tabs on the beam spring to push the two arms of the fly spring outwards, allowing the beam spring to pull the flyplate up. Once the flyplate is raised, the arms on the fly spring snap back into place, holding the fly plate in place. Two studs inside the switch housing limit the upwards travel of the fly plate. As the distance has changed between the fly plate and the contacts, the capacitance has decreased (down to 0.4 pF, according to the patent again). One of the electrical contacts is connected to a sensor in the keyboard's controller, which registers the decrease in capacitance as a key press.

Releasing the key causes the plunger to raise itself to its default position, which in turn pulls up the beam spring upwards, reverting it back to its rest position (making another audible snap in the process) which turn lowers the fly plate back into it's resting position.

The switch has an actuation force of about 50-55g, and the patent specifies that it had been shown to work after over 100 million key presses. The keyswitches were usually entirely covered over with a layer of insulating rubber, and the keycap slotted on to the mechanism over the rubber, guaranteeing spill and dirt resistance (despite the fact that the mechanism itself would be hard to damage with liquids or dirt). Due to this insulation and the heavy duty construction of the keyboards, they were relatively quiet compared to some later IBM Models, being described as quieter than a Blue Cherry keyboard. Without the various layers of insulation, the switches are quite loud.

A dissection of a Beam Spring switch can be seen here. A general discussion of beam spring keyboards, and pictures of disassembled keyboards using them can be seen here.

[top]Original Buckling Spring Switch

The original IBM buckling spring mechanism was patented by Richard Hunter Harris in 1972 as the "Catastrophically buckling compression column switch and actuator". The patent can be seen here. It was never used in any known production keyboard, but is included here to show the progression of the buckling spring design.

The patent described three related mechanisms based around the buckling property of a spring for use in a keyboard switch. The first one consisted of a spring in a sealed housing (barrel) tensed between the keycap and a fixed point directly underneath which was electrified. The barrel surrounding the spring was conductive, and hooked up to a circuit. Depressing the key would cause the spring to buckle, and it would make contact with one of the sides (in such a simple mechanism, the spring could buckle any way) and thus completing a circuit. This mechanism would be very simple to implement, but the electrical contact mechanism proposed would be prone to bounce issues.



Consequently, the two subsequent variations proposed more sophisticated electrical mechanisms. In the second variation, the spring instead of being flat and straight as in the first description was slightly pushed forward. At either side of the spring, the barrel was closed to prevent the spring buckling sideways, and a wedge was positioned behind the spring to prevent it buckling backwards. In addition, the wedge was conductive, so that the electrified spring made contact with it to form a circuit. When the spring was buckling by the depression of the keycap, it made contact with another conductive surface, making yet another circuit. Thus there was a circuit for when the key had been pressed, and another for when it was depressed, and this would have provided a more desirable electrical switch at the expense of complicating the mechanism.



The third variation was similar to the second in terms of the fact that it ensured the spring buckled forwards, but it did away with the electrification of the spring. Instead, when the spring hit the appropriate section of the barrel it hit against some sort of sensor which detected that the key had been depressed. A number of suggestions are provided in the patent for what should be used, including a piezoelectric sensor, a photosensitive sensor, and most notably a capacitive sensor - which was what was ultimately used in the refined buckling spring mechanism patented six years later. It's interesting to note that if the capacitive sensor was based around detecting the distance between itself and the spring, it is in some sense ways similar to the operation of the Topre capacitive mechanism.



The patent promised a simplified switch design which would combine the tactility, actuation, and hysteresis (i.e. the ability of the switch to return to it's rest position when released) into the one spring, but ultimately it seems the means of actuating the circuit was difficult due to the unpredictability of which way the spring would buckle, and mechanisms to control the direction of the buckling added complication and potential unreliability to the design.

[top]"Capacitive" Buckling Spring Switch

Patented in 1978 by Richard Hunter Harris, this was a significant improvement upon the earlier buckling spring design, and was used in the Model F series of keyboard and some of IBM's electronic typewriters. The patent officially referred to it as the "Buckling spring torsional snap actuator". The patent can be seen here.

The main problem with the mechanism described in the original buckling spring patent, and other proposed keyswitch designs based around the buckling action of a spring, was that in a mechanism where a straight spring was attached to two fixed points, and compression, occured (i.e. it was pressed down on) the direction in which the spring would buckle was indeterminate, and this meant that extra elements had to be added to the switch to ensure that it buckled in the right direction. This was undesirable as it added complication and potential sources of failure to the switch.

Harris' solution was to change the design so that the spring was tensed between the keycap and a pivoting hammer that could only move in the desired direction, and that the spring was slightly bent forward in the desired direction of travel. These two measures ensured that the spring could only buckle one way.

But more importantly, instead of using the spring itself to actuate the contacts, the purpose of the spring was to drive the hammer forward. The hammer was made of the same material as the fly plate in the Beam Spring switch, and in terms of how they register key presses, they are somewhat similar - at rest, the hammer is raised over two capacitive contacts. When the key is pressed, the spring buckles forward, pushing the hammer onto the plates, and causing a change in capacitance which is recognized as a key press. In fact, is is possible to use this mechanism on the PCB of a Beam Spring keyboard, and vice-versa, except for one major difference - they work in reverse. The Beam Spring registers a key press when the hammer is pulled away from the contacts, whereas this buckling spring mechanism recognizes a key press when the hammer lies right on top of it.

