Skip to main content

Rotating something to alignwith something else (again).

Comments

36 comments

  • Jacant

    Create a Sketch plane on the top plan view. Project to Sketch the 100mm Arc and the edge of a tooth. Draw a circle from 0,0 to intersect the edge of tooth line. Draw a line from 0,0 to the intersection of the Arc and the Circle and another line to the midpoint of the projected edge tooth line Use Move using the direction of the  horizontal line, with the Orient to Object of the other line. Or measure the angle and move/rotate the gear. With the Move Anchor at 0,0

     

    0
  • Me Here

    Thanks, that worked (using Orient to Object method). measuring from the intersect point of the radial and circle, to either tip of the moved tooth, and both are within 1/10000th mm the same.

    The distances in this image are overlayed and hard to read, but they are 14.0495 and 14.0496, which is as clos eto perfection as I could hope for.

    However, if I measure the angle of the arcs from the vertical out to the same intersect point, and use those measurements to drive the rotations, and then measure as above, the results are different by several hundredths of a mm.

    Despite that I have angular precision set to 6 digits:

    The measure seems to round arc angles to 2dp:

    All teh methods I tried used the measure tool to obtain the rotation angles, and all resulted in similar inaccuracies, which meant that my gears did not mesh properly.

    I guess you can call me convinced by the Orient to Object rotation method, despite that it seems weird to use a translation handle to rotate :)

     

    (I still wish I could Alt click a point on the object being rotated, then select Upto and click some other point in space and it would rotate the alt-clicked point on the object until it aligned radially with the Upto point. But ... )

    Thanks again. Another step forward after many forward and back.

    0
  • tab...

    Hi Buk.

    The 'primary precision' seems a misnomer to me ( but perhaps i don't understand properly)

    Say i set it to 1 and i enter a length of 1.123456789, should i expect to get a length of 1 ? -no apparently not.

    What i get is an enter box display, with a 1 dp value... i type 1.123456789 regardless...

    I measure the line using 'e', with display set to 8 in measure options dialogue, it reads 1.12345679 it has rounded up the 9th dp , displaying to the 8 dp.

    So, i think it's the same with angles...

    same with face movement... ( Primary precision set to 6 ) after entering, dialogue reduces to .123457 I still ponder this...

    0
  • Me Here

    My bad. I'd never even noticed the Measure tool options.

    (But why oh why don't they take their defaults from the Units configuration screen!?)

     

    0
  • Jacant

    The 'Angular' precision can be changed to 6 decimal places, as with the 'Primary' precision of 6 decimal places. 

     Degrees- Minutes - Seconds.

    Degrees - Minutes.

    Radians.

    0
  • tab...

    According to me, it doesn't effect the accuracy of modelling - maybe if you capture the display result somehow to use as an input. TB investigated further when i find the energy.

    No, it seems a measurement info dialogue can't be selected ( copy / paste etc).

    Test: angle display set to 2dp, face rotated 0.123456789 degrees, then move>face> 'e' initiated with 'e' measurement between faces, value 0.12345679 returned, 0 entered - faces moved parallel 0.00000000 ( 8dp)

    The Primary Precision is meaningless wrt modeling accuracy ( i think ?)

     

    0
  • tab...

    I don't think it's only the max displayed value either...

    Same faces, 0.123456789 apart... annotation dim changed from 6 dp to 15 dp. Now where did that extra 0.000000000180636 come from?

    Below, faces and points copy moved and a line drwn between moved copied points - see annotation note. it's a very small error.

     

    0
  • Me Here

    Tim:"No, it seems a measurement info dialogue can't be selected ( copy / paste etc)."

    The arcs are drawn in sketch mode; the rotations are done in 3D mode; How to discover the angle? Use the measure tool. It won't let me cut&paste; so type it.

    Not good enough, but better than eyeballing it. Right.

    Wrong.

    Of course, then come the questions, why can't you cut&past. Why does the measure tool lie...but no point. Noone is listening.

    0
  • tab...

    For plane rotations in a sketch, angle on the sketch can be measured relative to an intersecting plane intersection when adding the measure function after a move tool axis is selected, this can then be modified to 0 to align, or set as a new relative amount.

    So, perhaps ( as i haven't studied your particular situation ( sorry) ) by placing a plane on the referenced 3D geometry ( somehow) prior to any 2D sketch manipulation, this might allow accurate exact measurement inputs...any faces etc on the insecting plane that intersect the sketch plane get a measureable virtual line.

    a virtual intersection line with face ( on a plane ) that intersects the sketch plane.

