Perfect Ehiopia Natural Roast and My New Favorite All-Around Profile




“I’d like to see quite a lot of expansion without too much forceful rupturing of the seam; in other words, i want the beans to get much bigger, but also not change shape that much. The face of the beans shouldn’t have become too convex for example.”

I think that this is important and something I notice in many Pro roasts.

I have had a lot of roasts though where not all of the beans expand like the majority do (maybe approx 25%)but the roast tastes good which is a bit puzzling given all the beans were under identical conditions.

A lot of Single Origins can be more than one varietal which probably explains some of it. Some are simply processed better than others are I suppose. Naturals tend maybe to be the worst.

I find that quieter 1C’s tend to prolong more than the sharp ones. Sharp ones are maybe more common with more aggressive profiles or catching 1C while still ascending?

Recently I’ve been preferring the prolonged crack type profiles where 1C is on the final segment of the profile as it declines without actually noticing until you posted this.


Et tu Brute? Lol jk, I like 'em all, just no burning or bitterness please. A flavor I haven’t really come across but looking to find is buttery. And I think my favorite accent flavor is honey. What were we talking about?..oh yea, naturals…i like them too. Apologies for a non contributing contribution.


Yea, sharp cracks come from more aggressive profiles I’d say. That can mean a lot of things; basically, energy can be applied or relieved at any point in a roast, and it will always have a different effect than any other point being changed all other points being constant. Start changing all the points around (btw - new point available, am I right? cheers to that.), and the possibilities for manipulating crack become truly endless. Not to mention moisture loss and presumably some other slight physical and chemical changes in the seeds over time as variables in crack control :D.

I’m, at the moment, a fan of slightly ascending finishes on the temperature line, but only in conditions like the profiles above. I feel like it’s different from how i used to use ascending finishes because I use a much longer line at a much (overall) hotter temperature as my finishing line. My present theory about why this improves my coffee has two parts: 1. it reduces RoR variations around first crack, causes a more even and progressive browning and first crack, and due to actually using a slightly lower finishing point, causes less caramelization and/or pyrolysis. 2. I think it reduces the “baked” defect in contrast to my old profiles here which seemed to enhance it. it presumably does this by reducing dips in temperature at any point. My concern with having a descending finish is not the descending finish but the high points you need to have above and before it in order to make it descending. Those higher points, i say speculatively, make the actual bean temp more liable to rise too early making the comparatively lower inlet temperatures later enough to flatten out the RoR a little bit too much. Though i also say speculatively that I like the RoR to get about as low and slow as i can before it does start to bake out the coffee. I dunno.


i think getting around the bake defect is more or less how we manage the moisture in the bean. But something that might help shed some light on the production of flavor is the first post in Fruit for Thought the link for “Investigations on the hot air roasting of coffee beans” by: Schenker, Stefan on page (of the pdf) 139-140 (pgs. 126-127 in the thesis) lists the production of certain odors/flavors that are produced more with the two roasting styles tested.


I’ve been doing some experiments with declining RoR using the Ikawa Shakiso Natural profile as a base. I’ve always liked this profile but been somewhat puzzled by it, specifically as it has 2 distinct sets of increasing and decreasing RoR finishing with a long decline.

I have removed point 1 and moved point 2 forward in time but kept the same temperature. Looks like this

I’ve had some very nice roasts out of this. No baking detected. Why does the dip work? Presumably because it gives energy to the final decline. The initial decline slows it down then we ramp up again. Doesn’t this go against Rao’s theory?

I’ve run experiments removing point 3 thus the dip. Like this

I thought that this shouldn’t work due to baking but it does seem to if you like very light roasts. If you get 1C around 5min and stop around 6min. Or longer if you run a moderate to high flat fan. Again I’ve had some very good to superb light roasts from this, especially Kenyan and Ethiopian washed.

Maybe I’m not good at detecting baking but it seems you need to take it further before this happens.

Finally I’ve been experimenting with a more “normal” gradual decline like this

‘Rocko#nat | 50g’ freshly roasted by @ikawacoffee, here’s the recipe #IKAWAHOME -

I’ve had some superb roasts from Burundi and Congo washed on this. OK on the Ethiopian natural I tried it but not stellar. No baking detected so far, 1C 6 to 6:30 stopping 7:45.

Trialling the same shape but with a lower maximum temperature for some Costa Rican honey process. Not yet tasted.

So to cut a long story I’m still puzzled by this roaster and wish we could get more feedback from it about RoR!

Kenya AA roasting profiles?

