Micro Switch SW Series

Contents

• Overview
• Design
• Encoding and logic
  • Encoding grid
  • DTL or TTL encoding
  • Shared tracks
  • MOS encoding
• Options
  • Keyboard profiles
  • N-key rollover
  • Two-key rollover
  • Key repeat
    • Bi-level (double action)
  • Lock keys
  • Secretarial shift
  • Illumination
  • Tactile feedback
• Product groupings
• Switches
  • Packaging
  • Schema
  • Subseries
  • Plunger colour
  • Model interpretation
  • Models
• Timed repeat
• Keyboards
  • Schema
  • Subseries
  • Documented
  • Discovered
• Keycaps
• Other components
• Maintenance
• Documentation
  • Circuits
  • Integrated circuits
• See also

Overview

SW and SN Series are Micro Switch’s original series of Hall effect pushbutton switches. Solid State Keyboards use SW Series switches, which are mounted into special frames within the keyboard, and cannot be used separately. By comparison, SN Series switches are fully-self contained and can be snapped into a panel or attached directly to a PCB, depending on the model.

SW Series keyboards are comparatively uncommon. By 1976, Micro Switch had introduced a replacement series of keyboards—SD Series—and these are far more commonly encountered. SW Series keyboards continued to be produced until at least 1995, and the final update to the switch drawings was in 1999.

Via his son Scott, co-inventor Everett Vorthmann from Micro Switch has indicated that “SW” denotes wired solid-state keyboards, i.e. fully-assembled keyboard units with solid-state switches.

US patent 3596114 “Hall effect contactless switch with prebiased Schmitt trigger” describes the design of the integrated circuit found inside each switch. As patented, the chip contains a Hall element, a Schmitt trigger and a Darlington pair amplifier with dual independent outputs. The patent suggests that considerable effort was expended on the silicon implementation.

Design

SW Series switches are all open at the base. The return spring protrudes out from the bottom, and the Hall sensor IC (“lead frame package”) is a slide-in fit. These switches clip into mounting frames placed onto the keyboard matrix PCB. These frames have locating nubs for each return spring, loops to secure each switch into position, and holes to allow the terminals of the Hall sensors to pass through to the PCB. Micro Switch give the reason for the mounting frame arrangement as being a way to prevent keystroke impact from reaching the solder joints on the PCB, although this seems to be contradicted by the fact that the frames are riveted to the PCB.

The design rationale for having open-base switches (instead of fully-enclosed switches) is not known, but one explanation would be serviceability. Since the switches are contactless, they would typically only need servicing to replace a fouled housing or broken plunger. In such cases, there would be no need to supply or replace an expensive Hall sensor: all that is needed to replace a broken switch is to swap out the housing assembly, without the need for any desoldering. Replacement switches bought singly do come with a fresh Hall sensor, as well as an instruction leaflet (except in later years) and a pair of pry tools, suggesting that service technicians would not buy boxed replacement parts as they would soon have a mound of unused sensors, tools and paper.

The return spring is not part of the bill of materials for the switch: the switch models cover only the physical design and choice of Hall sensor. When sold separately, the spring is included, and these parts have separate numbers indicating that they are replacement modules and—for models with a spring weight other than the default of 3 ounces—the weight of the included spring.

Switches are marked with an arrow on the top, which points towards the back of the switch. Confusingly, the corresponding arrow in SD Series switches points towards the front.

The diagram below depicts the general construction of an SW Series keyboard, starting from the printed circuit board (PCB). The PCB is often dual-layer, and consequently it is not uncommon to see some form of insulating material between the PCB and the mounting frames. In one keyboard this seems be to pieces of PCB substrate, while in others it is more like self-adhesive paper.

Metal mounting frames are placed above the PCB and secured with rivets. Each row of keys has a separate mounting frame, and these frames hold the switches in place. Replacement switches on eBay can be found both with the Hall sensor fitted and left separate. It is more likely that the switches were placed into keyboards with the sensor already fitted, but once the IC is soldered in place, the switch module can be extracted without removing the Hall sensor. This allows for easy switch repair.

