deskthority

I've seen people converting their model M to F. They use a new pcb and flippers and springs from the F. But the M also has flippers, they're just smaller. So why not use existing flippers? They should change the capacity just as well, right? Am I missing something?

According to pandrew, F flippers are made out of carbon loaded plastic, so that helps them register, and normal abs ones wouldn't register. Thanks for pointing this out! I had no idea.

In this thread, fricked tested an F flipper made of non-conductive resin, and reported it doesn't work. But I think it might be due to xwhatsit expecting a larger change in capacitance than is provided by a non-conductive-resin flipper.

I have a feeling any plastic might make a difference of some sort that can be measured well enough.

See here: https://www.allaboutcircuits.com/textbo ... pacitance/

The capacitance of a capacitor depends linearly on the permittivity of the dielectric located between the plates. While graphite has a permittivity 10-15x that of air, abs has permittivity that's 2.5-3.5x that of air, so it should be possible to measure it, it's a pretty large difference. So while your capacitor doesn't change by 10x, it does change by 3x, which is still measurable.

My guess is IBM needed a 10x change in capacitance because in the 70s they only had very slow chips which just didn't have the time resolution to measure a difference that's 2.5x or 3.5x. I feel like nowadays we might be able to measure it.

However, if 2.5x-3.5x is not good enough, then maybe you could take copper foil, and stick it on your M flippers. That would definitely change the capacitance by much more than non-conductive abs alone.

Note that in all of this, I think it might be necessary for the PCB to be changed from what an F would use. The capacitor plates on the PCB would have to be smaller. But maybe they would work as-is - I have no way of knowing for sure without checking.

A simple way to check would be to:

1. take a piece of copper clad pcb, like the kind you'd etch stuff out of. Cut out a 1" square.
2. draw two rectangles that are the same shape as on a single key's capacitor plates in a model F
3. remove all other copper except two traces leading from those plates to the edge of the pcb
4. attach an inexpensive capacity meter
5. measure the capacitance of the capacitor as it is
6. move a Model F flipper to how it would be if the key was being held down. Measure the capacitance
7. do the same but with a Model M flipper. Measure the capacitance again.
8. do the same but with a Model M flipper that has copper tape stuck to its top surface. Measure capacitance.
9. Remove part of the capacitor plates to where they only have the kind of footprint as a model M flipper would cover. Measure the capacitance without a flipper.
10. Measure capacitance with an F flipper.
11. Measure capacitance with an M flipper that has no copper tape on it.
12. Measure capacitance with an M flipper that has copper tape applied to it.

I have copper clad pcb here and a capacitance meter, but I don't have any F or M flippers, other than inside my keyboards, and they're not bolt modded / disassembled, so I'm not going to disassemble them for that. If anyone has any extras you don't need I'd be glad to check.

The Model M flipper also has a small bump on the bottom for pressing the membranes together when it is rocked. To make it work without membranes, you would have to shave that bump off from every flipper.

You could cold-cast resin with metal or carbon in it, but it will become even more brittle and casting resins are brittle to begin with.
There might be some 3D-printing filament with conductive material in it.
But I think a flipper with a sheet of mylar tape could be a better idea. Capacitive foam-and-foil switches use mylar.

You can buy metal-infused printing filament, however the extent to which it influences capacitance is something I don't know. Plus, it's usually almost twice as expensive as normal filament according Google, so it might not be the most cost effective thing. Also the tolerances are so tight on both the flippers and the springs I doubt you could print to a reasonable degree of accuracy using a desktop printer. Your best bet would probably be coating the flippers in some kind of capacitive substance, like a conductive paint, and using that on a custom designed PCB.

The problem with using model M flippers is that the feel will be exactly the same, the only advantage is making the keyboard capacitive. Ideally, a bigger flipper results in a crispier switch, so you can just reproduce the classic capacitive flippers or play arround with custom IBM compatible designs.
About the sensing part, can't you take a very isolating plastic and make the capacitance decrease greatly? It would be sensed the same as a beamspring switch, where a lower capacitance means a HIGH level. On the other hand, you can make a flipper where a metal piece fits an internal engraving, leaving a bit of margin in the plastic to not scratch the PCB and affect the lifespan of the switch.

I'm going to preface this with it has been quite a while since I have done anything with electromagnetism so take what I say with a grain of salt.

The main issue that you would run into is the size of the flipper. Yes, the same principle does apply with the model M flipper as the model F flipper, however, since the model F's flipper is larger, particularly since it is longer, it will make a larger capacitance change assuming that all other factors like dielectric are the same. This is because the flipper can move farther away from the plates on the PCB. It is the change in the distance of the dielectric that changes the capacitance. So yes they, could be used to change the capacitance, but it would be a lot harder to detect. Also, the since the model M flipper is shorter the moving portion is closer to the non-moving portion. I think this would also reduce the capacitance change.

I agree with PlacaFromHell, I don't really see the purpose of the exercise.

Findecanor wrote: 14 Dec 2020, 15:27 The Model M flipper also has a small bump on the bottom for pressing the membranes together when it is rocked. To make it work without membranes, you would have to shave that bump off from every flipper.

it's enough to put a hole in the PCB

PlacaFromHell wrote: 14 Dec 2020, 19:17 The problem with using model M flippers is that the feel will be exactly the same, the only advantage is making the keyboard capacitive. Ideally, a bigger flipper results in a crispier switch, so you can just reproduce the classic capacitive flippers or play arround with custom IBM compatible designs.

it wouldn't be the same, you have the membrane pushing back at the flipper in the M which make it very mushy when rising up. If the flipper were resting against a solid object, like in the F, it would have the same crisp return action.

The press-in action between M and F is very similar.

Willy4876 wrote: 14 Dec 2020, 19:55 I'm going to preface this with it has been quite a while since I have done anything with electromagnetism so take what I say with a grain of salt.

The main issue that you would run into is the size of the flipper. Yes, the same principle does apply with the model M flipper as the model F flipper, however, since the model F's flipper is larger, particularly since it is longer, it will make a larger capacitance change assuming that all other factors like dielectric are the same. This is because the flipper can move farther away from the plates on the PCB. It is the change in the distance of the dielectric that changes the capacitance. So yes they, could be used to change the capacitance, but it would be a lot harder to detect. Also, the since the model M flipper is shorter the moving portion is closer to the non-moving portion. I think this would also reduce the capacitance change.

I agree with PlacaFromHell, I don't really see the purpose of the exercise.

It's not an issue. F flippers were large because that's how much they had space for, and they had antiquated 70s technology to work with. Nowadays we have 12-bit ADCs on the cheap which 70s engineers could only dream of, and we can sense at tens of kHz. Finally, you could modify the sense circuit to be an RC oscillator plus a counter, which could easily make this way more sensitive to capacity change than it is right now. Not that it's not already good enough, the difference might be about 2-3x capacitance, but really measurements need to be done because I'm guessing. This is not an "electromagnetism problem", it's a problem of engineering, and this has luckily been solved for us by 50 years of progress in semiconductors.

It's not an opinion, I have several spares and keyboards to compare and I can tell you that my "matless" boltmodded model M not only doesn't feel at all like my models F, also feels almost the same as a model M barrel plate and pivot over a PCB. That's not the only experiment you can do, grab any buckling spring clone (for example the Sycan TK-something) and the smaller the pivot, the shittier the switch.

cheater wrote: 14 Dec 2020, 23:24 It's not an issue. F flippers were large because that's how much they had space for, and they had antiquated 70s technology to work with. Nowadays we have 12-bit ADCs on the cheap which 70s engineers could only dream of, and we can sense at tens of kHz. Finally, you could modify the sense circuit to be an RC oscillator plus a counter, which could easily make this way more sensitive to capacity change than it is right now. Not that it's not already good enough, the difference might be about 2-3x capacitance, but really measurements need to be done because I'm guessing. This is not an "electromagnetism problem", it's a problem of engineering, and this has luckily been solved for us by 50 years of progress in semiconductors.

You are definitely wrong about it not being an EM problem. Capacitors work through electromagnetism, but that doesn't mean that you can't make it work or that you have to know much about it (My favourite part about engineering is that you can take all of the stuff that physicists have to work about and throw it out the window by making assumptions. This is where engineering jokes like pi = 3 = e come from).

As I said you can definitely do it, it will just be harder. If you end up doing it, this should be an interesting engineering project to follow. Personally the first thing I would do is make a PCB for a single switch and pick up something like the model M switch testers that orihalcon sells and replace the acrylic base with the PCB for testing. That way you can play with a single switch, which will be cheaper in the prototyping stage. The barrel plate will definitely affect the capacitance in addition to the the flipper so you will have to play around with it and find a threshold that works. The threshold will almost certainly change going between different model M chassis like the M-122 and the regular one. Also, keep in mind that things like proximity to ground planes, and external grounded metal also affect capacitance so something like a grounded metal shield behind the PCB or a full ground plane layer will help make it less sensitive to the environment it is in (BS like this is why I try to avoid anything EM or RF stuff my designs if I can help it).

Good luck with the project!

The switch testers are a good idea. I don't really feel like taking apart my working M's to test any of this.