Is gas the roasting nirvana?


Brad, your Control settings are similar to what I use, and resulting profiles, too, although I usually drop around 205 for DP, occasionally to 210. I’m gonna check to see if that Sidama Keramo is still available.

I have a v1.5 and the calibrated fan shows about 1074 at F3. I’ve never checked drum speed, and don’t vary it during roasts, at least not advanced to that point, yet.



As someone mentioned, there has been a lot of discussion of how some of the “RAO principles” may or may not apply to the Bullet. And of course he is but one expert of many with his own experience and expertise, but he is arguably one of the most well known and cited amongst the home roasting community. Whether or not you subscribe to his views in a fairly recent blog post he has tried to provide some clarity to some of the more popular ones that get misinterpreted.

One of which is that long roasts create baked flavors, he believes that it is a sharp ROR crash that creates baked flavors. Unless there is something weird with the scaling on your graphs or the way they are presented online, to me all of your roasts look like they crash at first crack. Of course could be my interpretation as well, but perhaps something you may want to rule out. I realizie that you were not necessarily complaining of baked flavors, but the crash could be having an adverse affect none the less.

For reference:


Hi Brad,

Thanks for your post and associated roast profile graphs. When I look at your profiles what I see are very short roasts that have around 2 minutes between the reported onset of FC and the end of the roast. In my own repeated and tested experience on two commercial quality roasters (1lb Probat knockoff sample roaster; Diedrich IR-1), this will not produce a satisfactory result with single origin Ethiopians used for espresso. It might well be great for brewed coffee used in other ways. And of course that is my subjective opinion, but I have tested it in blinded comparison testing with simultaneously prepared and tasted shots, many of them. The length of the interval after onset of FC is in my opinion critical, and ~2 minutes is very insufficient.

Are you using this coffee for straight espresso shots? If used in other brewing methods or in milk drinks, the results I describe may not be apparent.



Thanks very much for this article link, which I will look over later. I don’t know anything about Scott Rao, who certainly could be very knowledgeable about roasting.

I’m kind of an idiot savant type roaster; I only roast Ethiopians, mostly DP, and I only use them for espresso. Not all Ethiopians are suitable for use as single origin espresso so I only buy the ones identified as being suitable. I’ve found Tom at SM/Coffeeshrub to be pretty reliable on that determination, which is not to say that I love love love every Ethiopian he so-identifies.

Most coffees roasted for espresso are obviously blends of one sort or another. Most coffee that is roasted is not used in espresso, but if used in espresso, it is generally camouflaged into very large milk drinks. Now I do make a cappuccino every morning, but I believe that the milk dulls your ability to perceive important differences in the underlying espresso, so I don’t roast for cappuccino use, specifically. If I did I’d probably roast darker than I do.

I tend to discount broad sweeping conclusions about coffee roasting, because it isn’t clear to me that you should roast the same way for different brew methods. Different coffees also vary a lot in terms of how they roast. I can also state that I would put my straight shots up against anyone’s; I’ve tasted a lot of straight shots made from a lot of different coffees in a lot of venues over the years, and I like mine better than any I’ve been served elsewhere. Of course, that could just be puffery.

I did used to send out samples of my roast product periodically to people I know in the coffee community, and no one ever said that they thought my coffees were “baked.” I don’t believe they are “baked” for the intended usage. I would say that I have used my coffees for brewed/pourover coffee on occasion and I don’t much care for them, but then I don’t much care for brewed coffee in the first place. I’m not ready to buy into the notion that extending out the duration of FC to ~4-5 minutes produces coffee that is baked or has baked flavors when used for straight espresso shots, certainly not with the Ethiopians I roast.

Getting back to my original point – and this might be germane to the discussion, is that I believe that I am slowing down the roasts on the bullet more rapidly than is even possible on my Diedrich; again, the “piloting an oil tanker” vs a speedboat analogy might be apt. Even though I do cut the heat dramatically and increase the ventilation from ~50% through the drum to 80% on that roaster (Diedrich) at onset of FC, that roaster doesn’t respond as quickly as the Bullet does. So, what I might be doing is taking a nimble roaster like the Bullet and acting like I’m using a roaster that doesn’t respond quickly, so perhaps that’s the issue.

