Comparing 4 types of tact switch: ALPS unsealed, ALPS sealed, Omron sealed and TE sealed

This considers Omron and ALPS sealed tact switches as replacements for the TB-303 and TR-606, with some consideration of the TR-808 and JP-8 (and a few other Roland devices) which need a stem-less tact switch.

I decided that the Omron sealed tact switches were a better choice than the ALPS ones.

Josh Forgione in California kindly alerted me to some TE sealed tact switches, and sent me some samples.  I considered these as replacements for TB-303s and TR-606s and find them a possible alternative to the Omron switches.

I am also interested in replacements for the Cyclone Analogic Bass Bot (TT-303), which uses surface mount unsealed tact switches.  The TE ones have stems which are too small for the Bass Bot buttons.  There is a workaround for this, but the best option I know of at present is to adapt the Omron sealed switches for surface-mount (Omron doesn't make SMT sealed 12mm tact switches).

Tactile (tact) switches are momentary action on/off switches.  This page is a comparison of three types of tact switch which could be used in the TB-303.  This discussion applies directly to the switches used in the TR-606, and more generally to stem-less versions of the same kinds of switches, as used in the TR-808 and JP-8.  For more discussion of those switches, and for details of how to modify the Omron sealed tact switches discussed here, so they fit the TB-303 / TR-606 buttons nicely, please see:

Before mid-2010, starting with the first Devil Fishes in 1993, I installed the original kind of unsealed ALPS tact switch, with a "dust guard" – a thin mylar sheet with holes punched in it for the switch stems and LEDs.  (The plastic was from a 3M overhead transparency protector.)  This greatly prolonged the life of the switches – I guess by a factor of 5 or 10. 

The failure mode of these switches is that they become erratic and "bounce" – more than one open-close transition when pressing or releasing the switch, as detected by whatever de-bounce algorithm which is implemented in the TB-303's CPU's firmware.  As far as I can tell, the reason is that dust, I guess mainly 100% protein flakes of skin, get inside the switch and are compacted into thin layers on both the stationary contact and the moving metal contact which clicks down against it.  With a badly bouncing switch, the coating is thin and appears to be transparent.  It is visible under a stereo microscope, and easily removed once the switch is dismantled.  I have seen no sign of corrosion or actual wearing of the metal contacts.  The switches, once dismantled, can't be re-assembled, so this is not a method for fixing erratic switches.  They must be replaced. 

The dust guard reduces the amount of dust around the switch which can enter via the circular gap between the moving shaft and the metal ring in the body of the switch.  This lends support to my theory that dust is the cause of failure, since the dust guard can't affect any corrosive elements in the atmosphere, or the actual mechanical movement of the switch.  The electrical current in the TB-303 switch scanning circuit is very low – about 0.3mA, with an open-circuit voltage of about 5V.  I think this is probably too low to "self-clean" the contacts, by way of a little spark or similar.  Perhaps increasing the drive current, by altering the 15k resistors R223, R219, R218 and R48, say to 1k or 2.2k, would introduce some kind of self-cleaning current.  I have never tried this.  It is difficult to anticipate the long-term impact of this, since perhaps that process would also erode the metal surface, leading to corrosion or other problems. 

I was pretty happy with this dust-guard approach, since I think that an intensively used machine without such protection might need its switches replaced after a year or so, and it was clear that the dust guard greatly extended this time. 

However, there were some instances of these replaced switches becoming erratic after 7 to 10 years, at least with the machines which were intensively used.  Ideally, we would replace the switches and no-one would have to worry about replacing them for decades, even if the machine is intensively used.

In mid-2010 I evaluated two kinds of alternative tact switches.  I hadn't been aware that a sealed ALPS tact switch had been available for some years.  I purchased some Omron sealed tact switches and started using them.  These Omron switches had stems a little too wide for the TB-303 buttons, so I developed an elaborate jig with a small grinding wheel on a high-speed drill press to grind down the four sides of the top of the stem, evenly, about a thickness of human hair.  This was difficult and error-prone.  Later I devised a simpler technique of cutting the stem twice, to make it springy.  This is described in ../303-mods/ . 

During this time, I became aware of the sealed ALPS tact switches and purchased 100 of them.  I found they had a very poor feel compared to the original unsealed ALPS switches and the Omron switches.  There was little travel in the "click" operation, and little difference between the force required to activate the click and that required to hold the switch closed once it was in the down state.  (This was not a bad batch of switches. I was later sent some of these switches by someone who sells them, and they feel identical to the ones I tested.)

I kept two of these switches and returned the remainder.

I can't be absolutely sure these Omron switches will last a long time.  However, their datasheet specifies their life as 3 million operations.  There's no specification for what this means, but this is a high number for a switch.

I think that as long as dust is excluded, and the switches are not exposed to a corrosive atmosphere, then they will last a very long time.  The Omron switches are completely sealed against dust and liquids.

#summary

The short version is that the Omron sealed tact switches have a very good click feel, at least as good as the original unsealed ALPS switches.  They involve a somewhat higher activation force, but I think that is fine.

The sealed ALPS switches which we purchased had, in our opinion, a much poorer click action.  The travel was much smaller and there was less difference between the activation force and the holding force: that required to hold the switch closed in the downwards position before it snaps up.

According to the spec sheets, both the original and the sealed ALPS switches have the same travel: 0.3mm.  However, our experience is that the sealed ALPS switches have a much shorter travel between the point where they snap downwards, and the downwards position.

I intend to obtain some sealed ALPS switches from someone who is happy with them.  When I do, I will update this page to confirm or reject the possibility that the sealed ALPS switches we tested behaved differently from the switches that this supplier, and his customers, are apparently happy with.

If you accept our choice – if you lack a profound curiosity about the inner workings and behaviour of tact switches – then there's probably no reason to read further.

In early September 2010, I did some simple tests on the switches and put them up on the 303-mods page.  These tests, of activation and holding forces, were in accordance with our judgment about the clicking actions of these switches.

A few days later, I figured out a way of doing better tests.  The results of those tests are detailed below.  I removed the earlier test results from the 303-mods page.

The most important results can be summarised, but first a few terms I invented for these tests:

• Activation force.  The force required to almost bring the switch to the point where it clicks to the downwards state.  This is measured in Newtons, where 1 Newton is the force exerted by the Earth's gravity, at sea-level, on an object whose mass is about 102 grams. 
  
  
• By the time this activation force is applied, the stem will have moved somewhat from its zero force position.  This movement is what I call the "Initial Displacement".
  
  
• Holding force.  Once the switch is in the downwards state, the force which is required to keep it down.  This is less than the activation force, due to the snap-action hysteresis of the switch.  Once a force less than this is applied, the switch snaps back to the up position.  That new "up" position is not necessarily fully up, as with no force, and it may be marginally higher than when the activation force is applied in the Up state. 
  
  
• Delta force = Activation force - Holding force.  This is how much force must be removed after pressing the switch down with the Activation force, before it will snap up.
  
  
• Delta ratio, as a percentage = 100 * Delta force / Activation force.  This is the ratio of the Delta force as a fraction of Activation force.
  
  
• Click displacement.  The distance between the position just mentioned – with the Activation force, or just a little less, applied – and the Down position after the switch has clicked, and a force similar to the Activation force is still being applied.  The stem may go down further with greater pressure (probably only the Omron sealed switch would do this, since it involves a thin synthetic rubber sheet transmitting the force), but this "Click displacement" is the distance we feel the stem fall when the switch clicks.
  

Here is a summary of the most important measurements. "mN" means milli-Newton.

