Glass nozzles for extruders

I have been playing with the glass work. I have one basic nozzle that could be put on an extruder right now, but I want to play with some improvements. First, some pictures.


This is the tip of the glass nozzle, after using a stained glass diamond grinder to remove enough of the ‘blob’ at the tip to break into the funnel that forms when the tubing get hot and the surface tension pulls it together. This makes it easy to get almost any size opening you want, you just need a wire or probe that will just fit the opening you want, and continue grinding until it fits.


This is a side view of the tip. You can just see the funnel inside the glass pointing down towards the opening. Other people have posted that the greats resistance to plastic flowing out of the extruder is not the PTFE or the brass 3mm sections, but where it narrows done to 0.5mm. This automatically makes a very smooth, gentle transition from full width to reaching the narrowest section right at the end, so this should greatly reduce internal pressure and torque needed by the stepper motor. I am looking forward to putting one together to see.


Here is the full length of the first practice piece. All three images (with annoying captions) we made using my new, USB microscope camera. It is cheap, and has an INCREDIBLY narrow depth of field, so this photo was hard to keep from being blurring somewhere.


Here is a full length shot with a tape measure for scale.You can see that this piece is about 2 inches long, or a little over 5 cm.


So far, I have scrounged most of the tools from other jobs. When I found that a propane torch could not soften the boro-silicate, I first clamped an old, used oxy-acetylene torch to the table, and worked the glass in the flame by hand. I tried a hammer to roll it against the steel table, but that did not work as well at making the funnel as just holding it vertically and heating the bottom. I also tried a screwdriver the shape it, but also got nowhere.


Someone suggested using MAPP gas. I don’t think MAPP gas alone will melt the boro-silicate, but probably would work on standard window pane glass, or ‘soft’ glass. If you want to make your own glass nozzles, and don’t plan on making too many, this is an inexpensive way to get going. In the US, Home Depot sells a complete MAPP plus Oxygen torch package for $60. But it will only run for 20-30 minutes before you run out of something and have more cans of gas. A new can of Oxygen is $9 for 1.5 cubic feet, whereas to refill the welding tank, I think it was $35 for 40 cubuc feet. Obviously, the cheap start becomes expensive after awhile.

I plan on practicing some more, and have several improvements I want to try to accomplish. First is that the other end of the nozzle needs something to keep it from sliding out of the clamp when the filament is pushed into it. I plan on heating the other end and widening it into a flange, just like the plunger end of a hypodermic syringe. Next, I am trying to find a way to heat the tubing and embed the nichrome wire directly into it. This has not gone well so far. If the wire gets into the flame, it just burns up and disappears. If I try to get the glass hot and then push the wire into it, I quickly find I need 3 or 4 more hands! At the very least, I want to make some thermocouples and embed them in the glass so that the temperature reading will be more accurate. I may have to live with applying the nichrome wire with Kapton tape, just like on the brass nozzles.

Mike

Edited 1 time(s). Last edit at 08/01/2010 11:56PM by rocket_scientist.

Excellent. I really think glass is the ideal material for an extruder nozzle.

My concern for glass nozzles is the broken glass everywhe when it breaks.
I walk around barefoot

Great start, Mike. Looking forward to your progress.

Very nice work! Glass seems like an ideal material for extruders - it's slippery at the transition point, and the thermal conductivity is fairly low - only 4 times higher than PTFE. And it should be far stronger and more stable than hot PTFE. Nophead's experiments indicated that the friction from the transition point is much higher than the friction from the nozzle, at least for ~0.5 mm nozzles; glass could work better than PEEK in that respect.

I'm quite interested to hear how much force it requires to extrude with a glass nozzle and transition zone - any chance you would be able to measure it?

Looking forward to seeing some extrusion tests from these nozzles!

Wade

Because not all of us have glass-making experience, I'm searching around for scientific industries that can make glass extruders, with the hope of having a small production run to sell on RepRap stores or via Makerbot. Pegasus Glass (http://www.pegasus-glass.com/) indicated that a part like this is considered "precision" in the glass industry, and they wouldn't be able to do it. They referred me to Sandfire Scientific, who I've just contacted. So I'll see how that goes...

I want to drill a hole in the side and place a thermocouple junction actually INSIDE the plastic melt zone, and seal it up with more hot glass. The find some way to wrap the nichrome wire around it and wrap that with more hot glass to seal in the heat. And, I need to flare the other end into a lip that will keep the tube from being pushed out by the plastic. Then, hopefully I will be able to kludge together a test bad. Still don't have anything near a repstrap to run it on, and most of my tools are 2 1/2 hours drive from here for working on my rental house.

