Folks, I'm not sure how you high-end folk get any work done with this program. The mere thought of using this tool (if we are to call it that) for highly complex mechanical design scares the living daylights out of me. The performance and robustness issues it has are almost beyond belief. It's simply unbelievable.
As I write this, I've been sitting in front of the computer in total disbelief watching as SW is has been at 95% CPU utilization for 30+ minutes to delete 100 external references in a part. The part is a simple 1/8 extruded plate. The external references are on-edge references to rounded rectangles. This should take all of 2 seconds, if that much. Break the links and create new geometry to replace it. Or something in that vein. What the hell is it doing at 95% CPU utilization for that long?
I'd be more than happy to email anyone this simple part for a look-see. I could be doing something really stupid. However, if it is that simple to do something "stupid" that causes these kinds of performance problems with SW, well, my current opinion stands: This thing is a dog.
BTW, I got my VAR involved about a week ago. They recommend not using external relations. Well, what the hell is the point of using this sort of a tool if you can't relate features in parts to geometry in other parts? Without external relations, changes in one part would have to be propagated to other parts by hand. What's the sense in doing that?
Anyhow. Frustrated. Productivity shot. Very frustrated.
This is exactly why I dropped SW and went to Wildfire and Unigraphics. SW is terrible on large assemblies and it can't even come close to comparing with Wildfire or UG. I also make a lot more money doing contract work with Wildfire, Pro/E 2001, and UG. Solidworks contractors are a dime a dozen and most of them aren't very good. I still have yet to see larger companies such as Lockheed, National Instruments, Dell, Northrop, etc. use SW, but they are using Pro/Engineer.
Well Martin you seem to have made your mind up, on this one issue. Probably time to move on to some other 3d package, where I'm sure there will be many other "issues".
Generally I've found my Solidworks VAR will take on board any real problems & complaints and are relatively responsive to fixing them. This relativity is compared to previous contact with Autodesk VARS et al.
If that was the sum total of their response - they need a good kick in the posterior rectal orifice.
Maybe deal with another VAR, if you have that option or report direct to SWX on your experience. Our VAR here in New Zealand are very helpful and generally follow through on any problems, until it's solved, or an SPR is lodged with SWX.
I'd love to see real-world examples of complex work done with SW. Something like a car transmission. Something with lots of relationships from part to part.
As I've said in prior posts, I'm more than willing to accept that I'm learning, but I ain't that stupid. It's one thing to not do things optimally, it is an entirely different matter to not be able to use the tools provided due to bad programming and program instability.
My simple project is an example:
I have a keypad with some 50 buttons.
The buttons need to cut holes in the front panel and the chassis (two different parts).
I create a master sketch in the master assembly and use "convert entitites" to bring the button edges to the sketch.
Then use "offset entities" to create the cutout geometry (allowing some clearance for button movement).
This master sketch is then brought into the front panel part via yet another sketch and "convert entities".
An extruded-cut completes the task.
This leads to dog-slow performance and a whole lot of pain.
I've tried it without the intermediate master cutout sketch. No significant gain in performance occured.
Here's another interesting example of something I just ran into:
Create two assemblies. Each of these assemblies has a part called "base". The name is not important, just make it the same in both parts. The "base" part in each assembly should look different. Now, create a third assembly and bring in both of the above. You get an error message because you have two parts named "base". Now matter what you do, the design gets screwed-up. Now, what the hell? Each "base" should be referenced to the assembly it belongs to. Furthermore, if each "base" part lives in a different directory (a requirement if they are different files) the program should be smart enough to know that these are different parts and not replace one with the other. What if you get an assembly from a client and some of the parts are named as yours might be in other assemblies?
If you are dealing with highly complex and intricate designs. Lots of parts. Some that you may not be able to see perhaps. This sort of thing can hose you big time. Maybe SW is optimized for mold-making or plastics design. There seem to be lots of examples of folks doing fine in that domain. Or maybe making a few fancy looking parts for a motorcycle. I don't know. I'd sure love to know if there's a secret way to make this thing get out of my way and let me do my work.
All I want is a 1/2 inch whole and I'm still f***ing around trying to sharpen the damn drill bit I've been handed.
I've been around long enough to know that there are no perfect packages out there. Even software that's been around for a long time (ACAD, for example) has loads of issues). However, at the moment, SW is looking particularly bad.
The trouble with the moving-on scenario is that I don't have unlimited funds and I do have to get work done. Even if I could afford to purchase another package, there's a matter of staring from scratch. This represents weeks of work that I cannot afford to burn. Not something I can even begin to consider at the moment.
My VAR also suggested updating to SP1.1. I'm just in fear of what new maladies this might bring to the table. Yes, I can back out of an upgrade any time I want, but what about work you might have done under SP1.1 by the time a bug or problem is discovered? You simply don't install SP's in the middle of a project.
