Machine hysteresis

  • Filter
  • Time
  • Show
Clear All
new posts

  • Machine hysteresis

    Can someone explain to me in layman's terms what machine hysteresis is and how it has an effect on what we do.

    Thank you

  • #2
    that's when the machine is no longer able to have babies.
    Originally posted by AndersI
    I've got one from September 2006 (bug ticket) which has finally been fixed in 2013.


    • Smallsvillanova
      Smallsvillanova commented
      Editing a comment
      So....I....should put her down then?

  • #3
    When I made load cells (the functional part of an electronic scale), we would do hysteresis studies on cells.

    Power it on with no load, or just enough load to get the gages engaged, and let it sit for a week. Monitor the output.

    Since it isn't receiving load, that output should be zero. Since we don't live in a computer simulation (unless it is a REALLY F'N GOOD one), nothing is perfect, and it will drift.

    I've never done this deep a dive on CMM's, as I never had a hyper-accurate one, but if I did, any drift to the scales, construction and electronics while it is sitting un-used, have an effect on the measurement it provides while in use.

    My MSA data gets me using up about .0002" on (True) Position tolerances from machine and operator variation on our real world product line.

    I did do this testing with one person measuring only morning, one only mid-day, and one only end of shift to try and capture anything that might include.

    Beyond that, isolating hysteresis isn't that big a deal to me, as .0002" error on Positions, is roughly 70 millionths of an inch per axis, and both of those numbers are tiny fractions of the tolerances supplied by my customers.

    Even when I get a position of .001", if I give the report only .0008" I know the part is good with the uncertainty I exhibit in MSA. I don't need to look at individual causes of that uncertainty to eliminate it.

    If you are running hyper-accurate machines, or you are being forced to try for hyper-accurate results on a standard machine because those quoting your work don't care about your misery with the tools you have available to check the tolerances they get you work on, you might have to care about this.

    It could also be an engineer over-thinking the situation; or you have no MSA's so they are curious how much you know to gage a guess at your uncertainty in measurement.


    • louisd
      louisd commented
      Editing a comment
      0.0008" is an assumption that you have a solid confidence interval. if your MSA was limited in scope, your CI range will show additional uncertainty in your measurement method. To say TP measured at 0.0008 on your machine is high-limit acceptable is a serious stretch gamble.

    • Caemgen
      Caemgen commented
      Editing a comment
      That assumption is based on our standard product line.

      It is not, as you point out, an across the board statement of fact.

      Specific configuration of design would make such an assertion impossible to defend.

      As I said, this is on "our real world product line." Not on your product line, not on his, not on the product line we will possibly be making 10 years from now.

      Not on one specific part family we just took up, where I have an expected error of .0006" with a .0005" tolerance zone.

    • louisd
      louisd commented
      Editing a comment
      all i'm saying, is if you are conducting an MSA, take a look at confidence intervals. in minitab it's literally an extra box to check within the anova Gage R&R tool.

Related Topics