Congrats! ...pretty solid input you've got there, telling from deltaBT curve. What's your sampling rate and delta span settings now? Regarding crash&co. according to Scott Rao: you have a baked roast when you taste one, not necessarily when (you think) you've spotted one in your graph. I mean, there's obvious signs such as a real stall or decline in roasting temp, but especially around 1C start you need to develop a feeling for your machine whether or not you can trust readings. Because - and that's the dangerous part of high resolution logging - beans absorb quite a bit of heat prior to 1C to eventually release it suddenly upon bursting open. If one of these effects happen in direct adjacency of your thin and super sensitive probe, guess what you see on screen a second later? Thus, you should interpret readings relative to your settings and probes. For instance, if your sampling rate is 2s and your delta span is 6s, a delta of 0 means you had one or two negative values in your data set. But without knowing previous readings you cannot determine how much it'll point downwards 2s later. This is where the fancypants new idea of AUC comes in handy. But you wouldn't really need it visualised if you're paying attention [emoji56] Keep it coming mate!