	Original unsealed ALPS	Sealed ALPS	Sealed Omron
Activation force	1243 mN	1595 mN	1484 mN
Holding force	605 mN	1305 mN	764 mN
Delta force	638 mN	290 mN	720 mN
Delta ratio	51.3%	18.2%	48.5%
Initial  displacement	0.148mm	0.123mm	0.164mm
Click displacement	0.230mm	0.125mm	0.234mm

Please note that these tests were not done in a fancy laboratory and did not attempt to sample multiple batches of switches.   The force measurements for the sealed ALPS switches were based on two switches and the displacement measurements were based on one of these switches.

The figures in red show how different the sealed ALPS switches (at least the ones I tested) were from the other two types: 


I tested 10 unsealed ALPS switches for forces, and chose two of them with close to average forces for the displacement tests.  I used 10 sealed Omron switches in the same way.  I only have two sealed ALPS tact switches, and one of them I had partially dismantled.  So I used those two for the force tests, and the non-dismantled one for the displacement tests.

Even allowing for variation between batches, the differences between the ALPS sealed switches and the other two types are highly significant, and explain why we felt their click action was so poor.

Maybe some people like a lower Click displacement and lower Delta force, but we don't. 

The photo above shows the three types of switch.  The ALPS site http://www.alps.com and Omron sites http://www.components.omron.com have rather long and perhaps not stable URLs.  These are all 12mm snap-in tact switches.

• ALPS unsealed tact switch SKHCBEA010, previously known as SKHCAA.  These are the same as were used in the TB-303, and are still in mass production nearly 30 years later. URL
  
  Force: 1.27 Newton (~130 grams)
  Travel: 0.30mm
  Operating life (5V, 5mA): 1,000,000 cycles
  
  
• ALPS sealed tact switch SKQEAAA010.  URL
  
  Force: 1.57 Newton (~160 grams)
  Travel: 0.30mm
  Operating life (5V, 5mA): 10,000,000 cycles
  
  
• Omron sealed tact switch B3W-4050. URL Datasheet archives here: ../303-mods/Omron-Tact-Switches-2010-B3W_1109.pdf .
  
  Operating force (max): 160 grams (~1.57 Newtons)
  Release force (min): 30 grams (~0.29 Newtons)
  Pretravel 0.29 to 0.32mm
  Service life (min): 3,000,000 operations
  Width of top of stem: 3.8mm +/- 0.1mm. (But I measure 3.84 to 3.33mm in one dimension and 3.83 to 3.82mm in the other.)
  
  

The Omron datasheet contains this graph which relates force (vertical scale) to displacement (horizontal):


The lower curve applies to the B3W-4050.  I am pretty sure that these curves are an artist's impression.  The real displacement is less on the way to 150 grams: 0.164mm.  Then (according to our tests), the switch snaps and goes towards the bottom of its travel, which takes it another 0.234mm, for a total travel of about 0.4mm.

Here are some pictures of the insides of the switches.

They all operate by the centre of the spring disc (phosphor bronze or beryllium copper, I assume) popping inwards, inverting its normal concave shape.  Here is a 15 second video of popping some isolated discs. (Sorry about the erratic brightness.)


Each of the 4 videos on this page is available in two formats:  3GP and Real Video.  Both of them play OK with the free version of Real Player for Windows XP: http://www.real.com . Please let me know if you have any difficulty viewing the videos.

Here are some more photos of the switches.



The unsealed ALPS switch has its stem with a cylindrical flat-bottom plunger, which presses directly against the top of the disc.  The disc is shown in its normal orientation (curve up) on the left, and upside-down, to show the contact surface on the right.  While the centre of the disc pops in, the disc doesn't pop inside out.  So it is still being pressed against the two contacts at the top and bottom of the circular cavity, while, in its down state, the centre of the disc (which is plated with something more silvery) is also pressing against the centre contact.

The sealed ALPS switch on the right has its disc entirely enclosed by a thin adhesive plastic cover.  The plunger at the bottom of the stem has a hemi-spherical shape, and presses against the plastic sheet, which is stuck to the disc.  So the sealed switch has a more acutely focused force on its metal disc than the unsealed one.

