I recently encountered a weird problem while roasting coffee. This happened because I forgot to turn OFF the Auto Drop setting in the artisan software. While investigating that problem, I learned how to see the actual data being captured by the artisan software.
I was surprised by how random the data was. The rate of rise (ROR) curves that artisan produced for the Hottop B-2K+ roaster were chaotic compared to similar curves for the Hottop TC4C roaster. Some of the online discussion indicated this was due to the combination of the sampling rate in the artisan software as well as the low precision of the data produced by the Hottop B-2K+ roaster.
I searched through the artisan software blog to get more information. I got a good hint from this artisan blog post:
More on Digital noise
I did a bit of armchair analysis to understand the problem. I copied the data into an Excel spreadsheet, and tried to reproduce the problem. It is impossible for me to fully understand the effect of the numerous parameters in artisan that produce the ROR curves. I made some simplifying assumptions, and was able to produce similar results.
ZIP file with the Excel spreadsheet
I modified my Excel sheet to see the effect of changing the data sampling rate from the default of 1 second intervals. It appeared that a value of 3 or 4 seconds would give better ROR curves.
I recently changed the default sampling rate for my Hottop B-2K+ and Hottop TC4C to be 4 seconds:
This produced better (smoother) ROR curves. This graph shows the results for the Hottop B-2K+ roaster using a 4 second sampling rate, where the background curve used a 1 second sampling rate:
