Brassed off

My first attempt at an extruder had stainless steel bearings and a stainless steel drive shaft, more by accident than anything else. I wondered at the time how the bearings would last compared to the recommended brass ones. Obviously stainless steel is harder than brass, but brass should have less friction, so less wear.

The first drive shaft got retired because the bearing lands were off centre. At the end of its life the bearings and lands were still in good condition.



I ran it from 10/07 to 01/08 but up to that point all I had made was lots of HDPE test shapes.

I replaced it with a plain steel shaft that I bought from BitFromBytes. That had the big advantage of being solder-able.

It worked well for a long time but eventually the bottom land on the shaft wore down so much that the pump halves closed together when using undersized filament (2.7mm).



The bottom land has worn down from 3mm to 2mm. The bearings show a little wear but still have some life left in them. This is after running 5lbs of ABS through the extruder.



In the last few weeks I replaced the shaft with a new zinc steel one and switched to brass bearings in an ABS extruder with HDPE filament guide. That worked well until I noticed the pump halves closed together again. When I opened it up I found that the brass bearings had rotated in the ABS, but they had also worn down a lot, considering the short time I had used them.



The drive shaft lands are still fine though.



So it would appear that the best combination is stainless steel bearings and a stainless steel shaft, but I would have to find another way of attaching the nut.

Most metal bearings I have recovered from old equipment are bronze. I don't know how that compares to brass but it seems to be the thing to use.

Maybe it is time to look at ball bearings and an offset shaft like Ian Adkins' design.

Brain dump

I have added a page to the new Reprap Builder's wiki recording all I know about the original MK2 extruder. I make no apologies for including lots of links back to this blog and recycling some photos, but there are also many new ones.

Although the design has moved on, I have stuck with the original and tweaked it to fix the problems I have encountered. I don't think any of the later, more complex designs have better performance, except perhaps the anti-ooze version. That seemed like a nice idea but rapidly got too complex with a double nozzle and two heaters plus some tricky machining. I think instead stringing can be mitigated to a large extent by increasing the speed of moves between extrusions paths. I am hoping to go from 32mm/s on HydraRaptor to 150mm/s on Darwin. Also I think reversing the extruder motor a fixed amount measured by the shaft encoder, plus Enrique's oozebane and comb modules in Skeinforge should help a lot.

Spurred into action

Forrest's success at milling spur gears with Tommelise 2.0 got me wondering how well RepRap can make them with Fused Filament Fabrication. An obvious test case are the two gears on Ian Adkins' alternative extruder design. These are laser cut, but look just about achievable with FFF. I couldn't find a picture of Ian's, these are a similar design by Vik Olliver.



Here is how they came out :-



The infill does not meet the toothed edge very well but they seem very strong never the less. They mesh well even though the involute profile seems a bit rounded off.

In an attempt to get the infill to work better I tried running my machine with 0.3mm filament, 0.24mm layers. I didn't make the nozzle hole smaller, I just changed one number in my software so the filament is being stretched finer. The result was interesting :-



The infill fills the teeth and is also very flat. The hole for the motor shaft has come out smaller and the teeth are a bit asymmetrical. They lean clockwise, the direction the outline is laid down in. I think this is because I am extruding 0.3mm filament through a 0.5mm hole so the point it exits from can move about but 0.2mm. I expect using a 0.3mm hole will solve that.

I also had a go at making a smaller version of my screw top pot with 0.3mm filament. I used the same g-code, I just scaled all the coordinates by 0.3/0.5.



It came out well and the thread still works.

With the finer filament it takes about 3 times longer to make anything so I upped the speed from 16mm/s to 32mm/s. That had an unfortunate side effect: The wall of the pot is quite thin so after the outline and inner wall have been extruded, the infill is a very narrow zig-zag. That makes it a high frequency movement. Speeding it up made it more like an audio vibration which seems to resonate the frame of my machine, shaking anything loose off the desk.


HydraRaptor growling from Nop Head on Vimeo.