- Tactile feel classification
Tactile feedback in pushbutton switches provides a physical sensation in one’s finger to that the keypress has been registered. It is widely used in buttons, such as your television remote control or computer power button. In many cases, the mechanism responsible also generates an audible click sound; the use of click–tactile buttons within computer mice has led to the use of “click” as a verb to describe pressing mouse buttons.
Tactile feedback is widely used in computer keyboards. Historically, many switches offered a simple linear response (a smooth feel), but the vast majority of keyboards produced today are tactile. A key objective with tactile feedback is to provide feedback to the operator. Part of the idea is that you can stop pressing a key once you have received the tactile sensation, such that keys do not have to be pressed all the way down. However, with typical rubber dome over membrane keyboards, the overtravel distance is too short for this to make much difference; conversely, the soft landing of a rubber dome keyboard does at least soften the keystroke impact, which can be a hard stop in other switch types.
Linear switches can suffer from a “mushy” feel due to the continual rise in force as the key is pressed. Causing the force level to drop off during the keystroke, either suddenly or gradually, reduces the stiffness of the key. The “mushy” feel can be resolved instead by making the switch lower in force, but this brings with it the increased risk of unintended keystrokes.
In Omron’s article Hall IC Keyboard Switches Become the Leading Type (Yoshikazu Kitao, JEE, August 1978, pp. 27–31), the author said of the tactile model of B2H:
A sense of feel is fed back to the operator to help reduce keying errors and to prevent occupational diseases such as keypuncher’s disease.
No further elaboration is given on the benefit of tactile feedback.
There are numerous methods for producing tactile feedback. Depending on the design, additional friction may be introduced to the switch feel, which can be detrimental to the user’s perception of the product.
Although rubber dome keyboards are considered a comparitively modern invention (as a cost-cutting design), the use of rubber domes to provide tactile feedback dates back many decades. Rubber domes can be paired with a number of keyboard types to add tactile feedback, including photoelectric encoder (as shown in US patent 3465099 filed in 1967), capacitive (at least as far back as US patent 3965399 filed by Controls Research Corp in 1974), membrane, conductive and inductive.
“Buckling rubber sleeves”—rubber domes with a hole in the top, sometimes mounted upside down—are also considered to be rubber domes here. In fact, the 1967 and 1974 patent filings both show “buckling rubber sleeves” rather than the better-known rubber dome form.
Rubber domes are advantageous in that they can provide strong tactility without extraneous noise, and they can be tuned to provide a smooth tactile feel. Whether rubber domes are in any way inherently disadvantageous is not clear. In cheaper keyboards there is a risk that domes can split, and the silicone rubber material tends to stiffen with age, or lose all feeling. Rubber domes offered several simultaneous advantages: cheap to produce, cheap to assemble, quiet and good tactile feel. As such, they are more closely associated with cost-effective production than the more expensive keyboards in which they were also used, such as Micro Switch ST Series (conductive and capacitive membrane) and Corton DIN-compliant ferrite core. Tentative suggestions in various keyboard forum topics are that Topre’s rubber dome–based capacitive keyboards don’t exhibit as significant a stiffening of the rubber as cheaper brands, but more reliable data would be required to be clear on the lifetime limits of rubber materials and to what extent the target cost of keyboards results in inferior rubber material selection.
Several switch designs involve separating a permanent magnet from a metal plate to provide the tactile event. Typically, the pretravel of the switch is accommodated by a regular coil spring. When the plunger is pressed far enough, the pressure within the pretravel spring is sufficient to force a magnet below to detach from a metal plate, creating a drop in force.
Switches known to use this method include:
- Univac magnetic separation, example circa 1970
- Cherry series 201/202 reed switches, introduced circa 1970
- Fujitsu FES-5 reed switches, introduced circa 1972
- Unidentified Soviet copy of FES-5
- AT&T magnetic separation capacitive
Magnetic separation also generates a click sound.
