Problem with Tangent Plane on a cone.

[jsibre]

Hi,

I'm having a problem with Tangent Planes on a cone. 

I'm new to Fusion, so it's quite possible I just don't know what I'm doing.  I don't understand what the Reference Plane option does (I've played with it a bit, and haven't figured out what it's doing, and the docs here are not expansive on that detail: https://help.autodesk.com/view/fusion360/ENU/?guid=SLD-CONSTRUCT-TANGENT-PLANE).

In my actual project (as opposed to the test case project I've created), I'm working on something that resembles a colander (images on Google).  A flat bottom bowl that is perforated with a zillion holes.  The part that causes trouble is putting holes in the more vertical part of the bowl (I guess it's a conic frustum, with the narrow end 'down').  To put the holes there:

1. I'm attaching a tangent plane to a cone. 

2. I create a sketch on that plane, and draw a rect/circle a few mm up from the origin of that plane (landing the hole-to-be in the wall).

3. I then extrude that rect/circle to make a hole.

4. I then use Rectangular Pattern and Circular Pattern to copy that hole up the side to the top of the wall, and then copy it about the Z axis about 90 times (every 4º) to cover most of the surface.

I could be done at this point, but I want alternating holes to make it more 'diamond shaped' than grid shaped.

5. I create a second Tangent Plane in the same quadrant of the cone (I notice you can do it in each of four locations, 90º apart from one another), and use the angle to shift it by 2º (half of the 4º spacing I achieved by putting 90 around the circular pattern)

6. I create a sketch on that second Tangent Plane (TP), similar to the first, except that I make the hole a bit higher than the hole I made on TP1's sketch.  Now, this hole is a bit above and to the right of the first one.

You can stop there.  I went ahead and rect/circle patterned before I eventually noticed the problem, but if you know the problem is coming, and you look for it, you can see it at this point. 

The hole created on TP2 is too far to the left.  It *should* be perfectly between (but above) the first hole in TP1 and it's sibling to it's right, but it's not.  Also, if you *do* perform the rect pattern step, you'll notice that line of holes is *not quite* parallel to the first rect pattern done on TP1.

So, I've done a lot of messing with this problem, and created a test case for sharing (I'll try to attach it here), but I want to pause in case this is enough for someone to tell me that I'm doing it wrong, and I should do "xyz" instead.

I will mention though that it appears that when I rotate the TP, Fusion rotates it as if it were on a cylinder that was coincident with the line where the initially created TP touches the cone.  That is, pretend there's a cylinder there, instead of a cone.  If you do that, the rotation makes sense.  I feel like this is a bug, but again, I'm probably just doing something wrong. 

I found this post, and feel this may be somehow related.

So.  Test case.

I've spend a lot of time on this (an embarrassing amount - over 12 hours, probably).  I've built a simplified test case that demonstrates this with the cone, the two TPs, a greatly reduced number of holes, and 90º steps instead of 4º.  Why 90º?  Because then I can create a third TP in the next quadrant, and sketch a hole on it that *should* correspond to the hole on TP2.  and then you can see the wonky mismatch and angle deviation.

But again - I'm probably just doing it wrong.  Shoulda asked for help sooner. 🙂

Here's a screenshot of my test case.

TP1 is the first Tangent Plane, created in "quadrant 1" (arbitrary label).

TP2 is the TP created in the same quadrant, but rotated 90º.

TP3 is my extra (for test case only) TP created in the next quadrant over, and not rotated.  What i see there is what I expect TP2 to look like. Of course, this only works out this way because my test case uses 90º instead of the actual 4º I want to use in the real project.  The 90º is because it greatly exagerates the effect, making it super-obvious, and it also aligns with the next location I can create a TP at, which will look correct.

Here's one more screenshot where I've rotated to look straight at the right face.  You can see how the construction lines on TP1 & 2 are parallel, as if the rotation was around a cylinder, not a cone.  Odd that the angle between those construction lines and the imaginary line connecting the two origins is not 90º. 