Okay, this is something I’d love to talk about more. I’m a huge fan of Scott Rao’s roasting theories - I’ve read the Coffee Roaster’s Companion and all his blog posts many times and even talked to him a little.

It’s pretty depressing for any of us to be talking about RoR at all. I want to desperately, but we really don’t have that information unless we start using extra thermometers. So that’s what I sometimes do! If you wanna try, use a thermometer that goes up to 500 F and just wedge it in the exhaust pipe. It’s hardly the ideal measuring point, but it’s consistent - at least if your thermometer just stands there like mine does. Try to stick the tip of the thermometer beyond the cylindrical metal chimney piece if you do this. if you keep careful track, you’ll notice where it rises more or rises less - so far what Ive seen mirrors what ikawa said: an ascending line causes the bean temperature to rise faster, and a descending one causes it to rise, but more slowly.

Obviously there are limits to the “always rising” rule. I’ve been exploring these limits lately. To put it simply, I think the limit is about 234 C, which is easy to remember! And hey that’s pretty close to 456 F, so that’s my Fahrenheit limit! To put in a more comprehensive way, there’s a range of temperature that’s vital to getting good development and first crack to occur (I’m only talking about city and city+ roasting). the number above is just about in the middle of that range most of the time, and with most profiles I’d consider it a good temp to be near at the start of first crack. The lowest end of this range is the lowest temperature line that would make a bean progress through 1st crack if your roast profile was just that temperature the whole time - probably as low as 420 F/218 C on some beans or 450 F on others. The high limit of the range is around 465 F / 240 C. I use temperatures above 240 C for getting development beyond city + or for putting lots of energy into short light roasts.

RoR is a feature of bean temperature, so even with an extra exhaust thermometer, I only get some clues about it, not hard data. However, there are some principles we can use. According to Rob Hoos, a constant inlet temperature would in theory cause a perfect parabolic arc to form in bean temperature rise over time, but the many unique physical and chemical properties of coffee prohibit this, although it does, I think, come kinda close. According to Rao, one reason not to do this is to stretch out first crack in order to avoid baking. It’s very counterintuitive, but that’s what he says. The reasoning is that as beans crack they release a lot of moisture which, in that superheated environment, is especially capable of evaporative cooling, which quickly becomes a mass effect when a large percentage of the beans crack in rapid succession. Baking, according to him, is mainly caused when this escape of moisture at or around first crack cools the beans and causes a dip in RoR, often followed by a runaway rise in RoR - sometimes called the crash and flick. The dip is something I’m pretty sure I’ve picked up with the extra thermometer, and it was what made me skeptical of my Andino style approach after consistently tasting baked roasts with different Andino style approaches. I also got Scott Rao’s roast defect kit through the Regalia company, so I had baked coffee to compare to.

Nice lookin profiles btw! Looks like you’ve put some thought into them. I’ve often looked for ways to have a declining finish because it should in theory create the gentlest rise at the point where extra temperature is especially likely to break down and volatilize all the good stuff. I may try adding a little decline to the end of some of my profiles now that there’s another point (am I crazy about that?). I too have often liked the shakiso and it is basically my model for most roasts although I tend to lower all the early points to the point where it no longer has the descending finish! I think it some ways it does some of what Rao recommends for sure - it adds a lot of energy early and doesn’t make too many changes to heat settings for a minute on either side of first crack.

Both Rao and Hoos advocate for keeping a relatively low environmental temperature, which equates loosely to inlet temperature, and I think they both say max 500 F, but I could be wrong. I’ve been having good results with trying to keep the temperatures as low as possible. Having worked with a 15 kg drum roaster, I’ve seen first hand how a blast of too-hot air can severely damage the flavor of the coffee and impart all kinds of bitterness and astringency. Where I work, environmental temperatures around 550 F are probably common, boy is it bad.


Thanks for such a detailed post. Certainly helps me to put some flesh on the bones of my experiments.
It’s so easy to forget that we are dealing with inlet temperature and a decline in that doesn’t match the bean temperature.
Ages ago I tried sticking a digital thermapen in the exhaust and noted the same thing and the sudden temp rise on 1C. But it is not easy to do with the thermapen and something sturdier that can support itself obviously would be better and safer.


