Extruder nozzles

I thought I might post here my experience with finding and ultimately doing my own extruder nozzle design.

The story goes, trying to get a hold of 16mm dia PTFE rod and make the thermal barrier turned out to be near nigh on impossible, and seeing as I wanted to have extruder nozzles that can be unscrewed and switched,and also Tim Carr and Vik Olliver are working on RepRap kits for us kiwis, I started out to design my own, using the 16mm dia PTFE and M6 threads as a guide.

First problem, and it is a biggie, the mechanical drawings for the heater barrel, nozzle and thermal barrier have a massive shortage on the number of specified dimensions, because you really do need those to get a local CNC machining company to make you those bits, reason for a CNC company coming later.
I spent 2- 3 weeks figuring out the best and easiest way to get what I designed in Blender into a set of drawings with dimensions, in the end there was only one method to annotate the drawings to have dimensions on them, that being using an actual CAD package, and a lot of free ones don't work, funny that, I used both Solid Works and AutoCAD to annotate the drawings, Solid Works for linear dimensions and AutoCAD for angles.

The first set of drawings I sent out to get quotes on showed that the design of nozzle was going to cost a small fortune to make, maybe it was due to the text being cut into the nozzle itself to identify the size of the nozzle.

The second set had the text removed, and was able to talk with the manager directly, and he tells me something that I didn't think too much about.
I redid the drawings to utilize M6 DOME NUTs, and just waiting on the quote to come through. and i've sent the same drawings to a 3rd and 4th company to get quotes on, hopefully i'll get those early next week.

In the specification for RepRap, was to go down to 0.5 mm, with the near future going down to 0.2, so i've catered for this.

Here is the most important thing to know, which determines how anyone should make nozzles, in order to have a 0.3 or 0.2 mm hole means that the depth of hole is limited to 2mm. This is because the drill bits that come in that size need a shaft that is of bigger diameter to fit into their chucks, also because the size is so small the "wobble" of the bit when it comes into contact with the work is likely to break the drill if it were any longer.

So that's the compromise, dome nuts and not the current barrel with integrated nozzle design, will enable the smallest extrudes possible.

If anyone wants the drawings let me know, and i'll post them up somewhere, or maybe get one of the RepRap core group to put them into the repository

Edited 1 time(s). Last edit at 09/05/2009 05:19PM by Grogyan.

I actually wonder if PTFE rod is a good design. I have seen too many blogs writing about mechanical failure of the PTFE rod, and recently I saw quite a few blogs trying to use Stainless steel + Heatsink as thermal barrier (or create sharp thermal gradient).
Those are RepStrap though, not RepRap.

I thought PTFE rod design is optimized for home-made, one without access to machine shop. But if you do have access to a machine shop, there should be some more "friendly" and "better" design.

Last week I started building my RepStrap, I got no power tools (except for one very cheap hand drill) so I got all the components from the machine shop.

Here is the forum thread [dev.forums.reprap.org]. I attached my drawing at the last message. I hope that might inspire you.

I printed my first object with it, not blockage and mechanical failure so far.

I'm curious now, which blogs you're referring to with mechanical failure of the PTFE?

The biggest problem with the setup, is the adhesive, while there is little (i'm guessing here) heat transfer through the plastic, its the adhesive used that will break, epoxy weakens at around 100 degrees I have noticed at work, as I often use heat to chip off epoxy to get to where I want to get to for repairing devices.

If you use a heatsink, this is an even worse idea, as the heat transfer into the surrounding plastic will soften the extruder support bits and cause undesirable twisting and flexing.

I have got some Selleys fire cement from Mitre10 last month, and tested a bit out (might do another test now that you've peaked my interest in mechanical failures), to how brittle fire cement is, sadly it is very brittle.
I'll do another test to see how much heat is being transferred, i'll use a finger and a butane torch at either ends of a roll of fire cement.

Something i'd like someone who is prepared to find out for me, is whether fire cement extruded into a shape or moulded into the shape of a thermal barrier would be enough of a thermal resistance.

The alternative is a ceramic compound, and some are machinable.

Well I quickly created a wall (more of a blob really, but I did get the rough shape of a wall) of fire cement, roughly 15mm thick.



Don't try this unless you're prepared to get seriously burnt, this is extremely dangerous

With a butane torch on one side and my finger of the other, and I barely felt it go above ambient.


This means that it should be an excellent adhesive for the thermal barrier to the RepRap extruder parts

Blogs...

[hydraraptor.blogspot.com]
Swell and elongated.

[blog.reprap.org]
PTFE barrel is not very forgiving when thing fails. You need a new one.

