SUCCESS!! A toolpath cut normal to a cylinder (cylindrical cam)

Discussion in 'SolidWorks' started by Sporkman, Jan 7, 2005.

  1. Sporkman

    Sporkman Guest

    Finally I have found the proper way to create the CORRECT geometry for a
    toolpath cut normal to a cylindrical surface, and have been able to
    create my cylindrical cam accurately as a result. The thanks go to Dr.
    J. D. Mather, Assistant Professor in CAD & Product Design at the
    Pennsylvania College of Technology (www.pct.edu). He has created a
    number of tutorials that reside at:

    http://home.pct.edu/~jmather/content/DSG322/SolidWorks_surface_tutorials.htm

    One of his tutorials is for a cam that apparently represents a sine wave
    accurately. It's a VERY complex creation he's built -- perhaps more
    complex than absolutely necessary, but it's hard for me to say. The
    tutorial is definitely worth downloading and looking at, but the core of
    the technique that he (and I) used is the Feature Wrap function. Create
    a plane tangent to the surface of the cylinder and create a CLOSED
    sketch on it. You can create a curve apparently accurately representing
    the outline of an endmill cut by offsetting both ways a construction
    spline or series of construction elements (controlling the ends of a
    spline with tangency to line segments is useful) and closing the ends
    with an arc tangent to both curves. Exit the sketch, select it, and use
    Feature > Wrap, with the Deboss option selected. This will wrap the
    curve onto the cylindrical (or whatever) surface and make a true cut
    normal to the surface. The distance value is the amount the projected
    cut penetrates. Emboss (obviously) creates just the opposite of a cut.
    Magic!! Extremely powerful function that I did not realize existed.

    Kudos and thanks, Dr. Mather. And "way to go" SolidWorks Corporation.
    And I was cussin' 'em under my breath for not providing that easy
    functionality.

    Mark 'Sporky' Stapleton
    Watermark Design, LLC
    www.h2omarkdesign.com
     
    Sporkman, Jan 7, 2005
    #1
  2. Sporkman

    Sporkman Guest

    Nope . . . I think I was wrong -- false alarm. SHI_! Looking at the
    cut from a view normal to the outside of a cylinder I can tell that the
    cut is not precisely normal to the surface or perhaps (yes I think) the
    cut tapers toward the center of the cylinder. Feature Wrap with Deboss
    apparently does NOT create a cut that is like an endmill cut. The
    rather extreme complexity of Dr. Mather's technique is apparently
    necessary to create accurate geometry. And I'd have to say that just
    should not be.
     
    Sporkman, Jan 7, 2005
    #2
  3. Sporkman

    neil Guest

    did you look at the model I sent you Mark?
     
    neil, Jan 7, 2005
    #3
  4. Sporkman

    Jeff Howard Guest

    ......... the cut tapers toward the center of the cylinder. .........

    Mark, I'm not sure if I'm following exactly how you did it, but an embossed
    groove will taper toward the center of the cylinder, just as will a wrapped
    sheet metal representation. If you emboss a cut that is parallel to the
    cylinder axis and emboss to a depth of 1/2 diameter you should end up with
    a wedge cut out of the cylinder (it's all working on surface normal
    offsets).

    If you emboss to create a surface that represents the cutter axis path
    (should be a ruled surface; e.g. from any point on the surface draw a line
    perpendicular to the cylinder axis and that line should line on the
    surface) and then offset that surface cutter rad both ways you should have
    a good set of groove walls. (zat make sense?)
     
    Jeff Howard, Jan 7, 2005
    #4
  5. Sporkman

    jmather Guest

    The kudos go to Jeff Howard. He posted the solution back in OCT 04 or
    03 (I'll find the printout of the post when I get back to the office.)
    The key is describing the just the centerline path of the cutter with a
    surface. And then offset that surface for the slot walls. When cutting
    a curve with an endmill the tangency is continuously variable. Check
    the results with a follower pin in an assembly in motion. Let me know
    if you find any problem as we are about to begin the semester and I
    don't want to introduce an inaccurate technique. I need to be able to
    add different types of motion ie. parabolic, cycloidic and dwell.
    There is an example on the SWX website that I have not had time to
    analyze for accuracy.
     
    jmather, Jan 7, 2005
    #5
  6. Sporkman

    Muggs Guest

    5b is the one you want. Very enlightening!

    Yes, thank you Dr. Mathers and/or Jeff Howard.

    Muggs
    Change "home" to"comcast" to reply
     
    Muggs, Jan 7, 2005
    #6
  7. Sporkman

    Sporkman Guest

    Yes indeed, sir, and I thank you. I don't think that technique can give
    me this result, however.
     
    Sporkman, Jan 7, 2005
    #7
  8. Sporkman

    Sporkman Guest

    Well wait a minute . . . umm . . . I may not have looked at your
    technique close enough. That looks like it might have a lot to
    recommend itself. Did you say I could post this model on my Web host?

    I managed to create what I think is accurate (enough) geometry using
    Jeff Howard's / Dr. Mather's technique of creating offset surfaces from
    a Wrapped surface, and that requires a good bit of subsequent surfacing
    work (trimming and knitting and etc). Your's does also, but isn't quite
    as complex. It looks good and I'll look at it further.
     
    Sporkman, Jan 7, 2005
    #8
  9. Sporkman

    jmather Guest

    you will need to show all 8 decimal places to see the discrepancies in

    the angular measurement

    I make the tutorials available to all so that I might learn something.
    I am basically an over-educated machinist who got tired of dealing with
    designers who didn't understand geometry. The good ones were not
    afraid to come back to the shop and discuss why I marked up their
    drawings. The others never set foot in the shop again - at least not
    during my shift. I encourage physical (well now virtual) models to
    analyze geometry. If you can provide me with an alternative model I
    would appreciate it. In trying to do a more comples motion I became
    aware that I hadn't found the altimate solution yet. I think what we
    really need is a way to sweep one solid (endmill) along a path on
    another solid resulting in the intersecting body being removed - just
    like the actual machining operation.

