CAD Forum

I'm teaching engineering apprentices SW and need a list of top 10 most common mistakes newbies make in the core solid modelling - things like not placing the model central to the default planes, not understanding geometric sketch constraints, etc. , Cheers.

I would say things like:
- trying to make SW work like some other program, like Acad or Inventor
- not setting up a good file naming system - newbies probably need experienced help for this
- not getting professional help setting up Toolbox, templates, libraries, favorites
- becoming too reliant on configurations
- becoming too reliant on in-context
- becoming fearful of in-context without really understanding it
- becoming too reliant on multi-body techniques
- making sketches too busy
- using sketch patterns instead of feature patterns
- making extensive use of "regions"
- making sketch fillets/chamfers instead of feature fillets
- deleting things and recreating instead of editing (learn the editing tools)
- hack and whack modeling - using cut features to erase mistakes.
- watching too many youtube videos and assuming because it's on the internet it must be correct
- assuming there is only one way to do everything
- assuming there is a "right way" for every situation
- being impatient with file management
- not assigning a CAD Administrator
- assuming that it's best to use model dimensions on your drawing
- not having a real plan for model dimensions
- assuming you can do great CAD without understanding the manufacturing process that you are designing for

In addition to what @matt said:

-thinking "I have it all figured out"
-not listening to co-workers
-not listening to other, more experienced users (like here)
-not listening to floor personnel who have forgotten more about manufacturing than you and I together will ever know

Attitude makes a HUGE difference, probably more than understanding all of the finer points of CAD.

Misunderstanding the controlling relevance of context.

There are tools and commands for parts, others for assemblies, and others for drawings. Trying to do something outside your context will confuse the user: "Why is this command greyed out?"
Furthermore, the "toggle" contexts: while editing a sketch, when editing a component in assembly context, when making an exploded view, or when editing the sheet format, you are unable to do unrelated things until you close the active toggle context.
In all cases, there are usually relevant icons in the workspace or the tree to identify your current context.

matt wrote: ↑Fri Jan 28, 2022 10:11 am - using sketch patterns instead of feature patterns

- being afraid to use sketch patterns

Not taking the time up front to look at the basics and just jumping in and fumbling through it. This maybe more important if you've used other software.

Review what the mouse buttons do, the meanings of icons ect.

Randomly placing the first part in an assembly because you don't know to hit the green checkmark and fix it at the origin.

-Using assembly mates as a way to validate modeling precision rather than a tool for restricting degrees of freedom of components.

Using the Swamp to find answers?
Almost put that in purple...then thought better of it, because for once that's not sarcasm...

Using fasteners with screw threads.

bentlybobcat wrote: ↑Fri Jan 28, 2022 1:39 pm Using fasteners with screw threads.

Failing to defeature or roll back Mcmaster or other parts, in general.

1. Starting a model with the primary planes on a corner instead of center.

2. Dimensioning everything in a sketch instead of minimizing dimensions by using construction lines and relations.

3. Failing to keep sketches simple.

4. Using sketch patterns (sorry @Dwight and @CarrieIves) and not Mirroring sketch entities.

bentlybobcat wrote: ↑Fri Jan 28, 2022 1:39 pm Using fasteners with screw threads.

Something that I realized way too late:
Take the time to optimize your imported parts. Even the small ones like fasteners.
Put the main planes and the origin where they are useful, add reference geometry (& mate references if you want to), freeze features, add the correct weight, etc.

You don't ever want to touch those screws/imported parts again!

Changing something later messes with EVERY single assembly they are in.

I had so many headaches already with this - and so much time wasted.

Glenn Schroeder wrote: ↑Fri Jan 28, 2022 2:16 pm 1. Starting a model with the primary planes on a corner instead of center.
[...]

That depends heavily @Glenn Schroeder , I think.

berg_lauritz wrote: ↑Fri Jan 28, 2022 2:24 pm That depends heavily @Glenn Schroeder , I think.

Might be safer to say start with some relation to a stable reference. That reference might be the standard planes, or a reference / layout sketch, reference planes, etc.

