Aller au contenu principal

Is there a way to pull a shape and expand it at the same time.


18 commentaires

  • Jacant

    Use the 'Pull' tool with the 'Draft' option.

    Select 'Pull' Select the fixed surface, with the 'Draft' option. Pull one of the sides. Press Space to type in new angle.

  • Tim Heeney


    A few steps but about 30s of work...

    As usual, Jacants method is quicker - at least below you see another way with taper determined by distance offset instead of angle ( the more common method i'll agree). Retrospective action at any time( not undoing if you don't want) is possible either by edge offset or taper change ( quite nice again, eg.adding 5deg etc )




  • Robert Loper

    Thanks immensely.  I tried all the different techniques and got all of them to work for me.

    Part of my problem is I tend to click on the face of things and not the edges.  That makes a HUGE difference.


  • Buk

    I've also frequently felt the need/desire to be able to do a Pull->Scale in conjuntion with an Extrude or Sweep. This comes up a lot when modelling bevel gears:

    It has all the information to do what we want, there just isn't a way to tell it.

    The only -- but still limited and subject to uncontrolled variability -- approach to this I've yet found, is to use a variable sweep Pull, but the conditions required to make it work; and whether it produces satisfactory results are subject to the gods.

    Start with the surface you wish to extrude and scale. Define (place a point) at the point of scaling. connect a few corners of the surface to the scale origin with construction lines in 3D.

    You'd hope that you could just sweep part way down these lines to achieve the goal, but every attempt always fails:

    The problem appears to be that even though you only want to sweep part way, the fact the the Sweep Path guidelines converge stops it from working.

    What I've found does work (most of the time after some fashion or another), is to overwrite the lines with shorter ones that define the limits you actually want to sweep; and delete the converging ones. These used the mid-points of the original lines for a 50% scale factor:

    And then Pull Path selecting those lines and Full Pull, then fill the front edge gives:

    But -- and it nearly always happens for reasons that escape my logic -- whilst one side is perfect, the other side goes weird:But sometimes it works properly.

    I haven't discovered the rules for preventing that distortion!

  • Tim Heeney

    This works in V4...



  • Tim Heeney

    Add another group of scaled curves at mid height & rotation and a spiral is possible.


  • Buk

    "V4 Sorry Buk"

    No problem ;) With thte addition of a couple of lines,V2 can do that shape also. Using Pull->Path

    And the spiral version too:

    Unfortunately, a true bevel gear doesn't start with a flat 2D entity like this:

    But rather a complex, 3D entity more like this:


    Which is a whole different kettle of fish :)


  • Tim Heeney

    Well, i can always add some 45 cut pulls... but i that doesn't quite 'cut it' does it !

    Still, it looks believable but doesn't function at the moment i'll have to agree...

    Longhand method is probably the only way - many youtube examples esp. solidworks.


  • Buk

    "Still, it looks believable"  

    I looks are all you need that's great.

    If you need to be able to print them, have them CNC'd or perform FEA on them, you need something like this:

  • Tim Heeney

    Having never constructed a gear set from calcs...OMG - what a lot of work !


  • Buk

    Luckily, there are websites like that will generate highly accurate involute gear profiles including user specified profile shifts and radii for you to download.

    Once you converted the .dxf format to something DSM can import (.stp)

    you can then move on to trying to produce accurate bevel and spiral bevel gears.

  • Tim Heeney

    Thanks Buk.. Ive been on the KHK Japanese website. Calc programs available there to but i wanted to create the teeth~ gears from scratch to refresh my feeble memory and educate myself further.

    As a by product ( perhaps even more enlightening) - the 2D geometry construction / manipulation /measurement with pasting of results for lines lengths with snap point feedback /helpers/advisors ( perpendicular, tangent , endpoint , curve center etc) has been an extremely worthwhile and valuable  exercise in itself. I knew they were there of couse, but i've never really had to study / watch and perhaps reselect etc like this never before. I believe totally the accuracy of 2D construction - fantastic!


    Buk -Thanks for the tip...

  • Buk

    Once you've imported the profiles of your gear pair, I convert them to a pair of thin (1mm thick) solids and align their centres so that their pitch circles are at the origin. Note: do not mesh them at this stage.

    I also simplify the profiles prior to Filling and Pulling, by tracing over the involute curves and radii--  that the generator constructs with many short lines and arcs -- with a pair of splines and then Move->rotate->Pattern the two splines and two straight lines to regenerate the gear. For the 60 tooth gear above, that reduces the number of curve elements making up the gear profile from 6200+ to 240. It makes all subsequent operations much quicker.

