PREVIOUS < - - - - > CS 294-3 HOME < - - - - > CURRENT < - - - - > NEXT
Noah:
Use 3/8" endmill to cut 3/8" wide paths with 1/4" walls between them.
Cut path 3/8" deep (leaving 1/4" mid-bottom thickness).
Make through holes also with 3/8" diameters.
Suggested layout:
18x18 grid points for the channels and 1/2" border around it.
It probably would look nice, if the through holes are predefined in
pleasing manner, for instance all grid points u,v
where (u mod 3 = 2) AND (v mod 3 =3), i.e. locations 2, 5, 8, 11, 14, 17.
The maze algorithm handed out in class would have to be modified
so that in these locations the search proceeds through the hole to the other side
in a non-random manner.
No special starting and end-points are required;
the task could be to roll the ball from any corner to any other corner.
To make the corners more distinct positions,
they could be restricted to have access from only ONE direction.
Danyel:
Use 1/2" endmill to cut hexagonal grid of "junction" pockets on hexagonal grid
with ideally 0.866" center-to-center spacing, (so that adjacent circles
will overlap so that they will leave a channel of 0.25" width.)
Place "connector" pockets of 0.5" diameter at mid-points between
junction points -- where desired.
Cut junction and connector pockets 3/8" deep (leaving 1/4" mid-bottom thickness).
Make through holes also with 0.5" diameters.
Suggested layout:
It probably would look nice, if the through holes are predefined in
pleasing manner, for instance, so that all through holes are surrounded
by six junction pockets that do NOT go through the stock.
Thus if you can fit about 13 junction pockets in a straight line
into the available 1 foot work area, then all even positions could be
through holes.
As in Noah's case above, corner positions could be made to be "dead-ends"
so that they can serve as ready start ot target positions.
Yan:
Suggested approach:
Limit yourself to a maze that covers 4 faces.
The maze consists of paths that are 3/8" deep and 3/8" wide,
separated by 1/8" thick walls.
Reduce the 2" dimension to 15/8" and then fit 4 path widths into this range;
the two outermost paths are flush with the stock edges.
Reduce the 4" dimension to 31/8" and then fit 8 path widths into this range;
the two outermost paths are flush with the stock edges.
The paths along the stock edges are shared paths that belong to both
adjacent surfaces.
In the 3rd dimension, leave the stock somewhat larger than 4"
so that the piece can be properly clamped during fabrication.
However, making its complement, i.e., a HOLLOW path through space is a different issue altogether ! Obviously one could put a box around this structure and then reverse all the faces to have suitable part description -- but there is a problem with actual manufacturing: how does one get the support structures or the un-sintered powder out ? We talked about running thin holes through all the path axes through the whole cube, through which one could then push a long wire to coax the sand out of the passages. This would be an interesting experiment !
To make the maze not too hard to solve, it should probably just be one long winding path; -- or perhaps a contorted "Y-junction", so that one could put the ball into an "input" slot and then get it out from either a "YES" output or from a "NO" output.
Who is going to take on this experiment ?
One modification would be to modify the path profile, so that it has overhanging side walls with just a small slit-opening for a "ceiling". We would need a couple of access points where a ball could be inserted. Such a more complex structure with many overhanging features should probably be done with SLS where the support structure can easily be brushed away.
One other concern with these structures is price. The have relatively large volumes compared to the interesting part, the surface, where all the action is. Fabrication price depends a lot on build time, and thus these two structures would be relatively costly. Perhaps I will try one, if I have a pleasing design and can get a "free" run somewhere ...
Mike:
If you want to take a chance to see the torus being built,
remove the "dividing wall",
make a part about 3/4" thick and 3" outer diameter for SFF.
<-- Up to CS 294-3 HomePage