a little bit here, a little bit there...

Hi all,

after finally firing up my MPCNC, I am approaching the first useful tasks. One of the first is to make a new switchboard for my old sailboat, engulfing a number of holes and this and that. my machine is pretty good already in finding the place where to start its work. I.o.w., positioning without an end mill biting its way through some material works very satisfying.

Different situation when actually milling something, the test subject is 5mm plywood. Milling a simple circular hole results in circular holes, but these are constantly smaller than specified. For example 19,4mm instead of 20,0mm. The same error appears in simple rectangles, even when cutting just a slot in tool width. The borders are about 0,3mm smaller than the dxf file specifys.

When cutting multiple instances of the same subject, the spacing is absolutely correct, though.

Machine settings are triple checked, the GT2 belts are also not to be blamed.

One reason I can imagine: I am still using (historical reasons, long build time) the old version-one center array and carriages, of course also the old version of the tube holding corners. If any, the center array is somewhat sloggy. The direction to think and work for an improvement?

The spindle drive is the Makita RT0700. End mills are VHM 2mm 2 flutes, no twist.

The error is the same for co- and counterrotation.

Cheers, thanks for your thoughts in advance!

Two of the attached images show yet another surprise: Changing the acceleration parameter in Linuxcnc leads to an erroneous tool path - see the asymmetric corners. That is corrected already.

The third image is a feasibility test for the switch board. To make the insertion of the switch possible, I made the opening 0,3mm wider in the dxf drawing. Now it is a tight press fit.




That sort of error is tough to find. I would double check the diameter of your endmill (physically), and make sure you are using a finishing pass. The only other thing could be if you have a large build moving the material closer to the gantry can tighten up tolerances.

1 Like


thanks for your comment. I’ll track down the whole process one more time, also checking the gcode. That obscure corner error caused by the linuxcnc-acceleration parameter (set down to 250 mm/s² from 750) really surprised me. Constant tool speed beats mechanical precision in their goals, still can’t believe that. But after increasing the value to 1000 mm/s² that error was gone, the general geometry error remained there.

I’ll also try to intentionally set a too large value for the endmill diameter.

When I mill a single linear groove, the slot width is exactly the expected 2.0 mm.

The gantry “feels” tight, but…

Any suggestions for test patterns that point to certain error sources?