Encoding and logic

SW Series keyboards were designed to use two-of-N encoding. This arrangement avoids the need for (at the time) expensive matrix scanning circuitry. Each of the sensor’s dual isolated outputs provides one position of that key’s two-of-N code, and the fully-encoded output value of the key is available the instant the key is struck (made possible also by the bounce-free output from Hall sensors). The outputs of Hall sensors are provided by transistors, so inactive switches cannot pass current between outputs, and thus no diodes are needed to prevent wrong-direction current flow through the two-of-N wiring. The two-of-N coding is done entirely with PCB tracks and a small amount of logic, eliminating the need for a diode matrix. The physical encoding took multiple forms depending on the customer requirements and PCB design, as detailed below. The encoder can be DTL or TTL-based (medium-scale integration), using an assortment of separate chips, or MOS-based (large-scale integration), for a reduced part count and significantly greater flexibility in terms of output, such as permitting the shifted output from a key to be unrelated to the unshifted output.

Encoding grid

Keyboards with an encoding grid permit every key’s output to be defined independently using circuit pathways. A double-sided PCB has row tracks on one side and column tracks on the other side, with each row connected via a dedicated track to one of the two outputs of a switch. Every switch has each of its two outputs wired to two rows in the grid, separately from every other switch with no matrix. The grid columns form the N bits of the switch output, and this code is then converted to ASCII or EBCDIC (or other format) by the encoding logic. This overall implementation is shown in the diagram below, which shows a DTL/TTL logic implementation:

The diagram above is simplified, but shows the general functional layout of the circuit. The V_CC and GND traces are omitted for clarity. The strobe line signals when a key is struck; it is suppressed when two or more keys are pressed simultaneously to prevent the host equipment receiving invalid data. In reality, Micro Switch keyboards have a single double-sided edge connector, but here the two sides are shown adjacent to each other for clarity. With DTL or TTL encoding using binary encoders, at least 30 output lines are required (instead of the 12 shown in the diagram for simplicity), due to the number of valid two-of-N codes that are illegal inputs to the encoders (any code where both 1s are directed to the same encoder). A mere 12 lines is possible however with MOS encoding.

Provisions were often made, such as with models 53SW1-8 (in the Beehive Mini Bee), 70SW12-1 and 75SW12-2 for non-encoded keys. This took the form of columns in encoding grid that were routed directly to the edge connector and not via the encoding logic. Such keys were termed “function keys”, numbered from F1 onwards on the circuit diagrams. These function keys could be used to drive external logic circuitry directly, without needing to go through the keyboard’s data bus and ASCII or EBCDIC encoding. Function keys were also used for modifier keys, meaning that Control and Shift not only influenced the encoding process, but could be detected independently on the edge connector.

Model 64SW1-4 is a good example of a full encoding grid. Only the necessary connections are drilled and connected; the rest are left unmodified. Example circuit diagrams for various two-of-N models can be found in the Circuits section below.

The encoding grid can be fed into either DTL or TTL logic or into an LSI chip. DTL and TTL models define the final output codes directly within the encoding grid; with ASCII, the uppercase and control character output can be provided by bit manipulation logic circuitry. MOS-based keyboards had the option of placing the final output codes directly in ROM, for increased flexibility.

The ability or requirement to define the output of every switch via a dedicated pair of drilled and linked pads is both an advantage and a disadvantage: it is comparatively labour-intensive, but it allows a single PCB to supply any number of customers with no custom chips necessary.