When the key is released, the spring 'un-buckles' into its default state, which in turn pulls the hammer up and pushes the key back into rest position. The ingenuity of this mechanism is that a single spring provides the tactility, the movement of the contact element, and the hysteresis (i.e. the ability of the key to return to its rest position). Compare this with the earlier Beam Spring switch that required a different spring for each one of these tasks.

The mechanism is referred to as the "Capacitive" buckling spring to differentiate it from the later buckling spring design that used a membrane instead of capacitive contacts. The capacitive buckling spring mechanism is popular due to the crisp tactility and loud feedback it provides, which is widely considered superior to that of the later membrane-type buckling spring. It also has a lighter actuation force of about 60-65g of force compared with 65-70g for later designs. However, these advantages come not from any intrinsic superiority of the design itself, but rather the superior construction quality of the keyboards that used it. In terms of inherent advantages, the capacitive mechanism is far more reliable than membrane or electrical switch based keyboards as there are no contacts to be worn out - the keyboard merely senses a change in capacitance caused by the movement of the hammer. Thus the switch can take about 100 million keypresses before failure (compared with >5 million for a rubber dome keyboard)

[top]"Membrane" Buckling Spring Switch



This was a revision upon the earlier capacitive-style buckling spring, patented in 1985 by Edwin T. Coleman of IBM as the "Rocking switch actuator for a low force membrane contact sheet". The patent can be seen here. It is the most common form of buckling spring mechanism, and for all intents and purposes, when someone refers to buckling springs, they are referring to this mechanism in particular - it was used in the Model M style keyboards manufactured by IBM, Lexmark, Maxiswitch and Unicomp, as well as some electronic typewriters such as the Wheelwriter series.

From a technical viewpoint, there is not much to be said for them in terms of how they work - they are exactly the same as the capacitive type buckling spring, except instead of the hammer hitting capacitive plates to actuate the key, the hammer strikes down on contacts on a plastic membrane, which completes a circuit. In this regard, they operate in the same way as a regular rubber dome keyboard.

Multiple sources outline the sole motivation for changing from capacitive switching to electrical switching using a membrane - cost. The patent for the mechanism claims that the cost of producing a keyboard with a membrane based buckling spring was about half that of the capacitive one. The large reduction in cost was done at the expense of durability and reliability - the membrane buckling spring mechanism is rated by Unicomp for 25 million key presses. The membrane will also be ruined if liquids seep down into the membrane, whereas the sealed capacitive PCB of the earlier design can withstand liquids and be cleaned easily.

There are subtle tweaks to adapt the mechanism to use with a membrane. The hammer is much smaller, and the springs have a different number of coils on them, causing them to be more stiff, and the experienced tactility to be not as smooth. The patent talks about the force required to mash the two contacts on the membrane together, and it is quite likely that the springs had to be designed to ensure that the hammer struck the membrane with just the correct amount of force.

[top]"Brother Switch" Buckling Spring Switch



(Pictures from Qwerters Clinic, linked at bottom of section)

As discussed under the appropriate section, due the special requirements of Japanese computing, IBM Japan seemed to have a large degree of autonomy in making their designs. In addition, they seemed to outsource a lot of their production - one can assume that with the huge array of electronics companies in Japan that this would have made a large degree of economic sense. As can be seen from the Japanese section, a lot of their early PC keyboards were made by Alps.

When they released the -A01 keyboard as part of the DOS/V initiative, they decided to adopt a buckling spring design. Rather than go to the effort of setting up their own production facility of the US-style buckling spring, they commissioned Brother Telecoms to manufacture a buckling spring mechanism for them. In addition, by making their own custom buckling spring, they could make ones that could be easily adapted to their Alps-based keyboards casings and keycaps in order to simplify the changeover.

The Brother Switch (as it is known by Japanese keyboard collectors) operates like a membrane buckling spring, but differs primarily in the construction of the keyboards. In the Model M, the barrels in which the buckling springs sit are all formed on the one sheet of plastic. In the Brother Switch, each buckling spring assembly has it's own individual barrel, which can be removed from the keyboard without dismantling it.

It has been erroneously claimed that the Brother Switches are individual switching units, like Alps or Cherry switches. In reality, because they all share the same common membrane, this is not the case. Model F keyboards are actually constructed in a very similar way, except the individual barrels are covered and held in place by a metal plate, meaning they cannot be removed without completely dismantling the keyboard. While it might seem like a clever idea to have separate buckling spring assemblies to allow broken ones to be fixed, the reality is that the membrane common to all the removable switches is a much more likely point of failure. It is more likely that the individual nature of the buckling spring assemblies was more suitable to Brother Telecoms for manufacturing than the Model M design.