I might be able to measure the force required to extrude by putting the whole thing upside down, put a short piece of plastic in the tube, heat it up, and see how much weight has to placed on the nozzle assembly to a thin stream to come out at a decent pace. The weight of the nozzle assembly plus the added weight should show how much force is needed. People can then divide that by gear ratios to get required torque.

Mike

JB,
I am thinking about ordering more glass and doing a lot of 20 to see how they go on ebay. My goal is to have each nozzle already equipped with the in-plastic thermocouple, nichrome wire, and sealed in with glass. The first batch will likely have many variations, but if the sell, I can make more that should be more consistent.

I am curious to hear how expensive they will be for just the glass tube with orifice at one end and flange on the other.

Mike

Yes, I expect they will be quite expensive. Still, it's worth asking.

How long do you expect it will take you to make a glass nozzle, on average?

Also, as to how to mount the glass nozzle... I wonder glass would do in one of those compression fittings? Just as an example, McMaster-Carr 5949K61 (although it's slightly larger than ideal).

JB, that is possible. It would hold it very rigidly. But I would think a little cushion is needed to prevent breaking. I was think more like these mcmasters loop clamps. They will not be as precise, so maybe the best is a two piece RP holder with half the hole in the bottom piece, and the other half of the hole in the top piece and screw holes on each side.

Mike

To make just what is in the photos, a plain tube with 0.5mm hole, I think less than 15 minutes each. Adding the other useful features may move it up to an hour or more per piece. Still need to experiment and find out how much I can include in one nozzle.

Mike

Quote
jbayless
Yes, I expect they will be quite expensive. Still, it's worth asking. smiling smiley

How long do you expect it will take you to make a glass nozzle, on average?

Edited 1 time(s). Last edit at 08/02/2010 04:49PM by rocket_scientist.

Only 15 minutes! Wow.

I think I'd rather go with a compression fitting, because I think a rigidly mounted extruder is pretty important for having an accurate build. My concern is, yeah - would the glass withstand the pressure? Perhaps if it's not screwed too tightly.

The printing puts strong lateral forces on the extruder tip, so I think it's very preferable to have it rigidly mounted.

I just put a digital caliper on the first nozzle. The glass tubing is 8mm OD, and about 3.8mm ID. Try looking for a copper tubing compression fitting, the copper is softer and should spread the stress out with the glass better than steel.

The tubing I started with was the smallest ID that I could order from the vendor glass tubes in heavy wall thickness, which is what I think we need. The thinner walls will be less susceptible to cracking from thermal stress (counter intuitive, I know, but the thinner wall is more flexible, and can only have a smaller thermal difference from inner wall to outer wall) but will be more likely to break during use, and may be harder to shape and add parts as needed. I am ordering more materials to start making some to sell. Hopefully, I can work out the basics to get a complete nozzle assembly in glass.

Mike

Edited 1 time(s). Last edit at 08/02/2010 07:10PM by rocket_scientist.

This sounds really good. But if problems occur, these might be the reasons...

- Thicker nozzles might be somewhat worse from a thermal performance point of view - more heat will conduct up the length of the tube (lengthwise thermal resistance 44% less than a 6mm OD tube), for example, and less will conduct into the interior (widthwise thermal resistance will be 60% greater than a 6mm OD tube).

- Also, I'm a bit concerned about the 3.8mm ID - in my experience, with PLA at least, it's important for the ID to snugly fit the filament, or else you end up with a lot of backflow.

For brass tube fittings, there's 5272K297 which has an ID of 5/16" or 7.94 mm, which is probably close enough.

Here's a 6mm OD, 3mm ID glass tube. I think it would probably be strong enough (just guessing though).

[hotglasscolor.com]

Edited 1 time(s). Last edit at 08/02/2010 09:44PM by jbayless.

JB,
I concur. There is a risk that a thicker wall will move the melt zone further back. Although I still think the glass will maintain something like 200C per inch. I have ordered more material, including exactly the thin wall you suggest. I will see if that stands up to the stresses well enough. If it does, it will be even more resistant to cracking from large thermal gradients because it will be more flexible.

I am thinking that you and other might want to experiment along with me, so I will try to make 5 simple nozzles, just add the flange at the other end, and put them up on ebay. I think I can do nozzle plus shipping and handling for $5. Hard to go lower buying retail in small quantities and still groping my around in the dark on processes. You will probably have to add the nichrome wire and thermocouple with Kapton tape. A word of warning, if you try to melt the nichrome into the glass like I am trying to, my first discovery is that the wire burns away to nothing in flames hot enough to soften boro-silicate.

