3D cad question

Home Model Engine Machinist Forum

Help Support Home Model Engine Machinist Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I'm mostly self taught too. 2D Autocad v14 through evening college classes, then what was it, DesignCad? (AC 2d lookalike), then Rhino v4 3D. Then SW around 2010/11. I got a freebie course of the basics at the time, but most of what I've learned has been doing, instructional texts & some good online video courses made available through Lynda.com, now LinkedIn learning (free subscription through my local library). YouTube is a dangerous place, some good stuff & some.... less than good.

Rhino was a good example of what we are talking about. Like AC it provides no visual feedback that a sketch was 'correct'. Say 2 lines that are supposed to meet don't actually meet. Separated by a screen pixel. Because models have computational tolerance, it may still make a (albeit non-air tight) 3D model of sorts that also looks good from across the room. Things are going marvy as the features get extruded & manipulated. Then you notice this edge fillets & that one doesn't. Why? Then another click & the entire model blows up. Hmmmm... 2 issues here. 1) the imbedded sketch problem was never properly identified to the designer so its hidden. Maybe there are many more similar issues, identification is painful, no diagnostic debugger. 2) Rhino is not parametric like SW, so you cant even back up the tree to see where things have gone awry. (Rhino is 1/5 the price which explain some of the issue).
 
Last edited:
Yes, I hear what you are saying, and when I tell people I have never dimensioned or constrained anything in any 3D model I have ever made, they are aghast with horror.
I guess ignorance is bliss, because I don't dimension, and I don't constrain, ever, and I have never had problems.

Granted I may be doing odd sort of stuff with 3D modeling.
I basically look at photos of old engines, and work from photos to develop one sketch and one part at a time, for a unique engine design.
My parts always mate correctly, and my assembly simulations run perfectly.
For a somewhat more othodox design (such as a commercial project design), I can see where one may want to be a bit more cautious.

If one has been taught that you must use constraints and dimensions always, and one has never tried anything else, then it may be ignorance of what one really needs or does not need when modeling an engine that is the driving force.
I have seen professional CAD guys doing dumb stuff because "That is what they taught us".
Some folks get their mind in a box, and can't see out of it (the old adage "you can't teach an old dog new tricks").

I am self-taught in both 2D CAD and 3D modeling.
I don't use paperspace in Autocad, and I get far better use and efficiency by not using paperspace in 2D CAD.
The veteran 2D autocad users are also aghast that I don't use paperspace, but I have a much better method (XCLIP).

For airport work, where the drawings are very long (like a runway that is a mile long), sometimes I have to resort to paperspace, but that is the only exception where I found paperspace to be more efficient than XCLIP.

And I put all of my 2D work drawings in a single file (per project), so when I open a project file, I see every drawing on the screen.
I have had up to 100 drawings in a single file in AutoCad 2D, and this is normal for me, and highly efficient, since I can copy/cut/paste across multiple sheets, and always have an overview of the entire project, with instant access to every sheet.
Most people are horrified at the thought of having 100 drawings in one drawing file, but I do what is efficient, not what is generally accepted by others. Efficiency is money, and I can show you the money.

So perhaps a hobby person could try 3D modeling while completely ignoring constraints and dimensions, and if that is not creating any problems, maybe you don't need to be spending time on those.

I am an extremely meticulous CAD and 3D person, and so perhaps constraints and dimensions would force a less meticulous person to avoid problems.
I can categorically say that dimensions and constraints are totally unnecessary in a complex 3D engine model, at least for the engines I have modeled. I have modeled some complex engines.

.
I thimpfks you are naughty. You do your drawings for yourself? You make one-offs? Then all that is OK.
 
