Saturday, April 12, 2008


Some estimates...

Assuming that the convex surface of warping is spheroidal and assuming that Chris' measurements for warping on his 40x10x20 mm test block were taken at the extreme midline rather than the corners I come up with a radius for the spheroid of 425.3 mm for the warping of 0.47 mm for his 50% fill HDPE test block.

If that is meaningful then if Chris printed a 40x40x20 block of HDPE at 50% fill one would expect that the curling at the corners which would have a distance from the block's centroid of 28.8 mm would be about 0.94 mm. The warping at the midline would remain at 0.47 mm, of course.

(No additional content at

It's definitely not a spheroid because it has increasing radius of curvature towards the corners. It is more like a catenary.
Have you mapped it?
First off - I have no technical experience and this kind of thing is completely not my area, so the following may be complete rubbish, but here are some random thoughts anyway...

If you printed out a large flat surface and then turned it upside down wouldn't it demonstrate exactly the curvature you wanted?

And if it wouldn't (i.e. thickness + the weight of extremal parts affecting curvature across the object) then doesn't that kind of scupper the one-template-fits-all idea?

Maybe another solution would be to design an item-specific blank that could be printed with each item (basically the same shape as the base) turned over, coated with something to avoid bonding and then used to print on to.

I agree with nophead.

It certainly makes sense that the curve would be a catenary rather than a spheroid. In architecture, catenary arches are used to optimally distribute pressure through a structure like a doorway. It makes sense that this warpage is the material's natural tendency to equalize tension through the length of the block. So just as soap bubbles naturally form spheres to optimise for least pressure over minimum surface area - so RepRap's blocks are naturally forming catenaries.
Until you measure enough points on a variety of such printed surfaces the form of the curve, if indeed there is a consistent one, is pure conjecture.

I have no idea what I'm talking about, but anyway...

I was wondering if some of the warping might be due to the fill pattern. If you lay down a cross hatch pattern with the strands all straight then when they contract do they pull along their axis?

I thought that maybe if you used something like the Hilbert Space filling curve then when the strands contract they might have some extra springiness at the corners of the curve. Maybe some of the overall stress and strain would go into slightly deforming the curve corners on the interior of the object rather than pulling in all one direction and bringing the ends up the object up?


No I haven't mapped it but I could, I made a touch probe some time ago and have never used it. Assuming it works this would be an ideal use for it. Just too many things to try and not enough time.

The warping gets less with a sparse infill but even with no infill at all an open box warps in very much the same way.
Interesting fill pattern though, I would love to give it a try.
OK, this may be stupid but it has been bugging me for a while...

Presumably there is still warping even in no infill because the border itself is still in effect 100% "fill". What if along the Z axis gaps where left to short circuit the warping?

I am not sure if this will show up in ASCII art or not...


Something like this only with multiple (unaligned) gaps per layer?
Post a Comment

<< Home

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

Subscribe to
Posts [Atom]