[hydraraptor.blogspot.com]
See the tag / ring below of the PEEK thermal barrier? It's there to hold it in place and not popping out, or else when thing gets hot, it expand a lot.

My gut feeling is that it won't work for long, and when it fails replacement would be needed.

Plus, I have no where to know where I could buy JB Weld / Fire cement locally, so I just give up the idea of using PTFE. (I currently live in Shanghai China and without any idea what the Chinese name of JB Weld / Fire cement, I have no idea where I could buy any)

One more issue: I have absolutely no idea where I could by insulated nichrome wire. None of the merchant I contact knew. The most similar I got is are PTFE-wrapped nichrome wire, which melts and shorts at around 150 degree. Again, without the cement and such, and wrapping kapton tape proves to be too much trouble for me, this is a dead end.

So I settled with machining a Stainless steel tube.
[picasaweb.google.com]
It's extruding, yet the tip on top is just warm to touch

The stainless steel design is inspired by Nophead and the others:
[hydraraptor.blogspot.com]
[hydraraptor.blogspot.com]
[builders.reprap.org]

I remember reading a couple of those before, thanks, forgot about them.

I believe the problem lies with the fact that they are just assuming the rate stuff is being extruded, instead of measuring it, as there is a hell a lot of force being applied to the extruder to push plastic through.

I'm not sure yet of a way it can be measured reliably, but a work around for the bulging, is a simple one that I have seen Vik use on his Darwin, hose clamps.

Just did have an idea of measuring the rate of extrusion, you know how there is a place to wipe the nozzle, how about an attachment just next to it that uses an optical mouse encoder attached to a bearing, so as the extruded material is coming out, and being forced against the bearing, you can measure with the encoder (from an old ball mouse) how much plastic is coming out.

This is quite important I think to consider, especially with a nozzle size of just 0.2mm, that's roughly 2 widths of a human hair.

Good on you Sam getting your extruder going, but it appears that its more of a RepStrap extruder than one from RepRaped bits, excluding the heater barrel, thermal barrier etc.

Yes indeed it's a RepStrap. Because I had made a milling machine, so it's just logical to covert the toolhead from a drill to extruder.

Plus, I am not aware of any China, or even far east, reprapper! Check the google map [maps.google.com], I am the only one.

Which means getting someone to print the parts for me...I would have to eat the shipping cost.

Using skate bearing to measure the extruding material? I don't think that would work.
1. It will cool the plastic
2. or if hot enough, the plastic will stick to your bearing
Either case, it doesn't work.
3. We use nozzle to press aginst the extruded plastic to compensate for inaccurate Z/thickness.

I wonder if we can use any thing like Ultrasound Doppler effect to measure the flow (those are common used to measure fluid flow I guess, without any contact). But I would imagine that's beyond the budget of RepRap, plus it probably can't detect such a small thing and flow.

Edited 1 time(s). Last edit at 09/07/2009 07:20AM by sam0737.

True, so how about maybe using an optical mouse to read the rate at which the extruded material is coming out?

All i'm suggesting is knowing for sure that stuff is coming out and how fast, as this in turn determines how much filament needs to be pushed through.

... we had last year some talk about measuring the extruding-speed.

One possibility is marking the filament with colour-dots for visual sensing, another is embedding some magnetic microparticles and magnetize single segments and measuring with a highres Hall-sensor.

AFAIK sensing without marking was not possible as the extruded filament is very homogene in colour and occasional air-bubbles weren't enough.

Another problem is the location of the sensor: - you have to sense in or near the nozzle as the extruding speed may vary against the speed of the inserted filament (diameter against volume inconsitency and such).

VIktor

Oh wait. What's grogyan suggesting is interesting - the optical mouse.

The optical mouse that we use everyday now on unmarked table, or even glass for some model. The one that we though unachievable 10 years ago and requires a marking mat to use.

Two issues I can think of:
1. how are we going to mount it? There is no headroom below the nozzle because the nozzle needs to press against the molten filament.
2. There is too much math involve. What do you do if you find the flow rate is too slow or too high? If it's too slow due to slipping or blockage, spinning up the motor doesn't help. And relating this information to slow/speed up the feedrate is even harder Math that I wonder the microcontroller can handle or not.

sam0737 Wrote:
-------------------------------------------------------
> Oh wait. What's grogyan suggesting is interesting
> - the optical mouse.
>
> The optical mouse that we use everyday now on
> unmarked table, or even glass for some model. The
> one that we though unachievable 10 years ago and
> requires a marking mat to use.
>
> Two issues I can think of:
> 1. how are we going to mount it? There is no
> headroom below the nozzle because the nozzle needs
> to press against the molten filament.

