Today I received an email asking for advice on a couple points related to coffee roasting. One of these was a question about how best to start the roast for consistently matching a roasting plan if the readings of two probes at the end of roaster pre-heating are not consistent with earlier observations.
When something in the roasting environment changes, the first question should be why that changed. That could be something relatively innocuous like a change in the weather, but it could also indicate something more serious like a lapse in regular preventative maintenance. A dirty roasting environment (including the exhaust system) is an inconsistent roasting environment as well as a safety hazard. For the rest of this article I'm going to assume that everything is clean and in good working order.
I'm usually a little bit uncomfortable answering questions about ideal starting temperatures because I work in a large range of hot enough to get my coffee to do what I want and not so hot that I scorch the coffee. What that range is depends on the design of the roaster, how well the roaster is pre-heated, how large the batch is, physical properties of the coffee being roasted, and the specifics of my roasting plan, but at least on my roasters it's a pretty broad range, especially since I've developed an approach to the earliest portion of the roast where the exact starting conditions just don't matter very much.
What's happening at the start of the roast? The coffee gets hot, there's some water evaporation, but there shouldn't be much happening in the way of chemical changes. While it's certainly possible to go to extremes to blast through this portion of the roast as fast or as slow as possible, in commercial roasting equipment there's usually a broad range of times for the start of the roast where you will not be able to taste the difference in the end result assuming that once the coffee starts changing chemically you match your roasting plan exactly from that point through the end. That means that the exact starting temperature doesn't matter very much except that it needs to be possible to match the planned rate of change at the start of chemical change and retain the ability to control the roast from that point through to the end of the batch.
Adjusting the roasting data to match at the point where the coffee starts to change chemically is a way of analyzing roasting data that I've been using since well before I wrote Typica. I found it to be a very useful way of looking at batches that didn't match in flavor and figuring out why they didn't match, but this way of analyzing roasting data after the coffee was roasted never really caught on. It's difficult to explain, time consuming to do with paper records, and at the time I was doing this none of the software for coffee roasters had anything like this. Since most roasters don't have experience with this sort of analysis they tend to underestimate the difficulty of allowing a difference at the earliest portion of the roast to exist while matching later parts of the roast exactly.
When I started talking about this, roasters would come to me with roasting data and say here are batches that match the way you've described but there are clear differences in how they taste. In every case they had gone on a gut impression that the batches matched without really digging into the data to compare the batch data properly. When I did that for them, differences were both obvious and in exactly the temperature ranges expected for whatever flavor difference they were describing. Clearly, it's difficult to do this sort of matching manually so I decided to add a feature to Typica several years ago that makes this approach to the earliest part of the roast easy. Having this transformation of the data performed automatically during the roast changes this sort of analysis from a tool for understanding why two batches of coffee taste different from each other into a tool for preventing those inconsistencies in the first place.
If you can't do this, it's much more important to match up with your roasting plan as early as possible, but with Typica and properly configured profile translation I and many other roasters have observed that you can worry less about that and still get more consistent results.