Wednesday, June 14, 2006


Thinking about making filament again...

Okay, I have a couple of days till the PIC's arrive and Vik brought up the point of there being a developing need for filament production, so I've been thinking about it enough the I want to throw out what may be an approach for you all.

As you know, I was very excited about and did a lot of work on an auger extruder for filament. I got unexcited instantly when I found out how much better filament a simple piston design did the job.

The auger design I did requires powdered polymer unless you go up to the sort of auger diameters that Adrian used. As well, the auger design that I did wanted being broken down and cleaned practically every time you used it if you wanted good results. It also had problems with bridging of polymer powder jamming the feed hopper and any number of other maladies. It was basically German design, attractive but complicated and requiring rather careful use when what we need is Russian design, primitive, ugly and utterly reliable.

What I was using looked very much like a standard ASTM standard test rig for determining polymer melt index.
I got the illustration from a lab manual used up at McMaster University. You can see the full document here.

It's a nice concept except that it's a one shot device and looks to be a pig to use on a production basis if the user's manual is any guide.

Adrian was talking about using a pressure cooker some time ago and I had thought along the same lines as well. Taking the bare essentials from that idea, what would happen if we just tossed our scrap CAPA in a pot and heated it. Don't get lost in the details of how to heat the pot just yet, just assume that we can do it without scorching.... use a double boiler or something. :-)

Okay, so we've got this heated pot of scrap polymer. After a while the bits melt into a big puddle, the polymer slumps down (very slowly given its viscosity) . Now suppose we've drilled a hole in the side of the pot way down at the bottom well under the level of the molten polymer.

Then we start plumbing up a two stroke pump using Adrian's favourite plumbing parts. :-)

Here's roughly what the intake stroke looks like...

Pardon the illustration. I drew it on my white board and photographed it. That's not a great presentation method, but I'm in a hurry.

Your pot's on the left and the piston of your pump is driven by a gear motor that runs a threaded rod arrangement. I didn't get too specific on how that all fits together. At this point what is relevant is that we have what amounts to a linear motor that can deliver a LOT of thrust.

You can see two spring loaded back pressure valves. On the intake stroke the valve to the pot opens up letting the polymer melt be drawn in as the piston is pulled upwards in the cylinder.

Once your cylinder is full you reverse the rotation of the gear motor which closes the back pressure valve into the pot and opens the valve to the extruder tip.

That's basically it.

What's nice is that this system can operate VERY slowly to take into account the high viscosity of the polymer melt. It can also extrude specific amounts of polymer with a high degree of accuracy. Knowing the mass flow rate of polymer out of the extruder on a real-time basis lets us control a conveyer that takes the extruded polymer away.

That's my first cut at a filament extruder.

OK - so you want industrial scale - and you want 'Russian' engineering.

Here is a wild idea for you...

How about using an old car engine? Those ought to be pretty plentiful. I recall an episode of 'Scrapheap challange' (aka Junkyard Wars here in the USA) where they built a water pump from an old car engine.

Think about this...It's built to operate at boiling water type temperatures - so you can pump high temperature water around inside to get the entire pump hot enough to keep the plastic melted - it has inlet and outlet (aka exhaust) valves all there with nice smooth flow paths around them and all the mechanical timing stuff sorted out.

You'd have to change the valve timing to operate on a two-stroke basis,

Lay the radiator flat and stick a big propane heater under it - use the water pump as intended (although it's going to have to spin a lot faster than the engine - so you want a belt-driven pump).

You'd probably only need to run a single cylinder because one cylinder-full would be a LOT of plastic! Let's suppose you picked an old Mini engine - 850cc's spread over 4 cylinders - you'd get 212cc's of plastic every revolution of the engine.

That means that the engine would have to turn REALLY're going to need a lot of mechanical advantage to get the pressures you need. Maybe one revolution per minute or something.

