Filament

Quote

With rough feeding the filament beside the mirrors (e.g. two wire-loops) there's no need for aditional mirrors or ball-bearings - the IR-software is capable of 'ignoring' the displacement of the filament, as it's only measuring the relative distances of the contour-lines.

But the filament is oval in a axis that you have no control over. It rotates at will, and is too stiff to twist it to line of with the two measurement axes. I suppose with two mirrors you see three mesurements but if the major axis of the oval is in between two you get some error.

The webcam idea is fine if you are using a PC to control the machine but I am intending to make my machine autonomous like Forrest has done.

Peter, you are left with the problem of how to measure the liquid volume to a high degree of accuracy. A small ballcock with a shaft encoder on the arm? The liquid would have to be something that does not evaporate and does not cling to the plastic, or it would diminish with time.

Another crazy idea following on from liquid: A small chamber with a sound transducer that the filament passes through. It will have a resonant frequency that changes as the plastic displaces air.

---

[www.hydraraptor.blogspot.com]

Hi Chris,

... i think the error in circular/ellipsoid shape isn't so big that you can't estimte the correct metrics.

Picture an ellipsoid with two orthogonal viewing angles and calculate the diameter by { D=(d1+d2)/2 } - the error is biggest, when the a-axis of the ellipsoid is aligned to the camera-direction or 90 deg away. But it's a very small error, as the filament should be only slight distorted ...

Another crazy idea - maybe it would be helpfull, when we squeeze the roundish filament in a square shape? Easier to feed, measure and align?

Viktor

New image makes plastic filament (welding rods) in several shapes, mostly triangular and variations on that theme though not square.

[www.plasticweldingrod.com]

For myself, I'll be sticking with round since that is what most manufacturers are prepared to produce.

Edited 1 time(s). Last edit at 10/07/2008 09:14AM by Forrest Higgs.

Forrest, as far as I understand Viktor
with

Quote

Another crazy idea - maybe it would be helpfull, when we squeeze the roundish filament in a square shape? Easier to feed, measure and align?

he doesn't mean actually square filament but some square tubing to feed the round filament through.

That does make sense at least sometimes
If your filament becomes oval it will turn inside the square tubing in such a way that it requires least friction to pull it through.
That way you can align the filament automatically to be viewed by the camera

If the diameter gets too small for he tubing it doesn't affect the filament at all

'sid

... my idea was to squeeze the round filament with some pressure and heated rolls, so it would temporarily or permanent change the shape to a square with rounded edges.

So you have a defined shape for measuring and when the traction of the rolls should be enough to press the filament through the heated tube, this could possibly replace the screw-drive too.

... again - some more crazy ideas

Viktor

Is controlling the amount extruded not the aim of the various rotary encoders? Rather than measuring what's going into the extruder, is it not possible with the rotary encoders to work out what the extruder is emitting?

Hi Jon,

the rotary encoder measures "threads per turn" or with an estimated 1:1 transmission "millimeters filament per turn".

If the filament-diameter varies, then the transported millimeters are the same, but the extruded volume varies with the diameter of the filament ...

Viktor

Viktor, thanks for the explanation.

That being the case, would it maybe be possible to work out the volume of filament being delivered based on how much current the motor is drawing to deliver it? It would seem like more filament would result in a higher current draw...

... no, not really ... the exact amount of current depends mainly on the feeding-rate and changes in environment-temperature and -pressure, so you would have to measure much more variables with high accuracy and an immense overhead.

Because of this nested problems i'm searching a solution for room-temp-fabbing with low viscouse pastes for very small and precise parts - mostly ceramic bodies and/or 3D-structures with embedded conducting trays for micro-sensors ...

Viktor

If using the bitsfrombytes extruder, could you rough up the bearing which pushes the filament against the thread with sandpaper, then measure rotation of the bearing? Seems to me this would give a much better indication of flow rate than measuring motor rotation speed. I'm finding extrusion speed fairly constant at the moment, but this would be my first solution if it becomes a problem.

james glanville Wrote:
-------------------------------------------------------
> If using the bitsfrombytes extruder, could you
> rough up the bearing which pushes the filament
> against the thread with sandpaper, then measure
> rotation of the bearing?
>
I've thought of using a rotary rasp instead of the bearing for just that purpose.

[www.widgetsupply.com]

you thought of yousing that rasp for that purpose because you don't have any old PC cases left to shredder... so it's of no use at the very moment; at least be honest

'sid

James,
I think you misunderstand the problem. I have an accurate measure of the linear feed rate by using a shaft encoder on the motor. One turn is 0.8mm with M5 thread. The problem is that the filament diameter varies so the exact volume of plastic is not known without knowing the cross sectional area of the filament.

---

[www.hydraraptor.blogspot.com]

sid Wrote:
-------------------------------------------------------
> you thought of yousing that rasp for that purpose
> because you don't have any old PC cases left to
> shredder... so it's of no use at the very moment;
> at least be honest
>
ROTFLMAO! Since I started milling HDPE I've got more bloody shredded plastic than I know what to do with.