Joking aside, the fact that you can take third place at the world's without following Rao's advice would seem to suggest that you can achieve excellent results outside of his recommended parameters. I have no problem with saying, "this is one way to get good results," but presenting these things as "commandments," and implying that failing to heed them will produce unpleasant tasting roast defects, seems to be taking it too far.

Part of the problem, I think, is the feeling that a book like this needs to tell people "how to roast." A better approach might be to describe the effects that different variations in roast profile can have on the finished product, such as those that can be found in the Sweet Maria's library or in the archives of HB. Once people know how to achieve different results for particular coffees, they can decide for themselves what tastes best.

[creative nickname] wrote: I have no problem with saying, "this is one way to get good results," but presenting these things as "commandments," and implying that failing to heed them will produce unpleasant tasting roast defects, seems to be taking it too far.

I'm not sure that's what Scott Rao did, especially since he leads off the chapter with "Please don't take the word 'commandment' too seriously."

That's fair, and a more charitable way to read the chapter. (I haven't looked at the book since I read it a month or so ago, so my memory is getting blurrier about the details.) But I'm still surprised he didn't discuss the fact that some coffees can taste very good with a slow-start/fast-finish approach.

I'd like to be in this discussion a bit more, but I've been traveling and haven't keep up with my work.

Regarding the 3rd-place profile, it is likely the bean probe was either in a different position than it was for the other competitors (perhaps unwittingly, as they are known to move if bumped or not installed tightly) or on a different machine. I've looked at cropster curves from about 100 drum machines and the only ones that ever bounced up and down so much were ones where the probe was either too high relative to the bean mass, or moving around. It would be quite difficult (and undesirable anyway) to produce a curve that was legitimately so bouncy.

Also, all three curves shown "turned" at about 2:00. That alone is an indication that the probes were too fat (diameter too big) and/or were too influenced by the air relative to the bean temperature. When I've had clients with such laggy probe readings, I've always asked them to get a thinner probe (and/or move its position) and every one of them has had a revelation about their data quality.

All three curves were also generated using batches that were 1/3-1/2 of capacity... such small batches require careful probe selection and placement to read properly, and they rarely provide bean temp data of the quality provided by full batches in the same machine.

Likewise, I've been in the roasteries of some very high-profile US roasters, where they had no idea that ALL of their bean data was massively inaccurate, beyond the usual bit of thermometric lag. For example, one successful, cultish roaster had a 6mm-wide probe in a UG22 that was bent so its tip was 1/2cm from the face plate of the machine and the probe was covered in a 2mm thick layer of creosote. Their first crack temperature reading was off by about 20 degrees F and they didn't know their probe was a problem.

What I'm trying to get at is that you can't assume numbers are good unless you know the procedure and equipment used to generate them. And the data in those competition curves is questionable.

Also, please note that the "commandment" (which was meant light heartedly, as I tried to make clear) to achieve DTR of 20%-25% isn't meant to exist in isolation... for example, one could leave the gas way too low for the first 3-4 minutes of a roast, crank it up very high, create a terrible curve, and still hit 23%. If you read the book carefully, the recommendations are meant to work together as a system, not as a bunch of parts to be cherry picked (or discussed) in isolation.

Scott Rao

Scott, thanks for stopping by and engaging with us!

Any thoughts regarding my point above, regarding the viability of "slow-start, fast-finish" profiles, in which ROR is usually constant over long intervals, rather than declining? I've had a number of roasts made according to this method that tasted very good, sometimes better than a "3 Commandments" approach to the same greens. FWIW, I totally understand your worry that some published profiles may not be accurate, but I am very confident in the reliability and validity of my own probe data, having arrived at the right probe locations through lots of trial and error.

Perhaps this is an approach that works better with hot-air driven methods, like fluid beds and heat-gun roaster designs, than drums? If that is the case, maybe my observations are not inconsistent with your claims, because you did qualify that they were based primarily on experience on classic drum roasters. Regardless, I found it hard to deny the evidence of my senses that some flat-line style profiles produced results that tasted extremely good, even with less development time than you recommend.

Hi Mark,
It's hard for me to say without seeing your data.

I've had dozens of roasters show me coffees from curves with flattened ROR's (from drum roasters). All of them were pretty happy with the roasts, but every single one of them eventually decided the always-declining RORs were better. There are of course many factors beyond the "three commandments" at play, so it's not really possible to do two roasts, one with flat ROR and one with declining ROR while holding "all else equal" like in a proper experiment. There are always other variables that change.