I used to spell out the most stable objects in the CAD model for people. The SolidWorks training has always assumed that layers and layers of daisy chained references were always going to magically be stable. But it turns out that not all entities are created equally stable.

The "Rings of Fire" called out stable features. Slightly different idea, but same concept.

mattpeneguy wrote: ↑Fri Jan 28, 2022 2:10 pm Failing to defeature or roll back Mcmaster or other parts, in general.

I do a Parasolid wash (save the part as a parasolid and then reopen it and save over the file) right after suppressing extra erroneous features (like threads) on all McMaster-Carr parts (and similar vendor parts). This is because there is generally extra information in them that can slow down your assemblies when you have hundreds of fasteners in an assembly.

DanPihlaja wrote: ↑Fri Jan 28, 2022 2:56 pm I do a Parasolid wash (save the part as a parasolid and then reopen it and save over the file) right after suppressing extra erroneous features (like threads) on all McMaster-Carr parts (and similar vendor parts). This is because there is generally extra information in them that can slow down your assemblies when you have hundreds of fasteners in an assembly.

I have lately been using tools>insert>imported after the parasolid wash and/or if a part really doesn't need a parasolid wash.
Insert>imported makes sure you have clean custom properties as well.

berg_lauritz wrote: ↑Fri Jan 28, 2022 2:24 pm That depends heavily @Glenn Schroeder , I think.

There are at least two reasons to have at least two of the primary planes at the center of the model.

1. The most obvious is so you can take advantage of Mirror features, both in Parts and Assemblies.

2. Less obvious, often design intent makes it easier to edit later if that's the case.

Glenn Schroeder wrote: ↑Fri Jan 28, 2022 2:16 pm 1. Starting a model with the primary planes on a corner instead of center.

Glenn

I am currently working on a best practices presentation for our project modeling and am trying to articulate where a designer should put primary planes. I think Matt's idea that stability is the most important criteria is dead on. I am coming to a few conclusions as I create the presentation.

For our products, which are mostly floor-standing and benchtop instruments with a lot of user interaction, their relationship with the user is the most stable relationship. That's kind of funny, as I think of users as being the most variable and unpredictable part of the project, but it really seems to be the right approach for locating the top level assembly.

"Top" plane - the working height of the user interface is a critical aspect, so either the floor or the benchtop should be the "top" plane of our top level assembly. The instrument may evolve to be a little taller or shorter, and clearance under the bottom might grow or shrink, but we don't want internal components of the instrument to rise or drop once we have the human interface set.

"Right" plane - generally, users work at the center of an instrument, and that's where the "right" plane goes for the top level assembly. If sometime we did happen to design an instrument with an off-center interface, I'd center the "right" plane on the user, not the instrument. From there, the instrument might get wider or narrower, but most internal components shouldn't wander left and right if it does.

"Front" plane - lastly, we should establish a front face of the instrument that defines the depth of the user interface. If the front of the instrument is actually flat, then the choice is easy. If it's irregular, then the choice might be arbitrary, but that's still where our "front" plane goes.

That done, now the instrument can shrink (not likely) or grow without upsetting things too much. Add a few mm to the back to fit some more stuff in and generally things don't shift in ways that are a probllem.

Once the primary planes for the top level assembly are established, it is easier to set primary planes for components. Often you don't know what the mounting surfaces are going to be, so the functional location may be more stable. If you know the component is going to mounted on the right wall and move with it when the instrument grows, then model it's "right" plane on the wall. If it's centered on the user, as most things are, then put the "right plane" on the center.

Components with a stable orientation should be modeled as oriented, not flipped backwards or upside-down. It amazes me how many models I see that are not oriented correctly. It makes them harder to edit, more prone to mistakes.

Some components, though, may be used in many places in different orientations, so the instrument orientation doesn't matter. In that case, block-like components get modeled with their "front" plane as their mounting surface, built in positive Z, and probably centered for their other primary planes, though it depends on what seems most stable. Similarly, cylinder-like components should be concentric to Z and built in positive Z.