 

I feel like something is wonky here.

I hope this makes sense - it took a long time to figure out, and I'm afraid that it's too complicated to explain without a video.  (But also that a video will take another embarrassing amount of time!)

Thanks,

Jason

Can anyone tell me what I'm doing wrong? 

[davebYYPCU]

Before we dig any more rabbit holes.  Can check the file tomorrow.

Same size holes?

Bottom row will look squashed, compared to top row.

There appears to be some thickness in the bucket walls.

Similar problem are the holes straight, as if drilled, or conic tapered to centre line of bucket?

Presume it is to be Extrude Cut and not a Hole Tool hole.

How did you make the bucket? Revolve, Taper Extrude, Sweep or Loft?

Let me know.

[jsibre]

Hi,

Yes, the holes are all the same size (2mm x 2mm square in the test case, but at one point I had been using circles that were 2.5mm in diameter.)

Yes, top row will have more lateral spacing than the bottom row.  I'm starting at the bottom row, so it's defined to be acceptable, and lateral distance between the holes increases as the rows get higher.  Vertical distance remains the same.

The walls are approx 2.5mm thick - the dimension is set on the XY plane, so the actual thickness is a bit less since the walls are not perpendicular to the XY plane.  (If they were, it would just be a cylinder, not a cone.)

The holes are straight, not conic.  Yes, the hole is done with the Extrude feature vs the Hole feature.

The bottom of the bucket is just a circle on the XY that I extruded.  The walls were created by drawing a sketch of the wall's profile on the XZ, which I then Revolved about the top plane of the bottom part.

Thanks

Jason

Edit: fixed typo where I said XY when I meant XZ

[jsibre]

Interesting... I thought I'd had a brainwave after reading this post: https://forums.autodesk.com/t5/fusion-design-validate-document/how-to-create-a-plane-tangent-to-cone...

Based on that, I thought, "I should be able to define the plane I need by using Plane Tangent to Face at Point!"

Nope.  Well, yes, but it does the same wrong thing.  It's a bit baffling because in that context it really looks buggy.

[davebYYPCU]

Umm, 

We have a problem creating a diamond pattern with the current info.  2.5mm squares and 4 degree offset, are not compatible.  the 4-degree turn is too tight in my opinion.  What was your intended vertical spacing?

Demo - mock up to show what is a likely result without changes.

As to bug? in Tangent planes, cannot speak to that as I have always found ways to avoid them.

Might help...

[jsibre]

You're right - the test case didn't have the same dimensions as the original (I realized that after uploading). In the test case, the holes would ultimately overlap, but that isn't the issue. The problem is that the Tangent Plane, when rotated about the cone, does not maintain its orientation on the cone. Instead, it acts as if it's being rotated around a cylinder.

As to "found ways to avoid them", I'm curious: Is that because you've had trouble with them, too? Or you've just found other solutions that seemed less complex?

I've spent a few minutes pondering the fact that surely I can set up a plane with the desired orientation without using Tangent Plane or Plane Tangent to Face at Point, (i.e., three points, two lines, plane at angle, etc) but I haven't sat down at Fusion and tried to work it out yet. If you have a suggestion, I'm open to it.

Thanks!

[davebYYPCU]

Tangent plane on a face at a point, is difficult to create with stability.

Plane on Angle, on the Revolve profile line is so easy, no need to consider alternates.

The problem is that the Tangent Plane, when rotated about the cone, does not maintain its orientation on the cone

Construction planes are not geometry, so have very limited use when trying to duplicate them.  (Can be done but too complicated when simple alternates exist)

Your file has no rotated planes - these 2 are aligned.

Sketching on body faces is more stable and duplicating in patterns bodies or faces is very simple.

Back to the design, 

Start with two cutter bodies that are lower cutters for row 1 and row 2, in correct spacing orientation, pattern them up the wall, and follow with circular pattern.  (Max fit is 6 degrees)

Might help...

