RoastCoffee
It all started with a moka pot. That progressed into the crazy hobby of espresso.
The next reasonable step is, of course, to build an automatic fluid bed roaster with a control system that maintains a desired curve for repeatable experimentation with roasting.
Well, maybe that's just the engineer and maker in me speaking.
Meet roastCoffee, my automatic coffee roaster project. It is based on the FreshRoast SR500. I did get some tasty roasts out of it originally, but I couldn't help but think: I could totally improve this.
(NOTE: Do NOT play around with mains AC power if you do not know what you are doing!)
I replaced the original control PCB with one powered by an AVR microcontroller development board, mainly because of they are super easy to come by and easy to tinker with. A 1.3 inch OLED display adds nice potential for the user interface, and a set of satisfyingly clicky buttons add to the experience.
After hours of measuring, 3D modeling, and 3D printing, I had a drop-in control panel that fits right where the original went.
I also replaced the original TRIAC circuit with an off-the-shelf dual TRIAC module for precise control of the fan and heater. This now sits on the back with a big heatsink for power dissipation according to specification of the module's TRIACs. It only gets warm.
In conjunction with the TRIAC module, I used a pair of thermocouple modules for two bean probes in the chamber, using the average of the two for feedback control (or falling back to a single probe if one were to fail at any point during roasting).
To run the machine, I wrote drivers for reading the temperature from the thermocouple chips and controlling the TRIACs in a phase-delay configuration (this corresponded with the non-zero-cross output couplers that came with the module. However, zero-cross output couplers for integral cycle control are a better choice for their reduced impact on EMC and should still provide precise-enough control. I would want to swap the output couplers for future versions). For the control system, I implemented a discrete-time optimized PID controller. I opted to use the nifty U8g2 library for driving the display.
The attached video is of the first roasting before I had completed the enclosure. As for the temperature curve for the roast, I am following the advice presented by Scott Rao in his coffee roasting book. The resulting roast ended up slightly under-developed, but otherwise pretty tasty! It was a very peanut-y bean grown in Honduras.
I really enjoyed working on this, but now that I have the beautiful Kaffelogic Nano 7, I can focus on improving my roasting skills and move on to other fun projects. You can still check out the code here: https://github.com/vjapolitzer/roastCoffee