The side walls of our frames are tricky to locate in space. There is a great temptation to center the primary planes, and that usually creates problems. You are always going to have to tweak the location of the front edge, probably grow the instrument depth some, and it turns out that the center wanders something awful, and you don't want all the components you've mounted to the side wall to wander with it. So for the big frame panels, you build

side panels with "front" = instrument front and "top" = floor;
top and bottom panels with "front" = instrument front and "right" = centered; and
front and rear panels with "top" = floor and "right" centered.

I agree completely that usually a corner is the most unstable thing you can pick.

This post is a bigger than I usually want to read, so sorry for that. It did, though, help me figure out how I am going to present this internally.

Dwight

@Dwight,

It sounds like you've put a lot of thought into that, and for your industry it sounds like you've worked out the best way to approach it. I'm involved with modeling roadside signs, bridge rails, guardrails, etc.

So my Assemblies are generally modeled with the Top Plane at ground level. The same is true for Parts that reach the ground, like posts and concrete. Other components, like guardrail, hardware, and sign panels are modeled with the Top Plane centered, mostly because it makes mating so much easier.

Pretty much all my Parts, and most Assemblies, are centered on the Right Plane, both for mating and mirroring features and bodies in Parts, and components in Assemblies.

The Front Plane is generally centered on Assemblies, but it's not uncommon for it to be the front face, depending on the specific circumstances.

Glenn Schroeder wrote: ↑Mon Jan 31, 2022 10:12 am I'm involved with modeling roadside signs, bridge rails, guardrails, etc.

I like your approach to primary planes.

As a complete digression, I happened to come across the MUTCD and think it's amazing. What a well considered design project! I read the whole thing just because I was fascinated. My favorite is the requirements:

To be effective, a traffic control device should meet five basic requirements:
A. Fulfill a need;
B. Command attention;
C. Convey a clear, simple meaning;
D. Command respect from road users; and
E. Give adequate time for proper response.

Command respect! I never thought of that. Now I use that in all my project critiques.

Dwight

If you want an example of a TCD that does not command respect, search youtube for 11foot8.

One thing that seems to be missing with fresh CAD users is the concept and significance of "Model Scope".

I understand that nobody has a crystal ball so the future is unknown, but we can make some assumptions based on what has been. New users don't have the luxury of the experience of trying to modify a hack job of a model or had to fix several mates in several dozen assemblies and reattach annotations in just as many drawings. At least give some consideration to the life expectancy of your work, how likely is it to be copied for a similar part and what assemblies might use it. Then try to have an awareness of the implications and adjust your modeling methods accordingly. The fastest or easiest way to model something is not always the best method.

SPerman wrote: ↑Mon Jan 31, 2022 12:02 pm If you want an example of a TCD that does not command respect, search youtube for 11foot8.

Love me some "Can-Opener" video. Although it's not quite so amusing since they lowered the roadbed.

Glenn Schroeder wrote: ↑Mon Jan 31, 2022 10:12 am [...] I'm involved with modeling roadside signs, bridge rails, guardrails, etc.
[...]
Pretty much all my Parts, and most Assemblies, are centered on the Right Plane, both for mating and mirroring features and bodies in Parts, and components in Assemblies.
[...]

This makes total sense. Your general idea of parts is to be fairly symmetrical & I don't disagree with that general idea.

We (i.e.) have a lot of parts where the right plane (aka your mirror plane) does not make sense at all. - the parts are often asymmetrical & having the right plane i.e. at one side is more beneficial for mating/adjusting.


Cannot upload a picture currently. I'll try to add one later - I get a http error all the time.

berg_lauritz wrote: ↑Mon Jan 31, 2022 2:04 pm This makes total sense. Your general idea of parts is to be fairly symmetrical & I don't disagree with that general idea.

We (i.e.) have a lot of parts where the right plane (aka your mirror plane) does not make sense at all. - the parts are often asymmetrical & having the right plane i.e. at one side is more beneficial for mating/adjusting.


Cannot upload a picture currently. I'll try to add one later - I get a http error all the time.

I've been getting that all day too. @matt did a system update this weekend (https://www.cadforum.net/viewtopic.php?t=1659). Maybe loading pics got broke?

Managing data is probably the biggest issue for new/old users. New users become old ones pretty quick.
I'm not a big advocate for fully constrained sketches... imho, too much time is wasted there but teachers luv making it a issue.. it's not.
Although, creating layout sketches early and getting their bearings in how to manage the concept to move forward.
Probably letting them know they made a really bad feature/model and teaching them failure/acceptance and when to cut line and redo it or saveas a parasolid and get work done.
Have them embrace bodies and surfaces for reference geometry or again knowing when to cut line and break in context to get work done.

Glenn Schroeder wrote: ↑Mon Jan 31, 2022 2:51 pm I've been getting that all day too. @matt did a system update this weekend (https://www.cadforum.net/viewtopic.php?t=1659). Maybe loading pics got broke?

Pix should be ok. I added one earlier, here's another... oops. Http error. Give me a minute....


**Moving this error discussion to the What's Wrong thread. viewtopic.php?t=1664

zxys001 wrote: ↑Mon Jan 31, 2022 3:19 pm Managing data is probably the biggest issue for new/old users. New users become old ones pretty quick.
I'm not a big advocate for fully constrained sketches... imho, too much time is wasted there but teachers luv making it a issue.. it's not.
Although, creating layout sketches early and getting their bearings in how to manage the concept to move forward.
Probably letting them know they made a really bad feature/model and teaching them failure/acceptance and when to cut line and redo it or saveas a parasolid and get work done.
Have them embrace bodies and surfaces for reference geometry or again knowing when to cut line and break in context to get work done.

Paul,

With all due respect, it can be a big issue. When I took the SW Essentials course at our VAR years ago I remember the instructor telling us he knew of a situation where someone failed to completely constrain a sketch. Something got moved a little, or something happened that wouldn't have happened if the sketch was fully constrained (it was at least 10 years ago so I don't remember all the specifics), and it changed the Part just a little bit. It wasn't caught until the part went into production.

It cost the company several hundred thousand dollars, and cost the SW user his job.

I will agree that not every sketch element needs to be fully constrained, but new users won't recognize those situations.

Glenn Schroeder wrote: ↑Mon Jan 31, 2022 3:53 pm Paul,

With all due respect, it can be a big issue. When I took the SW Essentials course at our VAR years ago I remember the instructor telling us he knew of a situation where someone failed to completely constrain a sketch. Something got moved a little, or something happened that wouldn't have happened if the sketch was fully constrained (it was at least 10 years ago so I don't remember all the specifics), and it changed the Part just a little bit. It wasn't caught until the part went into production.

It cost the company several hundred thousand dollars, and cost the SW user his job.

I will agree that not every sketch element needs to be fully constrained, but new users won't recognize those situations.

Totally understand... but,.. this goes in other ways as well, say if your company/users only do direct modeling (no sketches).. catching geometry/fit errors is part of the job,.. and mistakes happen. So, just because a sketch (s) can be constrained does not mean they are correct or will the prt/asm be without errors.

Glenn Schroeder wrote: ↑Mon Jan 31, 2022 3:53 pm Paul,

With all due respect, it can be a big issue. When I took the SW Essentials course at our VAR years ago I remember the instructor telling us he knew of a situation where someone failed to completely constrain a sketch. Something got moved a little, or something happened that wouldn't have happened if the sketch was fully constrained (it was at least 10 years ago so I don't remember all the specifics), and it changed the Part just a little bit. It wasn't caught until the part went into production.

It cost the company several hundred thousand dollars, and cost the SW user his job.

I will agree that not every sketch element needs to be fully constrained, but new users won't recognize those situations.

Like just about anything else in SWX (or any CAD software, for that matter)........it depends.

Anyhow, we're really getting into the weeds now.

So, here's one related to the last few posts...

If your sketch is so complicated that fully constraining it is difficult, it's probably a bad sketch.

A little bit late to the party...

-- Trying to cramp everything in a single feature/sketch

I had come across a lot folks who "boast" that they can create a part with just a single feature/extrude/revolve.

If you need to use multiple feature (eg: extrude cut after creating a extrude to create a hole) so be it, provided they are logical.

Zhen-Wei Tee wrote: ↑Sat Feb 12, 2022 12:13 am A little bit late to the party...

-- Trying to cramp everything in a single feature/sketch

I had come across a lot folks who "boast" that they can create a part with just a single feature/extrude/revolve.

If you need to use multiple feature (eg: extrude cut after creating a extrude to create a hole) so be it, provided they are logical.

Good point. This notion can go both ways though. From what I've seen this comes from people that try apply "best practices" they've seen somewhere but do not understand why. If "avoid using extra features" is good then the metric is fewest number of features; no. We have people that do the same thing and boast about it then complain when someone uses a few extra features. On the other side there's people that are oblivious to what they are doing and will just tab, flange, hack, repeat until they wander into the shape they need.
Both extremes wind up being a bad model.

bnemec wrote: ↑Mon Feb 14, 2022 9:31 am Good point. This notion can go both ways though. From what I've seen this comes from people that try apply "best practices" they've seen somewhere but do not understand why. If "avoid using extra features" is good then the metric is fewest number of features; no. We have people that do the same thing and boast about it then complain when someone uses a few extra features. On the other side there's people that are oblivious to what they are doing and will just tab, flange, hack, repeat until they wander into the shape they need.
Both extremes wind up being a bad model.

I know what you mean. For years I've been using a stand-alone sketch to drive my Hole Wizard features. It just seems simpler to already have a sketch, so when I get to the second page of the Hole Wizard property manager all I have to do is drop points on the existing sketch and it's good to go.

I've had people criticize that method because it puts an extra feature in the tree. My reply is "So what?"

Glenn Schroeder wrote: ↑Mon Feb 14, 2022 9:46 am I know what you mean. For years I've been using a stand-alone sketch to drive my Hole Wizard features. It just seems simpler to already have a sketch, so when I get to the second page of the Hole Wizard property manager all I have to do is drop points on the existing sketch and it's good to go.

I've had people criticize that method because it puts an extra feature in the tree. My reply is "So what?"

I tend to make sketches that contain the defining features first. Such as mounting hole patterns, required bosses etc. Then a sketch after those that will define the first solid feature. Even this principle needs moderation; obviously there are many cases where it's better to place sketch on a face of the solid. I just struggle the with practice of starting out by extruding a random sized box or cylinder or tab just to have some shape to work with. That workflow seems backwards to me when the overall shape is usually defined by the features such as attachment points or whatever.

Edit: If you don't like using history-based modeling to your benefit then I highly recommend Solid Edge's Sync Tech.

Zhen-Wei Tee wrote: ↑Sat Feb 12, 2022 12:13 am A little bit late to the party...

-- Trying to cramp everything in a single feature/sketch

I had come across a lot folks who "boast" that they can create a part with just a single feature/extrude/revolve.

If you need to use multiple feature (eg: extrude cut after creating a extrude to create a hole) so be it, provided they are logical.

On the other hand, a first feature that's too sparse doesn't work for us. We place some importance on having important dimensions pop up when you double-click on a main face. If the designer starts with just a plain rectangle and follows that with cuts that could have gone in the first feature, you don't get easy access to so many dimensions.

Dwight

Glenn Schroeder wrote: ↑Mon Jan 31, 2022 8:16 am [...]
2. Less obvious, often design intent makes it easier to edit later if that's the case.

I think this comment is heavily underrated here.

Maybe the most common newbie mistake is thinking there is one set of rules that can be applied to all parts in all industries.

Maybe the most common newbie mistake is thinking there is one set of rules that can be applied to all parts in all industries.

This is also one of the most important things stated. With any CAD system there are a million different ways to draw the same box and circle, and no matter how it's created somebody will find fault with the way it was done. When it was just vellum and paper on a board the only thing that mattered was the end result.

SPerman wrote: ↑Mon Feb 14, 2022 11:25 am Maybe the most common newbie mistake is thinking there is one set of rules that can be applied to all parts in all industries.

