Tuesday, December 23, 2008

Extruder Ideas

So I have been thinking about the thermoplastic extruder lots since I read nophead's blog http://hydraraptor.blogspot.com/2008/12/sticking-point.html. My goal in this post is to decribe the problems and the abstract solution to the problem.

The big problem that nophead describes is the transition between molten plastic and solid plastic in the various extruder designs. The problem is that in some of these extruder designs there is a temperature gradient from the hot end to the cool end. Because these design have a gradual, nearly linear tempurature change from the "hot" end to the "cool" end of the extruder. This tempurature distribution leads to a semi-solid section of thermoplastic that gets stuck in the exturder. The first picture shows what the tempurature gradiant in the modified desoldering tool is problably like. Because the way the desoldering tool is made the transtion zone (that includes the glass transition tempuratuer for the thermoplastic) is large. This long transition zone results in a long portion of the filiment that will remain stuck in the exturder when you try to reheat it.

I propose that there is a very simple solution to the problem. Simply make the transtion zone shorter. This is shown in the second picure:
Reducing the length, or nearly eliminating this transtion zone (at least in the filiment) will prevent the filiment from becoming stuck in the extruder on reheat cycles. There are many ways that this can physicly happen. The current teflon sleve acomlishes this goal to some degree. If you look at the system as a set of lumped thermal masses, where each lump has as thermal resistance, then we can use an analogy of a voltage divider. The following figure shows this:The higher the thermal resistance of the transition zone, the closer to ambient tempurature the junction between the transition section and the cool section will be. Lets look at the equation that gives this thermal resistance: R=L/(k*A), where L is the length of the section, A is the cross sectional area of the section and k is the thermal conductivity of the material the section is composed of. So we want to increase the thermal resistance of the transition zone, right? Well to do this we can do three things, increase the length, reduce the cross sectional area, or choose a material will low thermal conductivity.

There are some tradeoffs in each of these ways to increase the thermal resistance of the transition zone. Increasing the length of the transition zone does indead increase the resistance but at the cost of a larger transition zone making the sticking problem worse.

We can reduce the cross sectional area of material in the transition zone, and this will increase the resistance of the transition zone. However we can only make the material so thin because it is in tension, and be need to make sure the stress in the transition material is reasonable.

Selection of material is at the heart of the mater. If we choose to use a metal, stainless steel would be one of the better choices, however it is expensive. Only a hand full of plastics and composite materials can handle the heat, hence the reason for the current teflon transition zone. However there are several other ceramic, and ceramic composites at are very cheap and meet the bill. An example of one of these materials is gypsum, conviently contained in drywall. More on this later.

Now keep in mind that what we really care about is tempurature of the filiment in the transtion zone. The very best thing we could do is eliminate any conection between the cool section and the hot section of the extruder except for the filiment it's self. Now as this isn't really possible, we move to the idea of removing the local connection between the transition zone and the filiment. I'm not saying this very well so I'll include a picture to describe what I'm saying:

I did a quick thermal analysis of this concept using an thermal FEA model. I assumed the whole thing was made of stainless steel with average convection cooling and heat flow. Here is the result:

I didn't model the filiment in the extruder, but will the low thermal conductivity the polymers there would be very little difference. As you can see the filiment goes from low tempurature to the feed tempurature in a very short difference. However this is only one way this can be done. Once again I just want to explain the concepts that will allow us to design a great extruder.

I'm all out of time for now, but i'll continue will this topic in the next couple of days.


  1. Congratulations.
    Visit and indicates to your friends mine blog.
    Good bye

  2. I don't think you can have a gap in the filament feed anywhere near the hot end. Liquid plastic flows back up the tube until it gets to the cooler section and freezes again. If there was gap it spill out.

  3. I may be wrong, but it seams to me that there is a gap between the drive rollers and the heating head in the Dimension FDM Machines... I'm going to do an experiment tonight to see if it is possible.

  4. It *might* be possible to have a gap, iff the filament is moving fast enough that it remains solid where it first contacts the hot zone (and acts as a piston, sealing the melted plastic from flowing into the gap. However, this is a chancy thing to count on -- if there is a pause in filament feed (or too much of a pause), then melted plastic (e.g. from just inside the hot zone) will spill into the gap.

  5. I've been thinking along similar lines re materials to "steepen" the thermal gradient, and (in my experience, buying small amounts), stainless steel isn't significantly (or sometimes any) more expensive than brass. (Unless you get 316, which isn't really needed here, since our env. isn't highly corrosive.) A cast-able ceramic might also serve as a small thermal break; I may try that soon if SS doesn't pan out.

    I'm not a heat transfer expert, but it seems to me that if the thermal conductivity in the transition zone were too low, one might again get the "freezing plastic in the transition zone" that nophead experienced using a (long) stainlesss steel extruder barrel. This might come into play at extrusion rates where the power needed to heat/melt the plastic was comparable to the heat conducted by the structure in the transition zone.

    -- Larry

  6. Not only that but when the filament is stationary heat will conduct back up it and melt the piece in the gap.

  7. I think that we could reverse the feed of the filament a little bit so that the tip is not longer in contact with the "heater" section when you are not feeding. This should prevent the filament from melting while stationary.