You need to look at the problem in 3 dimensions, so take Darwin for example, there is a hole in the print table for the nozzle wipe, and because the idea of using an optical mouse to read flow rate requires a "point of reference" ie a clear nozzle, its just easy enough to plant there.

> 2. There is too much math involve. What do you do
> if you find the flow rate is too slow or too high?
> If it's too slow due to slipping or blockage,
> spinning up the motor doesn't help. And relating
> this information to slow/speed up the feedrate is
> even harder Math that I wonder the microcontroller
> can handle or not.

Trust me, the micro can handle it, any modern micro these days can do it, ten years ago they could do it, now they are just better at it.

As I stated before, at the end of the nozzle wipe sequence, you begin to measure flow rate, you just apply a little more pressure, measure, repeat a couple of times and only applying slightly more force if there is nothing to read, usually you'll get a reading on the first attempt and that will give you good data.
So after say, 3 attempts there is nothing coming out, the print process stops altogether, and the screw drive reverses to ease off the pressure on the filament.

Oh. I think I misunderstood.

So it's just merely a check during wipe, instead of live monitoring while printing? This should be much easier.


I thought you want to do live monitoring of the flowrate, then feedback to slow or speed up the toolhead movement feedrate in ratio.

Though I am not sure about the usefulness. We should really fix the PTFE design if that was that easy to fail (Sorry I don't have any idea about this). Monitoring is just a bandaid for disaster prevention, but couldn't solve the blockage issue from happening.

Yeah, I meant during the wipe procedure sequence, that's the best time to do it, real-time monitoring would be pointless.

Unfortunately, PTFE is all that we can use, that is both cheap enough with good thermal characteristics.
I think I mentioned before somewhere that ceramics aren't very easy to machine, but there are a couple, its more likely to printout a mould for the thermal barrier and fill it with a ceramic compound.

The workaround is to basically use hose clamps, which are easily available at any hardware shop

Righteo, some news, got word back from various cnc businesses, and here is a real problem without more of a solution.

In order to get a 0.3 or 0.2mm hole just right to get roughly the same extrude diameter out requires that a dent in the end of the object to be machined, in my case its looking like dome nuts.
Putting a dent in the end causes a real problem, as the finish afterwards would mean that the extruded plastic no longer becomes ~0.2mm, but rather whatever the size of the dent is, which could be up to or even more than 0.4mm in diameter, and the extrude I expect, flattens out across that whole diameter of the dent plus some due to surface tension in the material.

Anyone here whose done a 0.3mm (or 0.2mm) extrude and actually measure the diameter of it?

With my extruder it depends on how close the head is to the layer it is printing on as well as how fast the head is moving.

Frank Davies

I'm just referring to the extrude when the drive train isn't moving and not actually printing anything.

Could you measure it for me please, and let me know what the diameter is, and the size of nozzle that you used too?

Thanks

Well I don't have a very good measurement tool, but if I run my extruder into air, coiling itself down onto a surface about 25 mm lower than the tip, thefilament is about 0.68 mm (+/- 10%) wide.

The hole that it comes through was drilled with a size "75S" PCB drill into an acorn nut. Wikipedia shows that this has a diameter of 0.533 mm. Of course the hole will be slightly bigger. I bought a box of 10 bits because I knew that I would break about 1 bit per hole, which proved to be true.

I picked a relatively big size because it would be easier to drill. I also knew bigger filament would print faster and I am impatient.

In my opinion, if you know how fast you are pushing filament into the hot part of the extruder, and you are confident that hot plastic is not being accumulated there (equilibrium), then you can figure out what is coming out. In practice, I think that you would have a steady feed mechanism (like a pinchwheel driven by a stepper), and adjust the parameters of the slicer (skeinforge) until you got good results. Of course the fact the dribbling or oozing is seen means that there is some storage (non-equilibrium) but it is clearly limited.

I really like the idea of measuring the speed of the filament going into the heater directly, with a mouse sensor, or maybe by how fast the bearing opposite the pinchwheel moves. This is because I know that my pinchwheel slips slightly sometimes, and can dig a hole in the filament and stop pushing entirely. I would really like to know if this is happening, particularly in the middle of a big print.

Thanks,

Frank Davies

I have made a nozzle with a 0.3mm hole. The diameter of the plastic extruded depends on the type of plastic, and the speed of extrusion, due to die swell. I simply set the head speed to stretch it back to the size I want. With a 0.3mm nozzle I have built objects with 0.3mm and 0.4mm filament.

With a 0.5mm nozzle I have built objects with 0.3, 0.4 and 0.5mm filament.

---

[www.hydraraptor.blogspot.com]