Adventures in Topre-land

[matt3o]

Sorry if I'm going to say utterly obvious things to most of you, I always like to hack my way into a new topic and the following is my experience with topre.

I spent the last week trying to hack a Leopold FC660C, I failed (due to inexperience and lack of measuring instruments) but I grew and insanely respect for Topre switch and Topre keyboards.



First of all. Is it mechanical? Now the debate makes sense to me. The answer it clearly: no. I'm not saying that it is membrane. I'm saying that Topre is the exact antithesis of a mechanical keyboard and it actually forms a category of its own.

I'm probably going to use wrong terminology due to my very low electronics understanding, but I hope the overall concept will be clear.

A Cherry MX is an on/off switch. The circuit is either closed or open. There's not such a thing in Topre switches. Under each key there's an electromagnetic field and the variations in this field are used to determine the position of the key. Note that I used "position of the key" not "if the key is pressed".

Here's the culprit.



Mr. AD5258, a 64bit (!!!) digital potentiometer. Theoretically it would be possible to know the position of the key along all its travel. Topre arbitrarily chose an actuation point, but we could have 2/3/4/5/... actuation points for each key, or we could use the keyboard as a weighted piano (harder strokes make stronger notes). The actuation point seems to be hard-coded in the HHKB, but it might be possible to check that on the Leopold (admittedly I might be completely wrong here).

This is the FC660C header schema



The Leopold has 2 pins going from the potentiometer to the keyboard controller. This maybe means that the potentiometer can be read. On the HHKB the controller is not connected directly to the potentiometer and the actuation point is hardcoded into the eeprom. (Paradoxically this makes the HHKB more sophisticated and also explains the higher price point).

I don't know who had the idea to use such a complex system on a keyboard, it's like taking the Ferrari to go around the corner to buy milk. But I'm glad it exists.

So why rubber is such a chic switch? As far as I understand the spring duty is just to vary the electric field, and of course to insulate the field we need to use a material such as rubber (the field would be altered by other materials). My guess is also that it would be practically impossible to change the spring (say with a stiffer one) because the slightest material variation causes a variation in the field altering the actuation point.

Speaking of price.

Have a look at the number of components (and the quality of the soldering) in a Topre keyboard (the Leopold which is a cheap one)



And this is just the PCB which takes care of sensing the capacitive switches and matrix. The keyboard brain is in the daughter board (unfortunately the mini USB is very low quality, though)



For a comparison, have a look at all a cherry mx keyboard needs:



The one pictured above is a FC660M (the cherry mx counterpart of the FC660C, very lame products naming here). Weird enough the mini USB is higher quality on the cherry mx keyboard, everything else is of course not even comparable with Topre PCB.

I believe I can hardly argue about Topre price from this point on, even though admittedly higher volumes would lower the pricetag, and I hope the Novatouch could help us in this task.

So. The only physical/tactile feedback on a Topre comes from the rubber. Deal with it. Can it be considered a membrane or even a mechanical keyboard? Hell no!

Is it worth the price? If you look (or even just think) at the complexity of the capacitive system: hell yes it is!

My journey so far has been: mechanical (in the early days) > membrane (in the 90s) > cherry mx/alps/buckling > topre > (the future)

I really do hope one day I will find a clicky switch worth replacing Topre, unfortunately MX Blue/Green are just surrogates (decent ones but still surrogates).

And now for your viewing pleasure I present to you the Franken-Leopold





55g membrane on a Leopold FC660C. With a stronger actuation point it is now (arguably) one of the best keyboard I had so far (shame for the lasered keycap legends).

Sorry for the wall of text.

[Bramster]

That is quite a look into a keyboard... Nice!

[matt3o]

I just want to add the the Leopold (unfortunately) has a higher (closer to the top) actuation point compared to the realforce. I'm not a huge fan of switches that are too easy to press. The 55g dome slightly soothe the problem but still if I could hack the controller maybe I could also vary the actuation point.

[002]

Great post! Interesting comments about the actuation points...it reminded me of a Topre prototype board with a giant dial on it which was used to alter the travel...or actuation point - I can't remember now:

knobforce.jpg (517.09 KiB) Viewed 12835 times

[matt3o]

Indeed I saw that picture on GH! So it seems it would be totally feasible. Maybe I'll try again to hack the Leopold controller in the future...

[Muirium]

Getting ideas about variable actuation points across a single keyboard as well as weighting…

I'd love to try that dial! One of my favourite things about Topre is the shallow actuation depth. Way different experience to the hard depth of buckling spring.

But if this is really analogue sensing… goodness. Perhaps cap sense is what we need for our force / travel experiments?

[matt3o]

BTW I discovered why the left control is wobbly on the realforce.

The slider is castrated.

[Muirium]

It shipped like that? Boo!

[matt3o]

It shipped like that? Boo!

I believe it's to make room for the (fucking) LED.

[Muirium]

On Control? Why would there be an LED all the way down there?

[matt3o]

On Control? Why would there be an LED all the way down there?

you can swap ctrl with capslock

[Muirium]

Oh, right. Quite thoughtful of them putting an LED on both. But not to your advantage!

[tlt]

I always assumed that Topre choose rubber domes in the design because they think that it feels better and has better ergonomics than springs. I don't see way you should not be able to make a capacitive switch with a spring. I think many enthusiasts like mechanical switches before of the quality of Cherry MX not because of the springs themselves, look at cherry MY they got springs an get no love.
If you really like to go analog precision on your keyboard I think hall effect is the most accurate sensor.

[matt3o]

I don't know if a spring used for force-feedback could be also used for the capacitive sensor. Maybe it would make the whole design too complex.

[tlt]

I'm not sure but isn't that how model F switches work?

[matt3o]

isn't that buckling spring?

