I've been home roasting now for more than 6 years, with almost 5 years experience using variations of my present set up, a 1lb commercial sample roaster:



While I've made steady progress in the quality of my roast product, only in the last few months have I "put it all together" to the point where I'm genuinely happy with my results and seldom disappointed. To the extent that anyone can learn anything about roasting from an online post, I'm hoping that some people may find the following information useful in their own home roasting odyssey. Unlike most of my posts, I'm going to try to keep this brief and will list what I think are the most important points in order of their importance. These points if ignored will probably prevent you from producing a roast rivaling what you might buy from a good roaster, other than by chance.

(1) Use the best green coffee that you can afford for the intended use of the coffee. Not all beans will be good for all uses, but you cannot have excellent results with mediocre beans.

(2) Accurate and reproducible thermometry is absolutely essential to good and repeatable results. The thermometry probe must be located in the same place every time, or your measurements will mean almost nothing. And, you can't just look at the probe readings, you have to act on them!

(3) Final roast temperature is important, however the way that you get to the target temperature is just as important. 440F reached in 10 minutes will not be the same as 440F reached in 15. Some portions of the roast cycle will have more importance than others, and it pays to learn which are which, and you can only learn these by experience and by tasting your results!

(4) The speed of temperature increase that yields the best results will probably NOT be a constant uphill slope at the same velocity. There are at least two different portions of the roast that deserve different treatment. In my roaster, my experience is that the impact of what I do before first crack has markedly different results than what I do after first crack starts and before the roast is terminated.

(5) The natural tendency is to want to have things "happen," e.g. to experience first and second crack, but your best results might come from terminating the roast at a lighter roast level, e.g. before the onset of 2nd crack. In order to do this, you are going to have to learn the thermometry of your roaster past the onset of 2nd crack, in order to anticipate it and to end the roast before it starts.

(6) You must approach roasting as though it is a necessary skill that must be mastered in order to become a "complete coffee person." There is nothing that I do with coffee that has as much of an impact on the drinks I make, as the quality of the raw material I use, e.g. the roast products. Everything else, including barista skills, espresso machine temperature stability, grind setting, you-name-it, pales in comparison to what you produce as a roaster. If you approach roasting as a PITA (which I used to do) you will get only so far as a roaster and no further. If you constantly evaluate what you are doing by the criteria of your results, you can continually progress. In the end, equipment and beans permitting, you should get results that are good enough to rival a good commercial roaster, but these sorts of results do not appear out of thin air, they require a lot of effort on your part.

(7) Get in the habit of weighing your roast product, and comparing the before and after weights of the green and roasted coffee. You can learn a lot from this, and it will help to keep your roasting style more consistent. If you have a batch where there is a big difference in the weight loss, this points to the likelihood that something has changed -- either the beans themselves or something that you did.

As an example of number 6, above, I performed a little experiment with my roasting before Jim Schulman arrived for his most recent visit this past summer. I took several coffees and roasted them all in 2 different ways, to the same final temperature but with either 2.5 minutes or 4 minutes between the onset of first crack and the end of the roasts (e.g. the speed of temperature increase after first crack was altered). I hadn't always been consistent in the length of this "intercrack" period and wondered what impact it had. The results were extremely obvious; the 2.5 minute interval produced dull coffee that was usable only if disguised in milk. The 4.0 minute interval produced some of the best results I've ever had. We compared simultaneously made shots from these two types of roasts (same beans, other parameters constant) in a blind tasting fashion and Jim and I both picked the slower (4 minute vs. 2.5 minute interval) coffees 100% of the time. It was that dramatic, but I would never have learned that without testing it.

Here is a description of the roasting parameters that work best in my equipment and may or may not work well in yours. Even if my parameters don't work for you, knowing that these parameters exist may help you to experiment a bit with altering them in order to find what works best for you. I have a drum and therefore if you air roast you will need to extrapolate to your particular set up, as the processes are different even if the results can be similar.

(1) Drum is loaded when the air drum temperature is around 350F

(2) Precisely weighed bean charge load of 454g is used every time. You will need to determine what is the best charge weight for your particular roaster. With my roaster, using 500g rather than 454g (1lb) produces distinctly inferior results, presumably due to unevenness of the roast.

(3) The very beginning of first crack is reached at 8 minutes, never much less and never more than 9 minutes. The speed of temperature rise is not terribly important but needs to be relatively constant and adjusted during the process in order to meet this timing.

(4) The heat needs to be reduced as 1st crack approaches and is reduced further after the onset of first. There is generally a period shortly after the onset of first crack where the temperature does not go up much or may even remain constant (BUT DOES NOT DROP) for a period of up to 1.5 or possibly two minutes.