[jsibre]

Thanks for the suggestion - I haven't tried it yet, but will.  I didn't know I could create a Plane at Angle on a circle (I think that's a paraphrase of "Plane on Angle, on the Revolve profile" - if not, then I didn't quite understand what you meant).  That would be more simple that what I DID try an hour or so ago.

What I tried was laying out some construction lines on the top surface of the bottom plate that gave me some lines I could use as the basis of a pair of "Planes at Angles". Actually, I guess it may be what you were describing above, but I added a couple of lines that were tangent to the circle.  Perhaps they were implied/assumed. 

That appeared to ultimately yield perfect results on the bottom row, but I did notice things "got weird" if I adjusted the angle (circled in image) after creating the planes.

By "got weird", I mean I started seeing evidence of the plane ... rotating.  Not rotating about the line it was defined on, but rotating on its own Normal (local Z, I suppose).  However, I got what appeared perfect after I created it in the correct spot, and didn't mess with it:

I haven't put the second hole through the pattern process yet, and it's at the same height as the first row to make the alignment easier to eyeball.  I measured with the measurement tool, and saw the spacing was only off by .001 mm.  I'm hoping that's just some odd rounding error.  I'd have expected it to be correct to a finer degree.

Finally, the reason I say "appeared" to yield perfect results is because I didn't complete the tiling.  You did, and on the bottom row it looks good..., but I don't know if you noticed, but at the top row in your image, you can see the holes have biased to the right (or perhaps left).  Suggesting the second column of holes was drawn on a plane slightly rotated:

Ok, after writing the above, I spent a few more minutes revising, and was able to put this together, based on the layout line approach:

Took a few measurements, and it seems correct. 

Just for grins, I've changed the angle in the layout to 15º (performing no other adjustments) and got this:

Now, obviously, the placement of the holes is wrong; ignore that.  Focus on the change in spacing from bottom to top.  The holes no longer go straight up the side!  This is what I was talking about when I mentioned things could still get weird.

I feel like that's a violation of the parametric nature of the product.  I should be able to go back and edit that angle (did I type 5º?  that isn't right; make it 6º), and it should 'play through'.  But it doesn't. Not perfectly.  It has side effects that are different from creating it that way in the first place.  But perhaps I'm not understanding something. I will cheerfully acknowledge I'm new to this product, and that it's view of a plane is actually more complex than "just a plane" - it includes a local origin, and a local definition of the orientation of the X, Y and Z axis. So far, I haven't figured a way to affect how those are defined, and I'm assuming that it's "just set" at creation time to something that seems sensible, based on the info used to create it.  Again, a blind spot in my understanding.

At any rate, I feel like I have a workable solution here, and perhaps I can just "get on with it." (Though I love to pick and tease at something that doesn't look quite right...)

Thanks for your help in learning a few more ways to look at a design challenge such as this!

[davebYYPCU]

Did your changes break my file? -I'll presume it's another version of one of yours - that I can't review.

"Plane on Angle, on the Revolve profile" leaving the most important word off - was 

"Plane on Angle, on the Revolve profile line" - the black line in my first picture.

However, my file did not use a Plane on Angle, or Tangent plane.  (Too complicated)

Your top hole bias pic green circles- yes that is my mockup coming back to bite me. 

Has to do with the widening spacing that I warned about earlier.  (I forgot)  

I Rectangular Patterned both "Holes" and I should have 2 rectangular Patterns.  One for each column of holes.

Which is directly related to the 15 degrees testing, the holes will have to be adjusted to being half of any angle change you want, which also relative to the circular pattern count.  My file is 6 degrees offset, pattern is 12 degrees for 2 columns, and pattern count is 30 for circular pattern.  (360 / 12)

Your numbers will be 7.5, 15, count 24.

Might help....

[jsibre]

I'm not sure what you mean about my changes breaking your file... I don't think it could, assuming the files aren't linked in the cloud (maybe they are?  I assume the .f3d file is self contained, and not some link to a version of the file in the cloud).