Various manufacturers have introduced keyboards with a snap-action spring system. This is generally where a flat spring or metal dome is curved upwards in the resting position, and is inverted under pressure during actuation. The transition from being curved upwards to being curved downwards tends to be a rapid change, generally accompanied by a sharp sound. The level of tactility afforded by such a technique tends to be lower than one might expect, however. Rubber domes work along the same general principle as metal domes.
An early snap-action switch type is IBM’s beam spring arrangement. Other notable examples are Olivetti snap action, Alps KFF Series, Alps KCP Series, Micro Switch SC and CT Series and the various Fujitsu leaf spring types. Fujitsu leaf spring switches involve a metal dome, the collapse of which can be either linear or tactile depending on the switch model. ITT ETL 18 is another switch with a flat spring tactile arrangement, although it lacks the snap feel of other designs.
The amount of deflection of the snap-action spring is typically less than the full travel of the switch, and there is little if any vertical motion of the snap-action spring prior to deflection. Generally, a coil spring is placed over the top. As the plunger is depressed, the coil spring takes up the increase in force until the force within the coil spring is sufficient to invert the snap-action spring. In this regard, the design follows that of magnetic separation.
Micro Switch SD introduced a tactile system based on a sprung peg. A peg protruding from the plunger runs along a track within the switch housing. This peg is pushed outwards by a coil spring. A ramp added to this track causes the peg to be pushed inwards against its spring for part of the travel, which makes the switch harder to press. The tactile event results from the peg reaching the end of the ramp, where it stops contributing to the force of the switch.
The precise chronology of this design is unclear. RAFI introduced RS 74 and RS 76 around 1975 and 1976 respectively, and these use an identical design of spring-backed peg in the alternate action mechanism, as does the alternate action mechanism of SD Series. It’s not clear which of Micro Switch and RAFI introduced this alternate action mechanism; we only know that RAFI did not introduce a tactile switch. The M and A charts for Micro Switch SD8 tactile were drawn in September and November 1976 respectively, which at present is the only available date information.
With a simple linear mechanical switch, the plunger slides smoothly along the movable contact as it presses it against, or releases it to move towards, the stationary contact. By specially shaping the movable contact or the plunger (or both) it is possible to introduce a period of increased force during the key travel. Cherry M5 provided a tactile option according to US patent 3715545 filed in June 1971; the design as patented used small notches added to the plunger cams. Tactile M5 and M7 switches are extremely rare and are not readily available for examination to determine whether the production designs followed the patent; they are also ultrasonically welded shut, preventing easy inspection.
“Siemens ST” switches used a similar arrangement to Cherry M5. The series name is not known, and neither is the date of introduction. Discussion in the Deskthority forum topic Siemens Transdata Terminal keyboard & Siemens T1000 switch suggests a date of introduction no later than 1975.
The series best known for this type of tactility is Cherry MX. The original tactile option was a similar idea based on a separate sliding collar, but Cherry would later file a separate patent in April 1987 (German patent 3713775) for a tactile arrangement based on simple plunger–contact interaction, a design that would become known as “Cherry MX Clear”. MX Clear simply involved an alteration to the plunger cams to introduce a tactile peak.
Datanetics DC-60, introduced in 1973, uses a very different arrangement. Instead of presenting an obstacle to be cleared, steps in one of the contacts add a downward force to the plunger’s separator bar.
Hi-Tek Series 725 was adapted at some stage to offer tactile feedback, seemingly by 1988. Inward folds in the contacts cause the separator bar to catch until sufficient force is exerted to clear them.