The Brother Switch uses a shorter spring than the Model M resulting in a stiffer and shorter travel. They also have quite a distinctive noise - a metallic 'twang' as opposed to the ping of Model M and Model F switches. Nonetheless, they have been described as very pleasant to type on, and superior to the Model M by some.

A great dissection of the mechanism can be seen at the Qwerter's clinic.

[top]Alps Plate Spring Switch



(Picture from Qwerters Clinic, from link at the bottom of this section)

Referred to as a plate spring, or leaf spring swich, this Alps switch was designed specifically for use in keyboards for IBM Japan, and was not used in keyboards from other companies, and is thus worth a mention here,

The switch could be described in some ways as a simplified Beam Spring switch, as it uses a metallic flat spring to generate tactility and actuate the contact mechanism. The design is far more simple though - the slider inverts a flat spring which mashes two metal plates together, completing a circuit. Upon release of the key, the flat spring inverts, releasing the pressure on the plates, and pushing the slider back up.

The mechanism provides an action that is described as particularly pleasant compared with regular mechanical switches by Alps or Cherry. However, it appears that the flat spring was prone to wearing out, affecting the tactility and ultimately causing switch failure. This switch was used by the 5576-001 and 5576-002 keyboard as well as the keyboard of the P70 portable computer, which was the only keyboard using this switch outside of Japan.

Further pics of the mechanism can be seen at the Qwerters Clinic.

[top]FAQs, General Knowledge and Miscellaneous

[top]What makes it click?

Primarily it's the buckling of the spring, and the sound that it makes when un-buckling, but there are a number of other factors to consider.

If you buckle a buckling spring assembly in your fingers, while it makes a click, it is not particularly loud. Yet, in designs like the Model F that let you remove the barrel, the spring buckling in the barrel sounds very similar to how it would when assembled inside the keyboard, with any difference accounted for by the additional acoustic properties of the barrel when located inside a closed keyboard assembly. So the acoustics of the keyboard's construction are actually more responsible for the volume of noise than the spring itself.

The sound of typing on a buckling spring keyboard has a sort of metallic 'ping' to it. If you slowly press down a key on a buckling spring keyboard in good condition, it makes a clean 'snap' sound. The release however, makes a ping. There are two possible explanations for this -

A) The patents note that the spring un-buckles at slower rate than it buckles, thus accounting for some difference in the sound.
B) When the spring buckles, it hits against the barrel. This presumably would absorb the vibration of the spring, whereas when it buckles, it is free to vibrate, which is likely why it causes the ping sound.

It has been suggested that the sound of the hammer hitting the membrane or capacitive PCB explains some of the sounds. In fact, this is not the case. If you take the buckling spring/hammer and removable barrel from a Model F (and presumably a Brother Switch keyboard as well) and position such that the pivot rests just on the edge of a desk or table, and the hammer hits nothing, the key makes almost exactly the same sound as it would if the hammer hit the PCB or membrane.

[top]Do these keyboards last forever?

[top]How can I make them less noisy?

Various quieting mods have been successfully tried on Model Ms:

Silicone grease mod

Dental floss mod

The Nuts and bolts mod isn't purely a sound mod, but it may help reduce pinginess. See also Ripster's American Version of the nuts and bolts mod.

[top]How can I make mine work with a modern PC?

[top]What's the deal with those plastic rivets?

[top]Keycaps

[top]Cables and Connectors

[top]Disassembly Guides


HTML Code:
<a rel="license" href="http://creativecommons.org/licenses/by-nc-nd/3.0/us/"><img alt="Creative Commons License" style="border-width:0" src="http://i.creativecommons.org/l/by-nc-nd/3.0/us/88x31.png" /></a><br /><span xmlns:dc="http://purl.org/dc/elements/1.1/" href="http://purl.org/dc/dcmitype/Text" property="dc:title" rel="dc:type">IBM Wiki</span> by <a xmlns:cc="http://creativecommons.org/ns#" href="geekhack.org" property="cc:attributionName" rel="cc:attributionURL">ch_123 (DMJB)</a> is licensed under a <a rel="license" href="http://creativecommons.org/licenses/by-nc-nd/3.0/us/">Creative Commons Attribution-NonCommercial-NoDerivs 3.0 United States License</a>.<br />Based on a work at <a xmlns:dc="http://purl.org/dc/elements/1.1/" href="geekhack.org" rel="dc:source">geekhack.org</a>.


Contributors: ch_123, microsoft windows, InSanCen, mcdonc
Created by ch_123, 05 August 2009 at 21:57
Last edited by ch_123, 1 Week Ago at 15:17
Last comment by ch_123 on 18 October 2010 at 21:11
105 Comments, 10,871 Views
(protected)

Bookmarks
Article Tools
Search this Article


Posting Rules
You may not create new articles
You may not edit articles
You may not protect articles

You may not post comments
You may not post attachments
You may not edit your comments

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

All times are GMT. The time now is 21:05.


template design by o2dazone
Powered by vBulletin® Version 3.8.6
Copyright ©2000 - 2010, Jelsoft Enterprises Ltd.