Re: [CAD_CAM_EDM_DRO] Re: 3D Contouring - Loss of Accuracy - Follow-Up

Hi All,

Thanks very much for some fresh inspiration and ideas! I don't feel so
alone anymore...

Firstly I'm very sure the problem isn't mechanical in the sense of loose
thrust bearings, pulleys etc. I've checked those things repeatedly and
if that were the problem, it should occur on a long single axis test
file. All shafts and pulleys etc. have lines drawn/scribed on them and
nails glued on as pointers so I would see if anything had moved. I
agree with others - the magnitude of the error seems way in excess of a
small thermal expansion so I have ruled that factor out of the search.
The spindle warms up a bit but the knee, saddle and table don't seem to
change temp much at all and my Z is on the knee - not the quill. It is
possible that the error I observed with the belts removed was related to
resonance as was suggested - I confess I did not re-tune the servos for
such a light (negligible) motor load.

FYI, Visual Mill does not use G02 or G03 - it breaks all contours down
into lots (and lots) of G01's.

Through a process of elimination and previous on-line discussions, I had
suspected that the problem was related to lost steps during direction
changes. I have only recently purchased the Rutex drives and so most of
the previous testing has been with G210 or G320 Geckos - which others in
this group had reported certain fussiness with repspect to direction
line change timing under some circumstances. I'm really glad to hear
that commercial machines don't lose any steps if set up right and use
glass scales to compensate for thermal expansion rather than lost
steps. Not knowing much about commercial machines - I was left to wonder.

To eliminate the suspicion of the Gecko's step/direction timing I fitted
the Rutex drives which are true edge triggered devices. If a step pulse
is high going low and 10uS wide, the Rutex drive samples the direction
line on the rising edge i.e. 10uS after the step pulse began. I may
have rushed a bit to begin testing the Rutex drives and will re-visit my
wiring and settings for step pulse polarity, timing etc. to make sure
the Rutex drives are indeed set up with correct polarity etc.

I do have the latest Flashcut V2 build and will explore the G27, G28,
G29 (not in that order) routine that Chris identified. It will be
interesting to see how that affects things however I will continue to
test for step/direction pulse timing issues. These tests should keep me
occupied over the Xmas break!

If anyone is interested in running a "problem" 3D contour test file on
their own system - contact me off list with some info about your machine
envelope, max speeds, inches or mm etc. and any other issues related
to the syntax of your g-code file. I will generate a suitable test
file. In performing the test, you will only need to cut air there's no
need to worry about the tool etc. It would be interesting to try some
of my problem files on different systems, including Mach1 and EMC.

Thanks again, everyone!

Cheers, Peter