DTL or TTL encoding

The conversion from two-of-N codes to ASCII or EBCDIC can be performed using decimal to binary encoders. The high and low nybbles of the output code are each derived from two 7 or 8 line to 4 line encoders; each encoder handles half the input values (1–8, 9–F) for each of the high and low nybble. For 7-bit ASCII only three encoders are needed in total, as the high nybble will only be in the range 0–7 (since ASCII only ranged from 0x00–0x7F) but models such as that used in the Beehive Mini Bee have all four encoders and thus full 8-bit output capability (see the Circuits below). Separate circuitry on the PCB is responsible for modifier key processing, as each key is only capable of a single output value by itself; the structure of ASCII makes modifier key circuitry a simple process.

Even with priority encoders, it is impossible to detect clashes (two or more keys struck at once) using the encoder logic because it is possible for all four encoders to be fed a single bit, a legal input for the encoders but an illegal two-of-N code overall. Instead, an electrical monitor detector (EMD) circuit detects an excessive number of active two-of-N lines using parallel resistance measurement, just as you would use in a self-encoding keyboard.

The logic-based approach can be seen in 64SW1-4. 63SW5-4 uses a MOS chip to process the encoding.

Shared tracks

MOS encoding removes the need for every key to have dedicated wiring: using modular arithmetic, switches that generate the same value for a particular output can have those outputs tied together. Thus, Q and W as adjacent keys would likely have one output on a shared PCB track. This arrangement can be seen in the Decision Data 8010 keypunch keyboard. Instead of a full encoding grid, the two-of-N columns branch out to multiple switches, and this reduces the size of the PCB. The conversion between two-of-N to output codes is handled via MOS.

58SW5-9 appears to combine a full encoding grid with MOS encoding, but a look at the reverse of the PCB shows that the encoding is custom to that model and unable to be modified.

MOS encoding

MOS-based encoding is a very compact approach. By feeding the entire two-of-N code into a single chip, clashes between keys can be detected by simply checking for too many active input lines, and thus the electrical monitor detector circuitry can be removed. The conversion from two-of-N to output codes can be placed into a look-up table in ROM, although not all models operated this way, with the MOS chip simply consolidating the encoding and collision detection logic. Using a ROM look-up table for the output codes provides the ability for modifier keys to generate output values that do not bear a bitwise relationship to the underlying key. For example, Shift+1 could produce “*” instead of “!”, and entirely custom modifier keys could be implemented. ROM look-up allows the use of a shared-track approach to the circuitry, as keys need only generate internal or intermediate two-of-N codes.

MOS-based encoding is able to use all possible two-of-N codes, without the constraints imposed by a bank of binary encoders. Thus, the number of encoding grid lines can be reduced to half or less than required by DTL or TTL encoding.

Suppliers of MOS encoders included Texas Instruments (TI), National Semiconductor (NS) and American Micro-Systems (AMI). All known examples are DIP-28, in both ceramic and plastic packages. Examples include:

As shown in the table above, encoder chips for SD Series keyboards were generally placed into SW Series although not universally.

The pinout for model SW-20314 is given under Documentation below.

Options

Keyboard profiles

Three profiles were advertised in the 1973 product brochure:

• Sculptured
• Stepped
• Sloped

In all three instances, a 13° tilt is depicted. This indicates thus that stepped switches have a 13° tilt to the plunger itself.

N-key rollover

The use of two-of-N encoding means that only one regular key can be active at once. If two keys were to be held simultaneously, the encoder would receive incorrect or invalid input: the output from the two keys would collide, with the encoder seeing the bitwise OR of the two scancodes (this would likely be an illegal two-of-N code). Thus, N-key rollover in SW Series keyboards carries a different meaning to how it is otherwise understood. SW Series keyboards with N-key rollover do not allow unlimited keys to be held: they simply avoid clashes from keys being struck in rapid succession.

SW Series keyboards with N-key rollover use sink pulse Hall sensors, whose outputs are only active for around 10–100 µs from the point that the key is struck regardless of how long the key is held. This prevents clashes arising, with a trade-off that key release cannot be detected. Before the second key is pressed, the first has already been detected and its sensor has shut itself off. Modifier keys still use current level sensors, and these must be wired separately.

