Hey everyone, not a ChiliPeppr question really, but I know there are a few Ox users here, so I thought I’d ask. Has anyone measured the position repeatability of their machines? Mine is within a couple thousandths on the X and Y axes, but only on one direction on each axis. It’s acting like it’s got about 0.01" or 0.3mm of backlash. For example, move right 1 cm then back, it’s within 2 thou. Move left 1cm then back and it’s short 0.3mm. Continue to the right 1cm and then back and it’s within 2 thou of zero again. I recently changed up the belt fastening and tensioning arrangement so I can adjust it and keep the belts tight. Tried a few different microstep settings, turned up the current, no luck. I’m running 425 oz motors. I’m thinking it’s gotta be mechanical, but it’s strange that the behaviour is the same and exactly repeatable on both axes. One of the pictures shows a circle test and the small circles in the centre don’t look very circular. Any thoughts?
Are you running GT3 belts? ($xtr=60mm)? Assuming you are using 8 microsteps, each microstep is 60/(200*8) = 0.0375mm, so your repeatability is on the order of 0 to 8 micro steps, which seems rather sloppy.
Are your pulleys good and tight? Do you have flats on stepper shafts so that set screws get a good bite?
$xpm=2?
Do you see the measurement change as the motors time out ($mt setting)?
Just FYI, using this test pattern https://dl.dropboxusercontent.com/u/50261731/ShapeOko_Calibration_Pattern_01b_cjm2_complete_2.nc, I was able to visually detect what turned out to be a 1 microstep rounding error tinyG was making in an early FW release. The Parallel lines and the human eye are a great combination here
I feel like you’re probably seeing fairly correct behavior. Have you measure your spindle head slop by putting your gauge right next to it and pushing on it with your finger?
Carl, yes those are the belts I’m using. I did have problems with the pulleys previously but I switched from setscrews to socket head cap screws held with loctite. I’ll take a look at that gcode file - thanks.
John, the thought did cross my mind that I might just have to live with it. It’s not super critical when building guitars, but when cutting inlays it’s nice to have maximum precision. The router mount does have a little bit of flex but even when cutting with very light passes small circles still look off.
If your shafts don’t have flats, you might want to try finding pointed (rather than cupped) socket head or Allen set screws. Or, you could try drilling a dimple into you shaft, so your set screws can really bite.
You might also try, as an experiment, reducing your Jerk value by x 1/2 or 1/5 or so. Nema 23s have a good bit of torque, and fast acceleration (jerk) will accentuate any shaft slip.
Tray and experiment like
G0 X100
G0 X0
Then
G1 X100 F100
G1 X0
Get similar offsets?
These are interesting tests; I have never tried same on my SO2, but will put this on ‘the list’
They do have flats on the shafts. The pulleys are definitely not slipping. When I had the problem originally I really torqued the screws and added thread locker.
My DW660 mounds flex a bit on mine…and depending on depth of cut and grain direction you’ll get more resisteance leading to missed partial steps. How deep of a cut are you trying to make in one pass? While the tool is capable sometimes the rigidity of our machines is not.
On that circle test I was cutting 0.7mm per pass using a 1/16 bit. Feed rate was 600mm/min. I don’t think that would cause it to miss steps because the motors are the big 425 oz/in ones.
Brian, now I am not sure - when you ran the back and forth test described at the top of the thread, was that in air (above work surface) or was the spindle running?
Hey Carl, that test was in the air. Did you see the video of the dial indicator that shows the behaviour?