"In this video, I demonstrate how to model a recessed knurled surface on a circular cap for 3D printing. Maybe it's not the best way to do it, so if you have a critique, please let me know! Thanks for watching!"
Hello KTEC DESIGNS. Very nice and informative video. As an undergraduate software engineer I can tell you that FreeCAD, as per documentation uses an OpenGL-based 3D API which means that it tries to either offload the calculations to your computer's GPU or, if you are doing this on a computer without a dedicated GPU (eGPU), uses the embedded GPU of your CPU. Now, eGPUs of modern CPUs are generally lower powered components designed, in most cases, for basic display output and are much less powerful than a normal dedicated GPU. In this situation FreeCAD tried to offload the 3D calculations of the model to your computer's GPU and straggled because it's a hefty calculation due to the geometric complexity of the knurled surface. I hope this is informative enough. Keep up the good work!
In this case, it is a combination of the OpenGL render after the OpenCascade 3D engine does the calculation. OpenCascade, as most older solid engines, are not using multiple cores. This operation, for solid model math is very compute intensive.
And, no there are no settings to change how the various libraries use the cpu or gpu. You can adjust limited functions of OpenGL in Edit>Preferences, but this changes how OpenGL addresses things like render order mostly.
Takes about the same amount of time with a quadro k40000. Commercial packages do this almost instantly. Its a shame, freecad is getting pretty awesome.
Nice method, especially compared to some of the more complicated approaches I watched first. With that said, I had an issue generating the subtractive helix. Turns out it was because I missed one fundamental step in the sketch... I didn't toggle construction for the circle and 45 degree line. As a first time user of the software, this was not an obvious thing for me to check. I actually re-watched the video numerous times before I realised my mistake. It might be worth adding an annotation in that section so that other newbies like myself don't make the same mistake. Thanks for the video!
Well as a complete novice, you want every .... detail, later you do not want to lose stuff what is obvious to you. Anyway, with tinkering around, you learn too. As a complete novice, maybe you should learn easier stuff beforehand.
This comment was very helpfull. I too am a novice and was getting an error of wire frame. After fideling around a lot and repeat watching the video multiple times, I checked the commets. Thank you saved me a lot of frustrating time waste.
Yes, very neat indeed! However, KTEC says at the outset quote “ if you’re new to FreeCAD….” and then quickly plunges into the process with little time given to explicitly enumerate the command options selected: following the mouse pointer in order to determine this was nigh on impossible. Yes, I get the principle, but - as the tutor says - I am indeed new to to Freecad, per se, and am finding the command structure - compared with my most familiar Sketchup pro - a little difficult to grasp without explicit guidance. FreeCAD is not, I find, particularly intuitive. I wish I could slow the video down so I could decipher the specific commands used to create this really impressive knurl, especially as I need to model a very similar cap and then create a 3D print from it. But yes, a really cool solution compared to other videos on this subject. PS: after re-running the video a number of times, managed to twig how the helix was generated and the polar array executed….phew!!
One of the things which make FreeCAD so difficult to begin with. The same stuff can be done in various workbenches and each one has specific limitations not obvious at the start...
Calculation Tip: To get the Turns required I created a calculations sketch (hidden, containing only construction geometry). Create a line at 30° Angle from origin. vertical length of line = part height. Then use a reference constraint to get horizontal length of the angled line. Divide horizontal length by `2*pi*(part diameter/2)` to get the value to use for turn of Subtractive Helix. To get the amount for Polar Pattern I used `[2*pi*(part diameter/2)]/[(knurl depth*2)*1.8]`. This calculated to 29.99 for me so I use 30. If the result is not up to expectations the last value can be increased/decreased slightly to get a better result.
I finally found my calculation. I used the distance formula (h = sqrt(a^2 + b^2)) to find the length of the square's hypotenuse (because the square is small, it's roughly equal to the arc length between the coincident points of the square that sit on the cap's circumference). I then used that arc length to find how many instances of the square that I can fit along the cap's circumference. That formula is "pi*D/h". Since this represents squares in sequence end-to-end, I divided the result by 2. So, "h=sqrt(2)" for a square that is 1mm x 1mm, and D=50mm, therefore the circumference "50*pi" divided by the arc length "1.414" is approx. 111. Divide that by two, you get 55.5. Rounded down I ended up with 55 instances.
I followed along, and I have a computer that is overkill for most operations. That said, this sucker still took about 2.5 to 3 minutes to calculate! Really nice trick though! :)
So now I see this after printing some dust caps for some solar filters. It would have been nice to have actual knurls. I used PrusaSlicer fuzzy skin to create a grip on the external surface.
great tutorial. food for thought. could you duplicate your helix patterns to a shorter length, and then apply your polar pattern to the shorter length helixes so you get a more rapid feedback? and then when you get something you like apply to the entire surface.
