Tim's keyswitches page

Keyswitch issues

Keyswitches are central to the feel of computer keyboards.

The Kinesis keyboard and the X-Keys devices both use Cherry MX soft tactile feel switches.

These switches activate before the bottom of their travel is reached.

The ML keyswitches (designed for laptops) appear to have reduced activation travel and reduced activation force - compared to the MX versions.



MX key switch

MX force diagram
 

ML key switch

ML force diagram

The upper lines in the force vs distance diagram show how the key displacement changes according to the applied activation force.

The height of the first peak gives an indication of how hard you need to press a key in order to activate it.

The both keyswitches 'bottom out' with similar forces to those that are required to activate them in the first place - which is not ideal.

Both designs bottom out in a manner resembling running into a brick wall. Not all keyswitches do this. IMO, Cherry need to rethink their approach here - if the keys are going to bottom out so close to the activation point.

From these force diagrams, the ML key switches look as though they would be better to use in a health-conscious keyboard - due to the lower activation force and shorter travel - and despite the fact that they bottom out even earlier than the MX ones.

However, having tried them, they seem to be slightly worse than the MX keyswitches. The constraint of small size has resulted in a number of compromises.

The ML keyswitches have incompatible keycap interface to the MX ones - and need to be board-mounted.

Some keyboard manufacturers seem to be using this type of laptop keyswitch in their products - e.g. the [Plum keyboard] and the [Devlin SpaceSaver range].

The [IceKey] keyboard also uses low profile laptop-style keyswitches as does the [TypeMatrix] keyboard - though these are both scissor/membrane type.

Both types of Cherry keyswitch have activation forces which seem to be harmfully high. The forces involved to activate them are far greater than what is necessary to prevent accidental keystrokes. Yet these are supposedly some of the best keyswitches on the market.

Keyswitch designers: please provide us with the option of reduced prestress and lower activation forces - without the users having to dismantle the switches and perform surgery on their springs.

Keyswitch activation forces

I've measured the average activation force for a number of my keyboards

They are ordered by activation force in the following table:

Keyboard Technology Force/cN
Tsunami City Runner 1000 Membrane 44
Advent Aluminium ADE-KBW100 Scissors/Membrane 45
Targus Wireless Keypad AKP01 Membrane 46
Cherry G84-4700 Cherry ML 52
Cherry G84-4100 Cherry ML 52
Kinesis Ergo Elan Cherry MX 53
X-Keys desktop Cherry MX 53
Samsung SDM4510P Membrane 56
Memorex TS 1000 Membrane 57
Genius Comfort KB-10X Membrane 58
Silicon Graphics SK2502U Membrane 63
Cherry G80-2334 Cherry MX 64
Targus Silver Keypad PAUK10E Membrane 66
Targus Black Keypad PAUK001 Membrane 69
IBM KB454UK Buckling spring 80

Minimising activation forces

The activation force is not the only factor that causes tissue damage with repeated use - but it seems to be a very important factor.

Other factors include: the activation travel distance, the maximum travel distance and the way in which the force changes as the key reaches the end of its travel.

This paper explains the situation reasonably well:

[Computer key switch force-displacement characteristics and short-term effects on localized fatigue] [PDF].

The basic idea is that damage is caused by the tendons sawing into the flesh that surrounds them, and - as with a saw:

damage = repetitions x sin theta x distance x force

This simple equation strongly suggests that tissue damage is proportional to the force in the tendons - which seems likely to depend on the activation force of the switches.

All the commercial keyboards I have used ship with activation forces which I find to be much too high.

It is clear that some people realise the significance of high activation forces in causing fatigue and cumulative tissue damage disorders.

[Datahand] produce low force devices, and [Kinesis] advertise a number of their devices as having low activation forces - e.g. see [here].

However it is equally clear that others do not understand.

This essay on [The Cause of RSI] suggests low activation forces are the cause of RSI.

Also, Cherry write on their web site:

Some specialists still believe that typing on an electronic keyboard is less damaging to health than typing on a mechanical typewriter since strokes need less force. This has proven to be wrong. Especially, work at a computer screen has shown that the force required for a movement is less of a problem - it is the high movement frequency that is a principal reason for RSI complaints. The minimization of the force needed to operate an electronic keyboard has enormously increased typing speed, and there is no longer a pause to execute a carriage return or insert a new sheet.

[http://www.cherry.de/english/service/servicedownload_rsi_servicedownload_rsi_4.htm]

This kind of material makes one wonder whether they understand the importance of minimising activation forces.

Low force microswitches

I discuss the use of low force microswitches as keyswitches on a separate page.

Tactile feedback

I rate tactile feedback highly. Auditory feedback seems to be of markedly inferior quality.

I hypothesize that direct spinal circuits can be formed to deal with relaxing upon encountering tactile feedback - based on existing withdraw-on-finger-prick circuitry.

By contrast, auditory feedback necessarily has to travel via the brain - which is slower - and thus less effective.

Often good quality tactile feedback seems to require high activation forces.

The G3M1T1PUL microswitches show that very high quality tactile feedback is possible while retaining a low activation force.

Other low force switches

DataHand keys are reputed to be low force (that produced by a 18-22g weight).

The Touchstream keyboard was low force (that produced by a 5g weight). However there was also very little tactile feedback.

The "Bloorview Miniature Keyboard" had a low activation force (that produced by a 30g weight).

The [CD Switch, model 90] has a low force (that produced by a 40g weight).

A membrane sensor called a [MicroNavRing] has a low activation force (that produced by a 20g weight).

A membrane sensor called SensorButton15 has a low activation force (that produced by a 35g weight).

Some IBM Model M keyboards are reputed to have some of the lowest activation forces available in a full-travel keyboard (that produced by a 35-40g weight).

However, my own IBM "buckling spring" keyboard does not bear witness to this positive reputation.

Some of the best switches I've seen in a commercial keyboard were those in the keyboard used in the Apple MacBook Pro laptop computer.

It seems to me that touchscreen devices - such as [Apple's iPhone] have considerable potential for decreasing activation forces further.

Spring surgery

I dismantled some of my keyboards and tried my hand at some spring surgery - in an attempt to decrease the keyswitch activation forces manually.

The surgery was successful. More details are on the keyboard surgery page.

Links

This site

From here you can go back to Tim's keyboard page.

The page about low force microswitches is related to this one.

Keyswitches

Cherry MX keyswitches (Cherry site)
Cherry ML keyswitches (Cherry site)
Photos of Cherry MX keyswitches
Photos of Cherry ML keyswitches
Photos of IBM buckling spring
Maltron on Cherry MX keyswitches
Spring surgery for Cherry MX keyswitches
Caps for Cherry MX keyswitches
Cherry catalog - with many keyswitch force diagrams
Keyswitch discussion

Keyboards with microswitches

General switches

Switches for sale - UK
LCD switches
Cherry distributors
Cherry switches - UK
  • Other

    Cherry Cymotion - Cherry ATK
    Cherry Office XPress - Cherry ATK
    Kingston Studioboard Mechanical USB - UNK
    DatadeskTech keyboards - UNK
    Mouse microswitch surgery

  • Tim Tyler | Contact | http://timtyler.org/