(5) I roast almost to but never past second crack (your taste may differ), and I try to keep the interval between onset of 1st and what would be onset of 2nd at around 4 minutes, never more and never less than 3.5 minutes.

This roasting paradigm has yielded the best results I have ever had for use in espresso (but also for drip, which is a secondary usage). The results appear to be highly repeatable. The weight loss percentage that I find with these roast parameters is around 16%.

You cannot have good and repeatable results in the absence of good thermometry. Roasting is not a spectator sport, it demands your constant attention if you are going to get the best possible results.

Hope this is useful to at least a few people.

ken

Matthew Brinski wrote:Ken, thanks for putting this up. You're the man ... even if you don't like the GS3.

Where'd you ever get that idea?

If I didn't already own 2 commercial home machines, and were in the market for an espresso machine right now, the GS3 would be very high on my short list. The only negative is that it is a bit new, and I prefer more mature technology, as I have a fairly limited toleration for problems with equipment.

ken

Matthew Brinski wrote:It was a comment made in jest since you never jumped on the "must have GS3" bandwagon (which I do understand your views from following your previous posts, and I also understand that you do regard it as a quality machine).

Seriously though, the information you and Schulman put up is great. Why don't you guys just write a book already?

Matthew,

You are extremely kind.

I think your comment, while true for Jim, is a bit much as applied to me. I do a fairly good job of testing other peoples' ideas, but unlike Jim I seldom come up with an original one.

ken

Fullsack wrote:<image>


To take this a step futher, pick the defects out before roasting. Insect damage, fungus damage and broken or chipped beans can give your roast a dirty, sour or moldy flavor. These little varmits were plucked from only 225 grams of a blend supplied by a well regarded purveyor of green beans for home roasters.

Advantage home roasters: we have the time to do this kind of thing.

Doug,

Those are some ugly looking beans! I've had some beans with a lot of defects, but the beans you photographed are worse than anything I can recall roasting, certainly within the last few years. Just on sight alone, and without regard to who was selling them, I'd make a point of not reordering any coffee that looked like THAT!

ken

mrgnomer wrote:
I had an iRoast2 for about 3 years before the Hottop. The Hottop I've had for about a year now. I'm noticing there's a stage between about 370F to 390F depending on how steep the heating ramp is to that temp where there's a sweet smell that comes out.

I'd really like to preserve that sweet smell in a roast to bring it out in a good espresso extraction. My rest from 1st to 2nd has been very short; usually under 2 min. From your points it looks like backing off on the heat at the start of 1st crack should bring the chamber temperature down to extend the rest without stalling the roast in 1st crack. You think I could get sweetness to come out more with a longer rest to 2nd?

One man's "sweetness" is another man's (fill in the blank). I find it a little hard to extrapolate what I might smell during a roast with what I'll taste brewed from it later, even if Schomer has a cliché out there about wanting espresso to taste like coffee smells.

My experience is (at least with my roaster) that if the period between onset of 1st crack to beginning of 2nd crack is short, like 2.5 minutes, that the flavors become multidimensional and "flat." Almost anything that tastes really good will have a whole lot of different flavors that coalesce to make a very nice, harmonious, impression. This is true of any great wine (where this characteristic is called "complexity,") and of most any good dish in a fine restaurant (or even the local Thai place). When this complexity is lacking, things taste flat, and that is what I get from coffee roasted with too short an interval between onset 1st and onset 2nd.

In my post I avoided using absolute temperature numbers since these will differ depending on what type of roaster one uses and where the probe is situated. My temperature probe seems to be located in an especially hot part of the drum and reads about 14 degrees F higher than it should judging by bean appearance at any given point in the roast. Because this reading is constantly high, it is easy to correct for it and to become used to the numbers which are not correct in an "absolute" sense but remain quite useful for controlling a roast.

The key point is to try to use thermometry if at all possible and to learn, by experience, what the numbers mean in the context of one's specific roaster. Once you have this, you can tinker around with your roast profile and then taste your results. If the results are good you can then easily repeat the roast parameters later because you will have real time data you can then impact by adjusting the heat source so that you get the same profile again at a later time.

ken


ken

Abdon wrote:I think (therefore I could be horribly wrong) that messing with the temperature before a good roaring 1st crack is dicey at best. All through half-way 1st crack, beans are for the most part soaking up heat; drop the heat and they will cool off rapidly. Depending on your roasting method, keeping up a target temperature around this point in time is tricky and opens the door to inconsistency. At a point into 1st crack beans expand and begin to radiate heat. Slowing down the roasting on this stage is fairly easy because the beans will be working with you, not against you.