For instance when drawing a steam engine cylinder with flanges.
If you don't reference the flanges from the correct surface, then if you change the length of the cylinder, the flanges become too thick or thin.
If you extrude the flanges correctly, they will remain the same thickness regardless of changes to the cylinder length.
Yup, I know exactly what you mean. That's the art of design. It comes with practice & experience. Make things easy on yourself, name all of the feature steps to something meaningful which is kind of like a form of visual documentation. The mantra is someone (including yourself) should be able to open the file a few years from now, see a logical series of notated steps to understand the gist of how the part was developed. Its like writing programs, 10,000 lines of pre code functionally works, but is may not be readily apparent to another programmer (including yourself if you wrote it) what the program elements do without useful comments along the way.

I also use the SW comment feature quite liberally. Right hand click on a sketch or feature entity in the tree, select comment option. You can hover over comments to see display what was written at the time, or pump them all out to a file, attach pictures... etc
 

Attachments

  • SNAG-2022-10-19 8.58.06 PM.jpg
    SNAG-2022-10-19 8.58.06 PM.jpg
    32.6 KB
  • SNAGIT.jpg
    SNAGIT.jpg
    29.2 KB
Last edited:
..., and when I tell people I have never dimensioned or constrained anything in any 3D model I have ever made,
Actually your sketch IS being dimensioned & constrained, but maybe behind the scenes without you realizing it. Just by drawing a simple line in space, it is snapping to vertical in this example (unless you deviate significantly in which case it lets you zing off at an angle). And its recording an incrementing actual dimension shown in box, which terminates whenever you click the end of the line. So if you wanted something like 0.3692345" that's perfectly fine. But more than likely you are making something in the real world and want to specify - its 0.375 or 0.300 or whatever. You can always go back & modify the dimension & validate what it is on the left hand panel properties.
 

Attachments

  • SNAGIT.jpg
    SNAGIT.jpg
    8 KB
Rhino was a good example of what we are talking about. Like AC it provides no visual feedback that a sketch was 'correct'. Say 2 lines that are supposed to meet don't actually meet. Separated by a screen pixel.
Solidworks is totally intolerant of a non-closed sketch.
For that reason, I draw lines that overlap, and then trim them off.

But as I said, I almost always draw the initial 2D sketch in Autocad, and then import it to Solidworks.
In 2D Autocad, I use either the snap-to function or trim lines off, so that there are no gaps or problems with the sketch.

If I had time to do a deep dive into the pros of constraints and dimensions, I am sure I could find some applications where they would be useful.

I work full time (and a lot more), and so anything I learn in the 3D modeling world is unfortunatly done in great haste.

I started using 2D CAD in about 1988, if my memory is correct, and that was an early version of AutoCad.
We had to use a $4,000.00 graphics card to get any performance out of an IBM-AT.

I changed companies and started using VersaCAD, which was ok in some respects, but did not have the all important XREF feature (imagine that).

Then the company went back to AutoCad.
Another company demanded that I use a third 2D CAD program (I forget the name), and I purchased it, and it was totally unusable.
I tossed the disk into the garbage, and stopped working with that company.

If they could combine AutoCad 2004 (2D) with Solidworks, I would be a very happy camper.
As it is, I do a lot of initial sketches in ACAD, and then import to SW, which is pretty easy to do.

Time is my problem; I don't have enough of it for work and model engine building.

.
 
This shows variations of fully defined sketch (all black lines). Notice in one I set the line as VERTICAL & 0.400" length. In the other I broke the perpendicular relationship & defined the angle as 102-deg, but still defined the line length as 0.400" Now the lighter grey dimension 0.391" is an internal calculated dimension telling me the vertical equivalent if that is useful. Both sketches are fully defined but I went about defining it differently as to what makes the most sense for that particular feature (design intent). All this good stuff helps you make more user friendly & bulletproof models.
 

