PTFE Tube Alternatives

[Ryan Carlyle]

I'm wondering if there's a better material for PTFE thermal barrier liner tubes. Printer manufacturers use them because they're nearly impervious to "cold zone jams" where retracted partially-molten filament solidifies above the heat break in the thermal barrier. The big problem with PTFE, as "everyone knows," is that PTFE's top practical temp is in the 240-260C range. This makes it unsuitable for most of the high-strength filaments on the market (nylon, PET, PC, etc).

One thought is simply using a higher temp rated perfluoro compound. There are tons of "PTFE-like" materials out there -- PVDF, PFA, etc. But I'm not aware of any that will reliably survive 280C as a solid. Maybe it's out there, I don't know. I bet it exists. I'm not a chemist though.

So an alternative that comes to mind is "filled PTFE" compounds. A solid powder or fiber filler is added to the bulk PTFE to improve the mechanical properties. (I use glass-filled and moly-filled PTFE at work for high-pressure seals.) The fill material provides dramatically higher creep resistance and improved thermal conductivity -- both good properties for a thermal barrier tube liner material. But it can also increase the friction and that would be bad. So it's an interesting material specification problem.

There are a lot of options here. Glass-filled PTFE has too much friction. Moly-filled, carbon-filled, or graphite-filled PTFE might work. There are also some interesting options like polyimide (Kapton) fill. It would be worthwhile for experimentation's sake.

[Ryan Carlyle]

Reading about that Griffin guy backflowing molten plastic up his Bowden tube at high extrusion speeds got me thinking. Pretty much every extruder reliability issue falls back to the fact that extruders are viscosity pumps. They build pressure by shearing viscous material. (Specifically they rely on a high-viscosity "cap zone" of semi-molten filament in the hot/cold transition area. This goopy plastic being sheared downward against the extruder walls by fresh filament feed is what contains melt pool pressure.) This causes issues:
- Most hot end jams occur because retraction breaks down the stable cap zone and pulls hot filament too high in the thermal barrier.
- PTFE tubes are added to make hot ends resistant to bad cap zone behavior.
- Oversized filament diameter jams happen because barrier tubes need tight clearances to provide enough filament cooling to keep the cap zone stable.
- Ooze control is challenging in part because cap zones only provide a positive pressure seal during filament motion.
- Even much of the JKN type speed/pressure compensation is necessary because of the cap zone moving up and down during extrusion rate changes.
- Lack of appropriate rheology to maintain cap zone behavior is a major reason why FDM printers don't print low-melting metal (eg solder) well. The metal transitions from solid to liquid too abruptly. All extruders in use today require a gradual viscosity change with temperature to function properly.

So, eliminate the cap zone requirement -- make the extruder stop being a viscosity pump -- and a lot of common printer issues may go away.

So what next-gen technology replaces filament-pushing viscosity pumps? The "obvious" answer is pellet extruders, but these are also viscosity pumps! The extrusion pressure is generated by a shearing action of the plastic between the screw auger and walls. An auger is just a bigger, slower viscosity pump. That's a step in the wrong direction.

Being a hydraulics guy, my immediate answer to inconsistent fluid flow is to run it through a positive-displacement metering pump. In theory, filament could be melted and simply pumped out the nozzle. But this is extremely difficult service for most small pump styles.
- I don't think a gear pump or vane pump could handle the viscosity without immediately "vapor locking" from insufficient suction flow.
- Normally for viscous-fluid service you would use a diaphragm pump or piston pump, but those are high parts-count and difficult to miniaturize.
- Progressing-cavity pumps would be conceivable, but would be difficult and expensive to build for this size and service temperature.

All of these would require "charging" with a viscosity pump to have enough suction side pressure to operate properly. So it's really a non-solution.

So then it occurred to me... Why not just replace the cap zone with an actual mechanical seal? Push filament into a melt zone like normal, but through a low-friction cup seal. High-pressure piston pumps often use very small Teflon or PEEK cup seals to seal around a plunger piston shaft. It's possible a 3mm bore high-temp rod seal already exists, although it would take a lot to find it. There are even flexible metal cup seals but the price is usually unreasonable.

Basically you would cool one side of the seal with some circulating liquid, and the other side of the seal would contain the melt pool pressure. Then the filament injected into the melt zone truly acts like the piston of a pump -- and the extruder is now a positive-displacement device instead of a viscosity pump.

Easy to build? No, not in the slightest.

[Chris P]

Here's an alternate approach (playing devil's advocate here).  What if we removed the need to zero out the pressure in the melt chamber by retraction?  Possibly by some sort of valve right at the nozzle tip.  I think some people are pursuing this...

If we don't need to modulate the melt pressure we don't need to retract and therefore the cap is not disrupted.

BTW I somewhat disagree with you about what JKN is compensating for.  Not only is it melt zone behavior, but the springiness of the filament in between the point of drive (eg the hobbed gear) and the melt chamber.  On a direct drive this distance is short and therefore the melt zone behavior dominates but the opposite is true in a bowden system.

[Ryan Carlyle]

I've heard of needle valves to shut off the nozzle. I'm not sure how you physically insert a needle into the melt zone without causing problems though.

Stratasys has an alternate model for extrusion flow rate in their patent literature. They model the hot end as an RC (resistance capacitance) circuit. Premise being, the current plastic flow rate is a function of melt pool pressure, and melt pool pressure varies as a first-order function of the difference between the current filament feed rate and the previous filament feed rate. Basically there is a capacitance-like time lag between input rate and output rate. It's comparable to applying a changing DC voltage to a capacitor -- the capacitor voltage does not change instantly.

How you physically interpret this (or JKN's quadratic compensation function) is an open question. Something in the system has some springiness. Whether that's Bowden drive axial filament compression or cap zone motion or something else is anyone's guess. Data could give us the answer but I don't see a lot of that floating around.