Deflected flat spring
One of the best known tactile methods is the deflected flat spring, commonly referred to as a “leaf spring”. (The term “leaf spring” is official: the SKCL/SKCM exploded diagram in the 1994 Alps catalogue uses the terms “リーフスプリング” (“leaf spring”) and “leaf spring”. However, “leaf spring” typically refers to the laminated spring used in vehicle suspension.) Unlike how snap-action springs are usually placed below the plunger, a deflected flat spring is placed alongside the plunger’s path, and the plunger pushes the spring away as it moves past. A step, hook or ramp in the formation of the flat spring applies a temporary increase in force while the key is being pressed, as the operator is pushing not only against the return spring, but also pushing the tactile spring out of the way.
Unlike plunger–contact interaction designs, deflected flat spring designs involve an extra metal spring whose sole purpose is to provide tactility. Such switches can—in theory at least—be converted between linear and tactile by simply adding or removing the tactile spring. (This conversion is not possible in Alps KCL/SKCL because the space where the tactile spring would be placed is reserved for an LED instead.)
The first switch type known to use a deflected flat metal spring is Fujitsu FES-8, introduced in 1976. The design was based on Micro Switch SD. Where SD Series had a sprung peg pressing on a rigid switch housing, FES-8 has a rigid post on the plunger that presses against a vertically-cantilevered flat spring. This approach requires more metal but would have been simpler to assemble and does not require exceptionally intricate parts.
In the early 80s, Alps Electric introduced a variation on the FES-8 approach for the new KCM Series, using nothing other than the bottom edge of the plunger to press on the spring. The flat spring in this instance appears to be derived from the actuator spring.
Around thre mid 1980s, Alps introduced two other variations on this design. Model SKCMAF (“ivory Alps”) uses a folded flat spring instead of attaching a spring to a plastic spacer, and SKCMAG (“blue Alps”) uses a nearly identical part that also generates a click sound when the key is pressed. The exact order in which each of these three Alps designs were introduced is not known.
The Alps folded spring design was widely copied.
The tactile models within Datanetics DC-60 and Hi-Tek Series 725 both use a similar arrangement where one or both switch contacts engage with the plunger in a similar manner.
Cherry MX is famous for using a sliding part referred to as a “click collar” or “click jacket” to generate a click sound. Although originally intended to produce a click sound, after the customer for whom the click feature was designed (Olympia) declined the product, it was marketed instead as a tactile mechanism, with lubricant added to suppress the click sound. The design also had the advantage of providing hysteresis. Some years later, a separate variant was introduced that restored the original click feedback.
Cherry MX was not the first switch type to use a sliding part to provide tactile feedback. Datanetics filed a patent in December 1971 (granted as US patent 3773997 “Key assembly diaphragm switch actuator with overtravel and feel mechanisms”) for a switch actuator for their elastic diaphragm switches. The lost motion mechanism was intended not just to provide tactile feedback, but to prevent switch jitter by preventing the switch from being held at the actuation point. Interestingly, the plunger can be rotated 180° to convert the switch into a linear type, simplifying manufacture. The force curve for the tactile feedback is included in the patent.
Stackpole also filed a patent on a tactile feedback system, in 1981, only a year before Cherry; this was granted as US patent 4361743 “Lost motion keyswitch”. The patent drawings are particularly unclear, and thus far a tactile variant of their KS-200 switch array system has yet to be seen. The switch design is targeted towards hysteresis to prevent teasing the switch contacts (holding them at the actuation point where they can jitter between open and closed), and tactile feedback. The design also produces audible feedback, but the patent notes that upon actuation the operator “can hear the contacts click together if the surroundings are quiet”, indicating that the click sound is not particularly strong.
The tactile models in Omron B2H series switches use a pair of magnets to generate the tactile feedback. One magnet is fixed to the plunger, and the other is able to move freely, albeit under continual magnetic repulsion by the plunger magnet. As the plunger is pressed, the two magnets move closer together and the level of magnetic repulsion increases until the magnets cross over, after which they attract and then repel again, pushing the plunger downwards instead of resisting the keypress. This method offers no additional friction, which in conjunction with the Hall effect sensing, yields a very smooth switch.
