What would be the ideal (theoretical) hotend/extruder combo?

I wonder what would be the ideal hotend... for either PLA or ABS. Could we come up with an idealized model what _should_ happen in the extruder independent of whenever this can be mechanically implemented.

What would be the goal of this exercise?
After thinking and reading about several variations on materials and heating I've come to the conclusion that there are just too many possible combinations what we could try. So I propose we work the other way around:

There are some constraints we have to assume anyways though:

The Filament is compressed to a smaller diameter using pressure one way or the other.
Ideally the material of the nozzle and filament should have zero friction.
In reality we do not have that but we can optimize the slope of the nozzle to be ideal for the material we are extruding. For example we could devise a model for calculating the most efficient shape using parameters like:

-) mean filament diameter & tolerance
-) friction of the material
-) viscosity & surface tension of the molten material
-) poisson's ratio of the solid material
-) glass transition temperature of the plastic
-) specifc heat capacity of all materials
-) thermal conductivity of all materials

... and much more
the list is pretty extensive all ready and there are several other parameters which could be optimized...


Geometry of the extrusion screw.
The optimum would be a rotating dough-nut shape which cannot be done but we could try to come closer than what we are using right now.
We could optimize the teeth of the extrusion screw(s) with these parameters

-) impact strength of the solid plastic
-) flexual strength of the screw & the plastic
-) desired speed
-) power needed to overcome the friction of the hotend.

Mechanism of heating
Right now we all use resistance heating but there are several other options depending on which material we choose for the nozzle.


-) Inductive heating
This would require some amount of ferromagnetic material in the nozzle but has the advantage that the nozzle could be made out of a non-heat conductive material like peek, glass or ceramic. All could be machined with a cnc mill and lathe and the latter even extruded to some extent.
One suggestion is to mount a tiny steel needle in the center of the nozzle which should poke the filament in the center. The material must then flow around it with a common orifice at the end.
Other alternatives would be a steel mesh and a flat spool. This would have the advantage of a very small heating mass and thus faster cool down times.
The most straightforward way would be to make the nozzle out of steel itself which should also work.

The advantage would be that the temperature gradient can be optimized using the with of the spool windings and a higher temperature gradient as a whole.

-) Peltier element
Would have the advantage of also providing cooling and a temperature gradient otherwise not possible.
The end of the nozzle could also be cooled to freeze the ooze momentarily with an extra element.

-) Electrostatic Heating
Highly speculative, but is said to be possible. It would employ tiny electrodes inside the nozzle and using high voltage to excite the molecules at the resonant frequency. In theory this would enable us just to melt the end of the filament completely using a set of electrodes at the end. They could be made with lithography. I don't know if this is viable for the whole heating process but I thought I'd mention the idea also..

-) Microwave Heating
As above,
a magnetron from a microwave oven could be used with a flexible waveguide ending inside the nozzle. This would at best go in hand with a bowden cable extruder mechanism, so we could reduce the moving mass as far as possible.



Which heating process should be best would depend on the above calculations, and it could well be that resistive heating is still the most viable option. But my bet goes on a system where a peltier element is used to rise the temperature to the point where the surface tension / viscosity holds the material in the nozzle and using electrostatic electrodes for the rest. Since ideally they would only heat the plastic itself it would stop any oozing as soon as the electrodes are switched off.

Edited 5 time(s). Last edit at 06/29/2011 02:54PM by ElectricMucus.

Ideally the fillament drive screw would only apply downward force with no twist. Any twist can cause issues with the fillament feeding from a spool. A hobbled pully on both sides with both being driven is a good approximation to this.

Jeremy

Yeah that would maximize the contact area, but might be very challenging to implement mechanically. Or did I understand you correctly? what exactly do you mean by hobbed pulley?

I think the major advantage would be the larger contact area. It is possible to use piezoelectric elements to move something in a similar fashion.
look at [www.pcbmotor.com] for an example. If there are elements all around the filament the force which could be applied would even be high enough, but you'll need several elements which might be as expensive..

But it would be very precise and probably more reliable...

