Can anyone link me that thing that could convert an F switch to a digital switch? Or put the right name in text at least? That thing sounds so promising (probably cost too much but still).
OK, might have found it: [size=78%]http://www.analog.com/static/imported-files/data_sheets/AD7147A.pdf[/size]
If I look at that thing, the only concern would be the delay. Depending on the value I look at, the conversion delay would be somewhere between 3 ms to 200+ms. Anyone know how to read this paper and get a "real life" delay out of it?
Nice thing is that it seems you can plug like 13 switch on 1 of these, so it could be reasonably priced for a whole keyboard.
the problem with this is that it's a capacitance _encoder_ not a capacitive switch detector. it's also too big for a single switch. this IC is for capactive touchscreens. the idea is that you segment your screen into sections and it digitall encodes the amount of pressure a user is putting on each section of the screen. hence, it's quite slow to update, because you don't need instantaneous or event-based scanning in this application.
my preference for capacitive switches is an IC that's extremely simple. ideally you would simply hand it a logic level voltage source and it would supply AC voltage to the capacitor/switch. when the frequency of the switch changed above a given amped threshold (because we're talking very very small capacitance values here) it would inject a single binary value into a bus or crossbar or switch matrix or something. this kind of IC would hopefully be very very small; internally, all i'm asking for is a something that's a bit like 2 opamps in a bipolar topology. this is really really to fit in an soic-8 form factor.
i do know such ICs exist; the touch sensors on eg my dell monitor almost certainly use such a chip. i haven't really done enough research to figure out what the best/cheapest options are though, since a) dfj and xwhatsit are making incredibly good progress on the problem, and b) i have way too many other things on my plate. it would necessarily have to come after getting a small production line up for the SSKrs (maybe 4-5 a month). however, i totally invite you to take a look at dfj and xwhatsit's work and ask yourself a couple questions:
can i find an IC that amplifies this signal cleanly enough that i can distinguish between off and on states? for example, the standard 8-bit micros we use; the avr series and the freescale 6800 socs all have very crude ADCs (you need a 2-bit ADC so not a problem) and acmps on them/
once you can distinguish between off and on states, you need to not only put a basic schem together for, let's say, a 10 key numpad, but, and this is a verification step for your solution to the last problem, can you actually lay this out on a PCB with a standard switch device matrix on it?