Heater 2.0: a (big) step in the right direction

Two weeks ago we wrote about our first heater design and some of the troubles that we have come to associate with it. In doing so, we also prompted some very interesting and active discussions about new heater designs in the CandyFab Hardware Forum.

Here, we describe a new heater and nozzle design, which (along with others currently under development) will serve as a useful reference design for the CandyFab project. It's still far from perfect, but it is none the less a leap forward in several ways. It costs under $20 and can be made without any unusual tools. Early indications are that the new design is quite promising.

Read on for the juicy details.

Our starting point is a commercially available sealed heating element. There are quite a few different options for this in a wide range of prices. We decided to go with one made to fit a soldering iron. Specifically, we have a 24V, 50 W ceramic element made by Hakko, $12 at SRA solder. It's model Hakko A1321, for the Hakko 908 soldering iron, and it is-- in all seriousness-- a real work of art.

It appears to be a thin-film heater grown on the inside of a slightly-translucent alumina tube; I don't know how they do that. The quality of its construction is simply stunning. Some other "comparable" heaters, in the range $6-$8 at SRA solder are a low-grade ceramic tubes with potted heating elements-- looking at one of those really proves how great the Hakko element is. Also note: I originally tried to use one of the less expensive elements, but it began to droop in performance immediately and died after about eight hours of print time.

Back to the topic at hand, we want to wrap several windings of 1/8" OD soft copper refrigeration tubing around the heater element. Tubing like this is easy to get and costs about $0.50/ft. (I got a 50 ft roll on Amazon.) The idea is that the air will slowly flow through the tubing, heating up as it goes, until it finally gets close to the very end of the heater, near the nozzle, at which point the air will approach its maximum temperature.

The soft copper tubing can more-or-less be bent by hand, but the stresses might be too much to directly wind around the ceramic element. Fortunately, it's a standard diameter (5/32"), and you can wrap it around the non-fluted portion of a regular 5/32" drill to bend it into a tight spiral of the right size.

Once the coil is wound tightly, the drill can be pulled out, and the ceramic heating element can be slid into place. (The heating element in the picture on the right is one of the low-end knock off heating elements, not the nice Hakko element.)

Ideally, the coil should be made tight enough that the windings achieve good contact with the ceramic, but not so tight that the heater cannot slide into place.

The next step is to insulate the heater and the copper tube. This is critically important. We want our heater to be modeled as an oven, more or less; an area that stores a lot of heat, and can tolerate a modest air flow through it without too much cooling. Without good insulation, you might imagine that the outside of the copper coils would instead get fairly cool by convection-- the heat would be wasted. The insulation will help us, with limited power, to make the turns of copper tubing close to the temperature of the ceramic heater.

The insulation that we used is ceramic fiber insulation with a foil backing, #9379K92 from McMaster-Carr. (It's very easy to work with, but ultimately not as good of an insulator as we'd like.)

Outside of the heater and copper tubing, we've added about 1.5 wrapped turns of the insulation, and it's all held together with a nylon cable tie. Yes, nylon can melt-- but it discolors before it melts, and now we have a telltale indicator should the outside of the insulation get dangerously hot. (Note that it does run too hot to touch, even with a tiny fan blowing gently on the outside.)

In the photograph above, you can see the very tip of the copper tubing sticking out the bottom of the bundle. (Alternate view here.) It's not nearly so nice looking as our other heaters and nozzles. However, we can work on looks once we have a more solid heater design. =)

So, that all seems good enough, but upon trying out this new design, we found that it wasn't actually getting the sugar hot enough. The problem certainly wasn't that the nozzle (end of the copper tube) doesn't get hot enough-- it can be made to glow red-- the problem was that there was not efficient exchange of heat from the copper tubing to the air moving through it.