    I then define a plane at right angles to the gear profiles on which to draw the construction required to generate the cones for the bevel gears. That construction consists of a rectangle and 4 lines:

    The rectangle length and width are defined by half of the PCDs of the gear and pinion respectively. In this case, the gear is a 60 tooth x 2.5m with a PCD of 150mm, hence 75mm length. The pinion is 12T x 2,5 hence 15mm width.

    The diagonal of that rectangle defines the reference cone for the bevel gears. The 15mm line is the facewidth, and the two 10mm lines at right angles to it define the inner and outer cones. ie. the cones at the inner and outer ends of the teeth.

    Pulling those 3 lines around each of the axis aligned to the rectangle, produce the cones that define the limits of the teeth of both gears in 3D space:

  • Buk

    I then rotate the two profile solids to make them at right angles to the reference cone diagonal:

    Then I project the profile of one tooth (plus adjoining root edges) onto the outer cone:


    Remember to set the projection direction to the diagonal.

  • Buk

    At this point, what you'd like to be able to do is select the tooth profile you've projected onto the outer cone and Pull->Extrude&Scale, to produce the tooth solid directly -- which is where this thread started -- but DSm doesn't provide that facility.

    So instead, I calculate the scale factor required to reduce the outer cone to the inner cone.

    Length of diagonal - facewidth / length of diagonal.

    76.4852927mm - 15mm / 76.4852927mm = 0.80388386485183706435629552085116

    And use that to scale the outer cone to the inner cone position. If -- having entered the scale factor -- instead of hitting Enter, you hold the ctrl key and click the yellow arrow, it will create the scaled surface as a copy, leaving the original behind.

    Now I need to connect the two tooth profiles with surfaces to form a solid. That may be possible using V4 Blend, but in V2, my attempts to use Fill either fail completely or produce very unsatisfactory results.

    Instead, I have to use Pull->Path with Scaling. To facilitate that, I connect 4 points on each of the two profiles together with (3D mode) lines. Also, because Pull->Path with Scaling tends to not pull complete to the length of the lines, I make these four lines extend beyond the cones at both ends.

    There are actually 3 lines in the following pic. The first two are drawn using "Define line from centre" , first from a point on one cone to the corresponding point on the other; And the second connecting the same two points, in the reverse order. I then overwrite the full extent of both lines with a single line end to end and delete the first two. (There may be an easier way of constructing a single line that passes through two points but extends beyond them (in 3D) but I haven't found it.)

    I then do the same for the other 3 pairs of points. You can then use Pull->Profile with Scale option to pull the tooth profiles to surfaces.

    Sometime you may be able to do this in a single step; other times (as above) you have to pull the straight edge at the bottom as a separate step. If you're lucky (I hate saying that about an engineering process!), then you can cap the two ends with Fill and end up with a solid

    I then use the Split Body tool, the Outer cone and Scaled outer cone to cut the oversized tooth back to the correct size:

    And now I use Move->Rotate->Pattern to replicate the tooth.

  • Buk

    There's a bit of work left to do to make those 60 solids Combine into a single gear, but I'm sure you can work that out. Then its rinse and repeat for the pinion.

    Once done, the final task is to rotate one of the gears by half of its tooth pitch so that they mesh:

  • Jayx .

    This is one of the best discussion on this forum recently. One question though (sorry for a bit off topic), Buk, what's the reason you still use V2?

  • Buk


    The short version is:

    • the hardware (gpu) limitations of my old machine. It won't run v4 due to the gpu/drivers not supporting the v4 DirectX version.
    • my new machine which has a AMD A10-9700 Radeon R7 apu -- which does support the latest/greatest DirectX/OpenGl standards -- does not like AMD's "Detect & Install" software**, which fails to detect their hardware, despite that the OS (actually 7 different OSs including 3 version of Windows and 4 of Linux) does detect it, as does CPU-Z and a bunch of other hardware detection software. I have an ongoing (for over 1 year) support ticket with AMD.

    (**This is not an issue unique to my machine; but rather an ongoing issue for many AMD users )

    I have also become somewhat sceptical as to whether v4 would actually fix my biggest issue with v2, that of memory usage. Many of my more complex models have grown so large that they are butting their heads against the 32-bit memory limit of v2.

    However -- on the basis of (mostly anecdotal) evidence comparing v4 memory requirements to those of v2 for similar models -- it isn't at all clear that v4 would be the panacea for my complex models that I hoped it would be. It seems that moving to 64-bit, causes the same model to require considerably greater memory.

    AMD promised -- just prior to the current global crisis -- to provide me with (another) new version of their drivers. Quite when they will be in a position to do so now is not clear. As and when I finally am able to install v4 here, I will take a long hard look at the memory usage.


Vous devez vous connecter pour laisser un commentaire.