Notes on CandyFab and heater design (1.0)

Tuesday, August 14 2007 @ 06:38 AM PDT

CandyFab has gotten a lot of attention since its debut a few months ago, and we have heard many, many responses to what people have seen. The vast majority of the reaction has been positive-- people like the idea of computer-controlled sugar sculptures. A small but vocal minority is underwhelmed, also for understandable reasons. Of course, you can't please everyone (like the guy who says that we should be selling complete CandyFab systems for $5-- or $10 including next-day shipping), and we aren't trying to. We have been perfectly clear from the start that we are not actually trying to compete with commercial fabrication systems. Those are designed as rapid prototyping systems; in other words; for doing work that you would normally do with conventional machining tools, but faster and (perhaps) with less expense. They usually start in the $20k-$50k range, and are able to produce small, detailed models of machine parts. Chief among the metrics used to judge machines of that type are resolution and dimensional accuracy. The CandyFab project has a different set of goals. We want to reduce the costs associated with three-dimensional solid freeform fabrication, and to promote the use of fabrication technologies for culinary, educational, and artistic purposes. We're not (just) interested in just producing small plastic parts with high dimensional accuracy-- there are plenty of other open-source efforts already working on that. We want to do what everyone else *isn't* doing. We want larger build envelopes, lower building cost (under $1k), and lower media costs. We want this to be something that you can make yourself and use it to make large scale, inexpensive sculptures, in a variety of materials. The CandyFab process is based on a low tech, low-cost, and fairly novel technology: selective hot air sintering and melting (SHASAM). And, our dirty secret is this: We haven't yet begun to scratch the surface of how well hot air *can* work. When we first were able to get 3D output from the CandyFab, about three months ago, the highest priority was get some objects printed to show of the capabilities of the machine-- since we hadn't made any objects with it yet. So, we printed a bunch of things with it, still using the very first heater design that worked at all: the "Heater 1.0" design-- a horrible kludge of a barely functioning hot air gun. In fact, all of the CandyFab output that we have displayed (thus far) was made with this hot air gun. There is only room for improvement. We have not done a good job of making it clear how much this heater design affects the quality of our output. Nearly all of the issues, difficulties, and criticisms that we've had with the CandyFab are directly traceable to the design of our heater. This distinction is important; when we discuss the limitations of CandyFab processes, we need to be clear which issues are fundamental limitations to SHASAM, and which are limitations of a particular implementation of it. We have recently been working on some new designs for heaters that alleviate many of the issues that we've had with the first heater design. And so it's at this time that we would like to bring up and discuss in detail the (few) good and (many) bad points about that first heater, and their effects on build quality.

Thus far, the effective resolution of the CandyFab 4000 has been approximately 5 mm per axis (horizontally) and 3 mm (vertically). So, what's limiting the resolution? The heating element is from a hot-air rework station. It's designed to heat a fairly high volume of air to high temperature, and relies on the incoming air flow to keep the "cold end" of the element (where the insulated wires come in) relatively cool. Think of this like a hair dryer: It's designed to heat up rapidly moving air, and it requires a certain velocity of incoming air to keep the coils from overheating. If you turn the fan off, or down to nearly off, the coils are sure to burn out in short order. When we put a "fine" (~ 1/16") nozzle on the output of our hot air gun, that increases the output air flow velocity to the point of absurdity-- you can blow sugar around from six inches away. If you try turning down the air pump so that the output air velocity is anywhere near usable, the heat gun overheats and shuts down-- there is a thermal cut-off switch in the "cold end" that tells us when things have gotten too hot. One mod that we made was to put a wad of fiberglass insulation between the heater element and the "cold end" of the element, so that we wouldn't get to thermal shutdown so easily. We also went to a larger nozzle (~1/8")-- an important compromise. Finally, we added a fan to the outside of the heat gun to help cool the parts of the heat gun that we don't need to be hot-- this was the critical step that finally made it so we could use it to effectively melt sugar. And it does work, it's just messy. Some of the detrimental effects of using the present air gun, which still has too high air flow, are as follows: We are using a larger than desirable air nozzle (~1/8"). That limits the effective resolution at that size. We are blowing the air around too much, which further degrades the resolution, by heating areas well outside the pixel that we are trying to address. We are blowing around the melted sugar. This leads to misshapen pixels that blur together and sometimes lurch where they shouldn't, and further yet degrades the resolution. Very thin layers of sugar get blown away before they can melt. This limits the vertical resolution. The thickness of the top sugar layer is changed by the blowing air, leading to differing melt times. Keeping the heat at a given point long enough to ensure melting everywhere leads to nonuniform color-- some spots become dark caramel. Sugar blows up and onto the nozzle, collects there, melts, darkens, and eventually drips down leading to a "chocolate chip effect" -- little black dots that show up on the sculptures. If one of those black dots ends up in a pool of clear melted sugar, it can turn it to dark melted sugar in a hurry. The output cannot be considered edible, since the heating element is not *known to be* free of toxic chemicals. Power is wasted in generating excess hot air. Now, it's not all bad. Note, for example, that we have been able to build 3D objects using even this inferior heater. Its coil design is actually quite a good heat exchanger, and does manage to heat the heck out of air. (Many simpler designs simply do not have adequate heat exchange with the air.) The second advantage of this heater design is that it is capable of producing a large volume of very hot air-- perfect if you want to build very large, coarsely grained sculptures. It would be better to produce this large volume of hot air at a lower velocity, of course. The steps forward are obvious, but take time. From day one it has been obvious that we need to replace the heater with one that reduces the air flow to that of a gentle breath that does not disturb the sugar. We believe that a better heater design can push CandyFab towards an effective resolution of 1 mm, when printing in granulated sugar, with relatively uniform white to light-amber output. We have have been working on a new design that (more or less) solves problems 1-7 above, but it's a hack. We'd like to hear your input and ideas about how to buy or build an inexpensive heater that addresses the problems above. In technical terms, what we are building is called a low flow rate process air heater, and it would be great if we could find an inexpensive one that was (more or less) designed for applications like these. Please leave your comments here, or in the CandyFab Hardware Forum.

Comments (2)

The CandyFab Project
http://www.candyfab.org/article.php/heater1