1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Needed Devlopments in Constraints

Discussion in 'Using Alibre Design' started by Lew_Merrick, Mar 14, 2018.

  1. Lew_Merrick

    Lew_Merrick Alibre Super User

    There are a couple of discussions going on that revolve around needed improvement in the system of Constraints that Alibre needs. Let's channel them into one "top[ic."

    Attached Files:

  2. JST

    JST Alibre Super User

    I will add one that I call an "approximate" constraint.

    The "approximate" constraint I would add is one allowing the constrained feature to be within some error band of the perfect alignment. Probably that means within a circular area centered on the perfect alignment.

    I suspect that this is really just a variety of the "align" constraint, and likely relates mostly to things like pins in holes, or hole-to-hole alignment. It has little relation to "mate", for instance.

    This mimics, for instance, the alignment of structural steel, where holes are typically made 1/16" oversize above the bolt size, for small bolts up to maybe 3/4". The hole can be out of position by half that in any direction and still assemble OK, since the bolt will still pass through the slightly misaligned holes.

    If it were easier to do a rectangular area, I would not necessarily hate that, but it would be less helpful for typical things involving round holes, which is a very common place for this to come up, at least in things I do.

    Maybe it would help if I showed an example of a typical structure......... The grating is shown in partly transparent form so that the underlying beams and angle clips are visible. Having all that accurate to six decimals is possible in CAD but not so easy if things are not all right angles.

    This has no sloped cover, or other item at an angle that makes it more difficult. Naturally, with a lot of X-bracing it simply does not come out perfect in the real world, and is rather difficult to make come out perfect in CAD. You may have to cheat a bit.
    ~base and platform.jpg
  3. bigseb

    bigseb Alibre Super User

    Why cheat in CAD? o_O We can draw things bang on. I get that the manufacturing process requires tolerances but the CAD itself doesn't need that.
  4. Thompson

    Thompson Member

    You absolutely have to cheat in CAD sometimes. I have occasion to design structures that have diagonal cross bracing that is secured with bolts at each end. There is absolutely no way that concentric mates will work at both ends of the brace. I'll use a concentric mate at one end and if the holes are very, very close, a tangent mate at the other. If they are too far apart and the holes in the brace are on its centerline, I'll use a 'width' mate (in Solidworks) that acts like a pin in a slot.

    If the holes in the cross brace were added by in-context editing, the brace would not be able to be used in several different assemblies easily. The dimensions will change a small amount and the mate will fail, even though it will work fine in real life.

    One of the things I miss most in Alibre is its lack of a symmetric 'width' mate in the assembly work space. I remember when I first started using SW, I thought "Interesting, but I'll never use that." I turns out that it's an incredibly useful and hassle reducing feature. I'm really surprised how much I use it. It's all about design intent. No more using a calculator to move hydraulics back to the center when a weldment is widened. And for example, it makes a 6 section telescoping rod work perfectly with zero effort.
  5. JST

    JST Alibre Super User

    We are wandering off

    The basic reason is that the program distance resolution does not let the actual measured-in-CAD distance between holes fit when that distance is entered as the distance between holes on the part it fits.

    Typically when the holes to be fitted are located at a diagonal, but their position is set by cartesian co-ordinates, as in lengths etc on parts at a right angle. Usually large parts, distances in metres. Then I measure the diagonal distance between and set that on the mating part..... and it does not allow an "align" at both ends.

    I see that Thompson has had that situation as well.

    CAD is "perfect", but only to its resolution.

    Apparently Alibre measures to 6 places, but checks alignments to 12 places (or whatever).

    I have seen equal measurements not align.

    And I have seen measurements that were set to XX.000000 later end up measuring as XX.000013. Usually it IS something like 0.000013 that is added, no idea why. Some oddity in the round-off routine, presumably.

    Let's not wander off too far.

    But tell us more about the "width" mate, that is relevant here. I do not recall that from using Solidworks.
    Last edited: Mar 15, 2018
  6. VoltsAndBolts

    VoltsAndBolts Senior Member

    Width Mates
    A width mate constrains a tab between two planar faces.


    Width references can include:
    • Two parallel planar faces
    • Two non-parallel planar faces (with or without draft)
    Tab references can include:
    • Two parallel planar faces
    • Two non-parallel planar faces (with or without draft)
    • One cylindrical face or axis
    To add a width mate:

    1. Click Mate [​IMG] (Assembly toolbar) or Insert > Mate.
    2. Under Advanced Mates, click Width [​IMG].
    3. Under Mate Selections:
      1. Select two planar faces for Width selections.
      2. Select two planar faces, or one cylindrical face or axis, for Tab selections.
    4. Under Advanced Mates, select one of the following constraints:
      Option Description
      Centers a tab within the width of a groove.
      Free Lets the components move freely within the limits of the selected faces or planes with respect to the components.
      Dimension Sets a distance or angle dimension from one selection set to the closest opposing selection set of faces or planes.
      Percent Sets the distance or angle based on a percentage value dimension from one set of the selection set to the center of the other selection set.
    5. Click [​IMG].
      The components align so that the tab is mated between the faces of the groove. The tab can translate along the center plane of the groove and rotate about an axis normal to the center plane. The width mate prevents the tab from translating or rotating side to side.
    bigseb likes this.
  7. JST

    JST Alibre Super User

    I like it.
    I want it.
    I gotta have it......

