Re: [CAD_CAM_EDM_DRO] Re: what the world needs now ... backlash aware stepper motor drive r

Having backlash compensation may be ok for light duty (i.e. low side
force) CNC applications like plasma cutting. My experience with
backlash compensation in my Flashcut CNC system is that while you can
completely compensate (in theory) for the backlash in the acme screws,
it still results in poor machining surface finish.

Before I installed ballscrews I was never happy with the surface finish
I could achieve for any 3D machined surface. Backlash equals slop and
with moderate side cutting forces, the table will take up the "slop"
rather than the cutting tool remove material (path of least resistance)
resulting in a poor quality finish.

What about taking the flex in the tool bit into account next? What
about thermal compensation? All these compensations can be done and are
done in high end systems, but it just complicates ones life
unnecessarily in my opinion.

I know the previous post was talking about DRO feedback to compensate
for backlash which can be implemented but I could see control loop
feedback problems appearing as position hunting, resonance and other
instability issues that could be hard to solve for mere mortals.

For milling metals I am convinced that you need ball screws with zero
backlash to keep things tight and accurate or keep your cutting forces
very low and then you can use acme screws and backlash comp.

Cheers, Peter



beer@... wrote:

>>>Such a beast would have dip switches to indicate
>>>at setup time what the backlash was (in steps), and
>>>upon detecting any reversal of direction, quickly
>>>counts up through that selected number of steps
>>>to take up the backlash.
>>>
>
> Uhh .. doesn't the software in most systems do this already ?
>
> Anyway, some time ago, I built and <mostly> finished the programming for
> ( and hence never installed ) a small unit with a PIC and an LCD to make
> a one channel DRO.
>
> Besides the quadrature inputs, though, there was also step and direction
> inputs AND step and direction outputs. There was one last output, a
> "pause" output to the software (CNCPro).
>
> Each step and direction pulse IN was echoed to the step and direction
> pulse OUT.
>
> The step pulses were timed, and compared against the DRO.
>
> Should enough steps be issued that resulted in no DRO activity, the
> PAUSE output was asserted and additional steps issued until things
> "matched up" again.
>
> It seemed to me that for best operation, the DRO feedback must be at
> least as fine a resolution as the stepper system. If the steppers are
> 4,000 steps per inch, then a .00025"/count DRO would be required to
> more or less "correctly" compensate for the leadscrew.
>
> However, even at lower resolutions of the DRO, it would still be
> useful, there would simply be an uncorrectable error, roughly the size
> of a quadrature step.
>
> Seems that this would be a better anti-backlash system ...
>
> Alan
>
>
>
>
>