Mechanical Enterprises Mercutronic
“Mercutronic” is a brand name of Mechanical Enterprises for their mercury-contact switches. The term “Mercutron” was occasionally used, perhaps in error. The series was advertised in Modern Data, April 1970 on page 192 as the “Mercutronic Coding Keyboard” which uses a “mechanical switching approach based on the movement of mercury in a sealed flexible tube.” That issue also mentioned that Mechanical Enterprises’ “Mercutronic Division” would be exhibiting at the 1970 Spring Joint Computer Conference (in booth 49007).
Few details about these switches are known besides what is found in the patents. It seems that at least four series of Mercutronic switches were produced.
All these designs work on the principle of a flexible mercury-filled tube, that is pinched shut by the switch to separate the mercury. The chief objective appears to be fast, bounce-free switching, as they were introduced around the same time as the microprocessor, when keyboards needed special circuitry to deal with contact bounce. As the contacts are sealed, the switches are protected against environmental influences.
US patent 3600537
US patent 3600537 (filed in 1969) is the oldest Mechanical Enterprises mercury-contact switch patent, for a conductive-liquid snap-action switches. At least two designs are featured, with neither being intended for computer keyboards. The first type is miniature snap-action switch.
Behavior Research Methods & Instruments, 1969, Vol. 1 (8) page 329 (PDF page 1) gives a description of these “Mercutron” switches under the heading “Subminiature Switches”. They are cited as having the “mechanical advantages of a snap-action switch with the electrical properties of a mercury switch.” They are said to have no bounce, and can switch “60 na” at 24 V DC. (Surely they meant milliamps?)
US patent 3707611
Shortly after the snap-action switch, Mechanical Enterprises filed a patent for a mercury-contact switch suitable for full-travel keyboards. This patent, US patent 3707611 (filed in 1970), depicts a keyboard switch with an external return spring. These switches were offered in two forms: coding and non-coding. The patent depicts the coding version, while all examples discovered in literature show the non-coding version. Coding switches generate the character codes directly onto a common bus, a short-lived idea that was considered favourable by some manufacturers at the time that this product range was introduced. Extensive details are known about the coding version from the patent, while the characteristics of the non-coding version remain unclear. The overall design is the same, but so far no details have been recovered showing how the non-coding types attach to, and connect with the keyboard.
The patent allows for the switch to offer tactile feedback, using a bowed leaf spring placed across the top. However, this feature is only possible in the single-spring version of the switch where the tube is pinched by the plunger itself. The depictions of the non-coding type all show the dual-spring type (also given in the patent as an alternative approach) where the pinching action is provided by a second coil spring, and this occupies the space where the click leaf would go.
Wireless World, May 1972 page 251 (PDF page 50) contains an advertisement for “Mercutronic” switches from Tekdata in the UK, with a illustration of a switch similar to that of US patent 3707611, but much lower profile (the base area is drastically shorter). Of these switches, the advertisement notes, “Encoding is performed by using m.o.s. chips or diode arrays”, indicating that they are the non-coding type (which can be determined also from the photograph). Based on a different advertisement from Tekdata (Electronics Today International, November 1974, page 7), it is clear that these are indeed supplied by Mechanical Enterprises. These switches are rated at 25 million operations, and are noted as being impervious to dust, noise and electrostatic charges. (The unspecified “Mercutron” type in the second advertisement gives a 50 million operation lifetime.)
Modern Data, April 1970, when listing Mechanical Enterprises as attendees of the 1970 Spring Joint Computer Conference, noted that “ME’s Mercutronic Division will display a series of interchangeable-key keyboards to be custom-built for the OEM user.”
Computerworld, Vol. V No. 6 (10th of February 1971) depicts the same swich as shown in Wireless World, in cross-section diagram form, and lists them as model MC-210, suggesting that they are MC series. These fit figure 11 of the patent, where the rubber tube containing the mercury is pinched by pulling it upwards by the hook end of a coil spring.
Sadly the document was scanned in very low quality, but the captions are just about legible. Note that the PDF itself is degraded beyond repair by a defective optimisation process; the image above is from the JPEG 2000 master images, and even that is horribly over-compressed. The illustration below from Electronic Engineer magazine (recreated) is much clearer but omits some of the switch details:
The non-coding type has a thin base, as seen in the illustrations above. By comparison, the coding type has a much taller base. The patent implies that the switch has twelve terminals in total. One of these is the common input, and the remaining eleven are output. The patent indicates that 2048 possible output codes are possible, being 211 according to the number of output terminals. In an article in the December 1969 issue of Electronic Engineer magazine (volume 28, no. 12) that pre-dates the patent, the bus is given as 11-bit and the output as 10-bit (for a total of 1024 codes).
The following design illustrates the general principle, simplified to seven bits of output, and with an eighth output line dedicated to collision detection:
These switches are entirely solderless. One of the objectives of the design was to allow for the code output of a switch to be changed easily, without needing to change soldered connections. The keyboard is required to have a low wall between key rows, and channels within this wall grip the retention prongs in the switch base. The keyboard design depicted in the patent does not use a PCB either. Instead, all the conductors are provided as parallel strips on the upper side of adhesive tape, which is then affixed to the bottom of the case. At the end of each row, the tape is folded through through 90° twice so that it is routed back along the next row, albeit with the conductor bus lines now in the reverse order. Each terminal takes the form of a pin, held under pressure against the conductor lines by a multi-leg flat spring that distributes current within the switch itself. The pins drop into a block, and this block is then affixed to the switch’s upper shell from below with two screws.
Being a coding keyboard, two-key rollover is impossible, as two keys pressed at once will cause a bogus character code to be generated. Each output terminal contains a diode, to prevent the switches interfering with each other (otherwise even inactive switches would jam the bus, in a manner analogous to ghosting in a matrix-wired keyboard). The recesses have room for both a diode and a resistor to be stacked, and this arrangement is used on a specific terminal to detect when two switches are connected at once. The patent does not give any further details, but when two switches are active, the total resistance of the power to that bus line would halve (by parallel resistance), and the current would rise with each additional concurrent actuation, allowing the keyboard circuitry to ignore any switch output when the switch bus is jammed.
Micro Switch’s KB encoding switches prevented inactive switches from jamming the bus by fitting every output terminal with its own pair of contacts, rather than fitting a diode to every terminal.
US patent 3845264
An advertisement in ELECTRONICS Australia, November, 1975 depicts Mercutron M-5 series. The design is almost identical to that of T-5 series, and the operating cam mentioned appears to be the same T-bar actuator as found in T-5 switches.
Sadly the PDF went through a defective optimisation process, leaving the image irreparably damaged.