re: backlash comp wasRe: EMC and lead screw compensation

Bill wrote:

>One way to avoid this problem would be to generate full arcs in 4
>quadrants so the control will read a new block at each compass point
>and compensate for backlash before executing the next quadrant. This
>method works very well on the Ahha software, I do not know about the
>other software packages discussed in this forum.

Most of the packages we discuss here have B/L comp. Problem is, in
anything but slow linear moves, the backlash comp often creates more
problems than it solves.

In arcs, by definition, backlash comp is done at quadrant boundaries.
This means that at least one axis has to completely STOP while the
backlash comp. is applied. Meaning cutter dwell and "witness" marks.

Often part finish is better with B/L comp turned OFF. If your
tolerances can handle it. The inertia of the cutting (especially with
a CVV (constant vector velocity) control will many times do more to
keep the part "in spec" than B/L comp.

There are some things programmers writing controls can do to improve
the "real world" functionality of backlash comp, but it is usually
implemented by "math" instead of logic. :-)

>While this is not a replacement for quality ballscrews and a system
>with ZERO backlash, it is better than having to live with no backlash
>compensation for arc moves in a low cost system.

ALL systems have backlash! Even "zero-backlash". Just to varying
degrees. :-) Physics. (Yeah, I know I'm nit-picking here, and it's
NOT aimed at Bill, who I basically agree with)

Hope this helps.

Ballendo