So you'd want to turn the transmission around so the engine would turn really slowly as you drove the drive shaft at higher speeds...possibly with another car engine!

What about a motorcycle, chainsaw or lawnmower engine? Well, those aren't usually water-cooled - and therefore would be harder to water-heat...but then you could heat them directly with a propane burner so long as you kept an eye on the temperatures inside.

On a yet smaller scale, how about a model airplane engine?
Just to add my two cents: there is something "German" in this idea too ;)
First, having a pretty long and complex apparatus going so far from heat source would probably lead to polymer congealing in it (and you'll need extra heat applied somewhere to keep it liquid).

Second, it seems overcomplicated to make the filament beforehead and then melt it again for extrusion. If you are thinking about the pot of liquid polymer already - it would be very nice and straightforward to channel it right into heated printing head, keeping it liquid all the way.

And, just to be positive, I'll try to give a bit of "real advice" ;) Basically, make that pot pressurized and use a simple aquarium air pump to keep it flowing nice and easy. Add a small stepper near the exit from the printing head and use it to move a small valve that would change flow rate or stop it altogether.

Oh, yeah, I'm from Moscow, Russia, so, hopefully, this would be a "Russian design" you are trying to achieve :-) Maybe not quite up to Kalashnikov level, but, hopefully, it would give some new direction of though to you. Keep up the good work - RepRap is a great idea!
I have to agree ont he complexity part. I'm familiar with concrete pumps, which are not as far removed from pumping plastics as you might think. In short, the valves will not close reliably. You need a swash plate.

I would like to speak in favour of a minimalist approach: use a screw thread to move a piston inside a tube of hot polymer - a giant icing syringe. Put the whole thing inside a block of thermal mass, bake in an oven, hook up to your motive power, then extrude. Refil, put back in oven, repeat. For higher throughput, put more than one in the oven.

I thought of making the syringe from threaded pipe and end caps. Use tank washers to hold the thermal mass in place - even concrete would do.

Joining filament is not as hard as I thought it would be. I just push heated ends into opposite sides of a 3mm hole in a PTFE block.

Vik :v)
I would really like it if we could use granular polymer in the RepRap write head itself and eliminate this secondary step. But the problem with that seems to be getting the mass down to a point where it can be moved. Vik's design doesn't need this at the moment, but for any machine that swaps heads and moves them about low weight is a must.

I agree on using compressed air in the chamber to create the extrusion force - that ought to work more simply than a piston.
You'd waste a colossal amount of energy heating up such a massive chunk of hardware, though if you were mass producing colossal reels of filament then I suppose it might be worthwhile.

However, I would suggest that the one-shot pump (where you'd remove the piston and refill the chamber with granules after each cycle) might be made more time-efficient by simply fitting it with multiple nozzles, so one pressing makes lots of filaments at once. Plus, they'd all hopefully be roughly the same length - perhaps the melt head could be fed by a magazine of short rods, rather than one long filament coming from a reel. Just a thought.
This comment has been removed by a blog administrator.
What I conceived wasn't all that massive. I had in mind that the swept volume of the polymer pump wouldn't exceed more that 4-5 cm^3. One of the earlier comments about a model airplane engine sort of fits what I had in mind insofar as scale is concerned.

At that scale one stroke would produce 500-700 mm of filament. I figure that it would be running VERY slowly since given the very high viscosity of polymer melt it would take quite a while for the dimple that filling the pump cylinder created in the pot of melt to fill in from gravitational force alone.
I once worked with a machine that had a two-stage melting pot for glue. The first stage had a granule hopper and a piston that pressed the glue against a hotplate. The plate was just hot enough to melt the first 3cm or so of granules into a glue pot. The glue pot was then maintained at operating temperature, and from there the glue was applied.
Perhaps something similar would be applicable to this situation, where the pot would liquify the scrap / granules, and as the melt entered the pump section it could be brought to operating temperature? That might cut down on the colossal amount of energy required.
Now THAT is a very interesting concept. It would certainly go a long way towards getting the air spaces out from between the granules and, if you used that large piston repeatedly, get rid of the dimple that extracting polymer melt from the pot would cause.