It was because of Rao’s “constantly declining ROR” and the lack of bean temp readings, that I started with the Marshal Etheo style of profile, I think it was by luck that it actually worked. If I looked at a drum raosters inlet temp the pattern of their inlet/MET coresponds to our inlet. I named the profile “Marshal” becasue of the Roaster who’s profile I was trying to mimick and “Etheo” becasue I misspelled Ethiopian.
I think where you may deviate a little in the understanding but it’s not a big deal. The sudden expulsion of water vapor is going to throw off the momentum of the reaction that is already catylizing, coupled with water which is/was slowing down the coffee materials from breaking down too quickly. If you remove too much water too quickly it’s going to be like cooking with a lazer. That high heat flux is going to create very off flavors. I think the flick is simply caused by the beans reflecting (having trasitioned back from the rubber to glass state) and put more heat back to the probe rather than the beans absorbing the heat when they contained so much water. Regardless of the understanding, mitigating the Crash and Flick is going to produce better coffee. I really look to caramelizing sugar when I think about this reaction taking place in real time…I wonder now what would happen if I cook green coffee in sugar?

I like having my oven thermometer above the beans, and for a $5 solution I highly recommend it. The only draw brack is that it’s analogue and so I have to write down the temp reading every 25*, which gets boring. But still very insightful, regardless that metal is much easier to heat then coffee beans and is made of a single material. I can see major heat changes, like how long it takes for the thermometer to start reading going from a relatively flat ROR to 30 or 40*/min.

I think we also have a lot of work around from not having a bean probe. For one, we can hear the cracks quite easily. There is also the point before 1C (depending on the profile) where you can hear a slight sizziling sound and see some more chaff leave the beans. And then as you mentioned the type of crack heard…poofing or sharp, all at once or spread out. We can “see” this ques by other methods, and then triangulate the approach. A little more basic and hands on (a little hipster IMO) in this new electronic wave of coffee with all the apps and probes.

I don’t know how I feel yet about this extra point. At first it was cool and now I’m like “what am I supposed to do”. Sure for the new Rao style curve I’m trying it makes it a little more fluid (lol), a little more curvy, but we will see if it makes a difference or not. “You’ll see, oh yes, you’ll see…” - Gollum

On the front of it, I want to find a lower temperature. But I don’t think that it necessarily will mean better coffee. My ET probe was reading 25 degrees cooler but that was becasue of colder weather. Without a bean probe, I’m not too sure what to think. I feel like we can mitigate a lot of the crash and flick just becasue we have a roaster that does so by nature. The real answer to the question I’m looking for is: how and when to apply the heat; where are the beans having trouble absorbing heat (evenly removing water) so I need to apply more heat to keep it steadily moving up with a smooth ROR? An IR thermometer I no longer think is good for this, I need a metal probe in the mass. Preferable coming from where that nut is in the chamber.

@stephen.pickering21, when I get in another Burundi or Congo I’ll give that profile a try. Can you share the link?


Sure. Edited original post to place profile under the screenshot.

I’ve lowered that profile by 10 to 8C (slight taper) to trial Central American’s. Early results on the Costa Rica honey geisha are excellent, bang on roasters taste notes.

Nice to get the story behind the name Marshall Etheo!
Android is still not updated, maybe pulled due to firmware fan issues? So not yet had a chance to use the extra point. Instead I’m using a 2 point profile. Going retro.


Do you find the first crack to differ in the temp range much and where do you find first crack usually? Were you able to update yet?


Yes. If you take the simple 2pt profile, 1C has occurred anywhere from 5:05 to 6min depending on the bean. So that’s approx 243 to 239C. Flat fan seems to delay 1C as opposed to a descending fan.

I find African beans take a higher 1C temp better than others. I’m experimenting with lowering these profiles to hit 1C in the 238 to 236C range.

I feel that to some extent this simple approach might be a useful sample roast type profile ( I know- on a sample roaster!) where you can slide it up and down in temp and quite possibly also in time depending on the bean.

I think you were alluding to something like that on another thread re. drum roasters. I’m not sure why every bean on this roaster seems to need a completely unique profile, the way we have been doing. This goes back to the beginning where Ikawa presented us with 6 very different profiles which sowed the seed of thought that everything needs to be unique. I think there ought to be something fairly generic that just needs small adjustments. Maybe that’s just hopeful thinking.

The interesting thing for me is that so far ( think I’ve run more than 10 different beans now) I’ve not had any bad roasts at all from this simple profile. With the caveat that we are talking light roasts here. But I don’t see why deeper roasts could not be more generic either.

The Android update hasn’t yet occurred.


Yea I would say that the closer the region and bean varietal the closer the shape of the profile from one bean to another. The profile I like to use that hits both 1C and 2C well and is listed below. I find that it’s good immediately to a day after but doesn’t do well with rest. And is like my sample profile. I usually get first crack at 526-527F/274-275C and second at 536-538F/280-281C. I drop on the dot of 537F for 2C. I hope you try it and lemme know how it does for you.