Extrusion Methods

Hi! About 6 months ago I purchased my first 3d printer. It was a printrbot simple, and I really like it. I think that 3d printing is really a cool thing, and I want to get more into it. The printrbot gave me great experience, and I think that I am now ready to build my own printer. I have lots of ideas about how I can make it really good and precise, but where I'm kind of drawing a blank, is extrusion. I have a few ideas, but I don't think that it's enough. Basically, my ideas are to have something that detects exactly how much filament is being extruded, and can adjust things accordingly, but are there thing that I'm missing? Is there anything else that I could do to help the extrusion process? Thanks for any help you can give!

Controling the amount of extruded material is the key to high quality prints. Afaik noone tries to accomplish this by measuring the actual output during printing. The reason for this is that it is practically impossible to measure the extruded amount while at the same time get any kind of precision in placement of the material.
What is done is that you have to measure the exact diameter of the filament (taking into account that it might be oval) and enter this as a correction factor in your slicing software. Of course your extruder has to be calibrated so it can extrude precise lengthes of filament beforehand. With these two factors, diameter (or more exactly the surface of the crosssection) and the length of filament the software can precisely determine the volume of material that is beeing extruded at any time.
This is much easier to implement than constantly measuring the output.

---

[www.bonkers.de]
[merlin-hotend.de]
[www.hackerspace-ffm.de]

Quote
Srek
Controling the amount of extruded material is the key to high quality prints. Afaik noone tries to accomplish this by measuring the actual output during printing. The reason for this is that it is practically impossible to measure the extruded amount while at the same time get any kind of precision in placement of the material.
What is done is that you have to measure the exact diameter of the filament (taking into account that it might be oval) and enter this as a correction factor in your slicing software. Of course your extruder has to be calibrated so it can extrude precise lengthes of filament beforehand. With these two factors, diameter (or more exactly the surface of the crosssection) and the length of filament the software can precisely determine the volume of material that is beeing extruded at any time.
This is much easier to implement than constantly measuring the output.

Thank you so much for your reply! You said that you don't know of anyone who measures the output of filament as the printer runs, mainly because "it is practically impossible to measure the extruded amount while at the same time get any kind of precision in placement of the material." Could you elaborate slightly more on this? What does the amount of filament being extruded have to de with the placement of the filament? Or am I missunderstanding what you mean by "placement of the filament"? I assume this to mean the location of where the nozzle is when laying a layer of filament onto a print...

So one issue is extrusion, but another issue is cooling/everything that happens once the filament leaves the nozzle and is laid down... Does anyone have any ideas here?

Thanks again!

For the print to work the filament needs to be deposited immediately after melting and very precisely. You just don't have any opportunity to measure it after melting. The amount of molten plastic at any time needs to be minimized to be able to control the amount of extrusion.
To me it seems you want to solve a problem that just isn't there (we do have very precise control over the amount already) by using a method (measuring the extruded material) that would prevent you from printing at all.

---

[www.bonkers.de]
[merlin-hotend.de]
[www.hackerspace-ffm.de]

A useful way may be to take two measurements:-
a) The filament demanded by the software. This could be by a counter connected to the extruder motor.
b) The filament actually drawn from the roll. A pair of pinch wheels connected to a counter to measure the length drawn is one way to do this.
Comparing those two counts and the measured diameter of the filament would give you a fudge factor for that roll of filament which could be used to manually change the feed factor. There are many caveats though, changing or even cleaning the hobbed drive wheel will alter the filament fed. Even the day to day temperature could have a minor effect.

For cooling the extruded filament I use an annular ring of only 12mm diameter surrounding the nozzle. This is fed by a positive displacement air pump rather than a fan. This works well, though honesty compels me to admit that some fan set ups seem to be at least as good.

Mike

Quote
leadinglights
A useful way may be to take two measurements:-
a) The filament demanded by the software. This could be by a counter connected to the extruder motor.
b) The filament actually drawn from the roll. A pair of pinch wheels connected to a counter to measure the length drawn is one way to do this.
Comparing those two counts and the measured diameter of the filament would give you a fudge factor for that roll of filament which could be used to manually change the feed factor. There are many caveats though, changing or even cleaning the hobbed drive wheel will alter the filament fed. Even the day to day temperature could have a minor effect.

What you describe there is a fail safe mechanism to detect filament slippage/blocked hotend. For a controlled extrusion the prefered method would be a simple closed loop drive. The reason for using open loop stepper setups is just a matter of prive. I think this will change in the not to far future.