Look, I'll use SW and will become better at it. However, the issues at hand have nothing to do with becoming better. I'm fairly certain of this after much experimentation. The fact that something as useful as external relations, for example, makes you pay such a penalty as to render this particular tool almost unusable...well, it's a problem that has nothign to do with ability and has everything to do with poor implementation. Yet another example is the one given in a prior post: deleting about 100 external relations took SW to 95% CPU utilization for about 45 minutes!!! I'd love to know how experience would improve upon that.
I'd take issue with anyone who'd claim that this is a solid and robust "production ready" package for complex, non-trivial, mid to large parts-count mechanical design. If I worked for a large company I wouldn't bet my job on choosing this package for mission-critical work. Not for a second.
If you are doing a bunch of designs where each part is done individually and things like moving a hole are done by hand on each and every part rather than through relations...then SW might perform very well. At the moment I simply can't see how one could use what the program claims to be able to do in anything other than very simple designs.
I keep remembering a post from a couple of weeks ago from a fellow who said that his company is switching to SW to work on assemblies of 10K parts. I just don't see how in the world that is even remotely possible. From my current vantage point this looks like a mistake of monumental proportions.
Sorry for being so negative at the moment. I am truly hoping that there's a secret I don't know about that will make some of these issues go away. I have a lot of money and time invested in this, and so, it would be more than great to discover that I am wrong and that SW is a powerhouse of a tool rather than the opposite. We'll see what comes out of my VAR this coming week.
In the meantime, I'm rebuilding all my parts to be "dumb" and forgoing all part-to-part and assembly-to-assembly geometry and/or sketch relationships. I'm using master alignment sketches and planes in the master assembly to locate everything. That will be the only connecting point between the parts and sub-assemblies. The bad news is that, if I modify a part that requies holes (for example) cut in other parts, I'll have to manually go through the affected parts and make the required changes. It seems the only way to ensure a certain level of productivity.
I haven't followed how you got to a situation where breaking 100 relations took 45 minutes.
I don't have lots of time to blow checking, but figured I would just see if a simple 2 plates with 125 relations could be quickly broken.
I made a 6x6 plate with an 11 x 11 array of holes. I mated a 2nd plate in an assembly and created that plate in the assembly with external relations all from the 1st plate. Worked fine in less than 5 minutes.
Then I picked the 2nd plate, chose "List External Refs" and then chose Break All. 2 seconds later it was done.
Thus, there is more to the situation, but it is NOT a problem with basically deleting 100+ relations.
If I had such a problem, I would experiment & then go to my VAR and then seek to have him get input direct from SolidWorks if the VAR couldn't help.
Likelihood is very high that you have something wrong with your SolidWorks setup, or with your system (and of course BlackViper's Windows Services information MIGHT be able to help there). Did you have your Anti-Virus software disabled when you installed SolidWorks?
Once in a while I find SolidWorks will get stuck on something and lock up, but it's relatively rare, and it's always because there is something going on behind the scenes which is not obvious. It may be a corrupted part, but those have been REALLY rare since SW2001+ came out. It's not that I believe SolidWorks to be perfect -- I complain about it as much or more than anyone -- but I find it eminently productive for complex work, including fairly large assemblies. No, it's certainly not a speed demon with large assemblies, but there are ways to optimize your approach to large assembly design that can help a lot. Developing the right work habits is important. One thing, of course, is to NOT create so many external relationships to begin with if you can help it. Certainly you're taking the right approach with deleting those relationships and making the part sketches independent (and fully defined). Also, delete InPlace mates, and constrain part locations manually. Another important thing -- don't create mates to components in a component pattern. There are many others. Refer to Matt Lombard's "Rules of Thumb", as noted in other posts.
I sometimes get frustrated with performance, but your situation sounds quite extreme and not typical. I suspect you have something funky going on with your system/setup. You need to get working with your VAR on this or find a new one who can help.
The company I work for uses SolidEdge and I find it slow to open and close large machine assemblies (500-1000 parts), user interface/buttons are non-distinctive and somewhat cryptic and it takes more clicks to complete a simple task than does SW. SolidEdges translators for STEP, IGES are slow and often can't finish the task given or finish with a significant amount of the data indistinguishable (haven't tried any others yet). The same files in SolidWorks 2004/SP5.0 on average open 6x faster and completes the task.
Getting assemblies to revolve the way you want in SolidEdge is clumsy and requires way to many mouse clicks. The list of measuring tools is long and wordy, SW is simple and quick. Nothing more frustrating than to having to constantly unsurpress part(s) and/or the assembly before they can be measured or revolved about their features. This can be automated through options, but now I have an assembly display that is slower than molasses. Picking parts quickly within an assembly is all but impossible. First you have to let your cursor float near or over the part you want, then three little dots appear, then a list of numbers appear to choose from. It is like not knowing what is behind doors 1, 2,3, 4,5 and 6, but you have to keep picking each door until the right part highlights. Trying to select parts in an assembly with the cursor is all but useless. because more often than not it picks everything but the one the cursor is over.