The plastic seal looks like it would do a good job of keeping out dust and any liquid which didn't dissolve the adhesive.  I imagine the hemi-spherical plunger would soon wear a hole in the plastic, but that wouldn't affect the operation of the switch, or the sealing against dust and liquids, since the entire bottom surface of the plastic is adhesive, and so it is stuck to most of the top of the disc.

(BTW, the groovilicious triangular graph paper is "L110X Isometric grid", by Gormack Graph Papers, Christchurch New Zealand.  I think it is from the late 70s or early 80s.)





This is the Omron sealed switch, with the one on the right modified so it fits TB-303 buttons.

The black circle is a disc of 0.4mm synthetic rubber, or similar.  The bottom of the stem has a circular plunger, which presses against this rubber, and the rubber presses against the centre of the disc.

This circular plunger would spread out the load more than the hemi-spherical plunger of the ALPS sealed switch.  I assume that this design will not involve erosion of the rubber sheet.  If enough force was placed on the stem, often enough, I am sure this would occur.  However, there's no need for people to press the buttons with much more than the activation force.

At first I thought the rubber disc doesn't seem to be pressed especially hard onto the circular edge of the main body of the switch, by the black plastic ring and the metal retaining clip.  However . . .




. . . this picture shows the underside of the rubber disc indented from being pressed into the circular cavity in the switch body.  It is pressed down by the black plastic ring above it, which is in turn held down by the metal frame.



From left to right: flat cylindrical plunger of the unsealed ALPS switch, hemi-spherical plunger of the ALPS sealed switch and the circular plunger of the Omron switch.




The centre contact of the Omron switch is circular, with a depression in the middle – very different from the broadly round centre contacts of the ALPS switches. 





Here is a close-up of the sealed ALPS switch, with its stem removed.  I have previously pulled this adhesive plastic off, so this is not exactly what it would look like.  There are three pieces of plastic to hold the disc in the right position.




This shows the underside of the disc of the sealed ALPS switch.

In order to test the switches, I needed to ensure the force was placed directly in line with the switch and that the stem was not able to move side to side, tilt etc.

I made a little test jig which mounted the switch upside-down, on a piece of old TR-606 front panel board, such that when I applied force to the back of the board, the switch stem would be moved without any rocking or eccentric forces.




The Haldex (Halda) LMV 1097 force gauge is presumably pretty accurate.  This one goes to 1.5 Newtons.

I used a digital scale to decide at what point of the arm the force should be taken from, and found it was about at the end of the tip.  (It should be about 102 grams per Newton.)  I attached a metal connector and drilled a hole in it, for the vertical bent paper-clip force transductor rod you see above.  It is important to keep the arm, which moves according to the force, at right-angles to the transductor rod, and to keep this rod aligned with the axis of the switch. 

The switch is inverted, sitting in the PCB, at the lower left.

Here are three videos of how I measured the forces.  Its best to watch them in order, since the first one explains the concepts.


The raw results are in results.txt .

The averages appear in the table in the Summary section above, and here.  The units are milli-Newtons:


It can be seen from these figures that the ALPS sealed switches (at least the ones we tested) have a much lower delta force and ratio – meaning there is not such a positive action of the switch staying down until quite a lot of the activation force is removed.

I used a series of digital camera (Sony DSC F707) images to measure the displacement of the stem under various conditions.  In the raw camera image, after rotation and unsharp mask (unfortunately the camera was not quite focused correctly) there are images such as displacement-ALPS-sealed-activation.jpg .  A zoomed to 0.25 version of that image is below.



In the original camera image, after rotation, 10mm is represented by 640 pixels.  So each pixel corresponds to 0.0156mm.

By pasting cropped versions of such images in a row, side to side, all in alignment vertically so their ruler part is exactly level with the left-most image, it is possible with Photoshop to measure the downward displacement of the PCB with the switch, in terms of pixels.  If you want to look, PNG versions of the  raw-camera-scale image, and a zoomed x 2 PNG version are named after each image.  They are not hyperlinks - add them to the URL of this page if you want to see the file.