Magsat Corporation developed a mechanical contact switch with a similar magnet arrangement.
Buckling spring switches angle each end of the return spring so that a slight S-shaped curve is formed in it. As the key is pressed, this curve causes the spring to buckle sideways instead of compress vertically. The bottom end of the spring is attached to a flipper that either forms part of a variable capacitor (IBM’s Model F keyboards) or operates a membrane switch (all other examples). Originally invented by IBM, buckling spring switches were also made by Alps, Brother and Cantech. US patent 3979571 filed by Oak Industries in 1974 suggests that Oak also made buckling spring switches years before IBM invented them, although these are yet to be encountered.
Tactile feel classification
Tactile feedback takes a number of different forms. The most common tactile behaviours are illustrated below with force–travel graphs. The terms used to describe the various approaches are not official terms from any manufacturer literature or any other publications, and were chosen out of need to provide a name for each classification. These graphs do not generally represent any particular switch; rather, they simply illustrate the general principle of each classification. In reality, there is a continuous spectrum of variation across the many decades of keyboard development and production; the classifications and graphs below are merely broad examples to help understand the principles involved. For reference, the plot of a linear switch is as follows:
The y axis of these graphs represents the force exerted by the switch at this position in its travel (or conversely, by the operator’s finger on the key). The x axis represents the travel (motion) of the switch, from fully-released on the left, to fully-pressed on the right. Actuation of the switch typically occurs around the tactile point, although this depends on whether the switch mechanism couples these by design, or simply intends for them to coincide without any guarantee that they will. Few switches do make this connection, with the rest having a separate tactile feel mechanism intended to signal actuation at the same time that this occurs electrically.
The basis for the diagrams below is the force curve measurements taken by Jacob Alexander (see Jacob’s force curves on Plotly) and SPARC (see SPARC’s keyboards page). Please refer to these pages for examples of force curves measured from a considerable number of different switch types.
The most common tactile feel is the “obstructive” method: the keypress is impeded by a rise in force that must be overcome before the switch will actuate:
Although switches with a basic obstructive feel can be highly regarded (in particular “blue Alps”, Alps SKCMAG), these designs carry a drawback that the tactile event fights against the operator. Where the tactile ramp is steep, the sense in one’s finger of overcoming it can be jarring, almost as though the switch is stabbing back. The plunger can also come to hard stop and require a significant and potentially difficult or uncomfortable level of force to push past it. The revisions made to the design of Alps SKCM switches around the turn of the 90s undid some of the fine tuning of their tactile arrangement, with the final form of “white Alps” (SKCMAQ) having a more cumbersome tactile feel over its predecessor “blue Alps” (SKCMAG). In some designs, the tactile peak is a temporary deviation from an otherwise linear force curve, most notably with Cherry MX, which earned it ire from Matias Corporation, when compared with Alps-designed switches and in particular their clones thereof.
Switch types with an obstructive feel include Datanetics DC-50, Fujitsu FES-8 and Fujitsu FES-9 (tactile models), Micro Switch SD (tactile models), Cherry MX hysteresis types (including MX White and MX Blue) and some Alps SKCM. (The overtravel portion of Alps SKCL/SKCM force curves is non-linear due to the nature of the actuator leaf spring. This can result in some of their linear types (including SKCC series) feeling tactile.)
The tactile peak varies widely in steepness and duration. The peak may have a linear period before it drops off, or it may take a parabolic form. Rubber dome keyboards typically have a steep parabola peak.
The severity of the tactile peak can be reduced by removing or significantly reducing the lead-in period:
Some force curves for Alps SKCM switches show this characteristic. Due to age and wear, it is difficult at this time to be clear what Alps’s intentions were, but their 1993 and 1994 catalogues clearly show a very low lead-in for SKCM.