My solution was to add a "preheat" coil as seen above-- that's an additional set of windings that effectively doubles the length of the air path through the copper tubing. There is a gap between the inner coil and the outer, which allows a slight temperature difference between the two. The hope is that after air travels through the outer tube it, it has warmed up enough that traveling through the final tube part should be enough to push the air up towards the nozzle temperature. Also, by having the gap, it should reduce the amount of heat conducted directly away from the central turns around the heater. Again, this whole structure is encased in the fiber insulation before use.

The downside of adding preheat coils is that they increase the moving mass and bulk of the printhead. A second issue is that the longer narrow tube resists airflow well, and begins to require a higher pressure (to maintain air flow) from the air pump-- the lowest end aquarium pump may no longer be sufficient; we may need to upgrade to the $10 pump.

Tubing
The next important modification that we made is to replace the regular silicone air tubing that goes to the heater with "crush-resistant" FDA grade silicone tubing, also available from McMaster-Carr, #3184K5. It costs about $0.75/foot, but the performance justifies the higher price. This tubing has a slightly thicker wall and keeps its shape quite well as the print head is moved around, leading to more uniform temperature and air velocity. The silicone tubing hooks directly onto the copper tubing that goes around the heater. To avoid overheating the silicone tubing, the copper tubing should extend several inches away from the heater at the point of connection, ideally exiting the insulation before doing so.

Room for improvement
We have tested out the use of this print head, and things do indeed look promising, but there are still several shortcomings of this design. First and foremost, the heat exchange process is still rather inefficient. The copper tubing is kept hotter than it needs to be, and the air does not get as hot as it should. The net effect is that even though the heater system is getting damned hot, we're not melting the sugar as well as we'd like to and the speed suffers. We are able to melt a much smaller "pixel" of sugar at a time, approaching 1/16", with very little blowing around of the sugar because the air flow now low enough. Melting this tiny dot of sugar takes about 2 seconds, typically, which makes it take a very, very long time to fab any object of appreciable size.

To melt faster for building larger objects, it would be better to use a larger nozzle with a corresponding increase in air flow-- still preventing the sugar from blowing around. However, that isn't really possible with this system. With this new heater we have gone from a regime where we have an excess of power and can't get the air flow low enough, to a regime where we have an excess of power but can't deliver it to the air as efficiently as we would like.

The solution is not to add more power-- there is enough heat; it's just in the wrong places. What we need to do instead is work on the heat exchanger. One way would be to add more fine-diameter preheat tubing. Another, perhaps better, method would be to increase the surface area that the air travels over by filling the tubing, perhaps with stainless steel wool or the equivalent. One variation that might work well is to go to from the long length of tubing to a thicker cylinder, maybe 1" in diameter by 1" tall, that contains a cartridge heater and is filled--almost densely-- with steel wool. As air is fed in from the top, it should have plenty of time to thermalize before exiting the bottom of the cylinder.

But how does it taste?
As a final note on this design, it's now reasonably safe for use with food. As you might recall, one of the issues with the old heater is that we don't know exactly what it's made of, or what chemicals our air will be coming in contact with. There's no such concern with this design; the air is restricted to flow through the copper tube and never even touches the heating element proper. There is one caveat to be aware of: The copper nozzle tends to become oxidized, and could potentially flake off bits of this material into food that is being fabbed. Heavy metal contamination *needs* to be avoided, so it would ultimately be better to use a stainless steel nozzle with thin stainless wrapped around the heater element. In some of the other heater designs that we're working on, the heat is supplied by a strip of raw nichrome wire. Nichrome is a very common heater material-- found in toasters, for example. It is worthwhile to note that nichrome has a much more serious issue with flaking cannot be used directly above food, unless mechanically sealed in place. Ultimately, that's probably where we need to go: there are a number of high-temperature ceramic sealants that would be suitable for protecting our heater elements.

What's next?
We are working towards reliable versions of the CandyFab hardware and software, and this is an important part of the puzzle. The reference design that we have shown here is obviously not the only way to approach the challenge of designing a heater and nozzle for CandyFab; it is also important to develop alternative designs as well.

We would really like to hear your comments, suggestions, refinements, and new ideas about this and other heater designs, and the CandyFab Hardware Forum is the place!