Two-key rollover

Even two-key rollover would be subject to clashes with current level switches. Micro Switch define two-key rollover as an electrical interlock that disables keyboard output while two or more keys are pressed at once. Presumably (as this is not stated) once the number of active keys has reduced back to one, the remaining active key will be reported. This implies that a continuous sequence of keys each pressed with the preceding key still active will not be understood, but occasional instances of a single key pressed too early would be recoverable and not result in a missed keystroke.

The Beehive Mini-Bee terminal keyboard uses TTL two-of-N encoding combined with an electrical monitor circuit. This circuit appears to detect when any two keys are held simultaneously, suggesting a two-key rollover arrangement. This may indicate how two-key-rollover was implemented, as details remain difficult to obtain.

Product Brochure SW reported that Honeywell’s Systems and Research Division found that use of N-key rollover reduced operator error by up to 30% over 2-key rollover. It may help to understand this claim in context: as late as 1967 (shortly before SW Series was introduced) and possibly later, Invac were still advertising their photoelectric keyboards with a mechanical interlock that made depressing two or more keys simultaneously physically impossible. Purely electronic keyboards (with no mechanical contrivances such as solenoids or motors) were still relatively new in the late 1960s.

Key repeat

Key repeat in SW keyboards was achieved by the keyboard repeating the strobe signal. The strobe signal indicates when a key has been pressed and tells the host equipment that there is a key code present on the data bus ready to be collected. There were at least three repeat options:

• Double-action switches, referred to by Micro Switch as “bi-level” (this was designed to mimic electric typewriters): with stage two actuated, auto-repeat is enabled
• Repeat key: a dedicated key enables auto-repeat of whichever other key you hold
• Standard auto-repeat, triggered by keeping a key held (by default, for half a second)

In keyboards with two-of-N encoding, timed repeat is achieved using a custom timed repeat signal generator chip (SW-10667) in conjunction with special timed repeat switches whose output ties into this chip. See under timed repeat for details.

Bi-level (double action)

Bi-level switches—found in 1SW200 Series—have a larger sensor IC, with two separate Hall elements, positioned at the top and the bottom. The plunger has two metal pins protruding from the bottom, each connected to a stiff spring. As the plunger is depressed, these pins make contact with the switch mounting frame. Pressing the plunger further requires these extra springs to be compressed at the same time.

Bi-level switches still have four terminals on the sensor; they will thus not have redundant outputs. Because these switches are used for auto-repeat, they must be level instead of pulse switches. It is not known whether they are current sourcing or current sinking.

Lock keys

Lock keys are achieved using one of the following methods:

• Alternate action switches
• Mechanical secretarial shift, as with electronic typewriters
• Electronic secretarial shift: here, setting and releasing shift lock is done entirely in logic, with no mechanical linkages

Secretarial shift

A single keyboard, catalogue listing 84SW12-2, has been found with secretarial shift. The left shift key uses 1SW80 and the right shift key uses 1SW86. Both modules have stepped plungers. 1SW80 contains a sink level sensor along with the mechanical linkage for left shift, while the right-hand shift key uses a 1SW86 switch with no sensor. The lock key uses a special plunger assembly that is not a switch. Additional modules support the bar that connects the shift keys together. The mounting loop that holds the lock module is used as part of the latching assembly. The lock keycap is mounted onto a metal pressing with a blade mount, and this uses two of the outer notches within the keycap that were also used with KB switches.

Illumination

SW illuminated switches are centre-lit using a small bi-pin incandescent lamp, the same as those used in Clare and Pendar switches. These have an ivory plunger. The plunger is tubular and rises up around the lamp. Special keycaps are required for illuminated switches. Some models were not fitted with a sensor; these were weighted at 33 oz, or just under 1 kg, to give the sensation of being immovable; these indicator types were used as status lights only.

