[X-Scale] has quit [Quit: HydraIRC -> http://www.hydrairc.com <- Po-ta-to, boil em, mash em, stick em in a stew.]
Bike has joined #homecmos
fltrz has joined #homecmos
kristianpaul has quit [Read error: Connection reset by peer]
kristianpaul has joined #homecmos
kristianpaul has quit [Ping timeout: 256 seconds]
kristianpaul has joined #homecmos
<fltrz>
Degi, I just read in some article about optical interferometers that "Conventional moiré-type linear encoders have stable
<fltrz>
readings but the resolution is limited by the large grating pitch"
<fltrz>
so the printer encoder is probably actually straight vertical
<fltrz>
and the spacing on the printhead slide is slightly different so you get vernier scale type moire patterns for the quadrature photodiode / led combos
<fltrz>
thats probably how they actually work
<fltrz>
also pretty genius
<Degi>
Vernier scale moire type?
<Degi>
Don't they just use two offset diodes by 1/4 pitch
<fltrz>
nope I now believe its 2 encoder strips, the long one like on ebay, and a short one with a slightly different grating pitch, so you see some kind of sinusoidal intensity variation due to moire pattern
<fltrz>
both are vertical
<Degi>
Hmh
<fltrz>
so you dont need to couple with a lens to the fine slit width
<Degi>
But I think in reality the second strip is just two slits on two LEDs and the slits have a n+1/4 pitch distance between them
<Degi>
Or maybe even without a slit, at the detriment of SNR
<fltrz>
I think a slit is possible, but then the precision the long encoder strips pattern need to be very high
<fltrz>
while with moire, you can spread and average the error over multiple stripes and the width of the strip
<Degi>
Why would a second strip on the moving part solve that?
<fltrz>
like calculating the center of a slightly blurred disk is more accurate if the disk is larger and thus more pixels could be used (not that theres any pixels here, but the same ahould apply)
<Degi>
Huh, the center?
<Degi>
I think they actually threshold the LEDs and process it digitally
<fltrz>
"<fltrz> nope I now believe its 2 encoder strips, the long one like on ebay, and a short one with a slightly different grating pitch,[...]"
<Degi>
Hm, I think they just have two thin slits in a block of plastic. The long strip touches the surface of that block of plastic, though maybe I can have a closer look at the one I have someday
<fltrz>
I think this makes the positioning of the 2 readouts less position critical, since the sinusoidal wave is larger, and if both strips are made of the same material (but differnt printed pitch), they have same thermal coefficient of expansion, so 1mm stays 1mm
<Degi>
Hm, its casted in a solid block of plastic, which should be relative precise and reproducible
<fltrz>
if you use say a metal plate or a different material, coefficient of expansion is a nuisance over the distance between 2 sensors
<fltrz>
also, wont be quadrature, but some other degree phase shift between sensors
<Degi>
At the 20-50 °C operating range? They're both made out of plastic
<Degi>
I think perfect quadrature isnt as important
<fltrz>
I still think its moire linear encoder, as the cost of a few cm extra is low
<Degi>
The whole sensor is < 1 m
<Degi>
*1 cm
<fltrz>
then its even cheaper?
<Degi>
I mean not having the moire strip would be even cheaper and the printers have every cost saved on them...
<fltrz>
having 2 slits is kind of having a moire subpattern, but I think the noise averaging is better, because instead of requiring each stripe to be super precise, you require the majority of stripes to somewhat precise
<Degi>
I mean the thing doesnt need to be super precise
<fltrz>
I guess
<Degi>
It needs to sense number of lines and direction, as long as their phases aren't super close, it should be able to do that. So like 20-160 ° shift should be okay I think
<fltrz>
but still more precise than 3-10 lines per mm
<Degi>
I dont think so
<Degi>
Yes, the printing resolution can be higher but that can just be interpolated from the velocity
<fltrz>
I think you can indeed lock to the encoder strip and use the mass of the carriage to smooth out carriage force errors, but it would need more time and length to lock in and stabilize after each carriage return
<Degi>
I guess using some second order interpolation its likely going to be fine. Maybe even linear interpolation, that tiny bit of jagginess wont be noticeable...
<Degi>
Or some weird tricks which make you wonder how it works to save as many costs as possible lol
<fltrz>
lol probably
<fltrz>
Degi, I read a lot in the context of moire type linear encoders that they are reflective, you had a linear encoder in front of you, would you mind shining a light on it at an angle?)
<fltrz>
typically chrome on plastic or smth
<Degi>
To see if the lines are reflective?
<fltrz>
could still be non moire of course
<fltrz>
yes
<Degi>
Wait apparently the encoder is somewhere else, but I have the strip
<fltrz>
since having light source and detector in the sensor makes things easier since the encoder doesn't need to be suspended
<fltrz>
I meant the strip
<Degi>
It looks relatively black and nonreflective
<Degi>
If I shine a laser thru it, I get diffraction xD
<fltrz>
hmm, perhaps it relies on whatever metal it is mounted on to be reflective?
<fltrz>
I was going to ask to shine a laser to check for diffraction orders :)
<Degi>
I think this one has LED on one side and PDs on other
<fltrz>
do you also see reflected reflection orders?
<Degi>
From one side, very clearly annd in an intersting patttern
<fltrz>
thats the transmission pattern or reflection pattern?
<Degi>
The 1st order is brighter than 0th order for some reason
<Degi>
Reflection
<Degi>
From one side it works
<Degi>
But its also modulated with a sine of a period of 5-10 diffraction orders
<fltrz>
not sure what you mean with from one side...
<Degi>
The other side of the foil just reflets a dot
<fltrz>
ah I see
<Degi>
But one side (I guess where the black liens are printed on) makes diffraction
<Degi>
Which would make sense even with apsorptive coating
<fltrz>
so probably the the metal is absorptive on one side and reflective on the other?
<Degi>
Hm
<Degi>
I think its black ink and the foil is reflectiveß
<Degi>
Hm yes, that seems likely
<fltrz>
like specular reflection on the transparent foil?
<Degi>
The side with no diffraction reflections has a brighter reflection
<Degi>
Yes
<fltrz>
perhaps absorption by the ink on one side preventing internal reflection when shone from a specific side?