I have heard conflicting opinions on using mates vs fixed mates.
There definitely are some conflicting opinions on this topic. It's a very similar argument when people decide to use the Fixed relation (the little anchor icon) inside a sketch. The problem is that when you fix geometry in a parametric modeling software, you're really defeating the purpose of the "parametric" name. When you fix geometry there is no logic involved in the definition of its location. Just like relations and dimensions allow geometry to intelligently react when other changes are made in sketches, mates allow parts to react when other parts move.
When you just fix geometry, it's just fixed in the random location it was last left in. No matter what changes are made to your assembly, the fixed parts are always going to stay exactly where they were. The biggest argument is that this is not a true simulation of how things are assembled in real life. It defies logic, and it ignores the benefits of using a parametric modeler.
I do have a lot of assemblies with both mated and fixed components.
Each has a specific use, benefits and drawbacks-
Mates are great for where you expect parts to move, or change shape, length etc . The downside is that they will add to the assembly rebuild time, especially if there are changes, or rebuild errors, that violate the mated faces, holes, etc
Fixed parts are just that- fixed in space- and will not move even if a "mounting pioint or screw boss is removed, moved, etc. or if you accidentally grab and drag the part by mistake.
I find that once I am happy with size, position etc of a part within a complex assembly, I may often "lock the part" out- so that it does not slow down the rebuilding of what I am working on (elsewhere in the same assembly). I do this by fixing the part in the assy, and sometimes even swopping it out for a parasolid "dumb" copy (if it is a large complex part).
The downside of doing this is that you will loose your references to the other parts in the assembly, and the affected parts will not update when you (or someone else) change something, and you will not get the red rebuild error flags to alert you to what has fallen over- but on the other hand you can significantly cut down on wasted rebuild time-
"Fixing" a part will suppress all the mates that locate that part in space.
If you choose, you can delete these suppressed mates. I sometimes do this to clarify the remaining mates - useful when you are using "view mates" to track down a mate that does not behave as expected.
Remember that you can always "float" a part again, and add or change the mates.
To clarify, I'm most interested on the performance aspect of this discussion. If you fix components (other than the first part), yes you get less mates, and less rebuild time, but do you also get a "tugging" effect in that the assembly fights itself. I get the impression that having mixed fixed parts is detrimental to the assembly performance. If you have no moving parts, and you Fix some parts, should you fix all the parts? Is the mate read time the main issue to large assembly performance (other than lightweight mode and part suppression). Can (mixed) fixed parts destabilize the assembly?
To broaden this discussion - What is the most optimal mating scheme?Lets assume you have a maxed out 32 bit system, and no more hardware/OS options.
How about mates like symmetric vs in place on planes? - I have seen assemblies where the user makes a ridiculous number of planes, with the thinking that planes are the most stable of all. Does the added feature outweigh the gains of the plane?
How does changed parts blow up the asembly mates? - Obviously if you mated to an edge, and the part change removed or renamed the edge, the mate will blow up. But some argue that any change to a face can cause the mate to blow up?
I think you are going to need to experiment for yourself with the models you have at your company. There are many variables involved and without any of us knowing what your companies modeling techniques are, or what you even design, it is hard to offer advice that will fit your situation.
There are general rules of thumbs, but they may not work for your companies modeling practices.
No matter what if you are having problems now with x32 bit systems it is not going to get much better in the future. I think you will need to bite the bullet and upgrade your hardware and operating system to match the desires of what you wish your software package to do for you and the size of models you create.
I would also get with your VAR and have them work with you to help you with your company modeling practices. If they can't do it consider bringing in a consultant to work with you to see if you can make improvements to your techniques when modeling.
With the number of posts you have created the last few days sounds like you are having performance and data managment issues.
Can you share with us your current computer hardware and the size and types of modeling you do. What is the engineering environment you are working in? Complete and detailed information will allow us to give you better, more targeted help.
I love doing animations, however the mates all seem to "add up" so what I do is create my models and assemblies with all the mates to define it.
Then I create subassemblies of the moving sections and fix everything in those subassemblies deleting all the original mates afterwards...that way the only actual mates that require computing are the useful new ones between the subs.
Other than that mates always trump fixing in almost every case in my opinion.
Retrieving data ...