    Maybe this will help.

    0
  • tab...

    Just what have i been doing...

    Something else i didn't know - any curve can be split ( allowing a point to be added) by the insection of any face acting as an infinite plane - tool tip doesn't say this.!

    I discovered this just yesterday...

     

    0
  • Me Here

    Sorry Tim, but no. Not even close.

    The arcs can only be drawn if the 2D sketch in which they are drawn, no other alignment of that sketch is possible. And the rotation (of a 3D object)  can only be done in 3D. And the only source of the dimesion are those arcs

    Besides, you're trying to address the symptoms, not the cause..

    The Measure tool (E) serves only one purpose: to inform the user of the (actual) size of things. Why would the user ever want it to lie to them?

    I can understand that the user might want less than the full precision to be shown in the dimensions on drawings --though weirdly, you can only adjust that between "the default" and the current configuration setting -- but displaying less accuracy when the user asks the program for the size, just makes no sense at all.

    0
  • Me Here

    Sorry Tim. Re-reading the below, it doesn't come across very gracious. I'm currently dealing with something of an existential crisis (least it feels that way.)

    The aftermath of using cheap components to build prototypes:

     

    Looking again at your first animation, you appear to be matching a sketch angle, to that of a pre-existing angle on a 3D object, that happens to be orthogonal to, and aligned with the sketch plane.

    This is alomost the exact opposite of what I am doing: namely, rotating a 3D entity, by the angle of an arc drawn in a 2D sketch.

    I cannot make head nor tail of what the second animation is meant to be showing me.

    0
  • tab...

    Sorry Buk,

    it's not intended as an answer to your particular problem ( as i said above), but just my 2 cents worth of recent discovery of adding planes to 3D faces / geometry to get virtual intersections thus allowing the taking of accurate measurements which might be useful when combined with the move tool - that's it...

     

     

    0
  • tab...

    Buk, sorry to for not apply my thought to your particular problem - addressed i hope this time.

    This is a way to get cut and paste measurements onto clipboard AND also to directly use in a move operation.

    I've added a 'Helper' face to assist the CAD 'logic' in measuring / rotating  - seems to be the answer if understanding the problem.

    Hope the pictures are clear - i'm sure you'll get what i mean this time.

    Make SURE the ankor goes to the real rotation axis and not part/ surface centriodal ( as it's not perfectly round. )

    In the below image - no move command , it's just a 'e' measure between the ' helper face' and the line to tooth center reference - copy to clipboard result.

    The above image/command uses the MOVE OPTIONS 'CALIPER' ICON ONLY command after  move tool positioned and axis direction chosen. Only this way / icon ( choosing method) is possible to work properly !

    sorry, pushed for time to check those links.

    I think the above works for your problem if understood - never certain.

    0
  • tab...

    So sorry Buk - i've edited the above post so many times to get right - hope it's clear now, i really do...

    More observations:

    0
  • Me Here

    Tim,

    Sorry for not responding to your latest posts earlier, but I have been waiting for 5 hours for DSM to do its thing with my latest attempt; and I cannot start a second copy of DSM to try out your methods, as the poultry 512MB of ram on my GPU is already overcommited as you'll see in the image below. (And just to add insult to injury, Amazon just informed me that the graphics card I ordered is on back order and won't be here until May.)

    As you'll also see (from all the red body intereference) in the image below, the attempt was yet another failure; which is why I am being such a pedant about getting these dimensions and intersection arcs accurate.

    I'll take a close look at your stuff above tomorrow.

    0
  • tab...

    Buk,

    No problem at all - i can appreciate those broken bearings can be alarming !

    Anyway, my last diagram - i hope this is all that's necessary, for successful rotations.

    I enjoy the 'paint' process...

    0
  • Jacant

    Hi Buk,

    Should the spiral bevel have the 'curve' going inward towards the center?  Be it left or right handed. Or does it not matter?

    The spiral is generally consistent with the curve of a cutter using it's diameter.

    http://www.sharifcadcam.ir/uploaded/4583269a-551b-4f70-83eb-aaab2232bfc8.pdf 

     

    0
  • Me Here

    I had a similar confusion and haven't found anything that diffinitively defines the handedness, so I drew this -- the image is from the Wikipedia Spiral Bevel Gear page:

    In the two diagrams on the left, the Beta line angle NW->SE; on the right, SW->NE.

    The top two, the arc draw with its centre closer to the gears centre; on the bottom, further away.