ALPS-non-sealed.png  ALPS-non-sealed-zoom-x2.png



ALPS-sealed.png   ALPS-sealed-zoom-x2.png



Omron-sealed.png   Omron-sealed-zoom-x2.png

My original plan was to use a thin piece of wire-wrap wire as the pointer to track between the images.  However, it became bent and was no longer horizontal.  So I chose features of the edge of the PCB instead as reference points.  I coloured the edge silver and then black, so it has some bright points.

To measure displacement, for instance for the second of the two Omron switches I tested, measuring the difference between the "Active force" position and the "Down" position, I used Photoshop to draw a rectangle between the same reference points on the two sub-images.  The Info window then tells me the height of this in pixels, in this case 30.  This is with the zoom-x2 file, where each pixel represents 0.0078125mm.  So 30mm is 0.234mm.





Here is a description of the displacements we might like to measure with a tact switch.  All these are with reference to the Up state, in which no force is applied:

1. "Active force" position.  How far the stem moves when the "Activation force" is applied.  This is the force which is almost enough to make the switch click downwards.  This would be a combination of metal disc deformation before it pops inwards, plus any elasticity in the stem and whatever lies between the stem and the metal disc.  The unsealed ALPS switch has direct contact from the stem's plunger to the metal disc.  The sealed ALPS switch is similar.  The Omron switch has a 0.4mm thick piece of synthetic rubber, which I guess would contribute significant elastic movement when something like the Activation force is applied.
   
   
2. Down position.  After the switch clicks, and we are still applying the Activation force.
   
   
3. Hold position, which may be somewhat above the Down position: how far down the stem is when we relax the force, almost to the Holding force, at which point the switch would click up again.
   
   
4. Once it clicks upwards, if we still apply something like the holding force, then the position may be somewhat higher then the "Active force" position, due to elasticity.
   

The photographs enable the measurement of 1, 2 and 3 above. In fact, I didn't quantify the Hold position, since it is clearly close to the Down position for all the switches.  The most important tactile characteristic of the switch, together with the Active and Hold forces, is the displacement between the Active position and the Down position.  I call this the "Click Displacement".

The initial movement from being fully Up to the Active force position, is less important, since it is a smooth function of the increasing force as the force rises towards the Active force level. I call this the "Initial Displacement".

The individual results are in results.txt .  The final results are:


With a lot more work, and switches from several batches, it would be possible to get more accurate figures.  However, these figures are consistent with my perception that the ALPS sealed tact switches (not just the one I tested here, but all the ones I tried of the 100 I initially purchased) have a very small "Click Displacement" compared to the other two switches.

The TE switches of interest are:


The TB-303 and TR-606 have holes for locating posts, but either type of switch would do, since the actual location is fixed by the four pins, with or without locating posts.  Here I will refer to the 1571219-2 as simply "the TE switch", however, at Digi-Key, the -1 is significantly less expensive and so might be a better choice.

TE Connectivity is a big company with a vast range of products.  Here is how I found  information about these switches in May 2015.  From this page:


there is "Tactile Switch" section with a link to another section with 176 switches:


and a javascript link which loads a PDF catalog which my browser was instructed to save to disk:  ENG_CS_8-1773450-9_SECTION_C_TACTILE_SWITCHES_0908.pdf.

Searching for sealed tact switches (long URL) I found the page for the 1571219-2 (really long URL) both of which illustrate a different switch, without a stem.  (The 1571219-1 page is here.)  These switches have stems.  The -1 and -2 pages link to the same PDF drawing, ENG_CD_1571219_F1.pdf , with a PDF date of 2008, from Tyco Electronics Harrisburg PA.  These pages are not very helpful, since it is difficult or impossible to discern which have locating posts.  The Tyco Electronics catalog just mentioned, also from 2008, is a little more helpful.  Page C40 has a chart of the various options with product keys and part numbers.  Here is the chart:



Digi-Key carry these 12 x 12 mm tact switches.  The prices are USD$ for quantity 100, 2015-05-06.



