This project was the first of many exploring how some of the basic building blocks of machines will look like when optimized for 3d-printing. The Instructable focuses on building a 3d-printable oven. Aside from the electronics and a few hardware components, this oven is completely 3D printed! To prove a 3d-printed oven will actually work, we use it to BBQ some meat! Beside the delicious tri-tip, we also made some unexpected discoveries about plastic ovens...
Step 1: Why Make a Plastic Oven???
Of course, I didn't start out with the intent of making an oven. Five years ago, I was a post-doctoral researcher at UCSD working in a lab which studied the immune system. A small part of my project entailed an building an incubator to keep cells alive while we were imaging them on a microscope. The first version of the incubator combined a 3d-printed housing over a metal body, and it looked a lot like commercially available "stage-top" incubators for microscopes.
The incubator design evolved over the next couple of years. During the first major revision, I stopped myself to ask why was I still drawing a box if I had a 3d-printer? It wasn't even a conscious choice - I just started with a box because boxes are the most common shape for hardware products. I realized boxes are basically burned into our subconscious design paradigm. And why are boxes so common? Because they're the easiest shape to make - just use a press brake. This is when I started to question more and more of my design choices. As I reevaluated the different aspects of the design over the next few revisions, the incubator looked more organic and the underlying functionality began to manifest in the exterior form.
The final version had a very complex 3d-printed housing over a simple metal body. The plastic housing incorporated plumbing for air lines to supply the incubator, a reservoir for humidification, connections for tube fittings, ports for refilling water, and spaces for integrated circuit boards. The metal housing was used to for the walls and floor of the culture chamber where the cells were kept alive. I used metal because of its high thermal conductivity to spread the heat evenly around the chamber.
I never questioned whether I should be using metal at all...
Step 2: Forced Air Convection
In traditional ovens, a high-intensity heat source like a flame or electric heater is used and the metal spreads the heat around the chamber. Incubators used in biology labs work similarly, which is why I was using metal in the first place. When I started my Artist in Residence at Pier 9, I wanted to explore how I could eliminate the metal components from my incubator design. It's not that I hate metal or something, but 3d-printing metal isn't very accessible at the moment and I was interested in making it as easy as possible for budding scientists everywhere to 3d-print their own incubators.
I focused on using forced air convection through 3d-printed heat exchangers, and this is where things started getting hairy. Since I was designing parts with complex internal geometry, I had to consider the 3d-printer I was going to use. The simplest solution was to use an SLA printer since that what I used in the past to make similar parts. Fortunately, the Pier had more than a few Autodesk Ember printers available so I started building with that in mind. Since the Ember trades build envelope for higher resolution, the design I made as an AiR was very small and became more of a proof-of-concept than a working prototype.
The proof-of-concept doesn't look anything like an oven at all. The body holds all the electronics and attached to the body is a rectangular "head" with the printed heat exchange structures. The electronics consist of a circuit board with tall, thin ceramic heaters (actually just power resistors) and a blower that circulates the hot air throughout the part. The idea is to heat the thin layer of plastic between the interior surface of the oven and the intrinsic heat exchange network.
During my residency, I worked through a number of problems - like building manual supports (the auto-generated supports weren't working for these parts), organization of the feature tree (or design history if you use Fusion360), and self-supporting geometries.
Unfortunately, my residency was only part-time and the 3 months were over much quicker than I expected. But I learned enough to work on a "real" prototype...
Step 3: Building the Prototype
After my residency ended, I kept working on this as a side project at my regular job at FATHOM, an advanced manufacturing company. We have a lot of 3d-printers in our shop, including a Fortus 900 from Stratasys. Right around the time my residency ended, Stratasys released a new material, ULTEM1010, for the Fortus 900. When I learned ULTEM1010 was NSF-rated for food contact (a first for 3d-printed materials!) up to 375°F, I decided to tweak the new design for FDM so I could actually use my "incubator" as an oven!
All of these files are available for download. If you'd like to join FATHOM's A360 hub and contribute, please send me an email.
- Top. ~59hr build
- Base. ~44hr build
- Impeller.~16hr build
- Fan housing. ~20hr build
- Fan bracket. ~4hr build
Total printing time was about 142 hours -- more than 5 days! This was a lot longer than I expected but I'm sure I could reduce the print time down significantly by eliminating material, many of the walls are unnecessarily thick.
- Motor (qty 1) : Turnigy 9014. This motor is complete overkill, but I tried using a smaller motor but it didn't have enough torque to spin the impeller. I wasn’t going to make the same mistake twice… ;-)
- Motor controller (qty 1) : RotoGeeks 12A ESC (w BEC). I tried a few different controllers, but found this one to be the easiest to program and integrate.
- Raspberry Pi2. 'nuff said.
- PiPlates DAQC plate. It has the two peripherals I needed (1) an analog out to the SSR and (2) a temperature sensor with a simple API and interface (1-Wire). It also happens to be, basically, an Arduino slaved to the RPI over SPI, so I thought it would be a solid decision as far as hardware architecture.
- Solid state relay (SSR): Crydom MCPC1225A. It's expensive, but it's worth it. Smooth linear control over ~10A of 115VAC is no joke because regulating AC power is significantly harder than most people realize. This SSR has a microcontroller inside that sets the power of the heaters proportional to an analog input. Without this SSR, controlling the heaters from the Pi would have been a lot more challenging. Eventually I’ll roll our own solution, but for the prototype, this was perfect.
- Thermal cutoffs: 240°C. ULTEM melts at 220°C so this will keep any thermal runaway from getting too catastrophic! FYI - I started with much lower thermal cut-offs while I was testing. The cutoffs were in series with the heaters (they just took the place of one of the heaters). You can see them in one of the pictures.
- Temperature sensor: DS18B20. Simple 1-wire sensor. It’s only good up to 150°C, so I’m currently working on replacing this with an RTD using the MAX31865
- Heating elements: 70x 5W 1 ohm power resistors. I run them a lot hotter than just 5W but they've held up just fine. Usually power dissipation ratings are in still air. If you wire them as 2 parallel strands of 35 heaters, you get ~700W of heating. If you wire them as 3 parallel strands of 23 heaters, you get ~1600W of heating. Choose wisely...
- Soldercup sockets: 72x 3-position sockets. These come in strips of 64 that you can cut up.
- 17V power supply, at least 2A. I just used a cheap bench top variable power supply to run all the electronics.
- Misc hardware:
- Bearing (qty 1) : McMaster PN 60355K506
- O-ring stock (10' length) : McMaster PN 5229T51
- Rubber bumpers (qty 4) : McMaster PN 9541K1
- Screws for attaching motor to impeller: M3x20
- Screws for attaching motor to bracket: M3x15
- Screws for attaching bracket to base: 6-32 x 1.5"
- Threaded inserts : McMaster PN 92395A113
After printing and cleaning the parts, it should be fairly obvious how they fit together. The wiring diagram is attached. The most time consuming part will be wiring up the heaters. There's quite a bit of soldering, it's tedious but not difficult.
Here's a link to the drive image for the Raspberry Pi2. There's plenty of instructions online for loading the image onto an SD card so I won't go into details here. The image includes Meteor.js as well as some Node.js modules and other other assorted libraries. The full source for the Meteor app is available on github.
Step 4: Enjoy the Fruits of Your Labor!
This is the easy part... A 1/2 pound tri-tip cooked in about 2 hours. I was flying blind without a meat thermometer, so I probably over-cooked the meat, but it was still delicious!