Attachments

  • SNAG-2022-10-19 9.08.52 PM.jpg
    SNAG-2022-10-19 9.08.52 PM.jpg
    15.5 KB
  • SNAG-2022-10-19 9.10.13 PM.jpg
    SNAG-2022-10-19 9.10.13 PM.jpg
    23.3 KB
Actually your sketch IS being dimensioned & constrained, but maybe behind the scenes without you realizing it. Just by drawing a simple line in space, it is snapping to vertical in this example (unless you deviate significantly in which case it lets you zing off at an angle). And its recording an incrementing actual dimension shown in box, which terminates whenever you click the end of the line. So if you wanted something like 0.3692345" that's perfectly fine. But more than likely you are making something in the real world and want to specify - its 0.375 or 0.300 or whatever. You can always go back & modify the dimension & validate what it is on the left hand panel properties.

Typically what I do when sketching in Solidworks is to draw a line, offset it some known amount, draw another line, offset it, with all lines overlapping, and then trim off everything.
Perhaps a crude method, but an exact method.
.
 
This shows variations of fully defined sketch (all black lines). Notice in one I set the line as VERTICAL & 0.400" length. In the other I broke the perpendicular relationship & defined the angle as 102-deg, but still defined the line length as 0.400" Now the lighter grey dimension 0.391" is an internal calculated dimension telling me the vertical equivalent if that is useful. Both sketches are fully defined but I went about defining it differently as to what makes the most sense for that particular feature (design intent). All this good stuff helps you make more user friendly & bulletproof models.

I have gotten use to using Autocad, where you just draw random lines on the screen.
Unless I have drawn a rectangle, I can just grab any line, drag it over, snap the ends of any other line to the new endpoints, and I am done.

Solidwork's associations (relations?) drive me crazy, because I grab a line and move it over a bit, and the entire model bends and warps out of position.
I have to turn off all relations on all things in order to be able to sketch in SW.
And thus the reason I generally sketch in ACAD, and then do a copy/past to SW.

I think in order to get some use out of the features you are mentioning, one has to wrap their mind around driving the sketches.
Driving the sketches is a totally foreign concept to me; call me dumb or whatever, I was raised on AutoCad2004.

My sketches are dumb, and simple, just like me.
I do have some complex 3D engine models though.

But as you say, the program has all the parameters stored, regardless of whether I dimension things or not.
I do know that the relations can be very useful, if you can understand how they work, and can draw and think about relations at the same time (I can't).

I have never noticed the sketches changing color in SW when they are fully defined.
Perhaps I never noticed that, or perhaps SW does not do that.
There is a snap icon when drawing a line, and if the line is perfectly vertical or horizontal, that icon will pop up on the screen, so I know I am vertical or horizontal.

I have sort of stumbled my way into SW, learning just enough to get by.
One can do a lot with minimal tools though in 3D modeling.

.
 
If they could combine AutoCad 2004 (2D) with Solidworks, I would be a very happy camper.
As it is, I do a lot of initial sketches in ACAD, and then import to SW, which is pretty easy to do.
My advice is to pull the Bandaid off & stay inside SW. There is absolutely no useful 2D command or function I can think of that AC offers that SW cannot do & probably better. Layers? don't need them, that's mainly a legacy 2D thing. Its not even really in the SW vernacular in that sense. Line colors? I think SW can do that, I don't bother. Drawing your sketches inside SW gives you the diagnostic features already mentioned & all dimensions are variables. That means design tables, equations, all kinds of other things you have now functionally omitted by importing. But the big issue importing sketches from AC into SW is the opportunity / likelihood to import buggy geometry as previously discussed.

The other issue is even a simple 3D part may have 10, 100, 1000 sketches applied to various surfaces as you work the model up, extruding, removing, revolving.. all that good stuff. So later on in the process you need to pick the appropriate model surface or plane to draw another new sketch & so on. Please tell me you aren't importing AC drawings in at that level? Ugh. LOL.
 
This is a typical 3D engine model for me, made with Solidworks.
Almost all the information taken from photos of my Dad's 7 hp Galloway.
No dimensions, no constraints.
No problems running it in simulation.


Galloway-Assembly-22.jpg
Galloway-Assembly-24.jpg
Galloway-ValveGear-01.jpg
Galloway-Valves-Head-Assembly-01.jpg
 
You are doing fantastic work. I am just offering suggestions to hopefully make your modelling experience & model robustness better yet. I'm no expert either, just a dedicated hobbyist.
 