    I really like to get perfect mathematically geometry in the CAD data
    set because then I know (or think I know) I am doing something right -
    but I am not too worried about 8 decimal places in the shop.

    J.D.
     
    jmather, Jan 7, 2005
    #9
  10. Sporkman

    Jeff Howard Guest

    That's it in a nutshell.
    Something that I think is pertinent: we are dealing shapes defined using
    NURBS math and 8 place accuracy isn't likely to be attained under a variety
    of circumstances.

    ---------------

    I think that a proof of groove geometery might involve looking at a section
    normal to the cutter path. If I remember correctly the problem is cutter
    path normal direction for any given ray off the cylinder axis changes as
    the radius from cylinder axis changes. It's been a while since I've looked
    at and tried to understand what's going on. (Makes me wonder about the
    swept groove I sent you, J.D. Could be inaccuracies in it; maybe small
    enough to ignore (??).)

    Other guestimate check methods involve running a pin in the groove and
    checking interference and doing radial hole cuts to compare with groove
    geometry. These are at least 3D checks vs. the 2D section cut.

    I'd love to see Neil's (thicken?) solution and any other possibles. Maybe
    we could get away with posting pics and STEPs (I don't have SW) on Sean's
    site (mcadforums.com).

    At any rate the cutter axis ruled surface offset is about the most accurate
    way I know to describe the walls. Groove bottom surface for an endmill cut
    is another subject.

    A question for somebody that might be involved in coding the cut; what's
    necessary to define the cut? Seems a curve that describes the cutter's end
    face / axis interection path and cutter diameter would be sufficient. Is
    the solid model topology really necessary or useful?

    Regards to all.
    ================================
     
    Jeff Howard, Jan 7, 2005
    #10
  11. Sporkman

    jmather Guest

    it doesn't appear to model what and endmill would cut.

    Put it in an assembly and animate it. The follower (or endmill)
    constrained to linear motion in one axis and the cam constrained to
    rotate only about the cylindrical axis. I'll fire up the RP machine
    on Monday and produce a physical prototype. But I can zoom in on the
    virtual assembly much closer than the tolerance of machining tolerances
    and it looks good to me. I'll let you know.

    John Eric Voltin indicated he had a better method. I don't have
    version 2005 to see his technique but I inserted a STEP model he sent
    me into an assembly with my model and they appear to be essentially
    identical. (not to imply that means the model is correct - I would be
    interested in seeing other solutions)
     
    jmather, Jan 7, 2005
    #11
  12. Sporkman

    jmather Guest

  13. Sporkman

    Jeff Howard Guest

    Likewise. John, could you email me a STEP (reply to sender - ignore the
    spam trap message - I'll fish it out) and (or post) a brief description of
    the method? Twould be appreciated.

    If Neil wouldn't mind going to the trouble; same same.

    --------------------

    Dale, I've run pin in groove interference checks in grooves created using
    the surface offset method and it seems to resolve to about 1e-4 inch; e.g.
    1e-4 under (don't remember; dia or rad?) clears and 0 interferes. ... If I
    remember correctly, could have been a larger value but still in the range
    of xE-4. Could also be that the groove will grow oversize, but I don't
    believe that should happen.

    ===========================
     
    Jeff Howard, Jan 7, 2005
    #13
  14. Sporkman

    Muggs Guest

    Muggs, Jan 7, 2005
    #14
  15. Sporkman

    Jeff Howard Guest

    ..... Sorry for the distraction.

    Not to worry. "Distracted" defines my normal state of existance. 8~)
     
    Jeff Howard, Jan 7, 2005
    #15
  16. Sporkman

    neil Guest

    yes post it if you like
     
    neil, Jan 7, 2005
    #16
  17. Sporkman

    Sporkman Guest

    I'd VERY MUCH like to look at that model, John. I'll send you an email,
    but a valid email address for me can be found on the main page of my Web
    site.

    Thanks very much,
    Mark Stapleton
    Watermark Design, LLC
    www.h2omarkdesign.com
     
    Sporkman, Jan 7, 2005
    #17
  18. Sporkman

    Sporkman Guest

    Unfortunately I don't believe it is accurate, Matt. The surface offset
    to the inside is what I tried with my cylindrical cam. It doesn't seem
    to want to offset accurately radial to the axis (or normal to the
    surface). See the pic I've got up on my Web site:

    http://www.h2omarkdesign.com/img/screenshot_cam.jpg

    It doesn't really work. Although I think it SHOULD work.

    'Sporky'
     
    Sporkman, Jan 7, 2005
    #18
  19. Sporkman

    Jeff Howard Guest

    ......
    Thanks a million, John. The message is on the Earthlink server (~1 MB).
    You probably got a spam trap message back (how I long for the "good ole
    days" when such things weren't necessary). I'll retrieve it and get back
    to you if it didn't make it thru intact.

    Thanks again and sorry to put you thru the trouble.

    ========================
     
    Jeff Howard, Jan 8, 2005
    #19
  20. Sporkman

    Sporkman Guest

    Anyone else besides me think that the Offset Surface oughta work
    (although it doesn't)? A radial surface offset radially really SHOULD
    produce the exact geometry we're talking about with an end mill tool
    path. The fact that it doesn't says something about what's going on
    under the hood with SolidWorks. In other words, something smells really
    fishy here.

    'Sporky'
     
    Sporkman, Jan 8, 2005
    #20
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.