Is there any good CAD software? 2

[EsoEng]

I was trained on PRO/E. Even the tutors didn't have a good grasp of the software. I later used SolidWorks - it's terrible. I've used Inventor - it's terrible. I've used FreeCAD - I didn't stick around long enough to learn how terrible it actually is. Now, I'm using Solid Edge - it's beyond terrible.

The purpose of this thread is two-fold: to complain about the diabolical standard of available general mechanical design software; and to ask is there is actually any that is not terrible.

Solid Edge is the current software that is torturing me. I gave-up on the other programs I mentioned. I will not be using Solid Edge again after I have finished my current project. It is inefficient, it is counter-intuitive; its graphical presentation (of the model being made) is utterly deplorable (it hides errors), and it lacks basic functionality. Things that should be easy and simple are either hopelessly inefficient (requiring esoteric knowledge and a dozen steps where one obvious step would have done), or do not exist.

At this point, I know what you are thinking: that I simply do not know how to use this program properly. Yes, you are right. But then, who does? I'll tell you who does: people trained following a lengthy and involved period of indoctrination (yes, indoctrination), and whom use the software frequently - every day - know - eventually - how to use it without having a mental breakdown each time. Me? I have not had specific training for it (only tuition for PRO/E), and I do not use it every day. What chance do I have? None! Well, maybe if I hired a tutor for a few £thousand and put aside a couple of hours a day to keep practising it, like I put aside time to exercise, use the toilet, eat, and generally respire.

The software is shamefully bad to the point that I can barely comprehend that it was not made the way it is with deliberate intent. Are real-life products actually designed using this garbage software that costs so much money? I am at a loss. I don't think there is any alternative to it. All the CAD software I've used works in very similar ways - just some very bad and others extremely bad. There is nothing good out there, or is there? Is there anything I don't know about? I Google for lists of this software, but nothing looks promising out of returned options I am yet to try (because it's either browser-based or made by the same companies that made the junk I've already used).

Photoshop. This is software I also do not use every day but it is far more intuitive than Solid Edge, and I do not get stuck with Photoshop as I do with SE. Why can't SE be more like PS with respect to basic operation? CAD programmers need to study Photoshop and learn how software can be intuitive and friendly to use.

Solid Edge is beyond frustrating. I actually feel hatred towards the company and people that made it. I wish they hadn't bothered. They have cost me months of my life with their awful, horribly made garbage software. I spit on them.

 

[SwinnyGG]

This speaks volumes to me. Solidworks was pushed hard during my schooling. That's ultimately what I learned most fully. However, I've also been exposed to NX, Inventor, and OnShape. And SolidWorks is the worst of the lot in many ways. Yet it's the one every company I've worked for chooses.

My standout complaint is precisely what you outlined above: SW is flakey. It lets you make whatever you like, but doesn't concern itself with robust relationships. Add in a linear, single-thread, modelling kernel... this is a recipe for disaster whenever you change anything. All the relationships between feature tree items, and dimensions, are explicitly linked to "the exact thing". If you delete a line, and redraw the same line... you will lose relationships and break features. Unless you use "undo" to restore your line. To me, this is beyond stupid for software. SW is parametric only in the sense that you can assign a parameter to vary that line, and update it successfully. But don't you dare delete that line to add a new one. Or surface/face/body. The solver cannot do anything with "new" geometry inserted into an existing feature tree. It requires user input to resolve the changes.

Inventor does better. It seems to have a better handle on relating to "the thing joining the other things" rather than "that one particular thing that is currently joining these other things." You can delete everything in a sketch, redraw it similarly, and things will not break catastrophically. It doesn't care so much if you do something to "this exact face", but rather is happy to do stuff to "the face that occupies this volume". As I understand it, the kernel does not require each feature to be generated completely prior to evaluating other features. To me, this is a better way to 3D model. It's not so dependent on maintaining arbitrary exact relationships.

But to refer back to OP... cheap, fast, good. Chose 2. Same as always. There's no such thing as perfect CAD software, but if you throw more money and training time at any of them, they will serve you better.

This is exactly what I'm talking about in my other comment.

If you don't understand or choose to follow the simple best practices of parametric modeling, you're going to generate bad models.

Manual control of relationships between features is, to me, a feature and not a bug. If I have a model built and I need to insert a variable feature to control something after the fact, I WANT to be making all of the decisions about how the relationships are built- in some systems the software just infers what relationships you want and going back in to modify them is more time consuming than just assigning them correctly in the first place.

 

[Brian Malone]

I have not found Solidworks to be as fraught as some have described. I cut my parametric modeling teeth on Pro-E 13-17 then 2000i before working with Soliworks, Inventor, Solid Edge, and NX. The key with Pro-E was minimizing parent-child relationships and try to use datums and datum axes as much as possible, and add in chamfers and rounds as late in the model creation as possible to limit tying geometry to the edge or surface of a feature that is not function-critical. This helped minimize what was called the 'measles' if a feature was suspressed or deleted and Pro flagged all failing features with red dots. I still model with the same methodology with the newer programs I have used since Pro and I do not find them to be particularly fragile. I use surfaces extensively in Solidworks and initially found its display less then useful compared to Pro-E (Pro having two color for insde/outsde of surface), but I have gotten used to SW surfaces and find the features fine. I have found Solidworks and Inventor each can be a nightmare if a proper modeling scheme is not used. I have had to untangle models that were constructed in a seemingly random patchwork fashion that would be more of an indictment of the skill of the engineer/designer than the software. And that lack of modeling skill could have been driven primarily by the guidance given to them when they were learning the program. In the sketcher mode, I like the auto constraints chosen by Solidworks better than those chosen by Inventor. But I spend 90% of my time running Solidworks and 10% on Inventor

 

[3DDave]

It seems to me to be a problem if a model can be changed so that afterwards the relationships that were put in place are quietly deleted leaving a disconnected pile of parts that are simply co-located.

I am reminded that while PTC tended to encourage users to fix problems when they arose, many co-workers believed in telling the software to ignore the problems and freeze the location of the unconstrained items, sometimes leading to parts being embedded in each other or some other problem creating result.

 

[Brian Malone]

In Solidworks, in an assembly, components can be fixed with no constraint to any other component and that will certainly cause grief - Inventor will let a component be inserted without hard constraints, too. So isn't this fragility an issue of use/user methodology than a failing of the software?

Occasionally, I will see Solidworks or Inventor create a poor section representation in drawing file for conponents that are physically intermeshed/press-fit but fortunately that is infrequent and for the use-case for the company I work for, that quirk is not a showstopper.

I send STEP model files and PDF drawings to all the machine shops I work with and I cannot recall having any issues with part errors due to poor geometry definitions. Their CAM programs interpret the exports from Solidworks and Inventor equally well.

I, too, originally started on a board with hand drafting, so I find no major fault with any CAD program.

 

[Brian Malone]

The older Pro-E versions required fully constrained and defined sketches before it would attempt to generate geometry which could be considered a proper algorithm but most modelers now don't enforce full definition, the sketch just needs to meet the requirements of the feature - extrusion needs a closed polygon/region, a surface needs a single continuous segment or closed polygon/region, etc. This flexibility is good because quick spatial envelopes can be defined and developed without getting held up on fine details. As long as the pitfalls of not going back and fully constraining/dimensioning the geometry is understood then there should not be an issue with a modeler having less strict controls. Just my 2 cents . . .

 

[SwinnyGG]

So isn't this fragility an issue of use/user methodology than a failing of the software?

YES! Exactly.

"This software sucks because it will let me hang myself if I use bad basic modeling practices" is, to me, an asinine thing to complain about.

 

[IRstuff]

"This software sucks because it will let me hang myself if I use bad basic modeling practices" is, to me, an asinine thing to complain about.

Dunning-Krueger? Although, there are claims that it's not real.

Nevertheless, it may simply be that despite one's desires, one's brain simply can't walk "the walk". Sometimes, people have to recognize that their brains simply aren't wired for what they want do. Dilbert referred to "it" as the "knack."

 

[Brian Malone]

Regardless of any training taken, tutorials used, books/references read, the expertise with any CAD program comes from using it. Us old dogs called it putting in 'tube time' prior to the advent of the LED flatscreen world of today. You won't get the knack if you don't dive in.

 

[btrueblood]

I and some colleagues inherited a SolidWorks design for a new machine. All of the sheet metal was developed in a top-down method: the cabinet was outlined and sheet metal parts were added, constrained in the assembly file to make things fit. I.e. the width of a panel, or the distance of a bolt hole to an edge, was not defined in the individual part file, but in the top assembly file. It was a fast way to design to a spec., but was confusing to work with. Opening that giant assembly file took seemingly hours, and then scrolling through a list of dimensions to find the one you wanted to change took a lot of time. And...once you changed that one dimension...all of the myriad linked dimensions on dozens of other child parts also changed, automatically, without you knowing it, unless you looked, painstakingly, at all of the other drawings too.

Again, not a failure of the software, nor really a failure at all. Until you go to production and the shop starts sending ECRs over the wall, the original designer is gone and the draftsmen are pulling their hair out because all of the drawing and part files are breaking. Fun times.

 

[ctopher]

I currently work with a couple "ACAD wired brain" engineers. They can't understand the concept of SolidWorks drawings referencing 3D models.
They will try their best to make the models, then try to draw the part onto a drawing.
Some people are not hard-wired to see 3D, or even understand how files are managed in general.

