Parent Chile Relations

Discussion in 'SolidWorks' started by umasemonoff, Nov 12, 2006.

  1. umasemonoff

    umasemonoff Guest

    Will someone who is knowledgeable on the subject of Parent Child
    Relations, give me a short synopsis on the actual function of it all.
    If the subject requires a lengthy explanation , and too involved for
    this forum, I will understand if there is little response. If the
    basic principle were brushed on, it would be more than appreciated.

    I am thanking you , in advance
    Uma Semonoff
     
    umasemonoff, Nov 12, 2006
    #1
  2. umasemonoff

    mjlombard Guest

    Parent child relations can seem a little misleading sometimes, the way
    things are represented in the FeatureManager. First, I'm assuming that
    you are talking about parent child relations inside a part, not between
    parts. It is easiest to understand if you look at it as you build a
    part rather than trying to look at a part that is already built and
    figure out the relationships.

    As you build a part, any time you build a new feature by using other
    geometry, you have created a parent /child relationship. The new
    feature is dependent on the pre-existing feature, so the new feature is
    the child. The only items in the FeatureManager which are not children
    of something else are the three basic planes and the origin. Generally
    speaking (how many disclaimers do I need to avoid getting flamed),
    everything else is dependent on the 3 planes and origin.

    The confusing part is that you usually think of Parents as coming
    before Children, but in SolidWorks, the way features and sketches are
    listed, the feature is a child of the sketch, but the feature is listed
    before the sketch. The FeatureManager is not a linear "family tree",
    rather it assumes that 3D features are more important than 2D sketches,
    so it makes access to the features easier by making them visible.

    You can investigate the parent/child relationships in a model by right
    clicking on a feature and selecting Parent/Child from the menu. This
    lists the parents and the children of the selected feature.

    Generally, if you delete or suppress a feature that other features are
    dependent on (the feature has children), there will be some effect on
    the children, such as they may get deleted, or the sketch may go
    dangling or things like that. Examples of things that cause parent
    child relations are:

    - using the face of a feature to sketch on
    - using convert entities on the edges of a feature
    - using an extrude "up to" end condition
    - dimensioning to the edge of a feature
    - dimensioning to the temporary axis created from a cylindrical face.
    - any time a face, edge, vertex, etc is used in the definition of a
    feature or sketch

    The good part of parent child relations is that when you change the
    parent, the child updates to match the original instructions with the
    new geometry.

    The bad part of parent child relations is that when you change the
    parent, the child updates to match the original instructions with the
    new geometry.

    Other levels of parent/child relations include topics like in-context
    references (relations between parts in an assembly), one part being
    inserted into or otherwise created from another part, also assembly
    documents are dependent on part documents, and drawing documents are
    dependent on assemblies and parts.

    It's not a simple topic, and not something that you can really grasp
    adequately without sitting down and using it a bit.There are many
    advantages and disadvantages of "parametric history based modeling",
    which cannot be approached carelessly. Everyone has a story about a
    model that they completely messed up when getting the hang of the
    concept.

    Many, maybe most people model things in a linear, sequential way such
    that there are many "generations" of dependencies, so like that book in
    the Bible where "...and abraham begat isaac and isaac begat jacob and
    jacob begat esau and esau begat..." that sort of "daisy chain" effect.
    This can have a bad effect on performance because all features must
    rebuild in sequence. To go with the family tree analogy, this makes for
    a very tall, skinny family tree, with few branches.

    The alternative, mentioned in a previous thread, is to build a
    structure based on a minimum number of sketches (two or three which are
    only dependent on the basic planes and origin) to represent locations
    and sizes of features, and a set of reference geometry (planes and
    axes) which are either dependent on the sketches or basic planes or
    both. This structure is referred to as the skeleton. Any sketch after
    the skeleton is sketched on one of these planes, and uses the sketches
    or basic planes for location and size references. In this scheme, you
    avoid making parent/child relations to faces, edges or vertices of
    solid geometry. Because everything is related directly back to the
    First Generation (basic planes and origin), Second Generation (skeleton
    sketches) or Third Generation (planes made from skeleton sketches), the
    "family tree" of this method is short and wide.

    Tall and Skinny or Short and Wide? The problem with a Tall and Skinny
    tree is that a problem in one feature is likely to cascade all the way
    to the end. In this model, errors propagate down the tree, (the tree
    being built from the top to the bottom) because there is a very high
    degree of interdependency. The Short and Wide method centralizes
    control in the skeleton features, and minimizes interdependency, so
    errors are contained more easily.In this method, solid features do not
    show up as the parent of any other feature. This can rebuild faster
    because the order of operations is not as important.

