We did some interesting calibration checks on the Shapeoko 2 at the space last night. Driver and viewer software for the USB microscope are now on the Shapeoko laptop. With the ‘scope mounted to the gantry looking at the lines etched on an old-school vernier caliper body we observed average backlash (difference approaching a point from different directions) of around 0.009″. Absolute accuracy varied from 0.003″ to 0.013″. That latter worst case error is about 1/3 the diameter of the holes we drill in PCBs, and so is potentially a problem.
There was some discussion of ways to improve the mounting of the drive belts. With some evidence of belt stretch, replacing all the belts is also under consideration. More details in this post on Jim’s project notes.
Thanks to Daniil, the SO2 now has a PWM speed control! We have high hopes of adding closed-loop speed feedback to it soon. We’re finally getting close to being able to machine plastic! (That’s about impossible at the native 12K RPM of the spindle.)
Our shapeoko has produced its first output under control of g-code! It’s taped up on the refrigerator at the space.
Inkscape converted a bitmap of the logo to a .svg, the gcodetools extension generated g-code, and vi did the final modifications. The .svg needs a little cleanup, but it was more than adequate for this first test.
We now have a profile that’s calibrated to within a few percent for X, Y, and Z, though there’s still work on max speeds and accelerations. This plot was made with a ballpoint pen in a very crude holder. The bitmap-to-path converter generated inside and outside paths for the lines, so the mismatch of the actual plotted paths gives us some insight into opportunities for mechanical improvement of the shapeoko/penholder system. While the penholder is responsible for some of the tracking errors, we still have a lot to do to tighten up the shapeoko. The plate joining the Y and Z axes wobbles surprisingly. But it’s starting to work!
Update 10/2/12: Using the very convenient test facilities of the axis setup in linuxcnc’s stepconf tool, I maximized travel speed on all 3 axes. The shapeoko1 profile is getting pretty usable. Here’s a little real time clip of it plotting. This one used a Sharpie, and even though it only stayed in one spot while the Z axis raised or lowered the pen, the paper bled the ink into very noticeable dots every time it stopped.