Saturday, 3 May 2008

Experimental extruder

I want to see how much of the Darwin design I can make out of HDPE as that is the plastic I have the most of and is the easiest to get hold of. It should also be the cheapest but I think I got a very bad deal with mine.

To extrude HDPE quickly, without losing accuracy, requires a fan blowing on the work piece while extruding at around 240°C. The PTFE insulator in the extruder starts to lose its strength under these conditions and it also extends about 0.5mm due to thermal expansion. The JB-Weld heater insulation also degrades rapidly. To address these problems I am working on a design using stainless steel as the insulator, which I first blogged here. Here is a second lash up I made to progress the idea :-

At the bottom is a brass nozzle made by the man himself, Adrian Bowyer, and is described here. It has already been superseded with the anti-ooze design shown here.

Above that is a brass barrel that came from BitsFromBytes, with my experimental Cerastil heater on it. I attached a thermistor to the barrel with JB-Weld.

The brass barrel is screwed into the end of a 1/4" stainless steel tube. The other end has been tapped with a 1/4" UNF thread and screwed into a small north bridge heatsink from a PC motherboard (40 x 40 x 15mm). I drilled through the centre and tapped it. To lock it in place and give a good thermal connection I made a square nut from a piece of 10mm aluminium bar. I spread heatsink compound on the threads.

The top of the stainless steel tube is screwed into an old PTFE barrel to join it to the pump. The barrel had swollen so that it wouldn't hold an M6 thread anymore, but fortuitously it seems to have swollen just enough to match 1/4" UNF.

This is by no means the final design, it is far too long and flimsy, it's just to test the concept using existing parts.

I also wanted to try insulating the barrel and nozzle with PTFE. I made an end cap that fits over the nozzle by plunging an 8mm end mill into a 12mm PTFE rod :-

The idea of this is to keep the fan wind off the nozzle and also give it a non-stick surface so that when filament curls upwards and will not stick to it. I also insulated the stainless steel tube with a piece of 12mm PTFE rod with a 7mm hole drilled through it. Here is the completed assembly :-

The gap in the PTFE where the heater and thermistor are and where the wires emerge is covered with fiber class wool. I hate the stuff, I only have to think about it for it to make me itch all over. It is a much better insulator than PTFE though, but I wanted something smooth and slender to not disrupt the airflow from the fan too much.

The wires are sleeved with PTFE insulation and then plugged into a floppy drive connector. So everything at the hot end is good for about 300°C.

How well does it work? Well it took me a long time to be able to get it to extrude HDPE semi reliably. Thermally it works well. With the fan off and the barrel at 250°C the heatsink only gets to about 45°C, easily cool enough to mate with HDPE, ABS and probably PLA and PCL as well. With the fan blowing it cools down to room temperature. The heater power goes from about 60% to 80% so the insulation works well enough. A better idea might be to lag the pipe with a thin layer of fiberglass wool and then wrap it with PTFE baking parchment to give it a smooth outer surface. Or maybe an outer metal pipe with fiberglass in between.

Mechanically it is not that great. It seems to a need lot of force to extrude. I had to open up the hole in the nozzle from Adrian's 0.4mm to my standard 0.5mm. I also had to up the temperature to 250°C. I think this is mainly due to where I am measuring it and how I calibrated the thermistor. Previously I measured the nozzle temperature and calibrated it with a thermocouple inserted into a hole in the nozzle. With this version the thermistor is in a notch on the surface of the heater barrel and I calibrated it with a thermocouple inside the empty barrel. Looking at the value of beta that I got I think that it is considerably hotter inside the barrel than the thermistor is outside. I am not sure how this is. With the heater on the outside of the barrel I can't see how the inside could be hotter. Perhaps the thermal connection of the thermistor to the barrel, via JB-Weld is not as good as it it could be. When sited in the acorn nut nozzle it was half buried in a hole.