Bringing in assemblies of automotive components results in the model swinging wildly off the screen while trying to revolve without first picking a vertex or edge. SW allows the user to revolve and zoom about the center regardless of where the origin is. If SolidEdge has something like this, I haven't yet discovered nor has our designers. I'm sure SolidEdge has its strong points in addition to the one you mentioned, but some of the everyday simple tasks are cumbersome and leave much to be desired.
I primarily use the software to view customer files, measure and create drawings as necessary to do my job. So I'm positive I don't know or understand how to use all the features efficiently. The ones I do have some familiarity with are not very impressive when compared to those found in SW.
Please send me the Assembly, Now that I have got my Solidworks to work quicker, this would be a good test for it. Also a copy of you set-up,( use Solidworks tools, copy options), and I will test it side by side with mine. Hopefully I will be able to give you some answers to you current problem. I will also make some RX files for you to look at, to see the time taken on each set-up.
In addition to Circular references, I found a new way to "break" references that flumoxed me until GoEngineer, my VAR, quickly set me straight.
I created two mating round parts in an assembly, where the assembly sketches in the assembly file started everything out. Part Files were edited in the assembly to take those sketches (Convert) & rotate to get the solids.
Later after the whole assembly was finished, I attempted to move one of the original parts, in order to remate to a very small ring surface. That caused the origin on the part to move and basically left the part unrecoverable except by reconstructing the original sketch. I recovered the "original" condition off of my DVD backups.
All of this was not SolidWorks fault. It was basically that I didn't realize the options available to me to take care of the issue at hand.
The strange origin movement also caused SolidWorks to freeze, which probably was SolidWorks fault for not error trapping this odd ocurrance which I created.
Well, as I tell my son "Never give up". I spent all night redoing some of the work to use less external references and more patterning at the part level. This made a huge difference in the workability of the design. Maintaining it is a different matter due to the fact that I now have two parts that have manually-created sketches and features that might require editing every time the parent part is modified. But, it's faster! I have a few more ideas from my VAR today, and might see if I can make the time to test them out.
I am also getting some off-list feedback from a couple of experienced users who were kind enough to take the time to look at my work. One of the comments is that I may be taking assembly sketches and in-context relationships too far. Clearly --and, as I suspected-- the issues are not so black and white. I know that there are performance and reliabilty problems, but I couldn't believe that they'd be so horrific when I see people posting here about 500 to 1000 part assemblies with reasonable frequency. And so, the title of this thread might not be entirely fair.
Maybe SW (or someone) needs to write a realisic study of how to approach medium complexity projects from start to finish. That would be useful. I keep using the car transmission example. Or maybe something like modeling a mobile phone with all of it's internal bits and pieces. Whatever it may be, it should contain a realistic and studied explanation of design approach options a the how's, why's an when's of the choices made.
You're not the first to see flakiness in part origins. A couple of years ago, I was playing with the "shape" command, and managed to *delete* the part origin. Google my post in c.c.s entitled "Goodbye origin" and the replies to it.
I remember you mentioned this approach when you visited a few months ago. After seeing Martin's assembly, and then your comment, I made a small example for Martin that used this technique, and demo'd a few others solutions / techniques to simplify his assembly.
I have to say that I will use that technique more as it really is a much cleaner way to ensure data integrity and design intent. Especially for concept modeling it seems great. What is a little tricky is remembering to un-check the merge solids as one makes features, or ensure you have the right feature scope (assuming one is not simply splitting the part after making one solid). But that just takes getting used to.
It also takes constant vigilance. SolidWorks has trouble keeping track of the feature scope. You can work away on a part for months without trouble, and then one day cherries will sprout all over your feature manager. Eventually you will track it down to one of the features having the wrong or no bodies in the feature scope.
Jerry Steiger Tripod Data Systems "take the garbage out, dear"
ah yes. life is random... should software be different?
And lest we forget the fun alternating extrude direction, or alternating mating condition. love those! Mmmm... Jerry'y cherries! Maybe we can get Ben and Jerry's to make a SW flavor? lemon sorbet with para-solids and Jerry's cherries! ;-)
I found a part which would loose its mates in an assembly. It was a series of mating parts that were amongst the first I did, as a trial run on an "old assembly" to learn on and get a quick Isometric blowup for literature in SWks 2001.
Literally 3 years later this check valve disk would suddenly change position each time it lost mates.
Investigation showed I had left many sketches as undefined in the mating parts and the check valve disk. Once I went through and fully defined everything, the "problem" disappeared.
Thanks for mentioning this one. I think it's a good reminder that fully defined sketches are certainly more desirable in terms of stability. When things are tied down, there is a better chance they won't get up and walk across the screen.
Here's a question for ya. Some SW features create sketches that, upon editing, are not fully defined. Say, a sheetmetal edge-flange, for example. Am I to understand that it is good practice to edit these sketches and fully define them?
PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.