When the thin wall comes in, do you want me to make 5 more simple ones? Or just wait and see how I do on putting the thermocouple right through the side of the glass into the plastic melt zone, and embedding the nichrome wire into the glass? I have two more ideas on how to get the nichrome wire permanently attached. One is to add little drops of very hot glass on top of wire pressed against the glass, kind like hot gluing it in place. The second, which anyone can try, is to tack down the mid point, then run enough current through the wire to make it glow, the pull it hard enough against the glass while I wind it on the melt into the surface. Not telling though is the nichrome wire will still have enough tensile strength when glowing yellow hot! If it works, the wire will be embedded at its maximum thermal expansion, and when it cools it will shrink quite a bit. This will either crack the glass, or reshape the wire. Then hopefully it will handle all the reheat, cooling cycles.

If we switch to fiberglass insulated or vitreous silica insulated Type K thermocouple wire, we can raise the tip temperature to 700C to 1100C, which is enough to melt aluminum. Now the problem is no the nozzle conduction heat back extending the melt zone, but the filament material itself! I look forward to playing with a molten aluminum extruder for extruding metal objects. Might be too runny, unlike a thermoplastic when aluminum lets go, it lets go all the way. We might have to pre-mix the aluminum with fine grained silica sand to make a slurry that is easier to control. Should not decrease the strength or conductivity too much.

Well, time to make some parts and learn how to sell on ebay. In the past, I have only purchased items.

Mike

I would also still like to wrap more glass around the entire hot zone to keep the heat in. But the wire is goin

Yeah, I was thinking of buying some of those tubes too, but I don't know how to work glass. (Yet). Sent a PM though.

I think I'll experiment a bit with compression fittings, so if you can make some flange-free tubes, thick or thin, that'd be good. =)

Embedding wire in glass - interesting. I wonder how they do it for safety glass, with those steel wires in it (although Nichrome is a far cry from steel). I was indeed just planning to use kapton tape, although if there is a reliable process to get the wire inside the tube, that would be quite elegant. I agree that, if there is a way to do it, it'd probably involve using the wires themselves to provide the heat. But I wonder if it's possible to do so, and still avoid shorts from the wires contacting each other.

I'm fairly sure the wire would stretch, rather than the glass breaking. But I don't have as much experience with glass as you do.

The thermocouple/thermistor would definitely get more accurate readings from inside the glass, but I wonder if it's worth it - I've had thermistors occasionally fail on me, or their hair-thin wires can break, and it'd be very hard to replace if that were the case. Although maybe the glass will help keep the thermocouple protected.

Going to really high temperatures could be dangerous - I wonder if the hot tip could set the MDF reprap parts on fire, for example. Plus, I expect that any molten metals (such as aluminum) would have very high surface tension, as well as absurdly quick heat conduction; the heat would conduct up the filament and into the metal printed part itself. I don't think I'd be ready to start investigating in those areas, but if you're going to, be careful with it!

I agree, plastic first, metal later. I was just noticing the specs and saw that it is possible, and seemed like a neat possibility.The current design could never go that hot because the PTFE would not only get soft but melt long before the aluminum. The glass would not only provide fairly good insulation, but handle the 700C easily. But I need an extreuder and repstrap first!

Mike

I did some more experimenting tonight. I worked on making a lip at the other end to prevent it from sliding through a clamp. My first efforts I tried too hard and messed it up.


sorry about out of focus, used the cell phone camera. These had way to much glass trying to make the lip, and using a battery powered drill/driver to rotate them and keep them uniform and circular actually made them twist and get very lopsided.



This time I just used the tip of a plumb bob to slightly widen the opening, then pressed it against the table to flatten it out some, just enough to prevent slippage. For the good ones, I then sealed the other end just like I have done before. They still need to have the sealed off end ground down until the right sized hole appears, then they could be used as nozzles if you apply the nichrome and thermocouple with Kapton tape.

Next is to drill a hole in the side for the thermocouple, and see if I can use hot glass instead of hot glue to attach the nichrome wire.

Mike

By the way, SandFire declined to provide an estimate for producing a glass extruder nozzle.

Seems like an ideal candidate for glass casting. Glass beads in a water cooled mold? I think you can do investment casting of glass in a kiln. Will have to look in to this.

I don't know if you could cast an 0.5 mm nozzle; at least, I'd be surprised if you could. I think the part is really ideally set up for glass drawing, but again, I don't have much glassworking experience.

I have actually not yet successfully down glass drawing. It is a very common, easy to do process, and used extensively in manufacturing. But I haven't gotten to that yet because the molten glass seems to keep doing for me what I want as long as I interfere with it minimally. But I want to try drawing some glass this weekend.

Right now, I get the pretty much ideal internal shape by heating about 8-12 mm of the end, hold it with the hot end down, and let surface tension draw the opening closed. That makes a beautiful funnel on the inside, and still keeps the wall thickness at the tip. But sometimes the funnel has a long, drawn out 'spot', allowing for precisely determining the orifice size, and some times it ends abruptly and there is a very fast change when grinding off the tip form small to large. I think that the abrupt ones may have less friction, but the longer spot ones may make a more uniform stream.