By the way, I neglected to share the .f3d of what I used to produce that latest batch.  Here it is, if you're interested.  I don't think you need to look at it - I'm only offering it in case you want to see what I was writing about 'first hand'.

There are some extraneous bits; the salient bits are the sketches named "Lines to help define planes", "Hole on PaA1" and "Hole on PaA2", and the Constructions named "Plane at Angle 1" and "Plane at Angle 2"

Thanks for your help.

By the way, I do still think this points to a bug or two.  The Tangent Plane doesn't work correctly when used on a Cone (IMHO), and the "weird" re-orienting behavior observed with both the Tangent Plane and the Plane at an Angle when the orientation is changed  doesn't seem right.  I think it the orientation of the axis on a plane can be changed, then that should be exposed in the UI in some obvious way. This "it works if you create it right the first time, but not if you have to make changes" doesn't seem like the intent. Both seem to prevent the parametric intent of the product. I'll see if I can't figure out how to submit the bug ticket.

[davebYYPCU]

I have not found a bug with the Plane on Angle in your new file version, they are set at 100 degrees, as expected, aligns to the 100 degree wall angle.  Did what you needed.

Let me know how to "Move" any Plane. 

You can parametrically change / select inputs.

Might help...

[jsibre]

Hi, the part that I think is a bug is that if I change the other angle (5º, 15º, the angle that has been used to move the plane a few degrees around the cone), it does not create the same end result as it would have if I created the plane AFTER that angle had been set. 

This bug can be reproduced for Tangent Plane, Plane at Angle, and Plane Tangent to Face at Point.  It may be present in other features; those are the only three I've used while on this little adventure.

I can send you samples 'proving' that, but I'd rather not.  It'll take more time.  The conversation above demonstrates it on Tangent Plane and Plane at Angle.  I saw it on Plane Tangent to Face at Point as well, but I don't think I provided any screenshots of that. 

[davebYYPCU]

Tangent Plane - can not speak to.

Plane on Angle - requires selection of a line, set the angle.  If the line is edited to a new position, the Plane updates.  Edit the set angle, the plane updates.  How would that be a buggy process?

Might help….

[jeff_strater]

---

@jsibre wrote:

Hi, the part that I think is a bug is that if I change the other angle (5º, 15º, the angle that has been used to move the plane a few degrees around the cone), it does not create the same end result as it would have if I created the plane AFTER that angle had been set. 

This bug can be reproduced for Tangent Plane, Plane at Angle, and Plane Tangent to Face at Point.  It may be present in other features; those are the only three I've used while on this little adventure.

I can send you samples 'proving' that, but I'd rather not.  It'll take more time.  The conversation above demonstrates it on Tangent Plane and Plane at Angle.  I saw it on Plane Tangent to Face at Point as well, but I don't think I provided any screenshots of that. 

---

I would like to see one of those samples.  I looked at your design a bit, but I don't really understand what the problem is in that design.  If you can demonstrate, on a simple design (preferably just a cone and a tangent plane) a case where the behavior does not seem right to you, please do share it.  Yes, it will take more time, but if there is a bug there, I'd like to see it get fixed, and that time will be appreciated.

Are you using a reference plane in creating your tangent plane?  If not, Fusion will use the pick point on the cone (which is not a stable entity, and almost impossible to re-create) as the point of reference.  You will get a much more predictable result using a reference plane.

[jsibre]

@davebYYPCU, the bugginess is that when the plane updates, it also rotates. However, if you create a new plane based off of that same line and angle, it will not be rotated.

[jsibre]

Hi Jeff, I'll put together all three samples then; I have a hunch that all three are related to the same underlying behavior.
Could be that it's not a bug, but I don't see it documented anywhere, and there doesn't seem to be anyway to influence the behavior (the rotation) directly. It only happens as a side effect.

[jsibre]

@jeff_strater,

Here's the first sample, basically just the sample from above, but stripped down to the original posting.

It demonstrates the problem on the Tangent Plane function.

I've started with my basic 'bowl' shape, and I've created three Tangent Planes on the cone.

