[sdiy] Two parameter mechanical controller thoughts....
rsdio at audiobanshee.com
rsdio at audiobanshee.com
Mon Jun 10 06:02:34 CEST 2019
On Jun 9, 2019, at 7:06 PM, Pete Hartman <pete.hartman at gmail.com> wrote:
> On Sun, Jun 9, 2019 at 6:22 PM <rsdio at audiobanshee.com> wrote:
>> Have you ever used a set of digital calipers? They have a pattern that is read optically, so the system knows where the slider is positioned at any time. 0-calibration is usually necessary on the affordable ones. I believe that the pattern is basically the same as the quadrature in a rotary encoder, except that it’s straight - not circular. I’m not sure how easy it would be to build something like this, but if possible then it would be a lot less problematic than something with long rubber bands.
> Google gets me lots of hits for "linear optical encoder" that involves transmissive sensors, i.e. it's clear with stripes and you have an LED on one side and a sensor on the other.
I’m fairly certain that calipers don’t work that way, but their sensor mechanism might not be available as a separate product.
> Not having as much luck yet finding anything where both the light source and the sensor are on the same side, but still digging through various links. This really seems like a good way to go, aside from the whole question of absolute positioning -- when I power on, where is it at? It can be saved in non-volatile space but I honestly don't trust that very far, in terms of long term tracking.
If you have a third bit, and an additional sensor to go along with it, you could sense the middle point, or the ends, or maybe multiple calibration points. Of course, that would still require the user to move the control to a calibration point after every power-up.
If you have 4 or more bits, then you can tell where the slider is without being stuck with the usual 2-bit pattern that’s only good for sensing direction of movement. I found some rotary encoders with 4-bit position, but they’re incredibly expensive compared to the usual 2-bit encoder. It’s possible to use 4-bit patterns and still have more than 16 positions, but then it becomes relative again, only without as much risk that fast movements in one direction will be mistaken for movement in the opposite direction (as so easily occurs with 2-bit encoders). The 4-bit versions need to use a gray code, but technically even the 2-bit encoders are gray codes.
p.s. I’m almost tempted to scavenge an optical mouse and aim the sensor at the rails of a long slider. You would either need to ignore one dimension, or separate them.
More information about the Synth-diy