hexagon logo
Register Sign in
  • Replies 22 replies
  • Subscribers 221 subscribers
  • Views 218 views
  • Users 0 members are here
Options
Related

Does Translating The Trihedron Ruin "True Zero"?

Bearpierro
Hello All!

I'm having a debate with a fellow programmer today on the effects of moving the trihedron away from the CAD's zero point.

This is in reference to a True Position measurement of a hole located at X0 Y14.5.


So there are two different methods to get our alignments and then dimension the true position:

Method 1
"Locate" The Part, by tapping a point on X, Translating the trihedron to that point, tapping a point on Y, Translating the trihedron to that point, etc for Z.

This obviously moves the trihedron around a bunch, and because of that physical features need to be used on the part to "return" the trihedron back to X0, Y0, Z0 on the drawing for the true position measurement to be taken.

Method 2
Leave the trihedron at X0, Y0, Z0 of the CAD model and NEVER translate it. Instead, tap points around the part and align without ever moving the trihedron.


Now, once dimensioned, both of these methods give me the same Nominal Measurement Values, X0, Y14.5.

Both of these methods use the same features to create their Iterative and Best Fit alignments.

Is there any difference in the end?

Please let me know if that wasn't clear enough, and thanks in advance for any help!
Parents
  • Technically, as written, the Y14.5 standard is written so that you align to the datum structure in your Feature Control Frame (lets say you have POS | Ø .014 MMC | A | B | C |).
    You constrain all degrees of freedom that A will constrain (lets say it is a plane, that is two rotations and a translation).
    If you are measuring two dimensional features (not points for itterative but circles or the like) you need to align that plane now prior to measuring the circles.
    Then you measure B, lets say it is a plane also, so it constrains the final of three rotations and another translation.
    This leaves one translation for C, it can be whatever you want. Doesn't matter.

    You now have zero zero zero. Name it ABC so you remember what it is and can recall it whenever you want.

    Now, you translate your origin (from your example) zero on X and 14.5 on Y.

    This moves your trihedron.

    You then inspect the actual position of your feature compared to this "TRUE" position you translated to.

    This doesn't help a machinist move the feature to fix it, but this is what you are to do, technically.

    Then you recall ABC and continue measuring.

    You do NOT re-measure the features of A B C and redo the alignment.

    Basic dimensions are just that, basic moves from zero zero zero to get to the feature. That is why they can be added and subtracted.

    Part says this hole has a .010 position to A B C, but the basic is from some random hole in a bolt pattern? YOU add/subtract as necessary to measure the hole compared to true position FROM A B C. The distance from that other hole means nothing. Only the distances and angles from A B C.

    It concerns me that you say you can't get back to zero zero zero.

    I measure 1500 dimensions on a manifold, zero A B C and recall it, bouncing all over the place for compound angle holes, and I get back to zero every single time.

    I'm more worried that your part is so far from perfect that it doesn't match the model enough to reconcile on the first alignment anyway (which might explain you doing an iterative to begin with), of you are re-measuring the datums but not using the same features/points as the first time. maybe?

    If you don't have angle holes, you don't have to move the alignment to check the hole.
    If you use ExactMeasure or GeoTol you don't have to move the alignment to check the hole.
    In both cases, though, the software IS moving the alignment to check the hole in the background and you don't see it.

    Make a ExactMeasure or GeoTol position and then click into it. You'll see a yellow trihedron pop on the screen.
    That is the alignment the software is using to measure the hole's position against true position.

    If you can't move it yourself and get back to where you were, you need to look at your alignments and see where you went astray.

    I said I do it on large, complex manifolds.
    I didn't say I got it perfectly right the first time, every time.
    It is very easy to hit the enter key before you clicked on the button to apply something in an alignment, or do an angle in the wrong order with a translation.
    If you are having a repetition problem, it's almost always in the alignment.

    I think there's a miscommunication here. I believe what I'm doing is correct, using features to create alignments should give me the proper XYZ coordinates to the drawing. What my coworker is telling me is that the XYZ origin (the original trihedron position set in the CAD) should be set when the CAD is created and not be moved. He claims that by moving it I am never going to have it accurately placed to the drawing. Is that right?

