Paresolid Kernel Interface Error involving Arcs in Geometry

[Scott Lent]

Hi,

 

I'm trying to extrude a solid and keep getting the following error:

 

Parasolid Kernel Interface Error

PK Error Code 547

Non-Manifold Body or Boundary.

 

My shape is essentially a rectangular block with a square notch on one side. The notch has a fillet radius at both inside corners.

If my geometry excludes the fillet radii, the solid is created just fine.

If I attempt to create the solid with fillet radii in the notch, I get the error.

Analyzing the chain reveals no issues (one chain, 8 entities, 0 overlapping entities and 0 direction changes)

 

Since the problem seems to be with the geometry, I've been trying different methods to construct the fillet radii, i.e. arc tangent, and fillet. The results are the same

 

Either way, If I attempt to have Mastercam join the entities which make up the notch (using join entities), the results are always the same 'Selected entities cannot be joined'.

 

Can anybody tell me what I'm doing wrong?

 

Thanks!

[Matthew Hajicek - Singularity]

What version, and can you put up a file?

 

Join is for merging lines to lines or arcs to arcs when they could have been split from the same original entity.  I don't think that's what you want here.

 

Are your entities all at the same Z depth?

[Scott Lent]

Using X9 here

Due to security policies here, I can not send anything out.

All entities are at the same depth. I've even XFORM Projected them to the same plane to to make sure.

[Colin Gilchrist]

Try using Screen > Combine Views. Sometimes an Arc can be on a plane that isn't the same as your extrude plane.

 

As an alternative, extrude the chain without the arcs, and add fillets to the Solid as a Solid Fillet operation.

[jlw™]

Non manifold body usually means your trying to add or subtract to/from a solid in a manner that will leave part of the solid not touching the original or an area of the solid that would be impossibly thin, such as the thickness of a line and is not possible in my experience.  Check that your fillet geo does not overlap any area at all of the line geo.

 

Also, if it works without the fillet, do it then go into solids and fillet there.  Just a thought.

[Scott Lent]

Colin I tried that and it did not work. I understand I could add the fillets using the solid fillet operation, but at this point, it's personal.

 

To simplify things for the sake of discussion, I've drawn a 4 x 4 square on the Z0 plane. I extrude a solid and all is well.

If put a 1" radius in each corner and extrude it I get the error.

I tried creating the square using 'rectangular shapes' and get the same results.

 

I've had problems extruding geometry in the past and managed the work through it after a while, but It's got to be something really stupid I'm doing wrong and I really want to figure it out.

[jlw™]

If you can't share a file can you use SNIPIT and post some pics of the offending geo?

 

 

Non manifold can only be a few things.

 

Go here and search for non-manifold:

 

https://www.plm.automation.siemens.com/en_us/Images/XT_Format_April_2008_tcm1023-62642.pdf

[Scott Lent]

I wish I could. Any File sent from here needs to be cleared by the export compliance office.

 

I have downloaded that pdf and will look it over. Thanks

[Guest]

So, when modeling, one rule of thumb I always follow is to draw EVERYTHING sharp. No arcs. Then I add fillets and radii as necessary. This minimized my modeling issues drastically.

 

JM2CFWIW

[Guest]

So, when modeling, one rule of thumb I always follow is to draw EVERYTHING sharp. No arcs. Then I add fillets and radii as necessary. This minimized my modeling issues drastically.

 

JM2CFWIW

 

That is good model building practice...fillets and rads last

[jlw™]

So, when modeling, one rule of thumb I always follow is to draw EVERYTHING sharp. No arcs. Then I add fillets and radii as necessary. This minimized my modeling issues drastically.

 

JM2CFWIW

 

 

Ditto.  I didn't know there was any other way to do it

[Machineguy]

jlw, Scott,

 There is, but its not the best/correct way of modeling in solids. Foghorn is correct. Radius/ fillets are the last thing you add to a model. Unless you can extrude them in the design.

Modeling is a real art to get good at. Never put yourself in a corner by only doing a part 1 way. Think OUTSIDE the box and leave options to do it differently if something wont work as intended.

By doing it as Foghorn suggests, you can rebuild a model easier when changes are made.

 

 

Machineguy

[Fred @ Slate Industries]

Whatever you draw / model in MC it understands.

It's now " Native".

[Guest]

I usually don't even extrude an arc. The ONLY time I will is if no other fillets blend to it.

One other thing I do that I just remembeted is fillet largest to smallest. Many times going the other way won't fillet.