Illuminated SW Series switches are rarely encounted; they can be seen however in a 101SW1 keyboard (possibly 101SW1-3-H or 101SW1-4-H). They were also used in 84SW12-2, which contains three such types: momentary, alternate actionand indicator.

Tactile feedback

Apparent tactile feedback is demonstrated in the switch model 1SW12-BL force curve. This switch is known from the IBM 3277 keyboard. The only characteristic visible in the photos that is different from normal switches is a pair of horizontal ridges across the back of the plunger, that look like they are designed to catch on something. The force curve also has the double step characteristic of Alps tactile switches.

No mention has ever been made of 1SW12-BL switches being tactile or clicky.

Product groupings

Product and part numbers are divided into two main groupings. Components use catalogue listings of the form “SW-” followed by five digits. This grouping includes magnets, shells, plungers, “spacers” (pry tools), PCBs, sensors etc. Switches, keycaps and complete keyboards have catalogue listings where “SW” is prefixed by a number.

Switches

Packaging

Replacement (-R) switches are supplied together with the spring and a pair of pry tools for removing the failed switch. Typically, the package also included an instruction leaflet for switch replacement; see under Documentation for a copy of the leaflet. Older switches were wrapped in thin brown paper and placed into a cardboard carton, while more recent switches shipped in a clear plastic packet, and often lack the instruction leaflet.

Before the switch can be placed into the keyboard, the return spring needs to be attached to the plunger. This is achieved by placing the spring over the boss on the plunger and rotating it anti-clockwise. It will then lightly grip the plunger and allow the replacement switch to be lowered into position.

The following photos represent neither a real unboxing nor the precise state in which the package contents were supplied. The sequence is a recreation using a previously-unboxed switch, where the box contents were returned to the box in perhaps a less suitable and less expert way than how they were placed originally. However, the instruction leaflet does unfortunately arrive just as creased as depicted.

Micro Switch PK-44600 standard shipping carton for smaller components; this was later replaced by clear plastic packets

The leaflet is placed around the inside of the box, and the switch is placed wrapped into the centre

Box contents: one switch, one return spring (to be fitted separately), two pry tools and the instruction leaflet

Unwrapped box contents

Schema

The part number schema for SW Series switches is as follows:

1SWM(-R(F))

“1”

Prefix for keyswitch modules

“SW”

Series name

M

Model number, in a range of at least 1–301

“R”

Denotes a replacement part

F

Operating force in ounces; where omitted, the operating force is the default of 3 oz

Subseries

Switches in SW Series fall into at least two subseries:

1SW1 Series switches appear to exist in pairs, with each odd-numbered model having a straight keystem for sloped keyboards, and each even-numbered model having an angled keystem for stepped keyboards.

1SW100 Series switches are not fitted with lamps: unlike 201SN Series, there are not pairs of part numbers depending on whether a lamp is fitted.

Plunger colour

SW Series was advertised in Information Display, Volume 5 Number 6 from November/December 1968. In this advertisement, the plungers were all shown to be grey. Leaflet PK 8503 2 (see Documentation, below) describes two plunger types: one grey (with a “hat shaped boss” for the spring) and one black (with a “plunger spring seat”). This document was drawn up in November 1971, but black plungers go back at least as far as 1970 (see under discovered keyboard examples). At this stage, all plungers of the same age appear to have been the same colour, either grey or black depending on age.

At some point in the mid-1970s, switches became colour-coded, primarily according to the sensor type. The oldest SW Series switch on eBay with a colour-coded plunger is from 1977, while the most recent one with a black plunger (of a type that would change to colour) is from 1974. Inspection of the SW and SN charts has allowed a reconstruction of most of the colour code. While the red and green colour choices appear to have been consistent (for sink level and source level respectively), it seems that sink pulse switches first gained white plungers, before blue was selected instead.