---

[www.bonkers.de]
[merlin-hotend.de]
[www.hackerspace-ffm.de]

I had more intended to address variability caused by material characteristics - the depth of bite of the drive wheel is significantly different between materials and even batches. While I use only observation of prints to get my "fudge factor" the last two rolls of black ABS measure up at 1.76mm but I set one at 100% feed and the other at 96% feed - the latter filament is detectably harder.

Mike

Quote
leadinglights
I had more intended to address variability caused by material characteristics - the depth of bite of the drive wheel is significantly different between materials and even batches. While I use only observation of prints to get my "fudge factor" the last two rolls of black ABS measure up at 1.76mm but I set one at 100% feed and the other at 96% feed - the latter filament is detectably harder.

Mike

I understand where you are coming from, but the teeth on the MK7 gears i mostly use are so flat that i doubt that there is any detectable difference in the amount of extruded material because of it. I never observed it anyway. I did have my doubts because of this when i used home made hobbed bolts for Wade extruders, but even with these, much more coarse teeth, there was never a problem like what you describe.

---

[www.bonkers.de]
[merlin-hotend.de]
[www.hackerspace-ffm.de]

Looking at the stubs of filament pulled out from my various extruders I notice not only a difference in depth of bite, but also flattening of the filament which would also give rise to a change in flow rate - though I am not sure in which direction.

I must also respectfully take issue with your assertion that closed loop servos would be a preferred method for extruders. In most cases allowing a degree of slippage would be preferable to putting in extra power to drive through excessive back pressure. If the closed loop was not from the drive speed, but from the linear length and cross section area of the pulled filament I think I could agree, but don't see this as being practical in the near term.

Mike

Yes, a closed loop controlled by the actual amount of filament transported would be nice and should be doable once you have a closed loop controller etc. at all.
Flattening the filament should not realy change the volume of extruded material. It will clog the hotend before it should have any noticable difference in the print. Myself i never noticed any real flattening outside of just plain maladjusted idler pressure when experimenting with new extruder designs.

---

[www.bonkers.de]
[merlin-hotend.de]
[www.hackerspace-ffm.de]

Ok so I feel that I really am then trying to fix a problem that doesn't even exist... So, I might just add something that stops a print if filament gets jammed or something, but other then that, I feel like extrusion accuracy is pretty good... My next question is more about stuff that happens after the filament has left the hotend, and has been deposited into a layer, then what happens? Cooling fan? Heated bed? Any other ideas? Is it better to have things cool slowly or quickly? Any ideas other than cooling ideas?

Thanks!

Different materials need different actions. PLA and PETG work best when cooled directly after extrusion, ABS is the other way around and works best with a heated chamber. There is no general rule.

---

[www.bonkers.de]
[merlin-hotend.de]
[www.hackerspace-ffm.de]

When the extrudate is first laid down it needs to be hot enough to partially melt the layer that it is being laid on. After this it needs to be cooled rapidly to a point where it is somewhat solid; if it is cooled too slowly fine details will be smoothed out by surface tension but if cooled too fast the adhesion to the lower layer may be poor. After this initial rapid cooling the whole of the print should be maintained at a temperature where the stresses will dissipate - somewhat like annealing metals.

There are many differences not only between different types of plastics (as Srek pointed out) but also between different manufactures, colours and batches. The shape of the print and how easily cooling can get between the parts can also effect the final result.

The whole field of 3D printing, particularly in the enthusiast's arena, is so empirical that the best idea that I can convey is to keep copious notes - labels with details of successful and failed prints on reels of filament, notes in text files along with the STL and GCODE files, Post-it notes plastered on your printers and on the wall.

Despite being a deeply scientific/technological person I still find that incantation, ritual and prayer seem to help - particularly when prepping a bed for a large ABS print.

Mike

Wow... thanks! I really didn't know how important the cooling aspect of things was... I'll to some serious thinking and some serious research on this, thanks!

Also, do you have any ideas on different things that I could maybe do to increase the quality of the prints? Do any of you guys have any special additions to your printers that affect how each layer is laid down that dramatically increases the quality of your prints? If so, please let me know what they are, if not, I'll just leave this here.

Thanks for the input, and I will be doing lots of thinking!

Sadly there is no magic gadget or button for quality. It is a lot of work and refinement to get optimal quality.

---

[www.bonkers.de]
[merlin-hotend.de]
[www.hackerspace-ffm.de]

Quote
Srek
Sadly there is no magic gadget or button for quality.

Right right... I'm just trying my hardest to think of a way that a printer could adjust at least some of these settings automatically to try to get the optimal quality each time...