Thursday, May 22, 2008


ABS Door handle

Adrian asked me to try his door handle design in ABS. Here is how it turned out: -

It took 2 hours to make (not including the raft) with a 25% fill. It seems like it would be plenty strong enough.

I added a 4.2mm hole for tapping for an M5 grub screw. I didn't teardrop it because below a certain size that is not necessary. It came out slightly under 4mm so I ran a 4.2mm drill through it and then tapped it M5.

Here it is with the screw fitted: -

ABS is a bit soft for tapping, so I think a captive nut would be a better design. We don't have any compatible doors so I did not try it out.

I think it came out very well, but being a perfectionist, here are a list of the defects: -
So I am pinning my hopes on anti ooze being the cure of all ills!


The quality on that is looking bloody good.

Vik :v)
So is the print quality good enough to make ball bearings or at least bearings using BBs?
Yeah, Nop has definitely cracked the print quality problem. :-)
Hi Gene,

Ball bearings are tough to make with fused deposition, because of the steps you'll get in one dimension. However, we can make 40 BC tech, roller bearings:

They're not as good as ball bearings, but we can make em now.. or at least once someone models one:)
Ball or roller earings and bearing races generally should be made of sterner stuff: rolling is only important at higher loads, but balls or even rollers would amplify the squishiness of the plastic, making the bearings less useful.

PTFE is good for journal bearings, but is entirely inappropriate for fused deposition. If you want to sinter powders (as additive PTFE manufacturing requires), bronze might be the better, cheaper choice.

And, as I've mentioned earlier, polypropylene is good for living hinges. Micromechanical systems make clever use of living hinges: I'd recommend looking through a MEMS gallery for inspiration if you're designing reprappable machines.

All that said, white cast iron is a reasonable material for bearings, and is amenable to lost wax techniques.
I think HDPE is our best bet for bearings at the moment, with a metal axle.

We might even be able to make a ball bearing like the one in the second picture here: -
In my view that is an industrial-quality rapid prototyped part...
That looks AMAZING.

One curious thing though - I see some brown tinting in the plastic in the photos - notably on the cylindrical section. I was wondering where that came from. If it were contaminated or overheated plastic - we'd expect to see stripes running through the object as one or other layer was affected. If it was some short-term random effect, you'd expect uncorrelated speckles. But it's hard to imagine an effect that would cause a curved streak that crosses many layers.

What is that?
As I explained in my third point:
"There is some brown discoloration, which is also indirectly due to ooze. When the extruder ends a filament run it turns off, lifts slightly and moves quickly to the start of the next run. Sometimes that leaves a string, sometimes it snaps and leaves a peak of filament sticking up. That rubs against the nozzle and picks up burnt plastic from it. Hopefully the anti-ooze valve will fix that as well."

It is not one feature across layers but actually individual blobs in slightly different positions on each layer, or more probably every second layer as the infill pattern alternates. They occur when the head leaves one run of filament and moves quickly to another. These points tend to happen in similar places on similar layers.

There is also a small bug in Enrique's fill pattern that sometimes produces a line of infill that overlaps another. That causes a bulge in the layer which rubs against the nozzle and also turns brown.

I have a nozzle with a PTFE cover which stops burnt plastic accumulating. I will give that a try an see if it fixes this problem.
The spatial resolution on this really is fantastic. Elsewhere I read a comment by Forrest that you're using a CNC table with a RepRap extruder. I had been assuming that the extruder was the reason why RepRap-printed objects didn't look as good as commercially printed objects, but this shows that I was mistaken. As an experiment to collect information about RepRap's spatial resolution error budget, this is very valuable.
I am using a table with 6um resolution but that is not the reason for improved quality. The 0.1mm resolution of the Darwin should be quite sufficient.

I think the reason my prints are more accurate is that my extruder has a shaft encoder to control the speed precisely, I have temperature control to +=3C, and I don't have any comms delays.

I also run my head faster so that I stretch the filament down to 0.5mm.

Molten ABS has virtually no spring to it so it is more compliant than PCL, and tends to make sharper corners, etc. PLA is also good in that respect.

All these things can be sorted out on Darwin so I expect its prints to be this good in a month or two.
Awsome ABS...

A quick question have you tried reducing ooze by doing a small reverse on the filament feed at each stop/non-printing-move to suck the excess back into the nozzle and reduce the amount of ooze.

Sorry I'll bet you have already tried this but I gots to ask.

Excellent pieces.....
No when I built my extruder with the flexible drive I realised it could not be reversed, so I made the electronics one way only. I could reverse now that I have switched to direct drive if I added an H-bridge, but it wont work as well as a valve. Even if you stop the extrudate it is still connected to the workpiece so when the head moves it pulls out more molten material.
Post a Comment

<< Home

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

Subscribe to
Posts [Atom]