<DocScrutinizer05>
rather two 90° phase skewed adjustable capacitors, or recently rather digital encoders
<eintopf>
but I need more granularity than 8 bit
<DocScrutinizer05>
when you need more resolution than 8 bit, then how about using a laser mouse looking on a ring mounted to the axis, and have an index marker of some sort (e.g. a hole in ring and a light path sensor) to keep track of the absolute position
<eintopf>
this sounds too complicated for me
<whitequark>
that's actually a very high-resolution method
<whitequark>
you have mice upwards to 8000dpip
<whitequark>
*dpi
<DocScrutinizer05>
of course you also can do a binary encoder, but that's for sure even more complicated. BTW usual potentiometers have a resolution quite a bit less than the requested 9 bits
<whitequark>
and you can make radius as high as you can possibly want
<DocScrutinizer05>
whitequark: exactly
<whitequark>
i.e. it essentially offers unlimited resolution
<DocScrutinizer05>
and shouldn't be any complicated to build either
<whitequark>
though at some point accumulating error will be a big problem
<DocScrutinizer05>
optical mice afaik have no systemic accumulating error
<eintopf>
what the hell is a light path sensor
<whitequark>
DocScrutinizer05: of course they do
<DocScrutinizer05>
a thing made of an LED and a light sensor, both form a gap
<whitequark>
it traverses 1mm, it says it traverses 1.1mm
pcercuei has joined #qi-hardware
<DocScrutinizer05>
that's not systemic accumulating error, that's maybe a calibration factor
<whitequark>
mmm
<DocScrutinizer05>
afaik the sensor works with pattern recognition and that has no _error_
<whitequark>
well, there are two concerns
<DocScrutinizer05>
when you move the target a million times back and forth for 10mm, it should still shop position:0.0000000
<DocScrutinizer05>
show*
<whitequark>
that'd be a different kind of error
<DocScrutinizer05>
everything else is not error but calibration
<DocScrutinizer05>
or error introduces by noise/jitter, but that's still not accumulating in this concept afaik
<DocScrutinizer05>
iow once you calibrated the ting to show 10000cm as 10000cm, it's as accurate as it gets: 1/8000 inch
<whitequark>
yeah, that makes sense
<DocScrutinizer05>
the rest are non-linearities but no accumulating errors
<eintopf>
you mean with index marker some point like 0 degree's and the mouse detects all movements?
<DocScrutinizer05>
yes
<eintopf>
this really working on long distance?
<eintopf>
s/this/do this/
<qi-bot>
eintopf meant: "do this really working on long distance?"
<DocScrutinizer05>
such concept obviously needs a setup procedure after power up
<DocScrutinizer05>
where's a long distance in this design?
<DocScrutinizer05>
the mouse sensor should hove a 0.5mm over your ring
<DocScrutinizer05>
hover
<eintopf>
ah, yes
<DocScrutinizer05>
or even glide on it
<eintopf>
I understand
<DocScrutinizer05>
after poer up you need one full turn of axis to make sure the index marker passed by
<eintopf>
s/I understand/I see
<qi-bot>
eintopf meant: "I see"
<DocScrutinizer05>
a second turn can auto-calibrate the mouse data, you know it's been 360.000° when index marker passing by a second time
<DocScrutinizer05>
when your mouse reported 36000 clicks until that 2nd index event, you know it has a resolution of 1/10 degree
<DocScrutinizer05>
err 1/100
<DocScrutinizer05>
such sensor would be *very* fine grained and accurate
<DocScrutinizer05>
and you can build it for 25 bucks from material you got in your garage ;-)
<whitequark>
25 bucks? how about zero
<eintopf>
okay, okay. This sounds like a nice solution. Thanks
<eintopf>
now I need to figure out how I can control the motor of my old scanner device and place a stick on it with some mass at the end. But maybe the motor control is too slow
<eintopf>
I am sure I could do that with libusb and some control messages
<eintopf>
for controlling a inverted pendelum...
<eintopf>
s/a inverted/an inverted
<qi-bot>
eintopf meant: "for controlling an inverted pendelum..."