how about 0.01 degree? see: Auto-Calibration Mods?
Earlier today I asked myself the same question. My Bosh acts like it’s accurate to 1mm… which perhaps isn’t good enough? I say ‘acts’ because when I put it on a solid surface and point it at a wall, it holds a reading, then when I slowly move it, it changes appropriately. I didn’t measure anything of a known length though… but I’ve used it to measure, marked with a tape measure, and things are usually perfect to within operator error. I can get you the model number later if you are curious.
take a look at https://www.youtube.com/watch?v=USa3HFLnrlk
with a metric tape measure, you should be able to get to .5mm, and possibly even
.25mm accuracy accuracy. It’s just a matter of what you measure against, and
measuring outside of motor to outside of motor seems pretty solid. measuing with
the laser would require you mounting some bracket one one side to bounce the
laser off of and on the other side line the laser up against something. I think
the variability in those too points is going to be worse than your accuracy.
If you have a laser, go ahead and use it, but we are primarily looking for what
we can do to get the motor-motor measurement more accurate for the
run-of-the-mill maslow user who doesn’t have a laser.
You’d be positioning a laser dot onto a target just under 10 feet away. Doing an arc tan or arc sin approximation, I’m getting ~0.06 degrees. Assuming 1/8 inch in 10 feet… Tan-1(1/960)
lasers are like $3 now… nbd.
that’s why I linked a much more accurate chip, (in the video).
no idea if the claims are true, but it sure sounds like something we could take a look at if we worked out a method of calibration.
a $300 Leica laser is claiming an accuracy of 1/16" or about 1.6mm
we do substantially better than that with a tape measure.
looking at granger https://www.grainger.com/category/distance-meters/distance-measuring-tools/measuring-and-layout-tools/tools/ecatalog/N-15y1
the accuracy ranges from 1/8 to 1/16"
The problem isn’t the accuracy of the sensor, it’s the accuracy of the mounting and the accuracy of the manufacturing of the mount (3D printers are NOT that accurate, things warp and shrink, which doesn’t matter most of the time, but when you are talking about sensors this precise)
I’d rather concentrate on how we’d use a sensor of this accuracy if we could.
The goal being reaching towards a self-calibrating machine…
it’d be relatively easy to replace ‘3d printed’ with ‘milled aluminum’, and I can’t imagine the per part cost would be outside of what the possible benefits would be.
These are the two “holy grail” measurements, which determine the accuracy of the machine?
I think the pinch roller on a low-strech belt/cable is a very interesting
approach. but I don’t see how a level helps that much.
As far as the sprocket position goes, we need repeatability, we could use a
simple optical interupter approach to detect a sprocket tooth going by to define
the equivalent of the 12 o’clock position. The only hard part is defining how to
get it exactly symmetrical on both sides (which may mean doing something on the
motor mount), especially if you allow different sprocket sizes.
I have seen some 3d printed parts on here to align the sprocket to exactly 0
degrees, but after just running through this part of the calibration routine
for the first time yesterday, I feel that it would be easier to figure out a
way to just use a tape measure to make this measurement, with the help of a 3d
printed jig. Here are some ideas, any thoughts?
- A Part that would slide on to the sprocket, and has a tab on top right over
shaft that allows a tape measure to fit in. Its twin would go on other side,
and you would use tab as guide to where to measure. Might still have problem
of having to line up sprocket to 0 degrees, but maybe less critical.
Actually, we are finding that we are going to need to combine a tape measure
measurement with a chain measurement, because this will tell us how much slop
there is in the chain.
Part that has same tabs to hold tape measure as 1, but instead hooks onto
edge of motor, or even uses motor mounting screw.
Easier now that motors horizontal on beam, but just measure from outside
edge of each motor, and calculate the difference between that and shaft.
This is the approach that will work, but it still doesn’t eliminate the chain
measurement or the calibration step.
Even if you assume that you have a perfect distance-between-motors measurement,
you still need to calculate the actual rotation radius (especially if it’s not a
pre-manufactured triangulation kit), and the yoffset value (which is pretty
trivial after you do everything else)