Which gets back to the “sharp ROR crash” point you make in your post. The total time I spend roasting the coffee after onset FC may be the same on the Bullet as on the Diedrich, but on the Bullet I may be making the change from rapid rise to tapered rise more abruptly and that may be an/the issue. I need to consider this the next time I roast and see how that comes out.


Agreed Ken, I think the agility of the Bullet to pull up short is at work here. Applying the front handbrake on a speeding motorbike is going to throw you over the handlebars, to drag in another analogy :wink:


Sorry to be monopolizing this thread, which I have intruded on in the middle . . . .

Just a note on why I am so adamant on the concept of having a long enough interval between onset FC and the end of the roast, specifically if one is roasting for straight shot espresso use.

My original “commercial” roaster was 500g/1lb sample roaster which was a knockoff of an old Probat sample roaster. It was purchased from which later became US Roasters Corp (maybe because the old name looked more like a gender reassignment office than a coffee roaster manufacturer). I had to modify by replacing the burner which was inadequate, and hacking in a thermocouple into the middle of the drum. Once I got this more or less perfected, I roasted hundreds of pounds of coffee annually on it, ultimately settling on single origin espresso and later concentrating on Ethiopians. The roast product coming out of this roaster was used in a bunch of blinded comparison tasting tests which were used to evaluate such things as the impact of pump type (rotary vs. vibratory) on espresso shots, the effects of freezing coffee, and even on roast profiles, the issue of roast profiling especially the duration of the roast after onset FC. There were a number of people who tasted the shots, generally in a blinded fashion.

As I stated earlier, one thing we tested blindly was the effect of the duration of FC after its onset, and we found that this was an easily discernible difference in that coffee roasted with only 2 or 2.5 minutes between onset of FC to end of the roast were grossly inferior for straight shots when compared to coffee roasted with a longer interval from onset FC to end of roast. All of this was documented contemporaneously on

Fast forward to when I bought my Diedrich roaster in 2011. I spent a couple of hours at the Diedrich factory with Stephen Diedrich, who demonstrated my new roaster to me, and told me how he roasts coffee and the profiles he uses. His recommended profile is a slow ramp up to onset FC with a rapid finish, in about 2 minutes, which is at least somewhat similar to some profiles posted in this thread.

I tried this profile repeatedly on the Diedrich with the coffees I was using at the time on my old Probat knockoff sample roaster. I found the results to be horrible for straight shot espresso usage, very grossly inferior to what I had been obtaining with my far less sophisticated sample roaster. I tore my hair out over this, played around with the profiles and still couldn’t get satisfactory results for espresso straight shots.

At this point I decided to try to replicate my earlier results from the sample roaster onto the Diedrich. I was able to replicate the same profiles with some experimentation, and voila, the results were again excellent. So what I learned from this is that the profile is more important than the roaster itself, assuming decent roasters, of course.

My error in transposing this profile, with which I have 15 years+ experience, onto the Bullet, is perhaps that I am slowing down the roast on the Bullet too quickly right after onset of FC; that hasn’t been something I could do if I wanted to do it on my other roasters, but with the Bullet it is easily done. So perhaps that’s the error in what I’ve been doing with the Bullet.

I would suggest to those of you who do enjoy straight espresso shots and who roast for that usage, to consider a little experimentation on your own – forget about Scott Rao and any other gurus out there, just do this little experiment with your own roaster and your own taste buds. Slow down the roast after onset FC; maybe not as quickly as I have done on the graphs I posted, but slow it down slow enough that you can extend FC out to last at least 3.5 minutes, preferably 4 minutes, but not more than 5 minutes. Hit FC before the 10 minute mark (9.5 minutes would be better), and keep your total roast time below 15 minutes (preferably 14 minutes).

Make some straight shots after you’ve rested the coffee 2 or 3 days and continue to taste the coffee over a period of 7-10 days post roast date. Compare these shots to what you have been producing with your faster profiles that don’t give you that “magic” 3.5 to 4 minute interval after onset of FC.

I’d be willing to bet that most of you, maybe all of you, will see what I’m talking about here and will see the improvement in your roast product. The difference is NOT subtle, it’s enormous, and the coffee will not taste baked to you.