A variation on this general principle is the obstructive feel with dip. Here, following the tactile peak, the force drops below where it would be on a plain linear plot, before rising up and returning to a regular linear path:
This approach is used with Cherry MX Brown and Clear, and it increases the tactile drop. It could be argued that this approach is also used with Cherry ML, although the ML force curve is quite unusual and does not follow standard patterns. The underlying ML tactile feel is superlative, but the high level of friction seemingly inherent in the design undermines what would otherwise be one of the best tactile designs.
In general, obstructive feel is not an inherent poor design, but careful attention is needed to balance out the forces, so that the switch does not feel like it is trying hard to stop itself from being pressed, and does not have a jarring feel.
“Stepped” feel is most associated with stacked-spring switches, in particular Futaba MA. It is also the behaviour of Alps KFF, IBM buckling spring, snap-action Fujitsu leaf spring, Olivetti snap action and the unintentional behaviour of linear Alps switches using their so-called “switchplate” contact modules. In “stepped” switches, the force increases linearly up to the actuation point, at which it drops instantly, and then resumes increasing linearly. The rate of increase may be greater after the actuation point that it was before it, as is the case in Alps KFF and Futaba MA.
Stepped feel tends to have a comparatively lower level of tactility versus other methods.
“Dip” feel involves a temporary reduction in tactile force. It is similar in princple to stepped feel, especially if there is a permanent reduction in force following actuation:
However, its scarcity makes understanding its relationship to other types impossible. This behaviour is known from Marquardt’s unidentified tactile type. It may also apply to Datanetics DC-60. Virtually nothing is known about the tactile form of DC-60: only the tactile type was advertised, but all known customers—in particular HP and Fluke—always ordered the undocumented linear types. The one tactile sample provided by Meryl Miller of Datanetics has a very strong and smooth non-impeding tactile feel, and it seems to be significantly more tactile than the force curve shown in the catalogues would suggest. Without an independently-measured graph available, or any other tactile examples, there is no way to determine whether this one switch has a force curve that matches that shown in the catalogue, or what the production switches felt like.
The parabolic force curve is one where the force reduces in a downwards curve, instead of increasing. Burroughs Opto-Electric is the only known switch type with a pure parabolic force curve:
Other parabolic switches have a linear overtravel curve. IBM’s beam spring, Marquardt “butterfly” and Omron B2H-F7W have a stepped drop in force as with pure stepped designs:
ITT ETL18 instead has a knee point into the linear overtravel:
Having the force remain continuous (non-increasing) or decrease during travel causes it to feel like it is “pulling in” the operator’s finger: the switch works with you rather than against you, compared to obstructive designs. This yields a “bouncy” feel. Parabolic force curves are rare, perhaps because fewer switch designs can produce it. It is not known to be used by any current designs (as of November 2020) but Input Club’s forthcoming Silo Beam switch is designed to replicate the feel of IBM’s beam spring, and will reintroduce parabolic feel into modern keyboard manufacturing.
Progressive rate is not generally an intended tactile feel, but simply the effect of certain designs of switch. In particular, RAFI RS 76 M illuminated exhibits this behaviour. At the point of actuation, the rate of increase of force itself increases:
A small number of switches have a hard stop during switch travel, after which the level of force is distinctly higher than it was before. These include Tokai MM9 series - Deskthority wiki, Robotron dual-spring contact, and RTA 80 Teletype (all determined from force curves measured by Jacob Alexander):
It is not known whether this type of tactile feel was intended, or was simply a consequence of the design of the switch. Jacob’s Stackpole Lo-Pro chart shows it to be double-stepped.
“Wave” feel is represented by a shape resembling a sine wave:
“Wave” feel is used by Topre electrostatic capactive, as well as being the officially-documented force curve for Cherry’s NTK, ATK and LPK rubber dome types. More typically, the “positive” portion of the curve is steeper and wider, and the “negative” portion shorter. Very few independently-measured graphs exist for rubber dome keyboards, so it is hard to understand these in the same level of detail.