LibreCAD
- Thread starter PeterDRG
- Start date
Help Support Home Model Engine Machinist Forum:
Ozwes007
Well-Known Member
This is a project I designed for a sawmill. Not the final draft btw. From this we created every part for Production.
https://www.thingiverse.com/groups/openscad/forums/general/topic:49365here is the code I use to export 3d to 2d from openscad. it requires you to pick each element to detail if you need and place it into the module area. then export as a dxf. Change the translate offset to suit your scale drawing. Bring into Librecad or your favorite 2d to dimesion and create production drawings.
Fairly quick and painless.
//dummy dxf draw
// 1st projection Side
projection(){
rotate([-90,0,0]){
draw();}}
// Second Projection Rear
projection(){
translate([720,0,0]){
rotate([0,90,90]){
draw();}}}
//Third Projection Top
projection(){
translate([0,-200,0]){
rotate([0,0,0]){
draw();}}}
// Original part
translate([700,-200,0]){
rotate([0,0,0]){
draw();}}
module draw(){
// place your drawing here
}
Here one from a different group
DoulbeCAD XL.
It works close to Autocad LT and TurboCAD as 2D.
Just try today and works on Windows 10 64 bit.
It automatically updated.
The PRICE IS GREAT TOO FREE.
DoulbeCAD XL.
It works close to Autocad LT and TurboCAD as 2D.
Just try today and works on Windows 10 64 bit.
It automatically updated.
The PRICE IS GREAT TOO FREE.
Polygonal tolerance (loosely stated) is typically defined in settings for different CAD programs. Points in space are straightforward, but curves and arcs are approximated in polygons. In Solidworks, you can get build histories to break if you change the vertex density, because a point on a polygonal circle is always an approximation. I've often seen polygons in F360 3D prints and triangular facets can be seen out of solidworks files as well, depending on how the export was executed. J codes in CAM/CNC are also approximations, as opposed to using boring (lol) tools. STL files degrade some.
For true tolerancing of 3D models, such as PMI for NX, you use the 3D space as your drawing and drag leaders onto surfaces for GD&T. ASME 14.41 covers the standards for tolerancing 3D models. It is overly expensive and will put you to sleep.
In Blender, I can get under the hood and crank up the polygon counts if I want, or constrain any vertex to a specific point in space to a micrometer (.000001). You could also define per 14.41 if you wanted to be nerdy about it. Blender is considered innacurate, but you can use it or others accurately like most other CAD applications.
Funny thing is, hardly anyone can read and interpret ASME 14.5 & 14.41, so dumbed down versions of anything are legit. As long as the machinist reads it or my printer prints it.
As for the original writer's request, FreeCad, F360 etc. should be able to handle those models in the original post just fine.
(Just goofing around in Blender below)
Last edited:
Ahh, I see what you are doing. But I'm thinking this does not add in any dimensions to the drawing, right?
Ozwes007
Well-Known Member
You are correct, that I do in LibreCad, both are free and the extra time required is similar to Any other cad program when producing production drawings. So the 3d is a concept drawing(with automation if required) and the dxf is the production or work floor plan drawings with tolerances etc added in.
Understood. I see how this could work well. The one thing I'm thinking might not be possible without extra work would be showing hidden lines?
I'm not sure I ought to post this here, but the talk about tolerances on drawings got me to thinking how we used to test tolerances in electronic design.
I'm talking about tolerances on parts values in a circuit, so what we'd want to know is how tolerances would affect performance. This is all rather expensive CAD used in industry; the last prices I can remember were around $35,000 per user.
The software would do what they called a Monte Carlo analysis. Users would specify a value, a tolerance, and a distribution of that tolerance for every value. The simulation would analyze the performance, like the signal level at some specific point in the circuit, while varying all of the parts randomly in the tolerance band by the distribution you gave it. We'd typically tell it to simulate a few thousand random combinations and it would tell us if the circuits would be acceptable over the variations.
I can see an analogy to mechanical designs by specifying a tolerance in a diameter, length or something and using Monte Carlo to determine clearances between parts as those dimensions varied.
Maybe this is more relevant to the guys writing the software than us users.
I'm talking about tolerances on parts values in a circuit, so what we'd want to know is how tolerances would affect performance. This is all rather expensive CAD used in industry; the last prices I can remember were around $35,000 per user.
The software would do what they called a Monte Carlo analysis. Users would specify a value, a tolerance, and a distribution of that tolerance for every value. The simulation would analyze the performance, like the signal level at some specific point in the circuit, while varying all of the parts randomly in the tolerance band by the distribution you gave it. We'd typically tell it to simulate a few thousand random combinations and it would tell us if the circuits would be acceptable over the variations.
I can see an analogy to mechanical designs by specifying a tolerance in a diameter, length or something and using Monte Carlo to determine clearances between parts as those dimensions varied.
Maybe this is more relevant to the guys writing the software than us users.
Ozwes007
Well-Known Member
Yes that would be correct in saying that. I don’t find it an issue and it is quite quick to do. The flip side is going from LibreCad to OpenScad. To generate a 150 PFC in OpenScad I do a dxf of its profile then import it into OpenScad with a linear_extrude function. Once you have a library of steel profiles it is very quick and painless. Apprentices like it because it’s free and they can use it on there home computers. Last Autocad HP computer I bought was over $7000 to run Inventor and Autocad.