 

[3DDave]

.
My company got grafted (think Dr. Frankenstein) onto a factory where they demanded that having a reference trimetric on complicated weldments was unacceptable. For a long time it was unclear why, and resistance to it was rather hostile, considering that such views took mere seconds to add and were always (being of the 3D model) accurate.

It finally came out that some guy in that factory had a niche - creating 2D constructions of trimetrics using ACAD. That half of that factory management shared the same last name should be no surprise. Imagine having years of experience and spending days of effort being replaced with "INSERT VIEW; TRIMETRIC; click"

 

[drawoh]

I and some colleagues inherited a SolidWorks design for a new machine. All of the sheet metal was developed in a top-down method: the cabinet was outlined and sheet metal parts were added, constrained in the assembly file to make things fit. I.e. the width of a panel, or the distance of a bolt hole to an edge, was not defined in the individual part file, but in the top assembly file. It was a fast way to design to a spec., but was confusing to work with. Opening that giant assembly file took seemingly hours, and then scrolling through a list of dimensions to find the one you wanted to change took a lot of time. And...once you changed that one dimension...all of the myriad linked dimensions on dozens of other child parts also changed, automatically, without you knowing it, unless you looked, painstakingly, at all of the other drawings too.

Again, not a failure of the software, nor really a failure at all. Until you go to production and the shop starts sending ECRs over the wall, the original designer is gone and the draftsmen are pulling their hair out because all of the drawing and part files are breaking. Fun times.

When I finalize a SolidWorks fabrication drawing, I go into the assembly model, and I convert all the assembly level parametric constraints to constraints local to the parts. This is easy to do if you have already done your fabrication drawings. Parametric modeling is a fabulous design tool, and a clusterf**k in manufacturing. Once you move a part into manufacturing, you cannot change form, fit and function, anyway.

In parametric modeling, you can set up a fabrication drawing that changes dimensions depending on what other files you have loaded. The possibilities are endless, with the exception of successful, quality manufacturing. 3D parametric CAD is not the least bit idiot proof.

Before any sort of manufacturing, you have to freeze your design.

 

[ctopher]

If I try top down design, users that are not properly trained, or don't care, will screw it up. Happened too often.
Unless everything is trained and at the same level, it's bottom up design for me.

 

[dgeesaman]

When I finalize a SolidWorks fabrication drawing, I go into the assembly model, and I convert all the assembly level parametric constraints to constraints local to the parts. This is easy to do if you have already done your fabrication drawings. Parametric modeling is a fabulous design tool, and a clusterf**k in manufacturing. Once you move a part into manufacturing, you cannot change form, fit and function, anyway.

In parametric modeling, you can set up a fabrication drawing that changes dimensions depending on what other files you have loaded. The possibilities are endless, with the exception of successful, quality manufacturing. 3D parametric CAD is not the least bit idiot proof.

Before any sort of manufacturing, you have to freeze your design.

I think that's the thing that I find frustrating with the latest parametric software - there are 100x more ways to walk away from a complete drawing and supporting models that are put together such that revising and re-using the underlying data is effectively impossible. The parametricity was only mildly leveraged during the initial design process.

My last 20 years is at a company where we do light custom work on top of our standard products. So for example we design job-specific machine adapters when one of the standard adapter options won't work. The adapter models and drawings that came from the product development team might regenerate and make a clean drawing, but if you attempt to copy into a new version and change dimensions you'd often get a mess and wish you'd started from scratch. The customs team sometimes built 'better' models and drawings that were reliable for being re-used and edited. Product development never had a requirement to provide clean, re-usable, revisable, model data and it showed. Releasing assemblies as-is that are related with top-down and circular relationships is truly passing down a nightmare.

I've also had time in this same industry working with teams who re-use almost nothing. Either the drawing and model comes from a programmed automation or it's made from scratch. Having decent parametrically-driven template models and drawings can serve both needs very well but they just don't do it that way and continue to struggle with both. <shrug>

 

[3DDave]

If I try top down design, users that are not properly trained, or don't carey, will screw it up. Happened too often.
Unless everything is trained and at the same level, it's bottom up design for me.

Yup. They can take 10X as long to go the hard way, but training is completely out of the question when they know the bare minimum to get by.

They do have one point - if they get done sooner, no one really cares. If they can bitch about how tough the job is then management just takes their word for it and avoids having to manage them. So they get rewarded for sloppy work and it's clear that when one rewards sloppy work, they get sloppy work.

Edit to include ctopher's later response:

It is also the case that the less capable the staff are the less difficult it is to find replacements. To amplify - it's not just that managers don't know the difference, it is that managers don't even want to know the difference. One can see the fear in their eyes when the subject is mentioned.

 

[ctopher]

Yes, they get rewarded either way, as long as the job is completed. Most managers don't know the difference.