If it would be only about the larger contact area I also found this concept:



This is used for extruding copper where the force for the extrusion must be several magnitudes higher... we could adapt this concept too.
The downside is the "scrap" material inherent in the process, but we could probably fix this if we guide the filament right into the nozzle.

Edited 2 time(s). Last edit at 06/30/2011 10:22PM by ElectricMucus.

I had another Idea of how heating of the filament could be done in a more efficient fashion.

The idea is to use a infrared transparent glass for the nozzle. There have already been successful experiments with glass hotends, but no good method of heating.
Using a red-hot heating filament around the hot-end and a aluminum reflector could provide heat where we need it.
We could use a steel spring though which we get a sufficient high current. Contact between the spool and the hot end should be avoided, since we want to heat the filament and not the nozzle. A toroidal aluminum reflector can focus the infrared to the right spot.

This would have the advantage that we do not have to heat the nozzle itself to this extent, we would get some heating in the nozzle since the molten material would heat it and the glass wouldn't be completely transparent. But we should still get a sharper temperature gradient than with a heater block.

Glass blocks IR.

IR grade fused silica would allow the IR to pass.

Any material will allow only a narrow band of wavelengths to pass through. The rest will either be absorbed or reflected.

That is true - I did overstate the case. IR fused silica only allows .2 to 3.5 microns or so to pass, so it would allow a significantly wider region of IR to pass than glass. I don't know whether or not that is sufficient to do what ElectricMucus suggests. I only meant to suggest a material that should work better than normal glass.

Edit: Well, it turns out that normal soda lime glass allows .3 to about 2.7 microns to pass, according to Wikipedia, so fused silica is only very slightly more useful. Never mind.

Edited 2 time(s). Last edit at 07/25/2011 01:21AM by madscifi.

According to this graphic there are several ir transparent materials, mostly salts (maybe we could machine a nozzle out of a solid salt crystal)


What would have to be done is checking the availability of these materials and whatever temperature (black body radiation) would be required to get the best transmission for each material.

A lot of those salts are rather poisonous, and most are sensitive to water (the water content of the plastic (and air, really) will quickly ruin the nozzle). Silicon or germanium look like they might be plausible candidates.

Problem with germanium & silicon is that their transmission spectrum isn't as wide, which is fine if you want to focus a laser. And they transmit only the long wavelengths which would give a very poor efficiency overall.

Well there is always table salt. (home grown crystals) or rock salt.
Too bad it would probably be too soft for our uses, and it's hygroscopic.
Otherwise it would be nearly ideal... and it would look neat, a softly red glowing nozzle with a aluminum mantle above it

The question is if it where feasible to construct a lens optic or not, otherwise glass would probably still better since it is strong enough that we could make the hot end much thinner.
But glass would require something like a toroidal halogen lamp or xenon flash tube for heating, not exactly cheap and reliable.

Maybe try to grow a salt crystal around a thin-walled glass tube & lathe it into a lens...? well wild brain-storming here

I think I've got it



119 infrared leds, if we assume 0.1watt per led we would have 11.9w power, this should be sufficient if focused correctly, we wouldn't have nearly that much power in form of infrared but with resistive heating alot is lost as well, so it might just work

... it will not...

if the transmittance of other polymers is any indication, they pretty much are transparent to most of the spectrum. too bad..

however: Different polymers have different absorption spectra with spikes in between.
I haven't found anything on PLA but here is a spectrum of ABS...


As it seems it has a spike at 704nm which could be utilized with 700nm leds but more research has to be done.

Another option might be trying to get to the other side of the spectrum in the uv range, at which most plastics are opaque. I doubt that the wavelength of uv leds would be sufficiently short besides them being expensive. But there might be circular uv lamps out there which could be used, again this also depends on whenever there is a sufficiently transparent nozzle material.

But I won't give up... somewhere in the electromagnetic spectrum has to be a spot which could be utilized to heat the filament directly..

PS: I find it quite difficult to get any clear data on the absorption spectra of pla or abs. Every image graph seems to be purposefully shrunk to prevent one from getting the data without paying for a download of the paper. Damn the third-party commercialization of academia...

Edited 5 time(s). Last edit at 07/25/2011 04:06PM by ElectricMucus.

