Typica

Sharing Roasting Plans with Typica

I've written a new article at Daily Coffee News that describes a method for creating roasting plans with one machine for use at a different one when there are differences in how these machines measure seed temperature. This is based largely on work that I did in 2012 as described in this video, but I felt the topic deserved an update now that I've been using the technique for several years, have more experience applying that to different machines, and because of a disturbing trend in the way people discuss rate of change data in coffee roasting which assumes that observations about this sort of data will be consistent across a broad range of machines when it can be demonstrated that this simply isn't true.

Linear splines work amazingly well for sharing roasting plans among multiple machines.

In this, I mention the use of linear splines to create a mapping function to take measurements from one coffee roaster and present them as if the coffee was being roasted on a different machine. This is a feature that's built into Typica and provides what I believe to be the best balance of power and ease of configuration compared with what's available in similar programs. This highlights one of the benefits of Typica being developed by someone who actively roasts coffee on a daily basis. Other roasters often notice that they need a similar feature, so they ask for the ability to provide a constant offset, not realizing that what they really need is the sort of non-linear mapping that feature was designed to provide, at least until after they try what they thought they needed and noticed that their measurements still aren't showing up right. If you're not actively using the software in a real production environment it's easy to take feature requests at face value instead of properly determining what people really need. Hopefully sharing my experience with this more widely will help developers of competing programs make calibration features that are genuinely useful.

I also note that these linear splines are only an approximation of some underlying curve. For those who want to dig deeper into this and derive a closer mapping, cubic spline curves are a great place to start.

In unrelated news, I'll be one of the judges at the SCA United States Roaster Championship competition at the Global Specialty Coffee Expo in Seattle in April. If anybody reading this is also going to that event and wants to set up some time to talk about coffee roasting or Typica, feel free to get in touch.