Q: Height of extruder nozzle above the print?

Greetings all,

I have a question for those of you getting good print results:
How high is your extruder nozzle above the print surface, for good printing?

Ideally I'd like to know:

Material / Diameter of nozzle opening / Height of nozzle above surface



The reason I ask is that I have (yet another) idea about measuring the rate of plastic deposition (so it can be controlled), but it'd need at least a little vertical height between the nozzle's exit and the (current) print surface.
Having a means to measure/control the rate of plastic output should make it easier to print consistently from granular feedstock -- and then we could avoid the up-charge for having plastic extruded into rod. I had been thinking about measuring pressure near the tip, but that's only indirectly related to the output volume rate. What (IMHO) we'd really like is to measure the rate (in mm/sec) that melted plastic exits the tip -- and (if we can) measure the diameter, to get a realtime measurement of volume/time.


Thanks in advance,

Larry

---

Larry Pfeffer,

My blog about building repstrap Cerberus:
[repstrap-cerberus.blogspot.com]

Irrespective of material and nozzle opening I always use a nozzle height of 0.8mm times the notional filament diameter. The diameter is set by the flow rate and the head speed.

I am currently extruding 0.375mm filament through a 0.4mm nozzle with a layer height of 0.3mm giving a filament width of 0.45mm.

---

[www.hydraraptor.blogspot.com]

Nop,

Thanks for your extrusion parameters. Looks like this idea probably won't work. Back to the drawing board. (Now there's a saying in need of an update....)

---

Larry Pfeffer,

My blog about building repstrap Cerberus:
[repstrap-cerberus.blogspot.com]

Assuming you're not leaking plastic, it's easier to just measure the flow of plastic in to the extruder in order to determine how much plastic is coming out. My old extruder used the magnetic encoder to keep track of the drive screw; if you're worried about slippage, mount a separate measuring wheel on the filament independent of the drive system.

With a good stepper based pinch wheel extruder, I don't have any noticeable slippage of the filament, and my calculated flow rates were good enough from the start - I didn't have to calibrate anything. There's a spread sheet on my extruder wiki page that you can look at if you're interested.

[objects.reprap.org]

This doesn't take into account variations in filament diameter, but I haven't seen a need for that yet - I have plenty of other problems.

Wade

Wade,

Using the input rod diameter for a realtime control of the extrusion volume rate, would work fine so long as one is using rod feedstock to extrude.

If one wanted to use polymer granule feedstock (much cheaper than rod stock) or recycled shreds, then won't be a rod diameter to measure. I was asking about nozzle height (above previous layer) as part of an idea to measure the extruder output volume, in the case where there isn't an input rod available.

Unfortunately, if the nozzle is typically as close as Nop says, then my idea (vibrating the exit orifice at a known frequency and observing the spacing of the ripples in the output melt), has yet another strike against it.

Measuring pressure as a proxy for extrusion volume is possible (modulo the difficulties of re-melting plastic in the sensed volume.) However, pressure is only an indirect measure of volumetric flow, vulnerable to changes in the orifice area (from partial fouling), or viscosity changes (temperature or composition variations.) Fouling is a tough issue for using recycled plastic; fine filters (AKA packing screens in injection moulding nomenclature) require large pressures to push melted plastic through them at anything beyond a snail's pace.

---

Larry Pfeffer,

My blog about building repstrap Cerberus:
[repstrap-cerberus.blogspot.com]

Ah, I see what you're getting at. Interesting! I had a look at some of my prints with a cheap loupe equipped with a mm scale today - all or them, irrepective of head speed or plastic flow rate, have a roughly 0.1 mm spaced ripples; which corresponds to my resolution (10 steps / mm). I think the head vibrates with the stepper timing.

Since the plastic tends to stick to the bed immediately, I think the ripple wavelength would simply be the head speed / excitation frequency, which would be a constant value, independant of the plastic flow rate.

One thing that does change, though, is the width of the extruded filament. Before I had a deterministic extruder flow rate, I would use the width of the extruded filament to guess at my flow rate. Maybe some sort of webcam vision system to check the filament width? Sounds very difficult. A lot trickier than a record needle.

Calculating the flow rate from pressure is tricky as well; it's going to be very non linear, and probably not stable either.

How about a far infrared webcam, digitally filtered to only see the most recent, hot filament? I think that might cost a bit too much though.

edit- one more stupid idea - put some sound transducers diagonally across the narrow section of the nozzle, and measure the Doppler shift of transmitted pulses. I used to use ultrasonic wind speed meters based on that setup. We'd need great timing resolution to pick up such a low speed flow though; the signal would prbably be lost in the noise.

Edited 1 time(s). Last edit at 03/10/2010 02:38AM by Wade.

Wade,

Thanks for your thoughtful comments -- and your (bonus) observation and measurement of ripples in your extruded plastic!