I'm not a programmer or particularly experienced at CAD in general but I'm not sure how you'd get an automated system to put tolerances in a drawing without major expense, the program would need to understand design intent not just of the part your currently building but of the whole assembly as you're building it. I use Alibre Expert but there's a Youtube channel that actually recommends Alibre Atom as an inexpensive CAD alternative and in his hands, very capable program that you own forever after a $149 USD purchase.
Draftsight was great.
I still have a free copy of draftsight in my computer for changing drawing to DWG files for AutoCad 2000Lt. I purchase AutoCAD for work in 2000 and still use today. I was using Draftsight till could install AutoCAD in Windows 10 64 bit. It about best I can find.
I did find a free CAD program that works just few days ago call DoubleCAD you may want to look at
http://downloads.imsidesign.com/ESD/DoubleCAD_XT_v5.exe
It is close match to TurboCAD
Dave
I still have a free copy of draftsight in my computer for changing drawing to DWG files for AutoCad 2000Lt. I purchase AutoCAD for work in 2000 and still use today. I was using Draftsight till could install AutoCAD in Windows 10 64 bit. It about best I can find.
I did find a free CAD program that works just few days ago call DoubleCAD you may want to look at
http://downloads.imsidesign.com/ESD/DoubleCAD_XT_v5.exe
It is close match to TurboCAD
Dave
Statistical tolerancing is covered in ASME 14.5. You typically want the most difficult part (cost, manufacturability) in a stack of tolerances to have the widest tolerance. The formulas are in basic algebra. There's a lot of snake oil sold in CAD programs too with CFD and fatigue analysis, but I could see something like that for electrical loads being very handy.
Thankfully none of these tools are all that necessary for home CAD use. Phew!
I have noticed that quite a few CAD programs include CFD and finite analysis. I would agree with you about snake oil as often they are a bit shy on the code origins. The lack of snake oil is one the features I like about FreeCad. It has a workbench for linking into OpenFoam, a very powerful opensource CFD package. The two programs complement each other as OpenFoam has no graphical front end, so my aim is model generation in FreeCad and use its CFD workbench to then feed in the boundary conditions to OpenFoam and run the solver. I should add that my interest is strictly home research. Being retired does not even allow thinking about propriety packages. Though I should say that even if I was still in the game OpenFoam would be very attractive as it is proving itself as good if not better than Fluent now.
Tim Wescott
Well-Known Member
- Joined
- Jun 3, 2018
- Messages
- 328
- Reaction score
- 99
I watched over the shoulders of the mechanical engineers at FLIR back at the turn of this century. At that point, AutoCad (I think that's what they were using) had a finite element analysis that couldn't quite accurately predict the frequencies of resonance of a structure, but they'd get them all off by about the same amount.
So if you whacked it with a hammer and determined that the lowest resonant frequency was 20% off, you could be pretty confident that all the rest were 20% off, too.
So if you whacked it with a hammer and determined that the lowest resonant frequency was 20% off, you could be pretty confident that all the rest were 20% off, too.
Tim, what you say about results being off highlights the snake-oil aspect. From what I have seen in my working life the proprietary vendors are not transparent on the mathematics encapsulated in their code. This another reason I like open source software. Finite analysis is certainly one of those areas where engineers can make massive mistakes, due to ignorance of the underlying principals. We kept our stress calculations as simple as possible to avoid relying on FE even though we had ANSYS.
The snake oil CFD comment came from a retired SR-71 thermodynamicist I worked under a dozen years ago. So I figure it's pretty legit. 
He did say statics are very accurate. Fatigue analysis is often spotty and "noisy", with empirical testing often showing unforeseen results. The FAA still requires static and fatigue test articles for certification.
In many instances, CFD is only about 80% accurate compared to empirical testing. High sampling rates allow for tests on model buildings which can be slowed down to scale. Same is true for tunnel to train impact testing and other neat stuff.
He did say statics are very accurate. Fatigue analysis is often spotty and "noisy", with empirical testing often showing unforeseen results. The FAA still requires static and fatigue test articles for certification.In many instances, CFD is only about 80% accurate compared to empirical testing. High sampling rates allow for tests on model buildings which can be slowed down to scale. Same is true for tunnel to train impact testing and other neat stuff.
I have had many years of experience designing and running physical fluid dynamics simulations, including wind tunnel work. This type of modelling I found, provided a very high degree of confidence to the clients of our consultancy. The CFD was quite complementary however in being able to put additional numbers against the physical analysis. It also allowed details to modeled over a range of scenarios rapidly. Yes I can agree on the 80%. This often comes from the computational grid size used. Make it too refined and the run times explode.
In retirement though, ones's options in fluid dynamics modelling are a bit limited, hence CFD becomes attractive. As far as I know FreeCad with OpenFoam is the only open source combination available. If there are others I would love to know.
In retirement though, ones's options in fluid dynamics modelling are a bit limited, hence CFD becomes attractive. As far as I know FreeCad with OpenFoam is the only open source combination available. If there are others I would love to know.
I was looking at OpenFoam + Blender and it looks pretty amazing. Looks like a deep rabbit hole.