Tangent planes appear to only be creatable on the four axis that correspond to the North, South, West, East portions of the cone.  Not sure of a better way to say it, but where you click on the cone will result in a plane that has snapped to one of four locations; it doesn't generally appear exactly where you clicked, unless it was on one of those four locations.

* Tangent Plane 1: Created on the portion of the cone which intersects with the YZ plane where Y is negative. I.e., the "South" portion if you're looking down on the object from above with "north" pointing up.

* Tangent Plane 2: Also created on the South side of the cone, but then rotated using the TP dialog by 90 degrees so that it's now on the East side

* Tangent Plane 3: Created on the East side of the cone.

All three planes have the same basic sketch on them. Notice that the sketch (and plane) for TP2 have been rotated (by 10º, I think, corresponding to the angle of my cone) relative to the sketch on TP3.

 I suggest that the sketch on TP2 and the sketch on TP3 should be the same.  There's no way for me correct the orientation of the geometry on TP2's sketch to match that on TP3.  The bug is that when TP2 is rotated by 90º, it appears to do so about a non-existent cylinder which is coincident with the cone on the axis upon which the plane was originally created (The South face). Notice that if you look at the right side, the angle of the geometry on TP2 is parallel to the south face of the cone.

 

Sequence of steps to produce this:

1. Create the base and conic form

2. create TP1 on south face (leave angle to 0)

3. create TP2 on south face (leave angle to 0)

4. create TP3 on east face (leave angle to 0)

5. create sketch on Tp1; draw geometry

6. create sketch on tp2; copy geometry from tp1 sketch

7. create sketch on tp3; copy geometry from tp1 sketch

8. double click on tp2 in timeline, change angle to 90º

At that point, you'll see that instead of the geometry on TP2 directly overlaying Tp3, it will have been rotated as well.

I'll follow this post up with similar examples for Plane at Angle, and Plane Tangent to Face at Point as I have time.  Doing this post and sample and screenshots took about 45 minutes.

EDIT: Wanted to mention, this one doesn't really have a work-around.  If you want a plane that doesn't lie on one of the four orientations you get by choosing "0º", you can't get it without the incidental rotation of that plane about it's own Z axis.  Further, subsequent edits to that angle will change that rotation.

[jsibre]

@jeff_strater 

Here is the sample for Plane Tangent to Face at Point.

The step are a bit different, but the result highlights the same unintended plane rotation.  Using this feature, it IS possible to avoid the bug, if you create everything correctly initially. The bug only appears when you go back and make changes to an existing plane after its initial creation.

Start with the same cone-shaped bowl.

1. on the top surface of the base, create a sketch containing two constructions lines that terminate on the outside surface of the cone.  These will be used as the "points" in the Plane Tangent to Face at Point.  I've created mine at South, and South-south-east (SSE, 45º counter-clockwise from the first one).

2. Create the first plane (PTtFaP1) at the southern side of the cone, using the point at the south side.

3. Create the second plane (PTtFaP2) at the SSE side of the cone, using the point as SSE.

4. Create the third plane (PTtFaP3) as the southern side.

5. Add the same geometry to sketches on the corresponding planes.

At this point, everything looks fine - no changes have been made to the planes since creation.

6. Go back and edit plane 3 (PTtFaP3) and change the point from the one at the south side to the one on the SSE side.

 

Witness the rotational artifact.

What's fun with this one is that if you then go back to your sketch with the construction lines that defined the points used for creating the planes, and change the angle between those lines from 45º to 22.5º, you can see the geometry for both PTtFaP2 and PTtFaP3 rotate to the left.  Now neither one is ... co-planar? I think that's what I want to say...  with the original Z axis:

 To me this is a very real use case, because you (or I) may not know exactly what angle you need between the two points on the cone initially when you create it - You'll have to edit it after the creation, and other work has been done to determine the correct point. (For example further up in this post thread, I apply a circular pattern, and the count of holes around the wall determines the ideal spacing.)