I have now found a source for the absorbance spectrum of PLA.



As it turns out the only part which shows significant absorption is in the UVC spectrum... so that rules out pretty much any other source except a circular UVC disinfection lamp. At 253nm this graph would be around 0.75-1 so we would have between 82% and 90% of energy going into the pla.
Well enough.

The good part is there are plenty of circular uvc lamps out there with sufficient power, for ex..



A potential problem would be that that ultraviolet could damage the polymer structure if used in this fashion... anyone knows more?

ABS and PLA do degrade rather quickly in UV - try putting a piece outside in the sun for a couple of weeks. I suspect enough UV to melt the plastic will cause problems, but that suspicion is not based on any hard evidence.

There seems to be not much left except testing it

From how I understand it polymer degradation occurs by the formation of free radicals which react with air over time. In this case the time in which this can happen is minimal and we do not have any air in the extruder. So this could even be beneficial if re-polymerization occurs in the molten material.

My assumption that a uvc lamp is the only source which can supply sufficient absorption was also incorrect. The graph above is for a 1 mil thick film of plastic, so a 1.75mm thick filament would be 70 times as thick. This means that there might still be other sources which could be used. The UVC lamp is a no-brainer but it might even better if a wavelength is used where the material is still mostly transparent so that absorption can take place further inside the filament.

Hi, I'm new here. I'm not an engineer, so my ideas will probably sound outlandish. Think of me as the "think outside the box" guy

For power delivery: How about fiber delivered 808nm laser (common diodes) to the hot end which is coated in a matt black ceramic material to efficiently absorb the frequency?

For modified feed, how about an internally threaded rod rotating on a bearing to drive the filament into the hot end. Or is the filament too uneven for that sort of arrangement? The beauty of a threaded rod is you could place several of them in a circle around the stepper and use a hobby servo motor to move a tertiary drive gear around to the desired filament.

... i have fibercoupled IR-diodelasers with 5, 9 and 25Watts @975nm through a 0.1mm fiber and another diodelaser with 25Watt @808nm through a fiber-bundle with 0.8mm diameter.

With this focussing isn't a problem - i cut/engrave with 0.2mm cutting width and are experimenting with sinterimg/melting of different powdered materials for laser-sintering ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

After letting those ideas sit for a while I now think most of those ideas I had are pretty worthless... in terms of efficiency.
While electromagnetic energy might be ideal in theory there are no practical sources which are efficient enough, basically everything is under 50% efficiency.

Also the costs wouldn't make it viable, expensive materials, lasers and so on aren't really something I am looking for.

On the other hand electrostatic heating might just work. But this might be dangerous.

Obtaining high voltages is pretty trivial and finding a nozzle material which complex resistance is opposite to pla which is also available shouldn't be hard to find. PTFE or glass comes to mind. And since we already use ptfe for our extruders and it is pretty much the best insulator there is chances are there is a frequency where pla melts and the teflon stays unaffected. Of course glass would be even better.

I think I will attempt to construct a high voltage function generator, put a string of pla between the electrodes and try to find out in which way you would have to sweep the frequency to melt it.

... the prices for high-power IR-diodelasers aren't so extraordinary!

You can buy comercially a fibercoupled diodelaser @975 or @808nm for around 300 Euros.

I sold the last salvaged 'naked' 5Watt@975nm-diode with atached fiber for 50 Euros ... and a complete set with optics and driver for 300 Euros ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

I saw a 3D printer today at a trade show that fed powdered metal through a small nozzle and melted it with a laser and deposited it in layers just like FDM. It produced very detailed metal objects. They did have a warping problem though due to differential contraction just as we do with ABS.

---

[www.hydraraptor.blogspot.com]

... they should heat the building chamber for reducing warping (and lower needed laser power), but with much higher temps than with plastic

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

Maybe I can help,

we have a wide range of extruders for ABS:
our lines: YJ PFR, YJ-PFRS, YJ-PS.

We recycle and sell machines.

Please see our webpage:
www.zaptech.eu

Adam

You can see some video here:

It's the most simple, 3 machines in 1 line:
[www.youtube.com]

or twin-screw:
[www.youtube.com]

Adam