Saturday, June 25, 2005


Strength of objects made by layered manufacturing

This may not be an original idea, but I haven't been able to find a reference to it.

Objects made by RP are anisotropic. In particular they tend to be stronger in the layer directions (say X and Y) than in the Z direction. This is particularly true for the extruded filament technique that we use. But it should be simple to program an RP machine to leave many small blind Z-axis holes in solids and then to inject-fill them using the write head when they reach the top of the solid. This should increase strength significantly, and also not slow the process down (the time taken to fill the holes should be less than that saved by not making them solid in the first place).

Just a thought.

What's the plan for getting the air out of the bottom of the holes?


Vik :v)
Objects made by FDM are slightly porous. The air should not get trapped.
When you are generating some kind of an overhang, it must be necessary for the plastic to harden very soon after exiting the nozzle - otherwise it'll droop down.

If the machine is designed to have the plastic harden that soon - then would it even be possible to get liquid plastic into the bottom of a deep hole?
I'm looking at a less viscous form of polycaprolactone from Union Chemicals. Apparently 1/5th the viscosity of Polymorph. It might reach the bottom more easily.

We could just build a shell and fill with Space Invader foam :)

Vik :v)
I was thinking that for a first iteration, it might be easier to build a pair of machines that when working together can self-replicate. An FDM to build structural pieces, and a router/driller to take raw cheap coper-clad PCB and spit out circuit boards. Of course, whether this will be a useful intermediate step depends on how easy it will be to create circuits with the FDM approach. (which we should find out soon I think)
Although not the Z direction, I suppose laying the lines that make up each layer in perpendicular directions when possible might also help strength. Perhaps that's just the norm though.

I guess there's always the possibility of composites too.
I wonder if slight variations in the size and shape of the holes from layer to layer would increase the strength. The poured pin would then have rough sides instead of being a smooth cylinder. Sort of like rebar in concrete. The natural variations in the deposition process might be enough to rough up the sides or maybe a bit of intentional error could be introduced to amplify the effect.
I thought I'd read something similar to this not too long ago. And what I'm now thinking may also be from something I'd read: instead of filling on subsequent passes, create geometry that allows a final object to be "filled" after the rp stage. This would be similar to how some of the metal rapid-proto systems work where they use a binder, burn it out, then fuze and or fill with another metal.
What about sticking a bolt in the hole?

Vik :v)
I was thinking of having a lot of holes - that's a lot of bolts...

I like Steve's lumpy idea.

(In response to 3d_geek, RP macines lay down support material for overhangs. We're thinking of using icing sugar; see the blog of a few weeks ago...)
I see a flood/fill nozzle coming along. It couuld even flood/fill in icing sugar.

Vik :v)
What's the steepest slope and thinnest wall that can be created without using support material? Could you, for example, create an object that's mostly blind Z-axis holes and fill them afterward? As long as you know the dimensions of each hole, over-filling shouldn't be a problem.

An algorithm that figures the optimal arrangement of holes and the volume of each shouldn't be too difficult. Is there any reason the holes need to be cylindrical? Rectangular, triangular or hexagonal holes would be easier to draw but might result in air pockets in the corners.

What I'm picturing in my head looks something like this:
1. draw the edges of the current layer
2. draw a grid to minimally fill the solid area
3. repeat for several layers with subsequent layer of grid draw on top of the previous grid, forming vertical walls
4. periodically fill the holes using a high-volume nozzle

The holes don't need to go all the way through the object, either, as long as they're filled before they're capped off. By varying the timing and amount of each fill, you'd be able to stagger the positions of the Z-axis blocks. Sort of like rectilinear Swiss cheese.
Alternately, it might be worthwhile to be thinking about such a grid-fill anyways - you could reduce the volume of material a part takes up when the full strength of a block isn't neccesary, and therefore lighten and cheapen the part - and have some of them that are kept smallish that you fill with your z-axis strengthening rods.

Perhaps define a standard pattern including both lightening and z-axis holes, to make design work easier?
But I suppose that's more a software trick, and my minor ambitions to work out a good human-readable language for the definition of parts, to be edited by IDE or text editor and compiled to code for the machine are showing...
How about honeycombing the interior and flood-filling with resin?

Unfortunately that would stuff the recyclable nature of the product.

Vik :v)
In answer to Steve's question about angle of overhang, our Stratasys machine will build a 45 degree overhang without support. That is the source of the teardrop RepRap logo that Ed Sells designed - it's the profile of a horizontal 'cylinder' that doesn't need support and which, in the vast majority of cases, works just as well as a circular cylinder for things like bolt holes.

Incidentally, the 45 degrees sounds like a number for algorithmic convenience rather than for physics to me, but I could be wrong.
the problem with making PC Boards is the parts are getting smaller and smaller, the lead spacing between parts is reducing.

image cutting traces that are 0.006" wide with 0.006" spacing. How about IC's with lead spacings of 0.016 inch

Now if we add the standard photo resist and uses the 3D printer as a drawing source - - - say with a nicely focus low power laser - - - that could work.
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to
Posts [Atom]