    Anyway, not many people model short and wide because it requires more
    discipline. I generally use a hybrid approach between the two methods,
    and usually wish I were more disciplined.

    matt
     
    mjlombard, Nov 12, 2006
    #2
  3. umasemonoff

    FlowerPot Guest

    Damn, another major blo-hard. And this one is a SW usage cop too! Must
    be nice to just sit in judgement all day saying thou shalt do this and
    thou shalt not do that. We've got enough preachers here to file for tax
    exempt status! Enough judges to make a Court TV show. Enough
    self-important blo-hards to go on CNN.

    Where do you guys dig up all these stuffed shirts?!?!

    Daisy.
     
    FlowerPot, Nov 12, 2006
    #3
  4. umasemonoff

    TOP Guest

    Parent / Child relationships in SW refers to the fact that each feature
    in the SW feature tree is based upon one or more pre-existing features.
    Not all of the features in the feature tree, however, provide geometry
    data. For example the very first feature in the feature tree of a part
    is the document itself which contains information of a global nature
    pertaining to the documnent. The second feature, Annotations pertains
    to the display of annotations, then there is material, lighting, etc.
    Then there is a second group of folders pertaining to the management of
    bodies, both solid and surface. These are lists that are affected by
    the current history/time state of the model and reflect the state of
    existing geometry on which subsequent geometry can be built. Finally we
    get to features that define the starting point for all subsequent
    geometry. There are four, three planes and an origin. Interestingly the
    origin is tied to the front or first plane, but the other two planes
    exist independently. These features which will always exist in a part
    or assembly provide the basis on which to relate all subsequent
    features. The preceding features are always there and cannot be deleted
    (except the document itself of course).

    When the first user created feature is made it must be related to the
    origin and one or more of the three planes. For example if a planar
    sketch is made it must itself lie on one of the front, top or right
    planes. The curves in the sketch must be related to the origin or to
    the other two planes by relations or to the dimension relation. This
    sketch might then provide the basis for an extruded feature or a
    revolved feature. If an extruded feature, then the direction of
    extrusion will be perpendicular to the plane that the sketch was made
    in. The sketch and the plane would then be parents of the extrusion. If
    the sketch were to be destroyed so would the extrusion because SW
    recalculates all the steps to the extrusion each time the model is
    regenerated. When the definition of the extrusion is gone, so is the
    extrusion. This is called history based modeling and is similar to the
    concept of putting a framework in a building. As we saw on 9/11 when
    the internal framework of a building is removed, so all that is based
    upon it will be removed pretty much in the order it was originally
    constructed. A parent then is a feature upon which features are built
    and a child is a feature depending on pre-existing features. The word
    time is used in this context to refer to a stage in the building
    process, i.e., you cannot have the extrusion prior to the sketch being
    made because the sketch is the parent of the extrusion feature.

    A feature can be based on more than one pre-existing features. A
    feature does not have to relate back to the original planes and origin
    through all pre-existing features. There are in fact parallels to
    creating a SW model and programming. Even the term spaghetti code from
    programming has a parellel in SW Parent Child relations. Unfortunately
    the one program that can really illustrate this point, AssyGator, is no
    longer available.

    Matt has given a synopsis of the tools for managing Parent/Chlid
    relationships so I won't belabor that point.
     
    TOP, Nov 12, 2006
    #4
  5. Pot,

    Do you actually use Solidworks ? or are you just an A-hole trying to stir up
    shit.

    Matt's been using it for about a decade, so have I. The information he
    supplied was correct. It may have been a bit lengthy, for a newbie (or
    incomprehensible for idiots like you), but it was right on.

    From googling your posts and responses, you haven't been around long. As to
    the content, YOU come off as a self rightous, arrogant, judgemental, BLOW
    HARD, in every one I read.

    You need to learn some manners lady.


    Mark
     
    Mark Mossberg, Nov 12, 2006
    #5
  6. umasemonoff

    ms Guest

    Pot = A-hole
     
    ms, Nov 13, 2006
    #6

  7. Matt tried to answer the question. I thought he did a good job, as did Paul.
    How did your response help the original poster? How did your response help
    anyone?

    When I was a kid, I was taught "if you don't have something good to say
    about someone, don't say anything at all." I have trouble sticking to it,
    but it's generally pretty good advice. So far I'm having a really hard time
    saying anything at all about you. That's not a problem with Matt or Paul.

    Jerry Steiger
    Tripod Data Systems
    "take the garbage out, dear"
     
    Jerry Steiger, Nov 13, 2006
    #7
  8. umasemonoff

    John H Guest

    I suggest we all just add FlowerPot to our "ignore" lists.

    By the way, are you getting the message yet, FlowerPot?

    John H
     
    John H, Nov 14, 2006
    #8
  9. nice explanation - phew, pretty concise too i'd say, considering what an
    enormous subject this is.

