Last time around I ran into two major issues.
- Tolerance between moving parts was too tight. My pivot pins on the inner ring were pretty much jammed into their holes…
- Fragility of the part. Using 1mm diameter pivot pins using the detail material (weak) was not a good idea. Coupled with the tight pin problem above, it meant I accidentally snapped a couple of pivot pins off the blades when I tried to rotate close the iris.
- Coarse triangulation of the exported STL which made the two rings too ‘blocky’ to rotate freely with each another. My fault, fixed by (labourously) tweaking/experimenting with Alibre’s Export.
So how did I fix it all?
This message thread on the Shapeways forums details a few experiments other people have tried printing ‘living hinges’ and moving parts direct from 3D print. Appears the magic number is >0.1mm gap between two components to avoid them fusing together…so I went for a margin of error and aimed for >0.25mm gap.
While reading it all, my previous design came as 26 individual components…which probably annoyed the hell out of Shapeways staff while shipping it. With the improved tolerances, I tweaked the design to combine all components into one assembly to be printed in one go hoping it’ll work ‘out of the box’. Yes thats right, printing in ‘one-piece’ and hoping it works!
A few other tweaks here and there; making it smaller to save cost, making the pivot pins thicker, putting a covering skin on it to hide/contain the blades and around an hour later…
One thing to note is that this only took an hour because I cheated and reused my previous Alibre Design files. Being a parametric modeller, all I had to do in most instances is edit some dimension values on the sketches I’ve set up on the previous design. Easy!
I also didn’t bother doing math calculating on how long the sliding slot should be, simply judged it by eye by sliding/rotating the blades in assembly mode to see approximately how much movement I’ll need to achieve full closure/opening. This is seeming more and more like an Alibre advertisement isn’t it?
The painful STL export again…
One rant though is with the STL export – its non-intuitive, the help file instructions were that helpful and it all came down to trial and error. I learnt :-
- Tweaking the normal deviation setting is a bad call. Making it smaller will result in either a humungous file or Alibre crashing out.
- Tweaking the surfaces setting is too painful. It either had a negligible effect on things or I was simply doing it wrong…
- Tweaking the maximum cell size worked! Well kind of. It basically asks the exporting to enforce a triangle plate size to be smaller than what you specify, so a small value generates alot of triangular cells. This is not optimal because even if you had a large flat triangular sheet that could be very well represented by one large triangle, it’ll now be forced to make it into 893475 triangles…
- Why kind of? Because the Shapeways design validator crashed/errored out of me when I had the cell size too fine (0.1). I had to nudge that back to (0.2) before Shapeways accepted the file.
Naturally I used Meshlabs to validate the triangulation and remove duplicate vertices again. This step is required to allow it to pass Shapeways validation, it also cuts the STL filesize by 80% typically.
Oh I forgot to explain why it was painful. Learning items 1-4 above probably took me oh over 2-3 hours? [Though I still remained productive by fixing my girlfriend’s Nintendo DS…].
Waiting for the shipping again
Actually waiting to find my wallet to click the order button 🙂 Lets see how this turns out!