Note that illuminated switches appear to all have ivory sliders, regardless of age and sensor type (just as they do in SN Series). This can be seen in a 101SW1 keyboard (possibly 101SW1-3-H or 101SW1-4-H).

Plunger colour	Era	Colour usage
	ca. 1968–1970	Possibly all switches
	Early to mid-1970s	Possibly all switches
		Mid-1970s onwards	Alternate action and dummy switches; bi-level; additional types
	Mid-1970s onwards	Sink level
	ca. 1976	Sink pulse, as found in 56SW5-2 and 84SW12-2, both dated 1976
	Mid-1970s onwards	Sink pulse
	Mid-1970s onwards	Source level

Model interpretation

Precise details on the various switch types are rare, in particular the output type of each model. However, some of these details can be derived from existing information.

Keyboard 54SW11-14 uses 1SW17 switches for most keys and 1SW11 for the modifier keys, which indicates that 1SW17 (sloped) and 1SW18 (stepped) must be sink pulse, with 1SW11 being a level type (sink level or source level). According to the SW and SN charts, 101SN1B1 and 1SW17 both used sensor SD-10023 (replacing SW-11504), and 101SN1B1 is demonstrated to be pulse type by Ed Nisley, as expected from the “B” designation shared with SD Series switch part numbers. This confirms the expectation of 1SW17 and 1SW18 being sink pulse. 1SW Chart 12 indicates that 1SW17 and 1SW37 are both sink pulse (by way of using the same sensor), and implies that both types have blue plungers (much of the data was previously deleted as parts went obsolete). Switch 1SW18 is confirmed to be blue, and the sensor inside is marked “B” as expected.

Keyboard 59SW9-1 has primarily blue switches (which will be sink pulse) with red switches for the modifier and repeat keys, and a few unexplained black switches. The red switches will be a level type. Switches 1SW51 and 1SW52 are both red types, and each have sensors marked “A”. This again corresponds with SD Series designation and would make these sink level.

This leaves green plungers, and source level switches, with the suggestion that these two correspond. 1SW41 is documented as being green, and as having sensor SW-11897. A number of types also have this sensor, including 201SN1C1, 201SN1C2, 201SN4C1 and 201SN4C1. “C” in SD Series denotes source level, which appears to confirm that indeed the green switches are source level as anticipated.

Although it is not possible to be completely certain about the above details, the various details all appear to fit in a way that makes sense.

Models

Switch model numbers appear to have started from 1SW1. 1SW1 uses a grey plunger, as shown in advertisements from 1968. The top of the plunger contains a circular hole, as also seen in some older Licon and Cortron switches, something seen in some later models. By around 1970, the plunger was redesigned (as illustrated in installation guide PK 8503 2), and it appears that the replacement models were given new numbers starting from 1SW11. 1SW1 is presently the only known original switch model.

The “R” suffix indicates a replacement switch sold separately in a box along with an instruction leaflet, return spring and two “spacers” (pry tools). The digits that follow the “R” indicate the weight in ounces of the spring supplied in the box and are only seen on “R” models. The exact spring weights are not yet determined, but taken at face value, “1.5” would be a half force switch (used together with a support switch on wide keys); “2” would be 2 oz semi-light, and “8” would be a heavy 8 oz switch. Testing with an R8 switch demonstrates that it is in the region of twice the weight of a normal switch, but proper testing is impossible without a spare mounting frame to hold the switches together.

Clearer details of the switches will be forthcoming, pending completion of collection and analysis of charts. The details below are a combination of the various charts together with the results of analysis and interpretation. The plunger colours listed are a combination of the documented colours (per the charts) and the observed colours.

A source of “↓” indicates that the switch is given in the keyboards table that follows the switches table.

Timed repeat

Timed repeat is one of the methods of achieving key auto-repeat. Timed repeat is an alternative to secretarial shift, where auto-repeat is triggered by keeping the key held down instead of pressing the key harder. Like with secretarial shift, the timed repeat system is only suitable for use with a small number of keys due to the extra circuitry required. Matrix scanned keyboards are able to implement repeat functionality entirely within the encoder.

