Hi Brian:
I think I have a fix for this. It will be in the next version. The problem
is a very long standing one that was thought to be gone. The problem is the
rotational axis are separate from the linear interpreter because they
typically cannot be based on feedrate without a radius compensation of the
angualr axis.. So when a move like G1X1A360 is made, Mach2 is forced to
recompute the feedrate based on an algorithm that compares the time required
to move X 1 to the time required to move A360. The move is then scaled and
the X movement feedrate lowered so that the X move now takes long enough for
the A axis to complete its move. This algorithm was created by myself and
Rab 2 years ago as a way to better control the rotational axis.
Unfortunately, though we scaled the feedrate, we did not scale the
acceleration curves. Till now, no-one had trouble as their A axis accel was
near enough to the X and Y that those systems worked fine. Now , of course,
with Mach2 growing in popularity, it was inevitable that some systems would
present themselves with a sensitivity to this oversight.
For the next version I have scaled the Acceleration by the same ratio
factor as the feedrate. This means however, that there may be a lack of
linearity in the acceleration curve tracking in different conditions, but
this would affect only mixed Linear/Angular moves. This fix is based solely
on the assumption that if a feedrate is scaled by 50% to compensate for
angular moves responce, then the acceleration can also be scaled back by the
same factor. It does make a world of difference in your test file, though
for some reason, my system is less sensitive to this than yours.
I will try to release by Friday this week with this fix installed. I'd
appreciate any feedback from you after testing this new algorithms output..
Thanks for the bug,
Art
www.artofcnc.ca