Hey Ken,

It’s all good, there are a few Rebel Alliance folks here who question The Curve. As I said, I have no beef with you wanting to extend the time from FC to drop to however long you wish so that your coffee tastes good to you for your extraction method. There may well be a different approach needed for single origin Ethiopian in espresso. As you noted, many espresso beans are in fact blends. And part of the “rebel cause” is tail wagging the dog - does a mathematically perfect curve dictate greatest coffee, which seems to be a quest that new (Bullet) roasters aspire to. I’m not sure it does.

I think as you say, the takeaway here is that the reduced thermal mass of the Bullet can lead you to come to a screeching halt and lose the momentum you built up with your slow ramp up to FC.

I have had drops that were 3:00 or 3:30 after FC started and the coffee was great to my tastes. I don’t think I have ventured out as far as 4:00 but then I also don’t tend to need/want a darker roast that may be required to cut through milk based espresso drinks vs straight espresso.


Hi Stuart,

I don’t roast into second crack; I always end the roast either at the end of FC or just before that. I don’t consider that to be a “darker” roast. My double shots are made from about 15g of coffee by weight, and the extracted espresso shots weigh about 30g.

That’s not an extraction designed to put into a ton of cow juice.

If one were intending to roast into 2nd crack, then getting a longer interval after onset FC would be much easier. Since I don’t ever roast that darkly (at least not intentionally), I have no idea what impact that would have on the taste.



It’s day 2 since I roasted the 3 coffees with the profiles I posted, so I thought I would try them.

The first one was Ethiopian DP Humbela Buku, which turned out darker than I would have liked and with an 18+% weight loss from roasting. Although more roasted than is my preference, it actually tasted pretty good:

Second was a WP Ethiopian Sidamo, not otherwise specified, given to me by a coffee roaster owner I know. The roast level was within my target range and the weight loss was about 15.2% Although not as complex as most of the Ethiopians I buy and roast, it was actually pretty good today.

Third on the list was a Guatemalan, also supplied by the roaster acquaintance. It was under roasted and the weight loss from roasting was only about 13%. This was one of the worst coffees I have ever tasted, worse than airplane coffee, and I spit it out into the sink without swallowing any of it.

As pointed out earlier in this thread, all of my three posted profiles resemble each other in terms of the slope of the curve right after onset of FC. It’s possible that the Guatemalan was “baked,” hard to tell with something that tastes that bad. As for the other two, they aren’t baked, and they don’t taste flat, again, as used for straight shots of espresso.


I fairly regularly roast a Red Sea blend (Ethiopian and Yemen) and it invariably needs to lie undisturbed after roasting for a few days before showing its true colours. In fact most of my Ethiopians I leave for at least three or four days to get their act together.

Maybe you just don’t care for Guatemalans as I have friends who can’t abide Ethiopian coffee :wink:

It has been an educational discussion and I thank all for their participation :slight_smile:


Like you, I don’t generally consume coffee until it has reached the 3 day point, however I find that I can taste it objectively at 2 days.

I don’t know anything about the Guatemalan since it was given to me as a gift. I think it was so under roasted as to be unpalatable, especially in an espresso shot.



Since I am only at the house with the Bullet part-time, my ability to roast on it is limited. I did do a few roasts on the Bullet last week, which I will post here:

Dry Processed Ethiopian

Wet Processed Ethiopian Sidamo

In both of these roasts I was able to reduce the speed of the ROR drop off from what I posted earlier, while still extending the roast out 4 minutes+ into first crack, without actually entering 2nd crack, which was my goal and the roasting profile I’ve been using successfully on 2 different gas roasters over the last 10+ years.

The results were pretty good, however I’m not terribly famiiar with either of the coffees and haven’t roasted the WP coffee ever on my Diedrich, and have only roasted one batch of the DP coffee on the Diedrich, which was months ago.

Sorry, no answer to the bigger question posed by this thread, as to whether “gas roasting” is better than electric, but at least I seem to be beyond the disappointing results I was having some of the time with the Bullet. Based on the limited number of roasts done, and the fact that I was the only taster, I would be very reluctant to ascribe any of this to avoiding an early “ROR crash” or much anything else.


I finally had the chance to do a roasting session on my Diedrich IR-1 yesterday, after a 3 month pause, mostly to be out of my main residence during the height of the coronavirus pandemic (it having been located in what was a “hot spot”). One of my justifications for buying the Bullet for my “business house” was that I was going to be stuck there for a while during the epidemic. I will also use this post as an attempt to get back to the original topic of this thread.