Here are my photos:

For reference, here is a photo of the insides of the unsealed tact switch from a Bass Bot.

ALPS unsealed SKHCBEA010          ~140 gm      65 gm     7 gm                  0.40 mm  

Before considering the feel of these switches, here is a discussion of the size of their stems.

I measured the dimensions of the approximately square stems of several switches:


The brand name of the switches originally installed in the Bass Bot is unknown.

In addition to replacing switches in TB-303s and TR-606s, I am looking for SMT (ideally) sealed tact switches to install in the Bass Bot.  In the future we plan to do Devil Fish mods on these, and the sealed switches which come with the machine will not do, since they will inevitably be contaminated by dust in the years to come, and so become erratic.  Below I discuss the suitability of the TE sealed SMT version.

The first question is how the sealed switches fit the TB-303 / TR-606 buttons:


The second question is how well these stems fit the Bass Bot buttons:

ALPS sealed: 

Omron sealed:

TE sealed:

The switches have somewhat different heights for the bottom of the button with respect to the PCB.  I think this is not critical for the TB-303 and TR-606 which both have a thin aluminium panel through which the buttons protrude, enabling the button to work freely over a wide range of mounting heights. 

However, the Bass Bot has a polycarbonate front panel around the switches on top of the molded ABS case.  The combination of the two is 3.5mm thick.  Since the distance from the rim of the button (which is inside the case) and the top is about 4.7mm, this leads to the top of the button being only 1.2mm above the panel, if it was possible to exactly align the entire front panel PCB at exactly the right distance that the rim of the button just gently touched the inside of the case.  This is difficult or impossible to achieve.  It may be possible to mill or rout (or maybe abrade with a bandfile belt sander) the inside of the case, at least around the button holes, to reduce the thickness and so to allow the PCB to be close to the case, with the top of the buttons higher above the front panel.  The height above the panel needs be considered in light of the travel of the stem as the switch is activated, and in the next sub-section I describe how the TE switches have a greater travel than the others.


Fortunately these heights are pretty close to each other.

I no longer have the test jig from 2010 - nor the patience . . .

The TE switches have a higher activation force (that required to cause the stem to depress and the switch to close) than the Omron or any of the other switches.  They have a very positive click action, with a long travel distance when this happens.  They have a higher delta force (how much the force must be reduced to cause the switch to spring up and turn off) than the others, the delta force percentage (100 minus the release force as a percentage of the activation force) is higher too.  I think they feel good - and I think the activation force is not excessive.

Here are fresh measurements of the switches, using a micrometer for travel distance and digital scales (with the switch sitting on the scales, pressed via a plastic ruler, to enable the elasticity required for the switch to click) to estimate the force of activation and release - with the delta being the difference.  For simplicity, the forces are in grams as force, not in the proper unit of measurement: the Newton, which is the force exerted by gravity at the Earth's surface on a mass of about 102 grams.
 

I only tested one Bass Bot switch.  There was a spread of values for the TE switch forces over the four I tested: 200/68, 200/86, 195/73 and 207/78.  The forces for the original ALPS switches varied considerably too.

I do not like the feel of the ALPS sealed switch.  The others feel OK to me.  I am happy with the force, feel, travel etc. of the Omron switch, but the TE would probably be fine too. It has a significantly greater travel, higher force and much higher delta force, so its click is very clear and pronounced.

We will keep using the Omron B3W-4050 for the TB-303 and TR-606.  They have worked fine for the last five years and I have a cutting jig.  The TE switches could be used, but I would probably need to file them down on two or four sides, which I think would be more difficult and error-prone than the cutting system.

We will keep using the B3W-4000 for the TR-808.

For the Bass Bot I will try using B3W-4050s with their leads cut and bent to suit the SMT arrangement.  An alternative would be to use the TE sealed SMT version, but this would require Teflon tape to keep the buttons in place, and I would need to check that the 0.43mm extra travel was not a problem, considering how little the buttons project above the Bass Bot's front panel.