Positioning Constraint slowness problem 2

[xmolddesigner]

I have some questions about assembly constraints. We have only been doing full assemblies of our molds for the past year or so, and had gotten used to the "old-style" Mating Conditions in previous versions of NX, and they seemed to work pretty well. Sure, you could have the occaisional "circular dependency," which would cause some frustration, but deleting a few mating conditions seemed to solve the problem. In October of 2008, we hired an NX instructor to come in for a day and show us some "tips and tricks" to make our assemblies easier to deal with. He showed us a new feature of NX5, the "Positioning Constraints" feature. He showed us how they worked, and they seem simpler and more intuitive to use. He also said they will eliminate about 95% of our circular dependency problems. We thought that sounded awesome, so we converted a few old assemblies to the new style constraints, and any new assemblies we create we use them as well.

The problem we have run into, however, is that the "Positioning Constraints" have almost crippled our systems. They are much slower to use, and making any changes to the constraints seems to take forever, even with "Delay Interpart Update" turned on. I contacted GTAC about this, and was given instructions on how to install a UGII_SIMULTANEOUS_SOLVE_CLEANUP variable to supposedly take care of the problem, but it really didn't seem to do any good. So now, we are faced with several assemblies that are a total pain to change or update, and no way to convert them back to the old Mating Conditions.

So my questions are:

1) What's the "conventional wisdom" on "Positioning Constraints?" Is this what everyone else uses?

2) Has anyone else experienced this condition?

3) Does upgrading to NX6 help improve and/or eliminate this problem? (We are currently at 5.0.6.3)

4) Does 32-bit vs 64-bit make any difference?
(We run NX on the 32-bit version because a few of our older 32-bit machines out-performed our 64-bit systems using "Positioning Constraints")

5) Any other hints/tips/suggestions on dealing with this problem?

Our hardware configuration is: Dell Precision WorkStation T7400, 8GB Ram, Dual Core Xeon processor, 3.3GHz NVIDIA Quadro FX 1700, Windows XP Pro 64 bit OS.

In conclusion, I like the idea of not having circular dependencies, but if it takes twice as long to build an assembly, we haven't really gained anything.

I truly appreciate any help you can provide regarding this matter.

 

[uwam2ie]

... just had problems / trouble with updating partfamilies with NX6.02 - going to nx 603 everything works fine. In NX5 I'd prefer to use the mating conditions. Nx 6 has tool (octopussy) to control the positioning constraints between components. The advantage are the fix and bond conditions.
Performance issues we haven't had sofar.

 

[hudson888]

Your hardware should be okay. I still have a lot of old mating conditions that I use mainly because I know that they work. Yes the constraints do get around circular dependencies but it seems at a cost if you want to constrain a lot of components.

Some people use large assemblies and they constrain or mate every single part. People working with very large assemblies as a rule usually try to model top down with components oriented to the absolute WCS. They often give up on mating/constraining everything because at some point maintaining your assemblies becomes a chore. So they position their components initially using either mates or constrains but then they delete them. Shocking I know but highly practical when you're working in really large assemblies.

Now there are a few problems I have seen with constraints. They are evaluated cumulatively and this will slow down your performance on very large assemblies quite a bit.

The second problem I found by accident was a question on the forum about deformed parts where I could obtain a measurement and then apply it to the deformation using mating but not initially at least using constraints. I think that one was pretty much a rare odd case out so hopefully it can be fixed and would very rarely affect any users.

The one thing that really bugs me with using constraints thus far is that the angular constraint won't do for me what I was easily able to do with mating conditions for my animations. I use such a mating condition with an expression that I vary with the Framenumber expression to rotate shafts. When I attempted this with constraints it seemed to have a tendency to get to 90 180 270 or 360 degrees and then flip the normals in the so the shaft either jumped back to the start or skipped out of phase. This meant that I've stayed with mating conditions because of that problem.

Anyway I hope this helps.

Best Regards

Hudson

www.jamb.com.au

Nil Desperandum illegitimi non carborundum

 

[xmolddesigner]

Thanks guys for the input. We are going to see what NX6.0.3 looks like - just not sure we want to make that move yet.

 

[hudson888]

Not sure if you can use mating instead of constraints for NX-6?

Best Regards

Hudson

www.jamb.com.au

Nil Desperandum illegitimi non carborundum

 

[JohnRBaker]

Yes, you can still use either Mating Conditions or Assembly Constraints through at least NX 7.0 and most likely NX 7.5 as well, but we are seriously considering removing the ability to create new Mating Conditions starting with the next release although we will continue to support the updating and editing of Mating Conditions for several more releases, but they will be reduced to a sort of '2nd class' status.

John R. Baker, P.E.
Product 'Evangelist'
Product Design Solutions
Siemens PLM Software Inc.
Industry Sector
Cypress, CA

http://www.siemens.com/plm

http://www.plmworld.org/museum/

To an Engineer, the glass is twice as big as it needs to be.