    I appreciate you going to this length by the way. It's a lot of info to take in.
Reply
  • Technically, as written, the Y14.5 standard is written so that you align to the datum structure in your Feature Control Frame (lets say you have POS | Ø .014 MMC | A | B | C |).
    You constrain all degrees of freedom that A will constrain (lets say it is a plane, that is two rotations and a translation).
    If you are measuring two dimensional features (not points for itterative but circles or the like) you need to align that plane now prior to measuring the circles.
    Then you measure B, lets say it is a plane also, so it constrains the final of three rotations and another translation.
    This leaves one translation for C, it can be whatever you want. Doesn't matter.

    You now have zero zero zero. Name it ABC so you remember what it is and can recall it whenever you want.

    Now, you translate your origin (from your example) zero on X and 14.5 on Y.

    This moves your trihedron.

    You then inspect the actual position of your feature compared to this "TRUE" position you translated to.

    This doesn't help a machinist move the feature to fix it, but this is what you are to do, technically.

    Then you recall ABC and continue measuring.

    You do NOT re-measure the features of A B C and redo the alignment.

    Basic dimensions are just that, basic moves from zero zero zero to get to the feature. That is why they can be added and subtracted.

    Part says this hole has a .010 position to A B C, but the basic is from some random hole in a bolt pattern? YOU add/subtract as necessary to measure the hole compared to true position FROM A B C. The distance from that other hole means nothing. Only the distances and angles from A B C.

    It concerns me that you say you can't get back to zero zero zero.

    I measure 1500 dimensions on a manifold, zero A B C and recall it, bouncing all over the place for compound angle holes, and I get back to zero every single time.

    I'm more worried that your part is so far from perfect that it doesn't match the model enough to reconcile on the first alignment anyway (which might explain you doing an iterative to begin with), of you are re-measuring the datums but not using the same features/points as the first time. maybe?

    If you don't have angle holes, you don't have to move the alignment to check the hole.
    If you use ExactMeasure or GeoTol you don't have to move the alignment to check the hole.
    In both cases, though, the software IS moving the alignment to check the hole in the background and you don't see it.

    Make a ExactMeasure or GeoTol position and then click into it. You'll see a yellow trihedron pop on the screen.
    That is the alignment the software is using to measure the hole's position against true position.

    If you can't move it yourself and get back to where you were, you need to look at your alignments and see where you went astray.

    I said I do it on large, complex manifolds.
    I didn't say I got it perfectly right the first time, every time.
    It is very easy to hit the enter key before you clicked on the button to apply something in an alignment, or do an angle in the wrong order with a translation.
    If you are having a repetition problem, it's almost always in the alignment.

    I think there's a miscommunication here. I believe what I'm doing is correct, using features to create alignments should give me the proper XYZ coordinates to the drawing. What my coworker is telling me is that the XYZ origin (the original trihedron position set in the CAD) should be set when the CAD is created and not be moved. He claims that by moving it I am never going to have it accurately placed to the drawing. Is that right?

    I appreciate you going to this length by the way. It's a lot of info to take in.
Children
  • in reply to Bearpierro
    Unless we are inspecting in plane or car locations, the model is often not in the same place in a CMM or on installation as when it was drawn.

    And, so far as I know, machines aren't capable of machining in those coordinates, they zero on a vice, or a tooling ball, or the corner of a block of metal. The CNC machine never (that I've ever witnessed) picks up G54 at the part's A-B-C zero.

    When the model is drawn, it isn't drawn in A-B-C zero coordinates. It is drawn off planes that are created all over the place, often incrementally, and the solid is added to and cut away.
    That is then mated into an assembly and run through some analysis (for structural stuff at least).
    THEN it is isolated back out and a drawing is made. That drawing probably has very limited tie-in to the way it was drawn.

    We draw perfect.
    The print is to let us make it imperfect while still working and controlling cost.

    You can import the model, take hits in probe mode, make the alignment and then run that, which will spin the part to match the way it is sitting in the machine
    -or-
    you can import the model, translate it so it is like in the machine and move the zero where you want it to make it easier on the guy to load the part so it looks like the screen
    -or-
    you can manipulate the part in some cad system and export it with those coordinates and leave it like that.

    Does not matter in any way, shape or form.

    Closest you get to an issue is with translating the model from its native CAD creation, and that is usually very minimal, other than the odd fillet radius trimming backwards.

    If I now understand what you are asking. Nothing you do with translation has any effect on the model's coordinates system for any features in the model.
© 2024 Hexagon AB and/or its subsidiaries. Privacy Policy | Cloud Services Agreement