Re: Re: EMC and lead screw compensation

Wally

They say there are times when one should keep ones mouth shut. -- like
when swatting flies on the ceiling. This might be one of those times.

> From: "Wally K" <cnc4me@...>
> Subject: Re: EMC and lead screw compensation

<s>
3 more questions.

>
> 1st do you or anyone else know what a reasonable measerment step
> would be for the leadscrew compensation file. The easy awnser is the
> smaller the step the better. But i have a feeling that at a certain
> point you get diminishing returns.

I was not able to find much about the size of steps but my guess is that
you can put as many as you want here. The worst case would cost some
computation time during the motion loop.

> 2nd, is it ok to have different step sizes in the file. Example, you
> measure every inch of the leadscrew and find a spot were it has
> greater error than the rest of the screw. Can you enter in
> measerments closer together in this region of the screw.

Again a guess. If they had specified a distance they would not have
included the first leadscrew comp column the way it is. Rather they would
have given it as first step, second, ...

What appears to happen when a comp file is encountered is that the
difference between any line and the next line is equally spread across the
distance of the step. This would mean that if you had a rolled screw 36
inches long that had a reasonably even expansion of .0003 per foot would
only require two lines in the comp file -- start and end. If your
leadscrew expanded during the first half and shrunk during the second half
you would need three lines.

I've worked with a few Mazak leadscrew comp files. These appear to be
equal interval and have few very small entries.

<S>

> 3rd question, isn't forward and reverse direction relative. How can
> we be sure we are putting the measured numbers in the correct
> direction column.

Relative absolutely. They are relative for the machine and the outputs of
the EMC to your motors, and they are relative to the return signals from
encoders. But they are absolute within the world model of the interpreter,
task, and motion programs themselves. I'm reasonable sure that this is
true of the leadscrew comp as well.

Emc doesn't care whether it's up, plus or minus, is really up only that a
commanded up shows a motion in the up direction of the returning position
signal. You predetermine which direction is up plus or minus by the way
you set several variables within the ini file that is read during a run.

I think that this has something to do with the round world theory
and the fact that there are machines running simultaneously in St Louis
and Hong Kong. ;)

I don't think that I'd count on leadscrew comp or backlash comp to take up
much of the slop of sloppy mechanics. As a number of others have pointed
out the better the mechanics of the machine the less effort you have to
put into trying to get the control to compensate.

Hope this helps but don't trust any of it.

Ray