Attachments

  • SNAG-2021-07-04 001.jpg
    SNAG-2021-07-04 001.jpg
    105.4 KB
  • SNAG-0009.jpg
    SNAG-0009.jpg
    79 KB
You are doing fantastic work. I am just offering suggestions to hopefully make your modelling experience & model robustness better yet. I'm no expert either, just a dedicated hobbyist.
I have stumbled into 3D modeling blindly, and no doubt I have much to learn.
Some of my method is due to being overwhelmed by features and options.
In order to learn Solidworks, I basically ignored perhaps 95% of the features available, and just learned some very basic fundamentals.

I learn new methods/features a la carte if I see someone doing something useful; ie: I go to the school of ytube and watch the videos on how to do things, like create a helical gear.

Its a ******* approach to learning/using 3D modeling, but it fits my tight time schedule and tiny brain.

No doubt that in a few years I will have learned more tricks and things.
One day and one trick at a time.

These discussions are good, and help fill out what I am missing.

Edit:
Nice radial !
.
 
BTW, you do know that you can import a background jpeg into SW & 'draw' on top of that to start the 3D model process? Even scale & integrate opposing views. Very useful for the work I see you doing utilizing using scans or pictures of legacy engines etc.

http://www.solidworkstutorials.com/how-to-model-audi-r8-car-in-solidworks/

Yes, I discovered you can trace over raster images that have been inserted into Autocad.
Solidworks may also have this feature, but I do it in Autocad.

And I scale the photo to what I think is full size, generally by gauging the flywheel diameter.
I draw all engines full size, and never to some reduced scale.

I also model engines exactly as they were originally designed and cast, and not how someone modeled/simplified them.

Occasionally I have to beef up a part a bit if I scale an engine down too much, and something gets too thin.

.
 
Solidworks is totally intolerant of a non-closed sketch.
For that reason, I draw lines that overlap, and then trim them off.
Yes but the intolerance is a feature, not a limitation. It is protecting you from making bad geometry inadvertently. An open sketch cannot be extruded to a (at least mathematically correct) defined solid. This is equivalent to the Rhino type crash I was mentioning earlier. If the computational tolerance parameter is set coarser than the open gap, it kind of makes a 3D model, but its not 100% correct & can cause issues down the line.

Trimming is fine, but closing geometry as you click-click is just easy, faster & arguably less error prone. Look for the little yellow symbol as you approach the termination point. This same feedback theme also pertains to 'waking up' the centers of circles & arcs by hovering over them
 

Attachments

  • EDT-2022-10-19 10.08.08 PM.jpg
    EDT-2022-10-19 10.08.08 PM.jpg
    40.9 KB
I should mention that I don't just pick random values for dimensions, such as on the Galloway above.
I keep a chart of standard fastener, round/square bar stock, and drill/tap sizes next to my desk.
Any shaft or fastener dimension gets selected from the nearest standard size in the chart.

Bore and stroke are also rounded to the nearest reasonable numbers, often whole numbers.

Etc.

.
 
Yes but the intolerance is a feature, not a limitation. It is protecting you from making bad geometry inadvertently. An open sketch cannot be extruded to a (at least mathematically correct) defined solid. This is equivalent to the Rhino type crash I was mentioning earlier. If the computational tolerance parameter is set coarser than the open gap, it kind of makes a 3D model, but its not 100% correct & can cause issues down the line.

Trimming is fine, but closing geometry as you click-click is just easy, faster & arguably less error prone. Look for the little yellow symbol as you approach the termination point. This same feedback theme also pertains to 'waking up' the centers of circles & arcs by hovering over them

I will have to play around with these features.
I am just not at that level of sophistication yet.

I create some complex models in spite of a lack of knowledge, not because I have great 3D knowledge.

.
 

Latest posts

Back
Top