Timed repeat makes use of a DIP-14 custom signal generator IC, part number SW-10667. The strobe signal—that informs the host when a keystroke has occurred and the encoded output is ready for collection—passes into this chip and back out (pins 13 and 14). When the chip detects a repeat condition, it cycles the strobe line to simulate multiple successive keystrokes. During normal typing, the strobe signal (generated by the encoder) passes from pin 14 to pin 13 unmodified.

The circuit for a single timed repeat key is shown in the diagram below:

On the right is the SW-10667 IC. A combination of a capacitor and two resistors connected to the chip controls the delay time until auto-repeat commences and the auto-repeat period. The resistor between pins 1 and 3 controls the auto-repeat period. Both resistors together determine the auto-repeat delay time.

The circuit was traced by “fricked” from the PCB of Micro Switch model 84SW12-2 (manufactured in this instance in 1976). The component values were also derived from this circuit. The details for SW-10667 were located by Deepak Kandepet in the Diablo Systems HyTerm Communications Terminal Model 1610/1620 Maintenance Manual from December 1978; a scan of the relevant page can be found under Documentation below.

Auto-repeat also requires special timed repeat switches. As shown in the diagram, these have the same four terminals as any other SW Series switch, and both outputs are connected to the encoder as with any other switch. The difference with the timed repeat sensors is that, after the initial sink pulse to indicate a keystroke (where the outputs drop from 5 V to 0 V for between 10–100 µs), one of the two outputs only rises to 4.1 volts if the key remains held, instead of back to the full 5 V. This adaptation allows a small amount of current to continue flowing so long as the key is held. 4.1 volts is close enough to the expected 5 V logic level that the encoder no longer detects the key and the switch is freed from causing rollover collisions with subsequent keys. However, by way of a PNP transistor, SW-10667 detects the active switch and begins its auto-repeat cycle. When the timed repeat key is released, both outputs will be at 5 V and the key is fully idle.

(The 4.1 level is given in the aforementioned Diablo Systems maintenance manual, for an SD Series keyboard. “fricked” measured closer to 3 V in 84SW12-2. The same SW-10667 chip is used in both. Although we have recovered many Micro Switch charts, none of these so far have been for the Hall sensors.)

SW-10667 also contains a pair of NAND gates that are unconnected to the rest of its circuitry; the reason for this is not known.

When a keyboard has multiple timed repeat keys, there is one PNP transistor for each encoder line. Switches that share the same encoder line will also share the PNP transistor and the related resistors. This can be seen in the diagram below:

Note that, as these details were mostly recovered from a single SW Series keyboard, and because timed repeat is fairly uncommon, the exact implementation details may differ between keyboard models.

Keyboards

Schema

The model number schema for SW Series keyboards is as follows:

KSWA-B

K

Number of keys (at least three, as “1” in this position denotes a single switch and “2” denotes a keycap)

“SW”

Series name

A

Unknown

B

Unknown

Subseries

There are at least two known subseries:

12SW Series

Current sinking non-encoded 12-station keypads (can be paired as a 24-station assembly)

16SW Series

Current sinking non-encoded 16-station keypads (can be paired as a 32-station assembly)

Documented

For more details, see Documentation below.

Discovered

The following is not a complete list of all known SW keyboards. It only lists keyboards where at least one switch model is shown, where it is particularly early or late production or where there is some other important distinction. In almost all instances of SW keyboards found online, the switch part numbers are not all shown, or are not shown at all (except for where it seems the switches were not marked in the factory).

Keycaps

Keycaps for SW, SN and SD switches all come under 2SW. Part Target list 326 separate entries for keycaps, and this list appears to be very far from complete, as presumably every combination of legend and colour has its own part number and, where necessary, its own NATO Stock Number (NSN). There is no single catalogue listing format for keycaps; the catalogue listing numbers are quite varied in their format, and at times very long and complicated (e.g. “2SW1-4C-M-4-2-1598C”). In Part Target’s search result, 2SW1 is associated with around 26 different NSNs.