Two pistons... hmmm.. I like it!
//just assume that we can do it without scorching.... use a double boiler or something//

How about a crock pot?
I believe getting weight down using a small heated pot should not be an insurmountable problem. Right now you have to carry with the printing head a pretty complex mechanism for advancing the thread.

Replace it with reasonably-sized heated pot (more like a small cup) and use _no_ mechanical parts on the moving assembly (aquarium motor would be located elsewhere and connected with a thin air hose)... The pressure probably need not be that much, so the pot would not weigh much more than a cola can ;-)

Perhaps I'm a bit too optimistic, but on the first sight it seems workable... It could probaly be even lighter than the current design (and hold more polymer too!)
Nice idea, but I've a feeling it'd be a real pain building an airtight port for refilling the hopper. You'd also have to ensure there was always a layer of molten material completely covering the exit hole from the pot, otherwise the air would just bypass the material and you'd end up blowing bubbles out of the extruder head!
I expect the most trouble would be with heating evenly and avoiding the scorching that plaasjaapie warned against. Could be solvable with careful and slow pre-heating profile and well-positioned heating elements...

Airtight port could be very similar to standard designs used in a lot of kitchens :) Just a ring of temperature-resistant resin between the cup and its cover, and a clamp or two holding it all together. The hose would be permanently attached to this removable cover by any means.

It would be reasonable to expect that the exit hole would be at the bottom of the cup - going to the head, so running out of polymer would not be that much worse than running out of filament in the current design.

It would be nice to estimate the amount of material extruded (in software) and not use the last few drops to avoid bubbles and simplify the initial melting of the added granules.
The melted pot on axis system w/ pressurized air is something I've been thinking pretty highly of for a few months, I've even got a little prototype with a copper pipe assembly with a bicycle tire type air valve firmly epoxied in. If the pressure is given by a piston, this might also help some of the drip problems that we are seeing with the current extruder.

If you're worried too much about weight, I also like the idea of having the z-axis directly move the platform with a screw-jack system, seems like to should add some stability to have x/y at fixed height, and z a mechanically separate piece.
I think, overall, I perfer the idea of having a 'spool of feed stock'... Two seperate machines, one that does the protoyping, one that creates the spools of feed stock...

You've got to store the material in some form, I think a spool of feed stock would store best (easier than pellets, powder or un/partialy recycled pieces...)

I would expect that a 'spooler' would extrude quicker than the prototyper (especially if we're talking something like 3mm vs .5mm), and could be built bigger/with a bigger capacity. This would equate to less manhours babysitting the machine... You could simply hook a spool up to the prototyper and let it run the spool dry... (big work surface, lots of items, maybe even a conveyor belt system to move away the completed parts or something)

From what I gather, it should be pretty easy to streamline a process of connecting the 'runs' to the existing spool, so long as the material is flexible enough you shouldn't have any problem spooling it up...

All in all, I think the spool is the best bet for 'continous runs'... Using sticks or a syringe or a pressurized vessel will inevitably limit the amount of material you can have 'ready' for the machine to use up... However 'big' you make it, that's the limit... With a spool system, you can just spin a bigger spool... make it wider, spin it more times... you could store all your CAPA (or whatever) on a single spool... you could have a different spool for different material/etc...
I was thinking rather than air pressure powered, you could use a heated bladder and mechanical pressure on the bladder to force the stuff out. you can pre-make these bladders, filling them with granules/scrap, putting it in a hot bath to melt it and sucking out any excess air. you then just pop off the spent one, pop on a new one and wait for it to heat up (or pre-heat it) and you are ready to go. pressure can be applied via a stepper controlled vice with some nichrome heating elements keeping it warm.
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to
Posts [Atom]