I've only worked on about 10 drum-less machines, so I'm not an expert. But I will share a bit of that experience: One consideration when using an air roaster is that you don't have a hot, massive metal drum to help move things along during the endothermic flash. So in an air roaster you may (I speculate) choose to allow the air temperature to drift higher around fc to compensate for the lack of drum and to help prevent the ROR from crashing. The air temperature in any roaster is ALWAYS influencing the bean probe... so, if your air temp is drifting higher than it would in a comparable roast in a drum roaster, that hotter air is dragging the bean temp reading up. So, the actual (not measured) ROR in the two machines may be identical, but the air roaster's measured ROR may look higher (or flatter, in this case) than the drum roaster's (declining) ROR looks.

Best,
Scott Rao

Thanks, Scott. To get a better example of what I'm talking about, here's a profile curve for a Kenyan I roasted last month that was pretty "flat," and had a much lower development time to total time ratio than you recommend:



After reading your book I tried multiple roasts that followed your guidelines, but none of them tasted as good to me as this profile. It had really snappy, juicy citrus acidity, a nice moderate milk chocolate body, and some pleasing floral notes. All the 20-25%, continuously declining ROR roasts that I did with this coffee cupped a little bit behind it (with blind evaluation), often having more muted acidity and fragrance.

The suggestion that air temp is screwing with my readings doesn't make sense to me, because my probe is buried in the deepest possible part of the bean mass. I moved it after some test roasts made it clear to me that I was getting to much interference from air temperatures in the first place I located it. I can see changes in the input air temp very quickly on my air temp probe, whereas there is a noticeable lag before it has any effect on the readings from the bean mass probe, which I take to be a good sign that I have isolated the two measurements as much as possible. FWIW, I never charge my roaster below the minimum weight of green coffee that will give me reliable readings, which is a lesson I learned the hard way!

As a general rule (and you can see it in the graph), my approach isn't to push in extra heat at FC, but instead to get things moving briskly enough beforehand that the beans will still be moving fast enough once they become more endothermic. In the roast you see pictured above, I had things moving so briskly that I actually dropped the heat quite a bit once FC got going, to avoid getting a finish that would be too dark.

Finally, it's a good point about how hard it is to isolate the different parameters. Apples-to-apples comparisons are hard to find in roasting. Change one thing, you change another. That's probably why it takes so much practice!

Thanks again for stopping by to chat with us. One of the coolest things about this forum is the way it facilitates exchange between pros and hobbyists.

Some of this declining ROR debate has me puzzled.

I'm typically holding the gas and air settings steady from the turn to first crack. When I do so, it's normal to see the ROR steadily decline throughout the roast.

I confess, I have been more focused on target times and temps than ROR. Prior to reading the book, that's all I knew. I always suspected there was more to this, but my greens were turning predictably brown and my coffee was/is occasionally pretty okay.

I am not an accomplished roaster and I make no claim to any roasting technique expertise. I'm reading and recommending Scott's book as an alternative to the blind groping in the dark I have been engaged in for years. Maybe not that much new to most of you, but collected in one handy place it's very illuminating to those of us still trying to figure things out.

Having said that, my question is what's so controversial about the declining ROR? It's just happens for me. Is it that unusual? Am I just lucky? (-these questions make me feel weirdly like I'm writing for Cosmo. LOL)

Depends on your gear, and what you are targeting with your profiles, I think. On a HG/BM roaster with continuously variable heat control, it is pretty easy to manipulate your power levels so that you maintain a constant, non-declining ROR for long periods of the roast. Similar things can be done in fluid-bed roasters. Some folks automate this by using ramp/soak controllers or other electronic wizardry.

I am a neophyte to drum roasting, so I won't say much about the ease with which these flat-line profiles can be executed on more traditional roasters. But I do know that some folks with drums have advocated the slow-start/fast-finish approach, which is more of a flat-line than a smoothly declining curve.

Just to keep it real, about the only time I do a slow start, fast finish roast is when I'm 25 degrees from first crack at 10 minutes. When that happens I increase the gas to ramp up the temp fast enough to salvage the roast.

That could probably be profile all on it's own, but that's not the way I'm roasting now.

As a matter of fact, when I have to recover a roast this way, I usually run to a FC+ and use it in a blend.

I can keep adjusting the heat and keep the ROR constant on my TJ-067. I simply don't want to work that hard.

This is all good info though. Maybe I'll learn a new trick.

Steve