Roasting temperature and event data was recorded and monitored through Artisan on my laptop via USB during roasting. The profiles I created were used while roasting and between batches in order to evaluate roasting behavior and make adjustments as necessary.

As the only variable I can control with the hot air popcorn popper is the heat and fan output through the AC dimmer, I was slightly limited in my ability to make adjustments. In a more capable system, one would be able to control the heating element and fan independently of each other, providing a greater range of possible roasting profiles.

In order for my roasts to be consistent and stable, as well as provide proper development time and a (generally) decreasing rate of rise, I set the AC dimmer dial to 25 from the following options:

My recorded roasts follow, where the dark blue line indicates the bean temperature, and the light blue line is the derivative of the bean temperature, which serves as a good indicator for roasting stability.

The first guiding principles for these roasts was to have a total roast time between 8 and 12 minutes. This was achieved by controlling the heat output of the popper with the AC dimmer, and is important to allow for greater control of the roast profile and to have a more gentle, and slower roast. Without the AC dimmer, roasting would take 4 to 5 minutes on full blast, which was difficult to control.

The second guiding principle was to create a bean temperature curve similar to a logarithmic curve so that the beans gradually reach their final roast level. Here are two examples for this principle:

In the first image, the temperature started out slightly lower than intended but also didn’t slow down properly towards the end of the roast, and perhaps was finished a bit more abruptly compared to the second image. The second image is still slightly below the plotted red line initially but does appear to approach the intended curve over time.

Whether these differences in roasting temperature data manifest in taste differences I am not sure, but in terms of consistency and reproducible between batches these principles were very helpful.