I'll add a corollary to this one...

"My way is the best way of doing it." - Prevents you from learning something new.
"Other's ways are better than my ways." - You lose the efficiency of doing it the way that makes sense to you.

The most important thing is to think about the primary planes and what makes sense...
If the thing you're designing is symmetrical, then one of the planes should be at the center of the part.
For example...
A tool cart.
The floor is constant.
You typically want everything a certain distance from the floor.
It's also symmetrical left to right.
I would rename my top plane to "Floor"
Front plane to "Centerline of cart"
Right plane could be the centerline of the push handle, or front of the cart, or whatever.
I also create planes at other "Fixed" heights... Caster mounting heights, for example. Then I can use those planes for sketch relations.
If we decide to switch casters to some with a different height, I can change the plane, and everything updated.

A piece of machinery that isn't symmetrical, but also sits on the floor.
Top plane gets renames floor.
Front plane might be some important part of it, or maybe the front panel, if it's enclosed in a sheetmetal box.
Right plane would be the another side panel maybe...

Dwight wrote: ↑Mon Jan 31, 2022 9:19 am Glenn

I am currently working on a best practices presentation for our project modeling and am trying to articulate where a designer should put primary planes. I think Matt's idea that stability is the most important criteria is dead on. I am coming to a few conclusions as I create the presentation.

For our products, which are mostly floor-standing and benchtop instruments with a lot of user interaction, their relationship with the user is the most stable relationship. That's kind of funny, as I think of users as being the most variable and unpredictable part of the project, but it really seems to be the right approach for locating the top level assembly.

"Top" plane - the working height of the user interface is a critical aspect, so either the floor or the benchtop should be the "top" plane of our top level assembly. The instrument may evolve to be a little taller or shorter, and clearance under the bottom might grow or shrink, but we don't want internal components of the instrument to rise or drop once we have the human interface set.

"Right" plane - generally, users work at the center of an instrument, and that's where the "right" plane goes for the top level assembly. If sometime we did happen to design an instrument with an off-center interface, I'd center the "right" plane on the user, not the instrument. From there, the instrument might get wider or narrower, but most internal components shouldn't wander left and right if it does.

"Front" plane - lastly, we should establish a front face of the instrument that defines the depth of the user interface. If the front of the instrument is actually flat, then the choice is easy. If it's irregular, then the choice might be arbitrary, but that's still where our "front" plane goes.

That done, now the instrument can shrink (not likely) or grow without upsetting things too much. Add a few mm to the back to fit some more stuff in and generally things don't shift in ways that are a probllem.

Once the primary planes for the top level assembly are established, it is easier to set primary planes for components. Often you don't know what the mounting surfaces are going to be, so the functional location may be more stable. If you know the component is going to mounted on the right wall and move with it when the instrument grows, then model it's "right" plane on the wall. If it's centered on the user, as most things are, then put the "right plane" on the center.

Components with a stable orientation should be modeled as oriented, not flipped backwards or upside-down. It amazes me how many models I see that are not oriented correctly. It makes them harder to edit, more prone to mistakes.

Some components, though, may be used in many places in different orientations, so the instrument orientation doesn't matter. In that case, block-like components get modeled with their "front" plane as their mounting surface, built in positive Z, and probably centered for their other primary planes, though it depends on what seems most stable. Similarly, cylinder-like components should be concentric to Z and built in positive Z.

The side walls of our frames are tricky to locate in space. There is a great temptation to center the primary planes, and that usually creates problems. You are always going to have to tweak the location of the front edge, probably grow the instrument depth some, and it turns out that the center wanders something awful, and you don't want all the components you've mounted to the side wall to wander with it. So for the big frame panels, you build

side panels with "front" = instrument front and "top" = floor;
top and bottom panels with "front" = instrument front and "right" = centered; and
front and rear panels with "top" = floor and "right" centered.

I agree completely that usually a corner is the most unstable thing you can pick.

This post is a bigger than I usually want to read, so sorry for that. It did, though, help me figure out how I am going to present this internally.

Dwight