hexagon logo

Midplane for Datums

Hey all, I have a couple questions about how PC-DMIS handles certain datum features.  I have a drawing where datum B is an external width dimension, meaning the actual datum feature would be the midplane derived from each side of the part.  There is a profile callout for each side of the part from which the midplane is derived.  I’ve been in this scenario many times and didn’t really think anything of it until out of genuine curiosity an engineer came up to me and asked me how it works.  His question was if the midplane is derived from the features that are being inspected to that same midplane (the sides I just mentioned), why wouldn’t the software just split the difference and have the profile results be equal on both sides?  This had me stumped.  I know that if we were to check this part manually, we would put it in a vise and use the jaws as the surfaces for datum simulation to get the mid plane, and zero our indicator out at basic to check the profile of each surface.  PC-DMIS is doing it right because it’s clearly not splitting that difference.  Basically my questions are:

  1. Upon defining that midplane as a datum, is it deriving that location based on the highest points of the surfaces as if the part was in a vise?
  2. In the 2009 and 2018 standard it states that the midplane is derived from two parallel planes at minimum separation. I see in the edit window for each surface that I can choose between LSQ and Min Sep.  Should I be choosing Min Sep?  And what exact difference will it make?

Any insight is appreciated.

Parents
  • You should use a 3D width instead of a mid-plane since it is defined as a width on your drawing.

    A mid-plane will just be the average of the two input plane centroid's and vector directions.  It contains no hits and is treated as a pre-resolved feature when referenced in the geometric tolerance command, essentially over-riding any datum fitting.

    A constructed BFRE 3D-wdith DOES contain all of the hits from the planes used to construct it.  It's centroid and axis represent the "mid-plane" but it functions exactly like your manual inspection analogy.  We expand / contract (depending on whether it's internal or external) two planes of perfect form until they make contact with the high points.  To your question about Min Sep / LSQ, if it's a 3D-Width and you're using the geometric tolerance command then it doesn't matter.  As long as you have DEFAULT DATUM MATH selected in the geometric tolerance command, it will take the hits and re-calculate the datum based on the specified math for whichever standard you are working to (ISO / ASME).

  • Thanks, Neil.  Using the width definitely made a difference.  Just for general knowledge regarding min sep, could you expound?  I read what's on the hexagon site in the best fit plane section but I'm having trouble understanding what it truly means.

  • I'd need to read through the standard (I'm assuming you're referring to ASME) but I think there's a little confusion over the language used and the name of the best-fit type in PC-DMIS.  MIN_SEP in PC-DMIS is what's also sometimes referred to as a MAX/MIN or Chebyshev fit which results in a feature who's max and min deviations are exactly equal.  When the standard refers to "two planes at minimum separation", that is not the same as using the MIN_SEP fit for each plane in PC-DMIS.  What they mean is that you are meant to optimize the width until it is no longer possible to move the planes any closer together - this is simulating clamping the part in a vice to use your earlier analogy.  Under ASME, all datum fitting must use a constrained L2 fit (see the Y14.5.1 2019 math standard which apples to Y14.5 2009 & 2018).  That is what DEFAULT datum math is in the geometric tolerance command when you have the ASME Y14.5 2009/2018 standard selected..

  • Is it possible to use (change to) the same lingo in PC-DMIS as in the standards? Changing the MIN_SEP to Chebyshev/MAXMIN in PC-DMIS?

Reply Children
No Data