On the other hand, the nice thing about using this feature (vs Tangent Plane) is that once you know what angle you need, you can delete the original plane, adjust the construction geometry, and then create a fresh one that's in the correct location, and Bob's your uncle.  I.e., there's a workaround.

[jsibre]

@jeff_strater 

Here's the sample for Plane at Angle.

Start off with the same basic bowl shape.

1. create a sketch on the top surface of the base again.

2. draw two "L" - one that starts from the origin (center of bowl) and goes south to the outside of the wall (25mm in this example), and then goes to the right at a 90º (right as you're looking down on it - left if you were in the car driving along the line 🙂 ) The other "L" is similar, but rotated 45º counter-clockwise again. (similar to how we set up the points in the previous sample, but we've added lines that are tangent to the cone at that point.) Those tangent lines will be used for creating the "Plane at Angle"s.

3. Create a Plane at Angle using the Southern "L", and set an angle of 100º to match the angle of the cone's wall. (PaA1)

4. Create a similar Plane at Angle (PaA2) for the SSE "L". Same 100º angle.

 

5. Create a sketch on each plane to serve as the reference geometry that we'll be comparing.

At this point, the geometry I've drawn on each sketch has a vertical segment that is co-planar with the original Z axis.

 So far so good.  Like the Plane Tangent to Face at Point - if we create our planes in the right location, and never edit them, we're in good shape.

Now, let's go back to that sketch where we laid out the two "L"s, and change that angle from 45º to 1º.

 And see what happened to the orientation of our reference geometry:

 The rotation has appeared. 

Just for funsies, let's continue:

6. Create another Plane at Angle on the recently moved "L" (#2, the one offset by 1º from the first "L"),

7. create a sketch on it,

8. and copy/draw our reference geometry to it.  I'll draw it this time, rather than copy it, but it doesn't matter - the result is the same.

 

 I've zoomed way in, so that you can see that the geometry for PaA2 and PaA3 share the same origin (all my geometry was drawn/pasted so that the bottom of the construction line is at the local origin), but that the geometry drawn on the sketch attached to PaA2 has been rotated (not sure if the sketch rotated on the plane, or it the plane itself rotated).

Oh, and here's something else interesting; a close up of PaA2 and PaA3 sketches, so you can see how the geometry relates to the local origins.

PaA2 - notice the drawing is square to the origin, but the faded image of the bowl is tilted left.

PaA3 - Still square to local origin, but now the bowl is level, too.

 

That's it for that one. 

EDIT: Forgot to mention, like Plane Tangent to Face at Point, this one is also work-around-able.  If you create the plane in its final location, and don't mess with it afterward, it will be fine.  It's only when you move it or change what it is defined on that you see the problem.

[jsibre]

@jeff_strater 

So, in summary, what you can see in all three of those is that it appears that a plane, if rotated about a cone after creation, actually rotates as if it were on a cylinder.  Geometry (and even the local origin of that plane) are rotated in a way that appears to be attempting to maintain parallelism between the original placement's Y axis and the final placement's Y axis.  I think.  Something like that - I may be getting a bit lost in the detail 🙂

I feel like it's a bit related to the sine of the angle of rotation, multiplied by the angle of the cone's angle (half the angle), but I haven't been able to confirm that.  Calculations contradict it, so far.

All three issues appear to share the same root cause, but the Tangent Plane is the worst offender, because there's no real work around.  If you want a plane that does NOT lie on one of the four options you get by leaving the angle set to 0, you're screwed. 

With the other two cases, you CAN get a serviceable plane if you create it exactly where you want it.  But moving it after the fact (or changing what the plane is based on) will 'corrupt' it, by causing that rotation that there is no way to eliminate.

I hope this is helpful - feel free to ask for more detail, but I'll ask that if you want more evidence, let's get on a zoom call (or similar) and do a screenshare.  I think it's much easier to see live than it is to try to tell the story with words and screenshots - and I'm not sure I could do screencast that wouldn't be boring and jump all over the place 🙂