Results 1 to 4 of 4

Thread: machining pro-tip: spindle error analysis vs runout

  1. #1

    machining pro-tip: spindle error analysis vs runout

    when building and/or maintaining machine tools for high accuracy applications, spindle error analysis is critical to both machine acceptance, and machine tool evaluation and diagnosis.

    most commonly, machinists and engineers will use the term "runout" and measure it as a method to evaluate there machines accuracy. this is incorrect. spindles do not have runout, spindles have error motion. and when measuring a surfaces run out, that is not necessarily linked to your spindles performance. excessive runout can be caused by several issues, mostly under the umbrella "chucking" but there can be other reasons as well. if you are measuring excessive runout on parts, look for chucking issues first (including alignment), then balance (for once per rev runout) and over-constraint issues (for multi-lobe runouts). hint: most of the ways traditional machines use to gain stiffness, is actually kinematic over-constraint, which is not good design, and will ultimately limit the machines accuracy. also pro-tip, using the same axis you cut the shape with, to measure the shape will always yield bad meteorology.

    to diagnose a bad spindle, or accept in a new spindle, or simply test the manufacturers claims, you will need to do spindle error analysis. one could try reversal as a method to do SEA, and it could work on a handful of micron level accuracy, but anything below that and reversal will be confounded by transient temp (spindles make great thermometers), backlash (rolling element bearings), and sensor stability.

    so a multi-sensor SEA is what is required.

    basically, by using multiple sensors, aligned at known angular locations, we can remove surface profile (runout) and look at the fundamental physics of the spindle itself.

    an example setup would be here:



    in the real:



    both of these rigs also include a sensor for thermal growth measurement, and the top one has a second set for XY measurement to include a tilt or 3d SEA setup.

    now, we spin our spindle, record our data ... now, what does that mean?

    well we have two sensors at known angles (most typically 90 degrees, but 90 is not required, its actually not even desired, but thats a whole different science lesson) and two sets of data from that. so phase shift one set to be the same phase as the other. then, you will see the surfaces runout, as confirmed by both sensors. the runout of the artifact will be easy to subtract out. if its a nice ground artifact, it will probably have a single lobe sine wave which can easily be subtracted out. it may have several artifact marks, these can also be easily subtracted out.

    now that we have done that, we can graph the actual spindle error motion, which looks like this:



    the nominal diameter of this circle, is the spindles synchronous error motion. this is motion that repeats with each revolution mostly predictably and reliably.

    the "thickness" of the circle will be the asynchronous error motion. this is the error motion that does not repeat with spindle revolutions.

    synchronous error motion can sometimes be compensated for, asynch, your are fucked.

    you can also now take your data and do and FFT on it, and knowing your spindle RPM, you can put that FFT in units of per rev, or sometimes called uppers:



    now you really get to see the meat and potatoes of your spindle. on rolling element bearings here is where you will see the ball count, the cage rotating, and the natural frequencies of both. on an air bearing or hydro static bearing, you will see your motor poles, windings, power supply stability or your air turbine blade count, things like this.

    there are several 3 probe ways to do this even more accurately, and using that two ball setup, you can do it 3d with a tilt component (esp useful for milling machines).

    using this information you can easily not only understand your spindle better, and diagnosis bad spindles. you can also improve spindle performance, and in some cases get "free accuracy" and yes, free accuracy does exist. you just have to be smart enough to find it. i wont show all my cards though


    maybe next pro-tip we will learn how you can use a 2x4 to compensate a linear axis. :evil:
    social conservatism: the mortal fear that someone, somewhere, might be having fun.

  2. #2
    Insider
    Join Date
    Apr 2012
    Location
    Knoxville, TN
    Posts
    2,303
    I have set up some Bently-Nevada systems on turbines that do the same thing. You can tell which turbine blade had a ding or a fatigue crack in it months before failure.

    Really fantastic tech. Applying it to a Machining tool like that is really sharp.
    Josh Coray
    J4 Paintball
    Lead Design
    www.j4paintball.com

  3. #3
    Quote Originally Posted by pbjosh View Post
    I have set up some Bently-Nevada systems on turbines that do the same thing. You can tell which turbine blade had a ding or a fatigue crack in it months before failure.

    Really fantastic tech. Applying it to a Machining tool like that is really sharp.
    FFT all the things
    social conservatism: the mortal fear that someone, somewhere, might be having fun.

  4. #4
    Insider
    Join Date
    May 2012
    Posts
    1,182
    i enjoyed this thanks for posting
    "So you've done this before?"
    "Oh, hell no. But I think it's gonna work."

Posting Permissions

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