Re: [CAD_CAM_EDM_DRO] Closed-loop steppers, was: breakout boards plus - hard choice

The Newall system scales are available in a 5 um resolution and would
probably work for this application (they're not optical scales, they're
based on capacitance I think). Also, I believe that the Acu Rite MillPWR,
their turnkey cnc system, uses servos with rotary encoders for gross
positioning but relies on feedback from their glass scales for closed-loop
feedback, as well as the 3-axis DRO component of the system.

On 7/31/07, Jon Elson <elson@...> wrote:
>
> scyvt wrote:
> > Jon
> >
> > You seem to be assuming rotary encoders mountes directly on the stepper
> shaft. That still
> > makes assumptions about the linearity of the mechanical drive
> components. What if a
> > timing belt breaks? Seems more like "partially-closed-loop" to me. What
> about feedback
> > from linear encoders reading actual position? More costly, of course.
> The problem with linear encoders is that the typical resolutions
> are 20 um and 10 um. Those are .000787 and .000394"
> respectively. That is pretty coarse resolution for CNC control.
> My Bridgeport retrofit with 5 TPI ballscrews and 1000
> cycle/rev encoders directly on the leadscrew has a resolution of
> 40,000 counts/inch, or .000025" (25 micro-inches). This is 31
> times the resolution of a 20 um linear scale!
>
> I am using only 3/8" XL series timing belts on the Bridgeport,
> and I haven't broken a timing belt in 9 years of use. And, I
> dont't use steppers, only servos.
>
> But, if you wanted to use steppers with linear scales, I'm sure
> it could be made to work. A good 31" linear glass scale for a
> Brisgeport X axis would cost at least $350. You might run into
> the problem where the motor has more steps than the encoder has
> counts, and that leads to some problems in how you manage that
> discrepancy.
>
> Jon
>
>


[Non-text portions of this message have been removed]