[tlt]

Yes. I'm guessing now so please correct me but I think the buckle both give a tactile feedback and a spike in the capacitive reading that might make it easy to detect and the actuation point more the same on all keys

[matt3o]

Yes. I'm guessing now so please correct me but I think the buckle both give a tactile feedback and a spike in the capacitive reading that might make it easy to detect and the actuation point more the same on all keys

the hammer at the end of the spring is either up of down (which translates into an on or off state). There's no inbetween. Also please note that the potentiometer in the leopold is 64bit. I don't know if they are all used (I hardly believe so) but potentially we could have a value that spans from −(2^63) to 2^63 − 1 (!!!!)

[Muirium]

Model F is indeed capacitative buckling spring. The "flip plate" (what the F stands for) is at the bottom of the spring and snaps rapidly between up and down position, controlled by the buckling in the spring. Topre's springs are there for sensing, not feel. While Model F's are there for feel, but not for sensing! Every designer seems to love contradictions…

As for 64 bit resolution, I've got to presume that's overkill and Topre simply uses a commercial chip that happens to do that anyway. No one could possibly sense the difference between 64 bits of physical resolution over a few millimetres. I doubt many of us could sense the difference between 8 bits! (Not me, I'm a spring buckler.) Besides, the active ingredient in capacitance is that conical spring. It has only a finite number of turns. Its actual response is probably quite jagged.

Genius design, though. I kind of want to invent my own switch with the Topre sensor and certain other magic ingredients for the resistance curve.

[Laser]

Wouldn't it be great to be able to modify the resistance curve, using a programming feature that alters at will the pressing force, using an electro-magnetic "spring" (to have a variable magnetic force instead of the spring)?

[matt3o]

Wouldn't it be great to be able to modify the resistance curve, using a programming feature that alters at will the pressing force, using an electro-magnetic "spring" (to have a variable magnetic force instead of the spring)?

that would be something. customizable force AND actuation point. The holy grail of the switches

[Laser]

I just did a quick search: http://gizmodo.com/283181/embotec-magne ... loats-keys



Keys levitate!

[matt3o]

really? news from 2007? http://en.wikipedia.org/wiki/Carbon-14

[Laser]

really? news from 2007? http://en.wikipedia.org/wiki/Carbon-14

Hehe

Well, it just shows that 1. the idea could work (on other site, it was advertised that the keyboard would even adapt the resistence to its user) and 2. that the Koreans already tried everything in making keyboards

[matt3o]

I would really like to try it

[adhoc]

Just to clarify - you can adjust the actuation point (= TRAVEL), not actuation force. I think some people mixed that up here. Actuation force is handled by the rubber, which will get higher with time as that rubber becomes stiffer.

Great write up, enjoyable read matt3o!

[Laser]

Not a bit of confusion here, just wishful thinking for a different switch.
Indeed, very nice article (even if only for exposing Topre internals)

[Muirium]

Exactly. We're talking about a whole new switch mechanism to work along with analogue cap sense, like Topre's.

I hear it's been tried a few times over the years (for many decades) but no one has aced it yet. There's a lot you have to get right.

[Daniel Beardsmore]

The bouncy things made of rubber are "rubber domes" (sometimes "rubber cups") or, if they are have a hole all the way through, "buckling rubber sleeve".

"Membrane" means "Any thin pliable sheet of material": generally something that is extremely thin, and flexible. A typical membrane keyboard is constructed thus:

Rubber dome and membrane.jpg (384.28 KiB) Viewed 12597 times

Keyboard membranes are flat sheets of transparent plastic of the same thickness as a normal sheet of paper, ca. 0.7 mm. Typical keyboards have three of these, and the outer two bear printed circuit traces, on the inside-facing surfaces. The centre layer contains holes, and the top layer is pressed down against the bottom layer by bending it through the holes. I refer to this as "pressure membrane". The pressure that brings the top layer against the bottom layer can be anything; your finger is fine. I'm estimating around 13 to 19.5 g force across 0.7 mm travel, which feels virtually no different to a resistive touchscreen:

Membrane measurement.jpg (794.74 KiB) Viewed 12597 times

Claims that you can "feel the membrane" in an Acer keyboard are pretty unlikely.

Historically, it was quite likely to be a metal spring (Cherry ML, Alps spring-over-membrane, Tai-Hao semi-mechanical, Acer etc) but these days it's normally a rubber dome, typically now a "rubber sheet", i.e. a sheet of domes.

Some keyboards use only one membrane, for example many vintage Mitumi keyboards such as the Apple Keyboard II:

Single membrane Mitsumi.jpg (771.57 KiB) Viewed 12597 times

I call this "conductive membrane", although that's a misnomer as the membrane always has exposed conductive traces. In conductive membrane, a conductive object is brought down on top of the membrane, for example impregnated rubber. That's what Mitusmi "rubber dome on a stick" switches are (which is a very odd name, as the dome isn't on the stick)

Rubber dome and membranes are completely orthogonal concepts. You can apply pressure to a membrane with other objects than domes, and you can use domes for other purposes that dome pressure.

(Bootnote: this is DEFINITELY something that needs a dedicated set of diagrams on the wiki from someone who knows CAD, as people mix up membrane and domes so many times. Calling bog standard keyboards "membrane" is not strictly wrong, but people keep thinking that "membrane" means the domes, and don't consider what it is that the domes are there for, and how you might go about detecting current through an insulating material … Also, capacitive switches don't need rubber domes — think foam and foil (foam and plastic) and buckling spring (plastic) — no rubber in either.)

[matt3o]

Also, capacitive switches don't need rubber domes — think foam and foil (foam and plastic) and buckling spring (plastic) — no rubber in either.)

nobody said that. you just need a not conductive material.