I have a similar problem as the number of faces increase. I made a plaque and it ended up around 1600 faces and everything I did after I padded the text just ran slower and slower, and I gotta say, from a design POV, knurling creates a multitude of faces.
It seems that doing Boolean Unions or Cuts in FreeCAD with a lot of intersecting geometries takes a while. I tried a couple of other approaches and it took about the same time. Try cutting in only one direction and see how much less time it takes in comparison.
We need a knurling workbench for freecad, as well as a thread workbench, this sketching shit has to go, when a feature can easily be automated, it should be.
Nice tutorial. I tried to follow it for another knob. But when i'm trying to make the second helix for the knurls, i get a error, face must be planar. Trying to learn how to use cad by mixing videos and testing for myself. You got any idea what could be my problem?
(FreeCAD 0.20) Upon adding the second subtractive helix, the first one got deformed somehow. I thought to fix that by using pitch height for both helixes, but upon waiting 10 minutes for the polar pattern to finish, I got an error that a boolean operation failed. So much for knurled surfaces.
I don't understand how the helix settings translated into the diagonal line, and how polar pattern of these lines made final knurling. Seems to just happen but how? The final result does look good.
Helix is applied as a pitch. Distance in the axis direction per turn. For a 10mm long knob, 1mm pitch would give you 10 turns. 100mm pitch would give you 0.1 turns.
When attempting to do the helical cut, I continuously am getting a " wire not closed" error. I've went over all the constraints & it all looks good but still getting the same error. Any suggestions? Thanks in advance
You need to calculate the perimeter of the cap as pi*diameter, that's the total length to cover. If your pattern has L length, then the number of repetitions must be the integer number closest to pi*diameter/L.
Could not get the subtractive helix to do what your video shows. I have watched this video several times and still cannot mimic your process I get "linked object is empty", also the helix parameters do not change when I select the z axis, height, turns,etc ANY tips?
I tried to accomplish this on my macbook and failed miserably. Any time he clicked and point to move it I was simply unable to do so. I just wasn't getting the same results whatsoever. Idk...
Had several attempts at this. Even tried padding my sketch instead of revolving it to produce a solid body, but to no avail: same error message pops up every time. “Wire not closed” when selecting the helix command. Anyone any ideas? Website returns: “There do not seem to be any answers to your search”, or words to that effect.
"Wire not closed" could mean that the sketch entities are not closed. By that I mean, the square might not connect together at every corner. You can check this by clicking and dragging each line in the square to see if it is able to move. The same with the vertices of the square. Another thing to make sure is that all other sketch entities besides the square are construction lines. If you are still having issues, you can email me a copy of your model, and I can see if I can find the issue.
@@KTECDesigns Success! had another go and encountered an oddity. Attempts to create a fully constrained rectangle/square were frustrated by the vertices not being selectable even with the body made invisible. The solution for me was to (effectively) flip the sketch by padding the body in reverse. Creation of a fully restrained square was then achieved without difficulty and I was able to create the opposing helical grooves and apply the polar array without any further problem. One further point. At the sizes used in your demonstration tutorial - approximately 1mm square - the sketch wasn’t displaying in model space after the constrained sketch was closed. By increasing the size: I used a 20mm square, the sketch was clearly displayed. Of course, at 20mm the results of executing the helix command were somewhat bizarre. However, I eventually settled for 5mm and all was well. Yes, I did maintain the depth of cut at 0.8, the “surplus” body, therefore, had no impact on the final result: it simply increased the visibility of the sketch in model space. It did occur to me that because I elected to use 20 occurrences in the polar array it also avoided the possibility of the larger sketch colliding with itself in the event I had opted for the maximum possible number of grooves? In SketchUp which I use for developing engineering design ideas: currently a sphere turning fixture for a woodturning lathe, I tend to create subtracting volumes, when executing boolean operations, that are (probably) a lot larger than they need to be, but it dies ensure the operation doesn’t miss anything. I fancy, by enlarging the sketch to a “generous” size in this knurling exercise, I was instinctively applying the same logic and successfully too. Would possibly be subject maybe, to the “collision” phenomenon when higher occurrence values are used in the polar arrays. By the way. I appreciate your responding to my plight, thank you.
I've tried to create your example, but I just can't get it to work. When I create the second subtractiveHelix it wipes out part of the body. Also I tried to polar array just the one subtractive helix and it won't create more than about 15 copies without wiping out the body. I get this error when I array it. Error: one transformed shape does not intersect support. I've messed with it for two days trying different things but nothing works. Any help would be appreciated. Thanks
Sorry to hear that. I'm not quite sure what the issue could be. If you send me your model via email, I could see if I could figure out the issue. In general, though, it's a computational intensive operation, so perhaps your computer is struggling to compute the result.
This is a real thing, and most shapes that get knurling are cylindrical. This can also be done on a flat surface. If this comment gets enough likes, I’ll do a flat surface example video.