Standard 2SW keycaps are 0.475″ tall.

The catalogue listing numbers were largely not depicted and varied in formatting; thus, they have been normalised according to the Part Target listings for consistency.

Other components

Maintenance

When fitting a replacement switch module, or returning a module to the keyboard, the return spring needs to be attached or reattached to the plunger. PK 8503 2 instructs repair technicians to “place the spring on the spring boss and compress and rotate it until the last coil of the spring expands around the boss”. For those who instinctively try to rotate the spring clockwise, note that in fact the spring must be rotated anti-clockwise. The leaflet recommends a pencil eraser for the task, but bare fingers are sufficient. Only a small amount of rotation is required for the spring to remain attached to the plunger, although this process can be a bit finicky.

Documentation

Honeywell have kindly provided a collection of SW series switch charts. Only a subset of the charts appear to have survived. The charts themselves were extensively redacted by Honeywell as switches went end-of-live, but they do have some older versions of the charts (as is the case with 1SW Series Chart 1), and with luck, they may find some more older revisions with more switch details.

Product Brochure SW “Solid State Keyboards” and Product Sheet 51SW12-1 “Typewriter Keyboard” are in item SILNMAHTL_31166 at the National Museum of American History Library (part of the Smithsonian Institution), who kindly provided scanned copies of both documents. The dates for these documents are derived from the three-digit codes placed at the bottom of the last page of each. Product Brochure SW—seemingly from March 1973—is mentioned on Electronic Design 17, page 118, from the 16th of August, 1973.

• IC operation keyed to Hall effect, Electronics, September 16 1968 (scanned by Bitsavers) — covers the introduction of Hall effect keyboards, and the preceding product ranges
• Micro Switch Product Brochure SW: Solid State Keyboards, March 1973
• Micro Switch Product Sheet 51SW12-1: Typewriter Keyboard, August 1973
• Micro Switch 1SW Series Chart 1, issue 11, created 1995-09-11, revised 1998-04-13
• Micro Switch 1SW Series Chart 1, issue 13, 1999-12-11
• Micro Switch 1SW and 1SW100 switch charts (surviving charts for individual switch types)
• Instruction Sheet, solid state switch module replacement (PK 8503 2), November 1971

Circuits

• Circuit chart SW-10 (6-bit mono-mode)
• SW-10259, 53SW1-8, Beehive Mini Bee
• SW-11339, 53SW1-8, Beehive Mini Bee

The Mini Bee keyboard schematics are from the Mini Bee Computer Terminal Service Manual, scanned by Bitsavers. The keyboard model (53SW1-8) is from the separate Illustrated Parts Breakdown.

Several documents included with a Micro Switch keyboard can be found in a single PDF on the MMcM/parallel-ascii-kbd Wiki on GitHub for a keyboard with PCB model SW-11234. This includes the instruction leaflet PK 8911 1, the specification chart, the PCB layout and circuit diagram, and the physical dimensions of 55SW5-2, even though the keyboard built on SW-11234 is a 56-key keyboard with slightly different keycap widths. The PCB layout gives a good indication of how the MOS two-of-N arrangement is laid out.

Integrated circuits

• SW-10667 timed repeat generator pinout (taken from the Diablo Systems HyTerm Communications Terminal Model 1610/1620 Maintenance Manual, December 1978, scanned by Bitsavers)
• SW-20314 MOS/LSI 2-of-13 keyboard encoder pinout (taken from the Diablo Systems HyTerm Communications Terminal Model 1610/1620 Maintenance Manual, December 1978, scanned by Bitsavers)

See also

• SN Series, the standalone equivalent to SW
• SD Series, the successor series for keyboards