What Material Do You Use?

I wonder if you could print a belt, then embed wire (or wire mesh) post-print by heating it.

What about using those PP strapping ribbons

they are pretty rigid (tearproof up to 300kg) and although almost inductile bendable.

maybe we can print directly onto that

'sid

... its possible and could be done in a fabbing-break and/or with finishing the back of the belt with some more layers of the resin.

Or imagine: - you fab the back of the belt with some V-grooves, insert thin steel-wires and finish the front with the teeth, so it's made with only one fixture.

For 'endless' belts you can fab on a big roll or elastic tube with the end-diameter of the belt ...

But it's never as strong as the embedded strings and fabrics in tech-grade compounds - they're made with high-pressure or with 2K-resins which are poured over the wires and then pressed and cured in shape.

What could be done is remelting the surface and then atach a prefabbed resin-fiber-compound, so you have a better material-homogenity through the belt.

Viktor

I was assuming you'd print the belt on-edge in a continuous, zig-zag loop. I figured printing flat pieces would limit the length.

Well Viktor, I thought more like "instead of steel wire"!

I tear apart my z-belt with my hands two days ago, no way that that was tearproof up to 300kg!
I tried the same with a strapping cord like that.. ending in searching my scissors
I think the real problem will be printing the teeth in such a way that they wont seperate.

'sid

I think I'll quit while I'm ahead since I don't see a practical way of fabricating a useful size belt on current machines! In any case to withstand the constant flexing I suspect that polypropylene would be a better bet than ABS - I think the belts use polyurethane and nylon facings for flexibility and wear resistance.

@nophead - My machine is a recent BitsFromBytes acrylic repstrap. I use the Arduino electronics with Gcode firmware, Skeinforge and send.py or ReplicatorG to send the gcode. The firmware is modified to include Zach's extruder PID adapted for single opto (no quadrature on the BfB extruder) and tweaked a bit.

The main problems I have are getting the rafts to stick and slight inter-segment pausing. The serial comms is really a bit slow although ReplicatorG is not bad at buffering the stream.

Using an 8mm/s extrude rate and 16mm/s movement has improved quality enourmously.

Regards

Peter

pjr Wrote:
-------------------------------------------------------
>
> @nophead - My machine is a recent BitsFromBytes
> acrylic repstrap.
>

So you did it on a BfB Darwin. That is a REALLY brilliant piece of work!

Hi sid,

... my post was the answer to Steve - it was nearly synchrone posting

Viktor

What's the most flexible material we can extrude? PCL? ABS?

I wonder how well something like Plasti-Dip (air-drying rubber) would do with a paste extruder.

PCL is most flexible out of (HDPE, ABS and PLA).

PLA is brittle.
ABS stress fractures if you bend it too much. You can make springs out of it, but not belts or hinges IMHO.
HDPE is quite flexible but only when very thin.

PP should be good, but its melting point is even higher than HDPE, so might be tricky to extrude.

Edited 1 time(s). Last edit at 10/16/2008 04:56PM by nophead.

---

[www.hydraraptor.blogspot.com]

Silicone caulking material should be easy to do.

I'm quite intrigued by the reprap fabrication of gears. I remember the day my C64 plotter stopped working due to a broken nylon gear that couldn't be glued. If there ever was a reason for this former 12-year-old to become interested in reprapping, it would be that. Just watching a plotter do its thing is mesmerizing enough, taking this to 3D making real objects is just a tantalizing idea.

nophead raised concerns about accuracy for making reprap's gears, stating that he'd have to get within 0.1 mm, a factor three better than his current (spectacular) results. My question is: can't we just use gears that are, say, 4 times larger, but with the same pitch. This would reduce influence of minor inaccuracies on the perimeter of the gear on rotation of the axle.

Of course, I assume out limit is absolute positioning error, not relative positioning. Assuming the screw rods themselves are accurate to well within our needs, it would seem that it is as easy to manufacture a 50mm gear to within 0.5mm accuracy as it is to manufacture a 12mm gear to that precision. If that's the case, we'd just have to use large enough gears. Am I missing something? ;-)

-Geert

We have target resolution of 0.1mm and use 200 step motors with half stepping.

The gears have to be small enough to give 0.1mm step with 1/400 of a turn. I.e. they need a circumference of 40mm which gives ~13mm diameter.

If a tooth was misplaced round the circumference by 0.1mm we would be outside our target and if the runout, i.e. radial inaccuracy was more than 0.2mm/pi we would again be off by more than 0.1mm.

---

[www.hydraraptor.blogspot.com]

What about using a larger pitch with the same/similar radius? wouldn't that make it easier to print a more accurate gear for a timing belt? I hope that's not too stupid an idea, I don't know much of anything about gears XD

I didn't realize how accurate those gears had to be!

Ah, I hadn't realized that we use the belt directly to drive the X and Y carriages, and wrongly assumed they were just used to synchronize the rotation of threaded rods driving a carriage through captive nuts. I guess that while it would be easy to get accuracy that way, it wouldn't allow for sufficient speed.

With M6x1 rod, you'd need 36 degrees for 0.1mm, or 20 full steps of the stepper.
For printing diagonal infills at 14mm/sec, we'd need to step both axes at 10mm/sec or 100 rotations/second. That seems indeed unreasonable and would be at quite a bit of mass at high angular speeds to stop when changing directions for example.

I guess I understand the importance of accurate gears and belts now for Reprap accuracy.

My gear was actually the 20 toothed one based on the one used by the BfB designs.

This means a circumference of 20 x 2.5 = 50mm and a resolution of 0.125mm with the current steppers and electronics. I am not entirely convinced by the accuracy arguments here - the belts are designed not to stretch, the flexible material of the belt will take up minor inconsistencies in the teeth and unless there is a pretty big error in diameter a rotation of 20 teeth in the gear means a movement of 20 teeth in the belt.

I guess I need to do a bit of experimentation to find out how sensitive things are to the gear accuracy.

I am also looking at using some micro-stepping driver modules to see if that improves things further.

Peter

Yes I suppose the belt will average out the positions of the teeth it engages with so I am being too pessimistic. It is actually the belt that gives the accuracy.

I actually have x10 micro stepping drivers on my Darwin but for some reason they don't seem to give even steps with the Keling motors. I haven't investigated with a scope yet but the same drives give very precise steps on HydraRaptor with different motors.

---

[www.hydraraptor.blogspot.com]

Wont the positioning accuracy ultimately be affected--and in a very complicated and stochastic way--by the build and the builder's skill? Because of this possibility, I still think Closed loop control is the way to go. Wouldn't it eliminate these issues?

Demented

Interesting.

I installed x8 micro stepping drivers on my X/Y axes a couple of days ago and found (after fixing an Arduino bug that was artificially constraining the step rates) that the movement was much smoother and less noise/vibration was generated.

Whether the additional precision will make a real difference is yet to be determined but the output seems 'cleaner' subjectively.

Ultimately a closed loop system could potentially give the best accuracy but it seems to me significantly harder to implement.

Peter