    I'd add a bit about how things appear in the tree: not only is the sketch
    shown under the extrude (example) so the 'parent' is under the 'child', but
    these days you can completely muddle the tree order by sharing sketches for
    multiple items, or drawing a sketch (sketch 1) then modelling a load of
    other stuff not using sketch 1 (sketch 1 remains at top of the tree) and
    then using sketch 1, suddleny sketch one is at the bottom of the tree! This
    isn't best practise by any means, but it is POSSIBLE, so it does happen!

    The 'wide' tree is definatly better for stability, but the nature of design
    means that it's rarely possible to model everything based on a few initial
    sketches and planes - the trouble with parametric modeling is that you can
    dragged down a direction where a simple change can become a days work....

    uh oh, I'm jibbering, better stop now.
     
    Lee Bazalgette - factorydesign, Nov 14, 2006
    #9

  10. What really makes this last example a pain in the posterior is when you go
    to edit sketch 1 after the above changes and the stuff you modeled is not
    available for reference in your sketch! You've got to delete the feature you
    built from Sketch 1, move it to the end of the list, and then add your
    feature again to allow it to reference the "later" features.

    Jerry Steiger
    Tripod Data Systems
    "take the garbage out, dear"
     
    Jerry Steiger, Nov 14, 2006
    #10
  11. umasemonoff

    ed1701 Guest

    Yes. Uggh.
    I often am tempted to use shared sketches or contour selection , but
    it just isn't worth it (except when disciplined - matts 'wide feature
    approach').

    My main layout sketches, I leave at the top of the tree, then start new
    sketches and convert the relevent entities into new sketches. If I
    share a sketch (and I do like to share) its only when I am pretty
    certain that it won't f*** me in the a** later (pardon my fr***).

    Yes, there are tricks to moving absorbed features up the tree where you
    want them to be (what I call the feature ladder - I should really get
    around to posting that sometime), but even those tricks are spotty
    depending on the tree structure.

    When, oh when, will we see Pauls flat feature tree? Oh When?
    Ed
     
    ed1701, Nov 15, 2006
    #11
  12. umasemonoff

    Bo Guest

    I vote with Matt for WIDE Trees when parts are complex. It keeps me
    from pulling my hair out.

    Since I design mostly plastic parts, and they can get complex easily,
    I've sometimes wound up creating a model where it is just simply
    impossible to change a dimension or two to optimize the part or
    versions without getting the chicken pox up and down the tree.

    Starting over meant creating planes off the primary planes and origin,
    and controlling ALL major feature starting points from those initial
    planes (& planes at angles from them & the origin) & axes derived from
    those plane positions, which help me a lot Dependencies from plane to
    plane are solid (If I can use that term; 'stable' if I can't).

    Once I redesigned my parts with a W I D E tree setup (as it has been
    called), I can now go in and alter an initial plane position or
    positions to get a new 'size' part, and everything behaves like James
    Dean's hair with Pomade.

    Bo
     
    Bo, Nov 15, 2006
    #12
  13. umasemonoff

    Bo Guest

    I vote with Matt for WIDE Trees when parts are complex. It keeps me
    from pulling my hair out.

    Since I design mostly plastic parts, and they can get complex easily,
    I've sometimes wound up creating a model where it is just simply
    impossible to change a dimension or two to optimize the part or
    versions without getting the chicken pox up and down the tree.

    Starting over meant creating planes off the primary planes and origin,
    and controlling ALL major feature starting points from those initial
    planes (& planes at angles from them & the origin) & axes derived from
    those plane positions, which help me a lot Dependencies from plane to
    plane are solid (If I can use that term; 'stable' if I can't).

    Once I redesigned my parts with a W I D E tree setup (as it has been
    called), I can now go in and alter an initial plane position or
    positions to get a new 'size' part, and everything behaves like James
    Dean's hair with Pomade.

    Bo
     
    Bo, Nov 15, 2006
    #13
  14. umasemonoff

    John H Guest

    I like the "chicken pox" analogy....gave me a smile!

    John H
     
    John H, Nov 16, 2006
    #14
  15. umasemonoff

    mjlombard Guest

    I agree about contour selection, its just too flaky through sketch
    changes, but if you have anything specific about shared sketches, I'd
    like to hear it. I find them to be far more reliable than converted
    entities or anything else. I think when contour selection first came
    out, the resellers showed shared sketches as being a function of
    contour selection, and it can be, but shared sketches can also be used
    with the normal method of sketching.
     
    mjlombard, Nov 26, 2006
    #15
  16. umasemonoff

    ed1701 Guest

    I have found them to be quite stable - I just have issues with shared
    sketches in the way they get absorbed, making reordering difficult.
     
    ed1701, Nov 27, 2006
    #16
Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.