    All four are distinct -- I copied each diagram and placed it over the centre of the other 3 in turn, and rotated in-place to convince myself.

    By my reckoning, my top left diagram is right-handed; and my top right is left-handed.

    Whether the other two ever arise, or are useful, I'm not sure.

    0
  • Me Here

    Tim,

    If, as you do in your images and animations, you have two lines that define the angle of rotation, then obtaining that angle using Measure, and then pasting it into the Move->Rotate is simple enough. Assuming Measure will let you Copy the value(*see below).

    And it works for whatever you are rotating. The need to converr lines to a surface before the in-Move Measure tool will deign to cooperate is a nonsense. Well done for finding the workaround, but for me, it just means it is  another of those icons that is essentially useless.

    But the main problem with your procedure is that it misses the fact that it was finding the angle at which to draw that line, that was the problem. That angle is defined by a the intersection of an arc with the prjection of a cone that doesn't exist. Further, the arc is at an angle to the plain of rotation and generating the cone just to find the point at which the arc intersects it, labourious and error prone.

     

    (*)But (anecdotally) about 75% of the time, hovering over the onscreen info fails to offer the copy button, no matter how much you waggle around trying to trigger it.

    I have found a workaround. If you right-click in the info panel, and then esc to dismiss the context menu, it (mostly) will then offer you the copy icon (and often the Create Flag Group icon, whatever the hell that is -- for another day). Sometimes it needs repeating this procedure 2 or 3 times, but it usually gets there in the end.

    0
  • Jacant

    Whether the other two ever arise, or are useful, I'm not sure.

    That's why I was asking. The gear you created has the curve the same as the bottom left image.

     

    0
  • Me Here

    I know! (Now, but it took me a while to see it)

    More to your liking sir?

    Don't ask about his mate :)

    There were also several other things wrong with the previous one, not least of which is that the cutting plane I used was (like your earlier one) aligned by the top edges of the inner and out tooth ends. This is wrong, the cutting plane has to be aligned with (tangent to) the reference cones.

    0
  • Me Here

    Finally, without any profile shift or other fudge factors, I think I got it:

    0
  • Jacant

    Excellent Job.

    Do you now have a consistent way to produce any size gear?

    Have you figured out how the central axle configuration will work and fit together, with all relevant bearings?

    0
  • Me Here

    jacant:"Do you now have a consistent way to produce any size gear?"

    I'll need to produce two or three more pairs  at least before I can say for sure; but I'm taking a day off to rest my eyes and brain. In particular, I need to try shaft angles >90°.

    With some combinations of teeth, once the angle gets above 90°, it leads to something called an "Internal crown" which is apparently much more difficult. I have only one reference to that, and its a comment in some source code. Searching hasn't found anything useful, so I've emailed the code author to see if he can shed some light on it.

    jacant:"Have you figured out how the central axle configuration will work and fit together, with all relevant bearings?"

    Not yet. I don't need the bearings in order to run a simulation to test the strength/weak spots of the gears, and there are so many possible variations (not just of ratio and how that ratio is achieved, but also the physical arrangement of the mechanism. Eg. shaft angle.

    It is also a possibility to invert the arrangement of the two pairs. Ie. in the above, the small blue gear becomes the static Reaction Control Member ( RCM. N1 in the calculations) and the large cyan gear becomes the output N4 (with the mating gears becoming N2 & N3 respectively).

    If as I suspect, the output pair of gears experience the greatest loads, swapping them over allows the use of a bigger module (2m or 2.5m). So many possibilities.

    Next step is to model one set of middle of the road pairings at a middle of the road shaft angle, correctly drawn and arranged with the coincident pitch cone points, and get a set of stress numbers/diagrams; and see where that leads me.

    0
  • Me Here

    BTW. Do you have any experience with loading .stls into DSM and cleaning up the mess?

    0
  • Jacant

    I have done it before. You have to remember that an STL file is an approximation of the outside surface of a model. It is split up into facets, each facet will be split up again into triangles. The more triangles the smoother the model. So when you 'Convert to Solid' - 'Merge faces' will only merge the triangles on each flat face.

    You can use the existing geometry to draw over the facets using Splines, Arcs and Circles.

    Upload an STL 

    0
  • Me Here

    Interesting. Is that a laminar flow nozzle?

     

    Upload an STL

    Can I send it to your gmail addy?

    0
  • Jacant

    I'm pending approval.

    It's an air nozzle for a laser cutter. Same principle.

    Just remembered, you do have my email. Sure send it there.

    0

Please sign in to leave a comment.