Yesterday I roasted 2, 2lb batches, of each of 3 different DP Ethiopians, for espresso, on the Diedrich, for a total of 6 back to back roasts, all taken to around 419 F bean temp on the Diedrich, which with that roaster at my altitude of about 5850 ft means the coffee makes it through 1st crack almost to the end but doesn’t enter 2nd crack. First crack was generally started around 375 F, after about 9 minutes 30 seconds, and then the roast was carried on for 4 to 4 mins 30 seconds further after which the beans were dumped.

As is my practice for many years, once I get within 10 degrees F of initiating FC, I reduce the gas heat input, and once FC actually starts, I further reduce the gas input to a great extent, plus change the ventilation to 80% through the drum (maximum you can get with the damper in this roaster). Once the rate of rise (ROR) slows precipitously, I gingerly increase the gas heat input and reduce the ventilation through the drum also, the latter only if needed. I paid particular attention to the bean temps over the next ~4 minutes of each roast, until the beans were dumped.

Unfortunately, I do not have any way to record the bean temps in real time or to put them on a graph, as the thermocouple in the drum goes to a digital read out but there is no way to interface with it as the roaster does not have an “electronics package.” Nonetheless, it is pretty easy to describe what happens as it is consistent and repeatable across all of the roasts I did with the 3 different coffees yesterday.

Once FC starts on this roaster, the bean temperature rises very rapidly as the roast becomes self-supporting, or “exothermic.” Even with the gas heat turned down to a minimum (in my case this is going down from a maximum of around 6.75 WC (Water Column Inches) to 2.0 WC, e.g. reducing the gas input by more than 2/3, plus increasing the drum ventilation from 50% to 80%, the beans very rapidly go from ~375F to 395F in the space of about 1.5 minutes. This is at least as fast a ROR as the beans are going through in the minute or two before the FC starts, although perhaps a bit slower than the ROR earlier on in the roast. Only after about 1.5 minutes of greatly reducing heat input and increasing drum ventilation, does the ROR slow considerably, and in some cases it slows down A LOT.

As I have mentioned earlier on in my posts on this thread, I intentionally slow my roasts down after onset FC to draw them out to 4-5 minutes after onset FC, which I have found works well for single origin Ethiopians used for espresso. I had some roasts with poor results on the Bullet attempting this profile, however it is now fairly obvious to me that what happened on the Bullet is not what happens on the Diedrich with the strategy I employ.

On the Bullet, whose drum doesn’t have a great deal of “thermal mass,” when you cut the heat input and you increase the fan speed, you get an almost immediate response in the “rate of rise” ROR, whereas in the Diedrich IR-1, this is delayed by about 1.5 minutes to my observation. So, IF the “ROR crash” as quoted here from Scott Rao is in fact a real detriment to the quality of some roasts, it also probably matters WHEN that ROR crash occurs. From my own informal observations on the Bullet, it seems that if the ROR crash occurs right after onset of FC, this is not good for roast quality. IF the ROR crash occurs later, in this case delayed by about 1.5 minutes, then it does not have the same effect.

Finally, bringing this back to the original topic of this thread started by Stuart: I would not compare roasting on the Bullet to roasting on a high-thermal-mass gas roaster. You may very well be able to obtain similar results, but the experience is totally different. The gas roaster is presumably designed to roast in a certain way, which the thermal mass of the drum dictates. It’s kind of like driving a Sherman Tank through a blizzard; you don’t need to overthink things, the tank is going to keep on going its way and nothing on the ground is going to stop it. The Bullet, on the other hand, is nimble and more like a sports car with summer tires on it, you can make it do what you want it to do, but that is going to take a lot of attention on your part. A commercial gas roaster is a very solid piece of equipment; the Bullet, in comparison, feels rather flimsy. One advantage of the Bullet is that at least in theory you could find a good profile and then just execute the whole profile automatically through the software. There are certainly big gas commercial roasters that can do that also, but you are talking an enormous amount of money for a gas roaster that can do that.

On the whole, I much prefer the Diedrich gas roaster to the Bullet, however I am more used to it, and I can’t say that with experience the Bullet isn’t capable of getting very similar results.