Muirium wrote: ↑14 Mar 2023, 21:15
That’ss QMMK to yyou, ssir!
LOL but it's kinda laughing at a cripple a bit. Thresholds needed a bit of tweaking, that's all. Didn't even have to increase debouncing window much - it's currently 4.
Muirium wrote: ↑14 Mar 2023, 21:15
Wonder how it handles bounce. I find Pandrew’s capsense amazingly good on my (hitherto barely useable) Beamspring. It may well take this in its stride, too.
Beamspring doesn't bounce.
![Image](https://github.com/dmaone/CSI/raw/master/BeamSpring/Waveforms/5251/cycle-quick-complete-rebound.png)
. There is a bit of rebound after you _release_ the key, but as you can see it's nowhere near full actuation level, so it's easy to ignore by setting threshold properly.
Buckling spring bounces even less - if anything, it overshoots when the flipper hits the PCB
![Image](https://github.com/dmaone/CSI/blob/master/BucklingSpring/Waveforms/ModelF/15us-fast.png?raw=true)
Spurious repeats with xwhatsit are due to the unfortunate design decision - namely, the desire to hold the sense line above ground, which leads to necessity to catch rapidly-decaying signals:
![Image](https://github.com/dmaone/CSI/blob/master/BeamSpring/Waveforms/xwhatsit/Key-pressed-5us.png?raw=true)
and truly horrible noise immunity. This is trivial to fix - one just needs a chip with 8 open-collector outputs - but apparently xwhatsit's sensor schematic is even more sacred than model F sense pads form factor, and nobody dares to touch it. But unlike the former, the latter is good enough - not great for 2023 state of the art, but good enough.