In my planned machine, the bed moves, but the extruder does not (except during infrequent Z-height changes.) Thus, I can avoid ripples from XY motion-associated vibrations (at least, so I hope.) If I were to vibrate the tip of my extruder, I might still get a ripple spacing indicative of the flow rate of plastic. However, the ripple spacing ought to be measured, right after exiting the nozzle, before it deforms in a very-hard-to-model manner, on top of the previous layer.

I suppose that if one had a good model for that deformation (or at least some good data for a look-up table), then your idea of a phonograph needle (or equivalent) to measure the ripple's frequency might work. Of course, the longer time between exiting the nozzle and the measurement, the harder it would be for using that info for controlling the volumetric flow rate. Significant time delays are NOT easy or fun to handle in servo loops -- especially when the delay is variable.

Doppler is theoretically applicable, but the ratio of plastic exit speed to sound's speed in melted plastic makes it very hard. (Order of 10 mm/sec over {probably} hundreds to thousands of meters/sec. So, order of 1e-4 or 1e-5!)
Similarly, the end conditions of the filament (in terms of reflecting vibrations) are beyond my meager ability to model.

By explaining some notions that probably won't work, I'm trying to stir up other ideas that might. I once heard a story that the brilliant scientist, Oceanographer John Isaacs, tried to start a _Journal of Negative Results_, so that knowledge gained during failures could be shared and used. Unfortunately, too many scientists lacked Isaacs' courage and candor in sharing his failures. A shame, really.

---

Larry Pfeffer,

My blog about building repstrap Cerberus:
[repstrap-cerberus.blogspot.com]

Yes, sounds like a tricky problem. I like the idea of the Journal of Negative Results - sometimes these forums feel a bit like that.

Just for kicks, I tried using a flashlight, loupe, and my cell phone camera to get a photo of the ridges. It sort of worked, but you only see the larger bumps in the photo, which correspond to the full steps at 0.2mm. The smaller bumps at the half steps (0.1 mm) don't show. The smaller gradations on the loupe are 1 mm wide.

[gallery.me.com]

Wade

Nop head said

Irrespective of material and nozzle opening I always use a nozzle height of 0.8mm times the notional filament diameter. The diameter is set by the flow rate and the head speed.
I am currently extruding 0.375mm filament through a 0.4mm nozzle with a layer height of 0.3mm giving a filament width of 0.45mm.

I said
Interesting. My extruder is usually 0.1mm - 0.2mm off the print bed to start will.
(I use a piece of paper for 0.1mm to calibrate the head and folded paper for 0.2mm)

My extruder puts out a 0.55mm diameter filament through a 0.5mm nozzle. This settles on the print bed with a height of 0.47mm and a width over thickness ratio of about 1.41.

A farout idea:
what about measuring the acceleration of the extruder head, i.e. the force exerted on the extruder head. It seems to me that the amount of extrusion leaving the head should somehow be related to the re-action force, as per Newton 3.
dForce = dVol * density * accel - Mg , Newton 1, where M = extruder-head mass.
dFlow * dt = dVol
dForce = dFlow * dt * density * accel - Mg
Is re-action force proportional to the integral of the flow-rate, plus linear time-function?
I know that the extruder head design is ideally "rigid", and any recoil force will likely be buried in a "sea" of environmental noise.
It seems to me that that provided the idea above is correct, it will still be contingent on the following:
1. obtain sensor sensitive enough to measure micro-/nano-scale movements?
2. apply some filtering techniques on sensor-readings, which may include:
2.1. obtain head acceleration, by placing one sensor on head, and another on mendel frame, "far" from head. Then subtract two reading, thus eliminating environmental/common "forces". This will reduce the need for extraneous signals to be perfectly random or "white".
2.2. Flow rates will probably be linear combination of step-wise functions with characterstic linear-combination frequency distribution, and therefore low-pass/band-pass filter can be applied. Assumption.
2.3. Sampling frequency should be done at Nyquist rate, to avoid aliasing.

I hope this helps.
Regards
Marius Botha
Pretoria, South Africa
[mariushermanbotha.wordpress.com]

Wade,

Thanks so much for the picture; I *really* appreciate your taking the time and effort! With the measurement grid, that gives me a much better idea about what the plastic is doing as it's being deposited.

---

Larry Pfeffer,

My blog about building repstrap Cerberus:
[repstrap-cerberus.blogspot.com]

BalanceSeeker said
what about measuring the acceleration of the extruder head, i.e. the force exerted on the extruder head. It seems to me that the amount of extrusion leaving the head should somehow be related to the re-action force, as per Newton 3

I say
This is exactly what adrian is working on with his 5D Gcode. The head acceleration is a function of the stepper motor and is controlled by the micro controller. Therefore it can be "modelled" without any sensor equipment. As the head accelerates the stepper on the extruder pumps out more plastic.