When to end a roast

If you were interested in roasting a medium light roast (as an example) on a new bean, what do you use to determine when to end the roast? Time post first crack? Bean temperature? Are there other options? Time options would seem to be hugely influenced by ROR. I’ve been using IBTS temperature with reasonable results but if I’m trying to medium light I may get medium or light (as measured with an Espresso Vision).

I get consistent results with IBTS; Espresso Vision readings confirm consistent results. Yes, I-RoR predicts time vs. temp at eject. So that has to be part of how you manage a roast. You can affect RoR to some extent using drum speed (slower causes RoR to increase and conversely faster tends to make RoR decline (has to do with contact-time between the drum and the beans).

You could try developing a Recipe to smooth the process and to get (usually) repeatable results. My Recipes are specific to beans & end-temp as well as batch size. So if I had 5# of something it wouldn’t be worth the effort… I’d be out of beans before I had a Recipe worked out. Since I’m usually roasting beans from the same source with the same processing and expect to continue, it is worth the effort. I haven’t had to discard a batch but it’s been close a couple times! Small changes can make a difference. I started with a profile that had good results and continued tweaking the Recipe to replicate the profile.

Recipes based upon Time haven’t given me the repeatable results I want, so I use IBTS readings. However, Recipes require a Time component even though I don’t want to have Time control the process. I use a Time value much less than expected. To keep the steps in order I also increment each Time value by 1 (Recipe-builder doesn’t have a way to keep things in order except using a Time sort, so this is my workaround).



There is a pretty good podcast called Coffee Roasting Navigated by Therese Brøndsted. She has a Bullet, and interviews several roasters, and roasts herself quite a bit.

She explores the various ways to figure out when to drop - smell, taste, time, temp, sight, and the pluses and minuses of each.

I usually (well, always) use development time. 20%. That usually takes me almost to the end of first crack, never all the way into second crack.

I have the problem that Bruce mentions above - I buy 5 pounds of coffee at a time, and never the same coffee twice, so creating a recipe would be a fool’s errand. (of course, for me, that is usually just in a day’s work) I like the dev time measurement, and will sometimes use the overlay to see how the latest roast compares to my last roast of the same bean.

Part of the fun of the roasting process.


Thanks for the replies. I think I also have the problem Bruce mentions. Maybe worse. I’ve been experimenting with small batch sizes and may have settled on 325 grams. I tend to buy five pounds of each of several beans. At 325, I can get seven roasts from a five pound bag. While I may roast the same way twice, I try not to. Each roast is an experiment (longer or shorter roast time, longer drying stage, different drop temperature…). That said, I’ve been working on partial “recipes”. I set them up so they start the roast (charge temp, any initial soak, initial power, fan and drum speed) and make scheduled changes to drum speed (currently none) and fan speed (one change just before first crack). I also have some reminders to mark yellowing and prepare for first crack. My current approach is to overlay a prior roast that more or less fits what I want to do. Once the recipe starts, I’ll make manual power adjustments. I use the overlay only as a frame of reference and use manual adjustment to do things like lengthen drying or increase roast momentum beyond first crack (if I want a higher end temperature) or whatever.

Bruce, thanks for the idea about using time value to organize recipes. I tend to use temperature thresholds and my recipes were a mess.

The Roasting Navigated podcast is great.

Thanks again. Carl


Looks like some interesting content in Therese’s site - thanks @dpiette

@BerkDirt - my 2 cents - I now use both time and temp. Originally I was mostly using time for most of my roast and drop based on sound, sight and smell of the bean, but I am learning that using target temps also during the roast to determine when to change P or F is also a good thing and gives me a bit more consistent control. What I’m seeing emerge in my data is for a given batch size, most beans are going to hit YP and FC within a temp range, then the question becomes how quickly do I want to get there and this is where settings will impact the timeline to get there. In the beginning of my journey I was all over the place to some extent, but at Bruce’s recommendation I was studying a lot of bradm’s profiles, contacted him to get some advice for 350g batches, and using his as my starting point to tweak. I’ve finally settled on about 3 different roasting plans that I use as my guide.

1 Like