[yummypaint]

I have used velostat for this purpose with a similar design, but i had too many problems with self-consistency of response across different buttons. There was a fair amount of hysteresis too, i thought too much to be usable as an instrument. I am wondwering if the felt is the secret sauce?

I am moving toward 3d printed pivoting keys with an internal mirrored surface that reflects variable amounts of light into a photoresistor depending on depression. Would prefer the velostat though if it can work reliably

[elihu]

My impression (I'd have to do more testing to verify this) is that velostat is not very consistent if you use it as a through-mode FSR (electrical contacts on opposite sides of the FSR) but works quite a bit better as a shunt-mode FSR (electrical contacts on the same side of the FSR as "interdigitized fingers").

The spacing of my traces for the interdigitized fingers is about 8 mils, and the traces are about 6 mils wide. I'm using JLCPCB, and they have no problem with that trace width.

I also use an op-amp (TLV274) in voltage-follower mode as a buffer in front of the ADC inputs.

Sensitronics makes a nice simple test board you can use to test various FSR materials with various trace widths: https://www.sensitronics.com/products-xactresponse.php

(Their website doesn't say so, but the board also comes with samples of their FSR material.)

Here's a (somewhat out of date) schematic showing approximately what I'm doing: https://github.com/jimsnow/microtonal-controller/blob/main/d...

(The keybed itself is a separate board that I haven't posted the schematic for yet, but it's mostly just a bunch more shift registers and FSR elements. FSR elements are notated as a squiggly line.)

There's a company called Wooting that makes gaming keyboards with analog key travel sensors. Their current design uses magnets and hall effect sensors, but I believe their earlier versions used LEDs, light sensors, and mirrors, which sounds similar to what you're doing.