[jlw™]

^^^^^^ man, sounds like I'm giving a class. ^^^^^^^

[Scott Lent]

Thanks all for your input. I redrew the geometry and after a few tries it worked.

I don't understand why it is so 'buggy'. I've had geometry that extruded just fine, but if I transform that same geometry up or down and extrude, I get the error.

Frustrating, but redrawing using different approaches has always eventually worked for me.

 

Thanks,

Scott

[Guest]

...

I don't understand why it is so 'buggy'. ...

Thanks,

Scott

 

I don't think "buggy" is the correct way to describe it. I think a more accurate term would be "particular" or "precise". JM2CFWIW

 

So, the skinny on solid modeling is that it is FAR more particular how and when things are done as compared to surface modeling. You can do some crazy blends on surfaces that would NEVER fly in a solid modeler. That's why Maya and Rhino still exist. Order of operations like in solving a math equasion is critical.

 

Draw EVERYTHING sharp, add fillets larger to smaller to suit and things will go well almost guaranteed (BTW, there are only 2 guarantees in life; Death and Taxes). Deviate from that and you may be successful, then again, you may not. I go with the high percentages myself. :

 

 

Today's lesson was brought to you by the letter "P", like Parasolid. :

[Colin Gilchrist]

One thing I would check Scott is your Settings in the Planes Manager. There are some "follow rules" that make the Planes change, based on your Graphics View of the part. I strongly dislike these options, and have turned them off in every version of Mastercam I've ever used. They are the "update C/T Plane with graphics view" and "reset Planes to Top in ISO Gview". These options may work great for some people. I think they are a hold over from the "Viewport" days, when people split their graphics screen into 4 quadrants, and had Top, Front, Right, and ISO displayed at the same time.

 

What happens now is that you might have drawn a profile to extrude in the Front Plane, and then you change to ISO, and the extrude fails because that annoying switch is turned on, and all of a sudden your Plane got switched to Top without you realizing.

 

I think the extrude command has some logic built in to essentially ignore the plane settings, and try to figure out what 'natural plane' the chain lies in, so that you don't have to fiddle around with Plane settings while creating Solids. I wonder if something like that could be the culprit here?

 

Also, when you've drawn your geometry, sometimes it isn't as precise as it appears. If your drawing lines and arcs in a system with only 0.00005 precision, and a display/analyze resolution of 0.0001, sometimes it is possible to not have a fillet be truly tangent, if you increase the resolution in Analyze. Say for example that your Arc is off by 0.00004. That is small enough to 'round down' when analyzed at 4 decimal place precision, but might be a large enough error when compounded with other errors in the model, to cause a failure.

 

Say your line is out of place by 0.00004 in the -X direction, and the arc is out 0.00003 in the +X direction. Adding the two together would give a value of 0.00007, and rounding the error now gives you a 0.0001 gap, during the extrude operation.

 

Of course it could also be none of these issues, and the function might have some quirky bug that just rears up from time to time, but I've seen some strange errors like this in my years of using Mastercamy.

 

Part of the reason we can get strange errors like this is because we as humans tend to believe that what we create inside the software is "dimensionally perfect", because that's how we entered the value in the dialog box. The reality is that computers don't store numeric information that way. They use Floating Point precision to be able to take both 'really large' and 'really small' numbers and convert them into a former that stores well in a Memory Address on your computer. So that Arc you placed at X1.25 and Y2.11, might really have the values X1.250000453232, and Y2.110002135439, when stored as a floating point value.

 

Again, none of these things I mentioned could be the source of your issue, but I think it is wise to challenge our assumptions frequently. Despite what we tend to be told, there are very few true absolutes in life. Much of our truths comes down to perception and interpretation...

[Matthew Hajicek - Singularity]

My favorite: positions of "-0".

[Fred @ Slate Industries]

Turn Grid on

Great training wheels

So you know where your actually drawing and on what side ofvthe block etc.

[jlw™]

grid

 

 

 

 

ew, gross

[Fred @ Slate Industries]

Its just like your Planes in SW

Only 99 times more obtusive

 

( Thats a big word for Fred) !

[Colin Gilchrist]

grid

 

 

 

 

ew, gross

 

The grid can actually be really useful. I never use it to "snap" to points, I just use it for "visualization" of where my Z depth is.

 

Alt + G

Make the Grid "visible", but not "active".

I set my XY spacing to like .03, and make the Grid Size between 2" and 50", depending on how large the part is.

 

Now, if you change your Construction (Z) depth, the plane shows you the "height" of your drawing, on whatever your active plane is.