Even with the nozzle removed it is quite hard to extrude 3mm filament by hand. Part of this has to do with how long the total barrel is and the fact that it has three joints. The inside of the stainless steel barrel is not as slippery as the PTFE. It might also be the case that the molten section extends further up the barrel causing more viscous friction. I plan to shorten the whole thing considerably: I will combine the clamp with the right angle bracket and take the tube right up to the base of the pump. I will support the heatsink with a cradle structure resembling an upside down table. More importantly, I will shorten the heater barrel by combining it with the nozzle and screwing the tube into it. Making it from aluminium, which is two and a half times a better conductor than brass and easier to machine, should make it easier to get a consistent temperature measurement.

As there is a continuous temperature gradient down the stainless steel, the point at which the plastic melts will be about halfway up so I think the heated nozzle can be quite short indeed. The limiting factor is how long it takes the heat to get to the centre of the filament with the very poor thermal conductivity and high specific heat capacity of the plastic.

Here is an HDPE version of the opto bracket with my best PCL version behind :-

I have no idea why it is so grey. It is not as neat as the PCL one but most of the errors are due to blobs forming when the extruder moves between extruding. These cause the nozzle to be displaced sideways when it gets close because it is so flimsy. Shortening it and supporting it properly will improve matters for sure. I also need to incorporate Adrian's anti-ooze valve somehow.


  1. A high temperature insulation alternative to fiberglass is rockwool. I have never used it, but according to:

    it has similar physical properties, but is not an irritant. The price for a 50mm x 6.55m2 rockwool cavity batt is 22 pounds, according to:

    I don't know if you can use batt insulation for the nozzle, but Fairalls:

    seems to have many insulating products and they may have a rockwool product than can insulate small stuff.

    If you're not going above 500C and nothing else not irritating works for the nozzle; you could try wool, which has an ignition temperature of 600C, according to:

  2. You should be able to go to a builders' supply warehouse or a construction site and get a scrap for nothing adequate to do half a dozen extruders. :-)

  3. "Adrian's anti-ooze valve"

    Have you got a link for that, please? I am building a reprap style machine with a 0.1mm nozzle, and the blobs are ruining the effect. It could be a combination of the reused plastic I ground up and placed in a hopper having different melting points... Which material do you use? My reprap design is capable of using rods, powder, and hopefully beads.

    By the way, where do you get your plastics from?

    Great blog, and you HydraRaptor is amazing!

    BTW based in Eccles, and will hopefully be appearing on the scene shortly.


  4. Hi Ginge,
    Adrian's first design is at but I think it wont be the final design.

    I am currently using ABS but I have also used PLA, PCL and HDPE.

    I got the ABS for, PCL from, PLA was a sample from Adrian and I think I got the HDPE from

    Sounds like you are building an interesting machine yourself.

  5. I would be a bit concerned about using ptfe if it's likely to over heat. Seriously bad fumes. If you find a shop that sells stuff for multi fuel fire etc they sell a number of things that might be more effective in rope and flat form. Fire cement might be effective too. Plumbers merchant and diy stores also do mats for soldering fittings on. These I think are a ceramic cloth.

    Just found reprap. I'm facinated. Hope this helps.


  6. Hi nophead, thanks for the great info.

    I hope that the bolt on is not the final design, but I do like the idea of the pin. I have my own spin on this idea that means I won't need another motor. The machine will have a pin or flap (testing needed) that engages when the drive thread is reversed 1/4 turn. This will create vacuum in the heating chamber stopping drips, and also mechanically engage the anti drip pin.

    My machine doesn't really have the reprap looks as I am finding a base plate that doesn't move more accurate than lugging masses of plastic up and down the z-axis. I beleive it is nearly impossible to calculate how much deformation and deflection there is in a flimsy frame under the varible loads. At least lugging the tool head up and down is a known quantity.
    It looks more like a CNC machine at the moment, albeit made from B&Q copper piping and masses of wires :)

    I will post up some pictures when I have the extruder re-assembled. They like to explode under the immense pressure at <0.3mm aperature.