. .
  1. #1

    Is there any theoretical reason or practical limitation to stop you using DROs to provide position feedback to a CNC controller? Even better if they're absolute DROs so that the DRO can provide limit handling and there's no need for homing?

    It seems to me that this technique would mostly remove any problems with backlash and missed steps - whereas 'closed-loop' steppers and servos address the missed steps problem but not backlash.

    Of course it would require a controller build to accept DRO input for feedback. Does anything like that exist already? Am I missing something here?


  2. #2
    Mud's Avatar
    Lives in Bristol, United Kingdom. Last Activity: 11-12-2020 Has been a member for 2-3 years. Has a total post count of 23. Received thanks 3 times, giving thanks to others 0 times. Referred 1 members to the community.
    Yes, use of linear encoders is possible - see for example LinuxCNC.

  3. #3
    I'm doing that right now on my lathe conversion. As Mud says, LinuxCNC and some other controllers can do it (excluding my Centroid Acorn and Newker 990 of course).

    The signals from linear encoders are generally like rotary ones ie quadrature A and B with a Z index, either single ended or differential.

    But I detect that old chestnut again here. You still need to get as much backlash out of your system as you can sensibly manage. PID controllers really don't like sudden nonlinear behaviour. I use the analogy or driving an old banger along a winding hilly track with half a turn of slop in the steering wheel. You (the closed loop controller) have to work bloody hard to keep on track. There's no magical solution that allows you to produce precision parts on an old dog.

    Same applies to backlash compensation in your controller - in some situations it can actually double the backlash error. And if you are missing steps, you need to find out why and fix it.

  4. #4
    JAZZCNC's Avatar
    Lives in wakefield, United Kingdom. Last Activity: 11 Hours Ago Forum Superstar, has done so much to help others, they deserve a medal. Has been a member for 9-10 years. Has a total post count of 7,785. Received thanks 1,338 times, giving thanks to others 87 times.
    Quote Originally Posted by Muzzer View Post
    I use the analogy or driving an old banger along a winding hilly track with half a turn of slop in the steering wheel. You (the closed loop controller) have to work bloody hard to keep on track.
    Good Analogy and reminds me of an old Transit van I used to have with a steering box, used to have to move the steering wheel +/-20 degrees either way from the center just to keep it in a straight line...

    Often better machines have both encoders on the servos/drives and linear scales on the axis that feedback to the controller which closes the loop.
    Many moons ago I nearly bought a very high-end 5 axis Mill made by a company called "JOBS" that was built for McLaren F1 team which cost 250,000 and that was 25yrs ago.
    This had Mitsubishi servos with absolute encoders and Glass linear scales on the Axis that both went back to the controller, the thing was unbelievably accurate but the complexity and setup was a nightmare apparently.

    To me, if you are going to this level of complexity then the WHOLE machine needs to match, because like most things in life can only be as good as the weakest link.
    -use common sense, if you lack it, there is no software to help that.

  5. #5
    Yeah, I wan't thinking about tarting up a dodgy machine tool so much as a way, using absolute encoders, to avoid homing and get 'limit switches' which I don't have now. Also I like the idea of controlling for the 'actual' position. I acknowledge that nothing's perfect and that the encoders have measurement errors and the mill will have some bendiness / alignment errors. I like the concept better though than relying on the stepper/servo's idea of where the position is plus a fixed value per axis of 'backlash'.

    My mill has no missed steps and about 40 microns backlash on X and 80 on Y and Z - although I plan to measure those more accurately soon.


  6. #6
    40 microns is pretty decent - about 1.5 thou. I think I might just about be able to get to that accuracy if I took care. My Japanese machine was originally claimed to be good for 10um when new but that would be with the gibs all spot on, spanking new ballscrews, following wind etc. Obviously that wouldn't be good enough for the willy waving Pootube warriors who claim to be able to "hold tenths". Yes and I'm Mary Poppins.

    For a real sickener, have a look at this video where a mate of mine describes what precision actually is. 1 nanometer is a thousandth of a micron ie a good 4 orders of magnitude better than I could hope for.

  7. The Following User Says Thank You to Muzzer For This Useful Post:

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Similar Threads

  1. Control box
    By Kev2960 in forum Gantry/Router Machines & Building
    Replies: 9
    Last Post: 03-02-2017, 11:14 PM
  2. Fitting DROs to SPG 2217 Mill
    By g6uwi in forum Milling Machines, Builds & Conversions
    Replies: 2
    Last Post: 05-03-2016, 08:59 PM
  3. DC motor control
    By phill05 in forum Lathes, Lathe Rebuilding & Conversions
    Replies: 3
    Last Post: 24-12-2014, 04:10 PM
  4. servo control vs stepper control
    By Peter. in forum Motor Drivers & Controllers
    Replies: 24
    Last Post: 17-09-2012, 08:36 AM
  5. Help please Steps and control?
    By totts in forum Motor Drivers & Controllers
    Replies: 9
    Last Post: 31-03-2012, 07:44 PM


Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts