Wednesday 26 August 2009

Fast extruder

I put together my new extruder controller, the worm pulley drive mechanism with the GM17 tiny stepper hack and the stainless steel extruder with heatsink and ducted fan to make possibly the most complicated extruder design yet!

You can see a better view of the drive mechanism fitted on another extruder base here: -

Here is a reminder of what the heater assembly looks like: -

The heatsink is cooled by a tiny fan. When run from 12V it is very noisy and way too powerful. With my new controller I can run it with PWM just a bit faster than its stall speed. That keeps the noise down and still gives more cooling than needed. I attached a thermistor to the heatsink by gluing it into a crimp tag with J-B Weld.

I can tell the controller to keep the temperature below a specified level by turning the fan on and off. I set the trip point to an arbitrary 35°C. It will even turn it on when the extruder is idle, much like the radiator fan of a car runs after the engine is switched off. This is needed to ensure PLA will never soften and jam in the cold part of the tube.

I run the tiny stepper motor at about 300mA to keep it cool enough to touch. It will take more current than that but runs very hot. A good design would use a single fan to cool the motor and the heatsink.

I ran the motor with micro stepping, so even though it has a 15° step, that gives 192 steps per revolution. The GM17 gearbox has a reduction of 228:1 giving a massive 43,776 steps per revolution of the worm pulley. That seems a lot, but the diameter of the pulley is 13mm, so one turn is 40.84mm of feed. That gives 1072 steps per millimetre. In comparison I have been using an 816 step shaft encoder and an 0.8mm pitch thread, which gives 1020 steps per millimetre, almost the same.

I started extruding ABS with my usual feed rate of 16mm/s for 0.5mm filament, which is 3.14 mm3 per second. I kept doubling it until it failed, which was 128mm/s if I have got the calculations right. At that point it mostly worked but something was slipping occasionally. I think it was the clutch in the gearbox. Backing off to 64mm/s it works fine. That is four times faster than the GM3 manages with a screw drive. It is too fast for HydraRaptor but I reckon my Darwin could go that fast. I have no idea what the build quality would be like but it would get the time to print one down to about 24 hours.

Here is a video of it spewing out plastic.

Fast Extruder from Nop Head on Vimeo.

It isn't mechanically compatible with HydraRaptor without making a new bracket to mount it on the z-trolley, so I haven't made anything with it yet.


  1. That's a damned good piece of work!

  2. Awesoooome!

    Assuming there were no pesky real world limits, with a 24 hour duplication time, everyone in the world could have a Darwin in 33 days:)

  3. Enriqu I think Dawin would be impressed at the speed of the evolving RepRap..

    Great work again.. I like the crimp connector idea .
    I would have instictivly put the thermistor at the screw junction of the welding tip and the alu heater block. Is this to do with the themal mass of the extruder / better temprature control as it like the thermal center of gravity of the extruder?
    The location of the thermistor is not where I would have put it.

  4. The thermistor in the crimp terminal is mounted on one of the heatsink screws to sense the temperature of the heatsink.

    The heater thermistor is embedded in the aluminium cylinder. The bang bang control has less overshoot and the plastic spends more of its time there. As plastic has such a high specific heat capacity and copper is such a good conductor, I don't think it cools much in the nozzle on its way out. I will add a PTFE sleeve to insulate the nozzle (and stop plastic sticking to it) before I use it to make something.

    I have had thermistors in various places and found this sort of arrangement to work the best.

  5. As always, great work!

    An idea for cooling the stepper moter. Competition RC cars have heat sinks for the motors. You might be able to find a round heat sink that fits your stepper motor at a large hobby store.

  6. Thank you for clarifiying the thermistor locations.
    I have seen these heat sinks they do look like they could be very usefull. The Model shop I found that has plenty of light enginerring bits and bobs like this is at Frodsham they do Mail order.. Thou I gues there might be a few model shops in Manchester.


  7. Say Nop,

    Couldn't you rotate that motor so it hangs down instead of up and then just print a cowling to make an air-duct from it that connects to the air-duct that comes off of the heatsink and then pipe both of them into the fan? That way you would be drawing cool air over both the motor and the heatsink with the single fan as you talked about and all you would need to change is the orientation of the motor--which I'm assuming is rotation insensitive--plus adding another small printed bit. Would likely extend the useful life of the motor by quite a bit if it gets very hot right now.


  8. Yes I think that could be done. The motor does not run very hot at 300mA so I would only need it if I needed more torque. I don't think cooling the motor affects its lifetime. Either it gets too hot and melts the insulation and burns out, or it doesn't and works.

    Capacitors and semiconductors last longer at lower temperatures but I don't think stepper motors do.

  9. Ah, I thought it was a DC brushed motor. In that case cooling will do nothing but make it a more robust design that can handle higher torque requirements which it is sounding like you don't need.

    In any case I've read this post two or three times cause I like how simple and clean the design is even at this phase of development. I can only imagine what it will look like with a few more design iterations. Of course, this is the result of a lot of learning already so seeing where it came from is also pretty impressive.