    I'd say add it to the list, but I think it is a subset of what has already been posted as requests.

    BTW.... in 2 dimensions, that could be similar to the "approximate constraint", if not constrained strictly to the centerline. Basically like the "Free" option, but in two dimensions.
    Last edited: Mar 15, 2018
  8. bigseb

    bigseb Alibre Super User

    Am I missing soemthing here? Why not just draw it correctly?
  9. bigseb

    bigseb Alibre Super User

    +1000 for this!
  10. Lew_Merrick

    Lew_Merrick Alibre Super User

    VoltsAndBolts (and others) -- If you set your Symmetry to a Reference Plane and use Align (or Mate) on such Reference Planes, you get the Symmetrical Relationship you are looking for. You just need to understand how to properly proclaim a Symmetric Relationship among Parts.
  11. Thompson

    Thompson Member

    Yes, you are missing something. Imagine two tubes each a little more than 10 meters long with a hole 1 cm from each end - holes are exactly 10 meters apart. Assemble them at right angles by thru bolting. The diagonal brace connecting the free ends will need to have its holes
    14.142135623730950488016887242096980785696718753769480731766797379907324784{more...} meters apart or the other tubes will not be held at right angles. If I place the holes at a reasonably accurate 1,414.21 cm, the mate won't work.

    If I have another structure with tube lengths of 12m and 18.55m, then the same diagonal tube (which must fit both structures without modification) will fit fine in practice, but the mate will fail even if it _was_ perfect in the first example. This situation occurs pretty frequently for me.
    JST likes this.
  12. Thompson

    Thompson Member

    Please let me fix my math error: In the second example the tubes would be 12m and 7.48m. Sheesh!
  13. bigseb

    bigseb Alibre Super User

    Just tried this and you're right.

    Well played, sir.
  14. oldfox

    oldfox Alibre Super User

    The 7.48 number is not resolute enough. You can however find out exactly (almost) the number it should be if you do the math.
    And since the sqrt(2) isn't a good ending number, then I doubt you would ever be able to get the right dimension for CAD work.
    It will, however, be close enough for that iron worker out on the floor drilling the holes. They will fit in the field. Especially if the holes are "sloppy" like JST alluded to.

    If you look at the assembly really zoomed in, check out the hole at the 7.48 point on the vertical member. It is visibly not exact.

    Attached Files:

  15. JST

    JST Alibre Super User

    As you found out, Thompson and I are not just blowing smoke.

    I just had one of those today...... bit not quite the same.

    I found out what it was..... The problem in THIS case was that I had set a distance to 4.000000 mm. When I checked the dimension, I got 3.999999 mm. And a 5.000000mm dimension actually showed up as 5.000001 mm.

    NO I do not know what causes this. I assume rounding errors that ended up coming into the displayed (and, I assume, used) decimal places.

    I know for a fact that I set both those to exact mm settings, and Alibre managed to add a bit, or subtract a bit in the process. I am told that Alibre actually uses something like 12 or 14 places internally, so this should be rare, if it ever happens. But I have run into several like this in the last few weeks.

    The issue that Thompson and I are mentioning, is like this, but is evidently happening in hidden areas. Either way, the parts end up not mating.
  16. TimoCAD

    TimoCAD Senior Member

    NX has it also, named 2 to 2, 2 to 1, 1 to 2. By the means these are the faces which get a midface constraint.
  17. bigseb

    bigseb Alibre Super User

    iirc it's 19 places. But you're right, this is a problem. Not something that comes up in my work so I don't pick up on it. And of course there are ways aroind it but a new constraint for this couldn't hurt.
  18. sz0k30

    sz0k30 Senior Member

    Exactly! Anytime you work with right angle VS diagonal dimensions.
    JST likes this.
  19. JST

    JST Alibre Super User

    Now I am wondering just what Alibre uses as a cutoff for checking alignment. It must be past 6 places, since I believe I have seen cases where the dimensions match to all 6 displayed places, and yet Alibre throws an error.

    As long as we are talking constraints, this is an important point. Alibre should NOT complain about any constraint where the lineup is good to all the visible decimal places. How in the world can you correct what you cannot see? If the distances match to the 6 displayed places, then the parts should work.

    And I have seen that dimensional change issue pop up into the 5th place, not just the 6th.

    I think you are right with the 19 places, I seem to recall that now that you mention it. It's probably in the specs somewhere. That allows for several multiplications before the roundoff gets up to the visible area, so it makes sense.

    We've gone and done it again..... Lew started a nice clean thread, and we have "somewhat related" discussion stomping all over it.
    Last edited: Mar 16, 2018
  20. simonb65

    simonb65 Alibre Super User

    Maybe the answer is that constraints just need a 'proximity' tolerance of a user defined amount (like healing)! I know it then becomes a not-so-exact science, but then again, real parts have tolerances (holes, fasteners, manufacturing, etc), but they always fit and align! ... or am I being too simplistic here?

    There will always be rounding and comparison errors with floating point numbers in computers ... it's just the way the number gets represented in binary form. The issue goes back to the dawn of computing and has never really gone away.
    oldfox likes this.

Share This Page