|
Post by scotttucker on Feb 17, 2023 11:14:55 GMT -7
Do you remember around February last year I posted a picture on the Amateur Planetarium facebook page of an aborted attempt to 3d print a pinhole star ball? The starball has been available since about 2018 on Thingiverse as an .STL file. It contained 8911 stars.
|
|
|
Post by scotttucker on Feb 17, 2023 11:18:43 GMT -7
When I downloaded it I had to abort the print because the inside surface was completely solid. Judging by the comments attached to the thingiverse file I was not the first to run in to this problem.
|
|
|
Post by scotttucker on Feb 17, 2023 11:27:45 GMT -7
I have occasionally returned to the file to try to understand why it does not work (in terms of printing proper perforations through the sphere). As my competence very slowly grows in Fusion 360 I tried to manipulate the .STL file. You can insert mesh (STL) files into Fusion 360 but meshes do not seem to lend themselves to easy manipulation like bodies. It was only after discovering the 'sectional analysis' feature in Fusion 360 that I could peer inside the mesh sphere and see the reason why the print was unsatisfactory: the holes did not actually penetrate the inner surface of the sphere. Hopefully these couple of screen shots show what is going on.
|
|
|
Post by scotttucker on Feb 17, 2023 11:58:09 GMT -7
I started on a quest to see how I could easily make them penetrate through the sphere.
I have finally managed it but not without going up a couple of blind alleys.
You can directly edit mesh files in Fusion360 but I did not fancy manipulating thousands of vertices by hand. The simplest way I could think of would be to convert the mesh to a body and then 'cut' out the inner layer by creating a simple sphere of the right size and using it as a tool to cut out the offending layer. Converting it to a body would also allow easier manipulation like adding features to make it look like say a Goto Eros ball. That is when I ran into problems - perhaps because my computer is old or perhaps because of the shear size of the mesh file Fusion crashed every time I attempted the convert mesh to body command. In fairness - when I installed F360 I did get warnings that my computer was not up to it! My first attempt was to put on some brave pants, open up my computer and insert an extra 8GBof memory (~£20). Fortunately the computer survived and is a bit faster but it did not solve the crashing problem. I spent a lot of time trying various work arounds - using planes to slice the mesh into smaller more manageable pieces and trying to convert them individually. I also spent hours trying to use the star rods (in the same place in Thingiverse) which is basically the inverse of the celestial sphere model - each star is a cylinder radiating out rather than a hole.
|
|
|
Post by scotttucker on Feb 17, 2023 12:28:59 GMT -7
The solution seems to be in the SCAD files that are also posted on Thingiverse that I had ignored up to this point (because I did not understand what they were). OpenSCAD is a free 3D modelling programme which is very different from Fusion360 in that instead of using the mouse and icons to make changes to a model, OpenSCAD reads a script file and constructs the model from it. It sounds horrendously scary but in fact is not that bad.
So I downloaded OpenSCAD and the starball SCAD file from Thingiverse - with a bit of head scratching you can kinda make sense of what the script is doing. You cannot change the model by clicking and dragging parts of it with the mouse you have to amend the script. The model is displayed on the right of OpenSCAD and the script is displayed down the left. If you change the parameters in the script the changes to the object appear graphically. Not sure what language OpenSCAD scripts are but it is not too difficult to follow. The script defined the base sphere with an outer radius of 100mm and inner radius of 95mm i.e. it was 5mm thick whereas the length of the rods was 3mm. All that was needed was to change the inner-radius of the sphere to make it a bit thinner and that was it.
It also became obvious how to change the number of stars (the star file is arranged by magnitude) - it repeats a loop that makes a hole and has a star count variable. I chose to simplify the night sky to about 3000 stars rather than the 8911 stars the model is based on. (Actualy I started with a 500 star sky to make the computing times more manageable while figuring out whether it would work). piccys will follow
|
|
|
Post by Ron Walker on Feb 17, 2023 13:00:58 GMT -7
How smooth are the holes. Could you just use the unfinished holes to guide drills to cut out the remainder of the holes and make a nice clean cut in the final layer of plastic?
|
|
|
Post by scotttucker on Feb 18, 2023 0:12:22 GMT -7
How smooth are the holes. Could you just use the unfinished holes to guide drills to cut out the remainder of the holes and make a nice clean cut in the final layer of plastic? Hi Ron - yes you could but it just seemed to be a shame to be so close to an 'effortless' star ball but still need to drill out 8900 holes. In any case I suspect there will be a lot of cleaning up because 3d prints are stringy.
|
|
|
Post by scotttucker on Feb 18, 2023 4:03:19 GMT -7
Here is a screenshot of the amended sphere in OpenSCAD with 3000 holes. You can see part of the script in the left and the picture on the right. I forgot to mention - with OpenSCAD once you have amended the script you press the preview button - depending on complexity it can take a minute or two to display a preview of your new object. Then when you are happy you press the 'render' button which works some magic and makes it into a mesh that can be saved as a .STL file. Perhaps it is my slow computer but the rendering took about 2 hours for the 3000 star version and 13 hours for the 8900 star version.
|
|
|
Post by scotttucker on Feb 19, 2023 4:17:34 GMT -7
Does anyone know how many stars a Spitz A2 projected - I am working on a way of adapting the model to a dodecahedron
|
|
|
Post by Ron Walker on Feb 19, 2023 11:39:08 GMT -7
Something like 500 stars down to the 3rd magnitude. Note that there a lot of 4th, 5th, and 6th magnitude stare that are represented to "fill in" constellations and they are done with stars much brighter then they should be. Take the Little Dipper for instance, while it contains fainter 4th, 5th, and 6th magnitude stars they are reproduced on the Spitz projector as 3rd magnitude ones. This is done so the entire constellation is projected and easily recognized.
|
|
|
Post by scotttucker on Feb 28, 2023 12:52:44 GMT -7
Thanks Ron - just a quick update - I amended the star rods from the Openscad file just to make them thinner and longer (the default makes the star holes way too big). Instead of using them to cut holes in a sphere, I have used them to cut holes in a dodecahedron. There are 750 stars to make a rough approximation to the 500 stars + some others - It will not be exactly as Ron describes but I hope it will make a nice Spitz A series model. Screen shot shows the star rod cutting tools and the base dodecahedron. Have started a test print will let you know how it turns out.
|
|
|
Post by markspoelstra on Mar 2, 2023 18:15:40 GMT -7
(just downloaded all that stuff myself) ...very interesting and cool project.
Is it you're intension to drill a hole in it and put a light inside?
|
|
|
Post by scotttucker on Mar 3, 2023 1:30:00 GMT -7
Hi Mark - yes. The intention to make a model of a classic spitz A1 or A2 projector along the lines of a Spitz Junior.
|
|
|
Post by markspoelstra on Mar 3, 2023 5:18:56 GMT -7
Hi Scott, Cool ! Can't wait to see...
|
|
|
Post by scotttucker on Mar 5, 2023 13:45:00 GMT -7
A quick detour from the Spitz model: I have spent part of this weekend taking my Excel files from the star ball based on truncated icosahedron thread in builders of optical projectors and using them within Openscad to try and 3d print some star plates. I have experimented with plate 5 (contains Orion)- I had a go at printing a 10cm diameter plate, with 400 stars on the plate (roughly ~1/32 of a 10k star sky). The first attempt scaled the stars between 0.2mm and 0.9mm holes. As you can see from the photo it is a long way from satisfactory but not completely devoid of promise - my 3d printer clearly cannot cope with the smaller holes - many were closed over.
|
|