I know nothing about glass casting from beads, it must be a fairly new process. But glass blowing, and related hot glass work if thousands of years old, and some very beautiful artwork, and all the scientific and chemistry glass tubing, stoppers, flasks, distillation columns and other 'mad scientist's lab stuff' was made by hand for centuries, so it is possible to get very precise, and very complex shapes if you know how. I can only think that with practice, and some classes and creativity, I can vastly improve what I have made so far.

I have more glass coming in the next few days, and if anyone other than jbayless would like some to experiment with, I think I can have a number of simple nozzles, with and without lips, ready to sell on ebay at %5 including shipping. They will only need to have the nichrome wire and thermocouple taped on with Kapton tape, and I am guessing that drilling a hole in a piece of wood or plastic, then cutting it in half to make a saddle clamp will be enough to start testing it in you extruder.

My hope is that we can replace two disparate pieces, the plastic thermal barrier and the metal nozzle, with a single stronger piece that will make the extruders more reliable. I have not seen a count of what hardware failures occur most often, but I am guessing the plugged, pushed out, or slipping extruder nozzles is up there. If the glass works as well as I think, this will move the extruder failures almost entirely into the stepper and gears, which we have much more control over.

Mike

... for thin nozzles you can emdedd a metallic wire with 0.5mm diameter (or 0.3, or ...) and remove it later with an acid solving away the metal ...

---

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

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

Good point! Pencil lead also comes in 0.5 and 0.3 mm sizes, and could probably be removed even more easily than metal wire. =)

Some of the glass blowing, bead making websites I have visited say that stainless steel is commonly used as a mandrel, and that there is a release agent that helps the glass to separate cleanly when finished. I think that if I made a SS mandrel that is 3 mm then tappers down to 0.5mm that I can either add glass to it, or try to stretch/shrink glass onto it. That might also allow for putting the nichrome wire on the inside of the glass, where it will heat the plastic better, and the insulating value of the glass will help minimize wasted heat. But it may be some time before I get the lathe up and running and try something that fancy. Standard nozzles built from tubing will have to come first.

Mike

If it helps, I have access to a lathe and could certainly make something like that, though I won't get the chance for at least a week.

Edited 2 time(s). Last edit at 08/06/2010 09:04PM by jbayless.

My concern with having the nichrome on the inside (Are you talking about wrapping it around the mandrel prior to shrinking the glass onto it?) would be that it would burn the plastic. Having it on the other side of a block of brass or aluminium helps to spread the heat more evenly so that the plastic is heated to its melting point, but not significantly more.

Perhaps you could include a SS insert that has a hole in the middle, and leave it embedded in the glass. Wrapping the nichrome around that would work nicely.

Edited 1 time(s). Last edit at 08/06/2010 10:03PM by Greg Frost.

I am also wondering about making lots of ground boro-silicate glass with the diamond grinder, then mixing it with syrupy sugar water as a binding agent, then applying it in layers to the mandrel and then cooking it in the flame to build up thin layers of glass. Then wrap the nichrome, so that the plastic does not rub against the bulges of the wire, then a few more layers, then shrink a tube onto it. Many possibilities, so little time!

The new order of glass arrived. More experiments this weekend. Including 6mm no lip 0.5mm orifice 8 cm long nozzles for jbayless. Anyone want one (some) to experiment with? Or just wait till I get the process more refined.



Mike

I couldn't wait! I played with some of the 6mm glass tonight so I made a whole set in the thinner wall, with the more precise , 3mm inside diameter


3 nozzles the standard way, one heated and drawn thin, plus two scrap to test the compression fittings .



this was the first, and the thinner wall did not want to funnel down. It just closed with a nearly circular end. This one will have the least friction, but the stream may come out at an angle. It is hard to see the internal shape after grinding down the tip to make it rounder and thinner. The areas that you see that are diffuse white have been ground down. You kind of have to imagine the internal shape underneath the ground glass of the tip.


The second one I heated a longer section in a cooler flame and got some tapper to it.


This one came out beautifully! Long graceful taper, easy to get any size opening. This will have a bit more friction, but should produce a very fine, thin stream of plastic! This one did not require any grinding, except to square off the end, so it is MUCH easier to see the internal shape.


I had a LOT more trouble with stretching the glass thin. Without a jig, the two ends, one in each hand, would get off center and pull the tip to the side. Also tended to pull too long and just get a streamer in between. This is where all the scrap pieces came from. The best of the lot is still bigger than 0.5mm orifice, but it is pretty straight, and looks beautiful.

Mike

Edited 3 time(s). Last edit at 08/07/2010 07:55AM by rocket_scientist.