How do I stop getting the "Unknown Filleting Failure" error?

[jakewells]

Hi,

Having just upgraded to PowerNurbs 2011, it seems the fillet tool is more tempermental. For the particular piece shown below, I traditionally would expect to be able to fillet the edge in question so long as the fillet size was less than the distance to the nearby edge. However, I keep getting the "Unknown Filleting Error" message. The geometry was created using boolean subtraction applied to two extruded PowerSketches, so I can't see what the problem would be ... any help will be appreciated.

Thanks,
Jake

http://www.jakewells.com/fillet.jpg

[jakewells]

Also note ... collapsing the geometry before applying the fillet does not help.

[nPower_Dave]

Also note ... collapsing the geometry before applying the fillet does not help.

Hi Jake,

I would guess that it is interfering with that other edge (it's very close). Are you sure that the fillet size isn't larger than the distance to that edge? Otherwise, I don't see any possible way that it would fail on a simple fillet of a round hole.

[jakewells]

Hi, I removed the chamfer to eliminate all doubt. You can see below that the circular handle indicates a fillet that is well shorter than the distance to any adjacent edge.

In addition to this, I have been having a lot of difficulty applying PowerFillet to STEP geometry imported using PowerTranslators basic (I consistently get the "unknown filleting failure" error).

I've posted the MAX file at http://www.jakewells.com/fillet.zip should you want to check it out.

Incidentally, it seems PowerNurbs now creates "memory nodes" in hidden layers. What happens when these are deleted?

Thanks,
Jake

http://www.jakewells.com/fillet2.jpg

[jakewells]

UPDATE:

The cylindrical shape that is being subtracted was created by extruding a Pwr_Circle primitive. However, if I create a Pwr_Sketch and use its circle tool to draw the circle (at the same size and position), I can fillet the intersection edge no problem.

I guess I can go with this workaround, but I was really hoping to instance circle geometry (I am adding a bunch of recessed bolt holes to a cover plate and was hoping to preserve parametric editing for hole diameter).

It seems arbitrary that one would work and the other wouldn't, so anything you can do to explain the mystery would be a big help.

Love your product!

[jakewells]

UPDATE 2:

I had been applying a negative extrude to the Pwr_Circle primitive. If I change it to a positive extrude and reposition the geometry accordingly, the fillet also works. So I guess the rule of thumb is:

** When boolean-subtracting geometry that was created by extruding a Pwr_Circle, filleting the intersecting edge is only possible if the extrusion is positive. (However, negative extrusion will work if you're using a circle created within a Pwr-Sketch.) **

A bit arcane, but I can live with it!

[nPower_Dave]

UPDATE:

The cylindrical shape that is being subtracted was created by extruding a Pwr_Circle primitive. However, if I create a Pwr_Sketch and use its circle tool to draw the circle (at the same size and position), I can fillet the intersection edge no problem.

I guess I can go with this workaround, but I was really hoping to instance circle geometry (I am adding a bunch of recessed bolt holes to a cover plate and was hoping to preserve parametric editing for hole diameter).

It seems arbitrary that one would work and the other wouldn't, so anything you can do to explain the mystery would be a big help.

Love your product!

Hi Jake,

Interesting. That case should not fail. We will take a look at it. So are you saying that you used a Max spline curve (circle) to make the hole first? And that fails on the fillet? But if you use one of our sketch curves (like our circle) it works? That would be very interesting. Both should work. By the way, did you try using the Power Detailer since you want to create a parametric array of holes?

[nPower_Dave]

UPDATE 2:

I had been applying a negative extrude to the Pwr_Circle primitive. If I change it to a positive extrude and reposition the geometry accordingly, the fillet also works. So I guess the rule of thumb is:

** When boolean-subtracting geometry that was created by extruding a Pwr_Circle, filleting the intersecting edge is only possible if the extrusion is positive. (However, negative extrusion will work if you're using a circle created within a Pwr-Sketch.) **

A bit arcane, but I can live with it!

Hi Jake,

This is all excellent feedback. It will really help us track down the problem. This is the type of feedback that makes it easy to find and fix problems. We appreciate it!

I think I misunderstood your previous post though. It's the Power Circle extrudes whose fillets are failing. Right? Thanks for the clarification.

[nPower_erik]

Hey Jake,

Your problem is quite a tricky one. I looked into it briefly
and saw that it is a debug/release inconsistency. Debug works perfectly and release throws the error you are seeing.

Additionally I saw that changing the fillet from circular arc to
G1-G3 seems eliminate the error. However, I'm guessing you would be happier with a lighter surface.

I'm going to look into it more but I just wanted to give you some feedback. I'll let you know when I figure anything else out.

Erik Toraason
erik@npowersoftware.com

[jakewells]

Many thanks, guys ... the glitch is easy enough to work around, it just took me a little while to figure out what it was exactly.

And Dave, you are correct ... it's only a negatively extruded Power Circle whose fillets fail when it is boolean-subtracted. I don't think it even fails in all cases (I tested random geometry with some success), but at least in this instance changing the extrusion direction and/or using a Power Sketch circle solved the problem.

Thanks!

[nPower_Dave]

Many thanks, guys ... the glitch is easy enough to work around, it just took me a little while to figure out what it was exactly.

And Dave, you are correct ... it's only a negatively extruded Power Circle whose fillets fail when it is boolean-subtracted. I don't think it even fails in all cases (I tested random geometry with some success), but at least in this instance changing the extrusion direction and/or using a Power Sketch circle solved the problem.

Thanks!

Excellent detective work Sherlock (Jake)! That will definitely help us track down and fix the problem. Then there won't be a glitch to work around any more.

[nPower_Michael]

Incidentally, it seems PowerNurbs now creates "memory nodes" in hidden layers. What happens when these are deleted?

Please be careful when deleting the memory nodes. They are there to work around a Max 2010 issue (I'm not sure what other versions of Max are affected) where deleting nodes would cause instability in the undo system. This instability could accumulate over a long period and eventually cause crashes. Since our plug-ins were creating and deleting nodes more often than most, many of our users were experiencing 'random' crashes on undo/redos after even fairly short periods of time (30-60 mins of heavy work). By 'hiding' the nodes instead of deleting them, we were able to stop these crashes, but at the cost of wasting some memory.

The fun part is, if there is no undo stack, it doesn't impact the user if we mess with and then flush it. So we automatically clean up these 'zombie' nodes from a saved file when it is reloaded (as there is no undo stack when you first load a file), and then re-flush the undo stack after deleting the zombie nodes to prevent any corruption introduced by them. So, if you want to clean up these nodes, just save and reload. You can manually delete them as well, but at the risk of crashing some time down the line, so it would be wise to save after the delete (and maybe even before), and then reload the file to clean up the undo stack. Since that reload would have auto-cleaned the zombie nodes anyway, just doing save+reload is the easiest way to clean up.

Please note, however, that we do maintain an extra layer specifically to hold these nodes, that does stick around through reloads. It is not recommended to delete this layer, or to move nodes in or out of it.