2011: An Extrusion Odyssey

When I had a look at the innards of the extrusion head for my FDM 1600, rebuilding the head (should I ever require it) looked doable, if daunting. It now looks even less daunting, thanks to Rob Falkenhayn over at Incredilution. Rob had a leaky liquifier on his FDM 2000, caused by a cracked liquifier inlet. With Stratasys wanting $5000 for a new head, Rob did what any sane person would do – he machined a new inlet and repaired it himself. His blog post is a great read for anyone who might have to face head repairs on an old FDM.

Since the previous post on my rapid prototyping journey, I’ve churned through all the paperwork needed to become a customer of Ashland Distribution, ordered my plastic, and lined up Chuck Hamley at Advanced Extrusions to turn the raw granules into the 0.070″ filament that my machine uses. The cost has not been cheap, but the whole experience has been invaluable – today I got to see my filament actually being made.

For small quantities, the raw plastic resin comes in a plain 50 lb. box, though I’ve also seen it in sacks.

The granules are just stubby little cylinders.  As this video shows, plastic strands come out of a multi-orifice extruder at the plastic production plant, and then get chopped into very short segments for storage and eventual shipping. So even though the raw ‘virgin’ resin is new, it’s actually already been extruded once already.  Before the resin can be extruded into filament, it first has to be dried to drive off any absorbed moisture (ABS is hygroscopic).

The hopper and funnel sit at the rear end of the extruder.  The dryer itself (the black unit to the right, with hot air supply and return hoses running up to the hopper) is a Conair (curiously unrelated to the hairdryer company, it appears) model D-75 and can dry 75 lbs. of resin per hour.

Here’s the extruder itself – it has 4 heat zones (zone 1 next to the input by the funnel, zone 4 at the output end) with the temperature increasing as the plastic is moved through the screw (located under the black housing).  In this case, zones 1-4 were set at 355, 387, 397, and 401° F respectively.  As Chuck Hamley explained to me, the resin datasheets will list processing temperatures, but these are either a very wide range or will be a temperature that worked for the manufacturer on their particular equipment.  No matter what, the extruder or injection molder will have to tweak these settings to get the best results.  Also, note the knob on the lower left of the control panel – this is the potentiomer that controls the screw speed itself.  The extruder needed to be dialed way down to nearly its slowest speed for extruding 0.070″ diameter filament.  Advanced Extrusions has a smaller extruder that might work better should I decide on additional runs in the future.

As an example of what temperature will do, at the bottom is how the MG47 was coming out of the extruder at the very start of the run.  The lumpiness is actually somewhat regular (more apparent if you look at a long length of it) and is an indication of the screw ‘pulsing’ the plastic out rather than smoothly feeding it.  By increasing the extruder temperature (and making adjustments to the puller speed), the lumps were smoothed out and resulted in the correctly shaped (though oversized) filament in the middle.  Further tweaks to the system yielded the filament at the top, well within the required tolerances.  As you might guess, this takes a few pounds of material to get the final product to size – after each adjustment, the system is allowed to settle down to a steady state and the resulting size is checked again.

This is the actual business end of the extruder, and you can see the white plastic filament exiting the die and proceeding into the water bath tank.  Note the rectangular band heater clamped around the die itself for heating the exit nozzle.  The pressure gauge displays the actual extrusion pressure – 2500 psi.

Here’s a little better view of the filament entering the long water tank.  It’s not easily visible (and even in person, you have to look at it for a bit to make sure you’re not seeing an optical illusion), but the filament is tapering slightly to a smaller diameter between the die and the water tank, as it is actually being pulled further down the line.  Since the plastic will be nearly ‘frozen’ to its final size as soon as it hits the water, adjusting the distance between the water tank and the die exit is critical.  With this in mind, note the handwheel at the lower left – this will move the entire water tank left or right, allowing very precise tweaks to be made to the resulting filament diameter.  Super clever.

Here’s a look down the length of the water tank.

After exiting the water tank, the filament passes over a long rack in the open air and is allowed to dry off. The big upright rectangular unit is a chiller for the water tank, but for such a small extrusion, cooling the tank wasn’t needed.

At the end of the rack is the puller, which true to its name, is what actually pulls the extruded filament through the production line.  For this particular run, it wound up running at a speed of 50 ft/min (do the math – extruding 5 lbs. of ABS takes nearly an hour).

The puller has two big orange rubber belts that drive the extruded material through.  There’s a rotary encoder on a swingarm at the top that actually measures out how many feet have been run.  The unit to the left is a flywheel cutter, and the filament is running through the cutter die.  The cutter was set to chop the filament into 18″ lengths or so during the setup phase so that the diameter could be easily measured without having to deal with a giant birdsnest of filament piling up.  Once the diameter was dialed in, the cutter was turned off, and the filament was directed into an empty gaylord where it was allowed to accumulate.  At the end of the extrusion run, the fresh free end of the filament in the box was pulled off and wound onto spools with an electric spool winder.

And here’s the very first result of all this work – 3 lbs. of filament (will be 5lbs. on the rest) on a cardboard spool, ready to be tested on my Stratasys, a 1.75mm RepRap, or an UP! printer.  The remaining MG94 ABS and 5308M HIPS will be run in a similar fashion in the next week or so.

12 comments so far

Add Your Comment
  1. super cool

  2. Wonder full , I´ll like to know the price , i am very happy whit this results and I saw a very good machines to do our filament . When you do the first model in your FDM 1600 stratasys ?. Is possible to use this material in this week?. Or need to waiting to dryer . Ok Have Blue thanks a lot for to show me this odyssey . I wait for your news .


    • Jose -

      I’ll have to wait until I know exactly how much material I have before I can determine a final price (the extrusion cost twice what was originally quoted, so some adjustment may be needed).

  3. Wow, fantastic post. Great to see you’re forging ahead with actually testing some plastic, after all of that conjecture, googling and reading of patents.

    I’m awaiting the test results with bated breath, and rooting for the MG94!

  4. Great Post! I need to check the ‘rebuild head’ blog as I’ve got a bad one on the shelf too.

  5. [...] Have Blue [dot org] » 2011: An Extrusion Odyssey [...]

  6. AAAAAAARGH!!!!!! Now you have to help me design this leviathan!

    • I’ve certainly thought long about it, and ‘leviathan’ seems an apt name for such an undertaking…

  7. Hello, I am currently trying to produce Abs and Pla filament with a similar extruding system like you have. I have some problems with the raw material choice and the water tank temparatures. Which specifications should have the Abs resin? And should i use a hot water tank before the cold one ? Thank you

    • The water tank was about room temperature – there was a chiller available, but we didn’t use it. The ABS resin that worked best was SABIC MG94.

      • Thank you for the info. Are you still producing filament?

        • No, it was too expensive for me to have done in small batches. Fortunately, there are now several suppliers that have begun offering MG94 filament, which was what I was hoping to spur in the first place!