This is an example of something that is going to vary depending upon the sort of roaster one uses. There are roasters that are "underpowered," which struggle to reach first crack before 10, 12, or even 15 minutes. My roaster was like that when I first received it, as the supplier (in spite of my making this issue crystal clear before purchase) failed to put a burner in the roaster that could supply enough heat at altitude. I live at almost 6000 feet/~1850 meters, and there are serious combustion issues when one lives appreciably above sea level. As originally configured, my roaster could hit first crack at 9 minute only if I had the heat on full bore and it was a warm summer day with ambient temps above 70F. Since I have to roast in cooler temperatures than that, and since I want to have control over the process, I was successful (with Barry Jarrett's help) in finding a replacement burner that more than doubled the available heat.

Roasting with my sample roaster is a totally manual process; I decide how much heat to supply so there is no issue with "messing with the temperature," rather I have to decide how much heat is applied for each and every second of my roasts, both before and after first crack. Each and every part of the roast profile can be controlled if you have control of the heat input (and/ or the airflow).

In a situation where one does not have control over this factor, either due to an inadequate heat source or lack of manual control, then this is an area that will demand compromises.

Abdon wrote:If you take first and second crack as milestones in the roasting process, that point after 1st where the beans suddenly begin to gain heat at a faster pace is another milestone that is just as important. This is where is the easiest to consistently slow down your roast between first and second.

My experience in a totally manual setup is different. My impression is that the beans rapidly gain heat because the heat source has finally gotten cranked up to full power, and it is easy to have a bit too much heat being introduced into the process and in that way to "get ahead of one's self." With my drum this happens because I've had to crank up the heat to hit first crack by 8 or 8.5 minutes (my typical roast parameter), and in doing so I am applying quite a bit of heat just beforehand, which translates into retained heat in the massive drum. Air roasters will act differently and can be adjusted more rapidly. I deal with this by anticipating what is going to happen next, e.g. I back off the heat in the minute or so before the onset of 1st so that this doesn't happen.

There IS a point in the roast where the beans become "exothermic," i.e. one can turn off the heat source and the beans will keep rising in temperature without additional added heat. With my roaster this occurs during the last degree or two or maybe 3 (F) before I end the roast, and it might actually be more that the drum is continuing to transfer heat to the beans, heat that was added beforehand. With an air roaster, if one left the blower on but killed the heat, you would probably not observe this behavior because then the blower would start to actually cool off the beans and work against this process.

Abdon wrote:For me there are two kinds of coffees, those that like to ease their way to the finish line, and those that improve by a slower roast with a sharp up turn at the end. They may both end up at the same roast level and even the same time, but will taste different.

I haven't observed this but then we might well roast different coffees, and to different roast levels.

ken

Abdon wrote:I have found (or have myself deluded into believing to have found) different nuances on coffees that can be roasted very dark. The most recent coffee I have been roasting like this is a Ethiopian Harrar lot #14-something from Sweet Marias. One way, there are roasted caramelly flavors with a hint of tobacco. The other less caramel with very strong dry tobacco flavors. You have a more consistent setup; could you run two batches, one with a steady march towards the end, the other with a slower gain, ramped up to speed in the last 30 seconds or so?

Roasting on my roaster is somewhat akin to maneuvering an oil tanker; it responds much more predictably and well to moves made in advance of when you want them to happen. I already go very slowly from the onset of first crack to when I terminate my roasts, generally a few seconds before 2nd would start. Since I'm striving for 3.5 to 4 minutes after the onset of 1st to the end of the roast, I'm not at all sure that I could even test your second profile without stalling out the roast. This may be an unchangeable behavior of drum roasters which have significant thermal mass in comparison to the bean charge. It is easy with my setup to go faster than I go now during this time period, but relatively difficult to go slowly.

ken

BradS wrote:Excellent subject and content, Ken & others. I have been playing with the very issue(s) you mention concerning time between 1st & 2nd, and in addition; cooling time. Both had been relatively short in my roasts, as I was usually getting about 60-90 seconds between 1st & 2nd, and, in my ignorance, originally tried to cool the beans (and remove the chaff) asap at the end of roasting. I'm hoping you'll take the time to critique the roast profile shown in the information below. This was roasted Saturday and I sampled the first shot this morning which seemed to have a lot more mellowness & personality than my previous roasts (this was also the first roast with a variac which I used to slow down the sprint from 1st to 2nd).

One aspect of my roaster did not make it into my original post; e.g., cooling. The cooling tray on my roaster is if anything over-ventilated, and as a result, the beans cool down to room temperature very very fast. I typically leave the beans on the cooling tray for 2 minutes, by which time their temperature is no more than 10 or 20F degrees above (and often equal to) room temperature. They then go into a wire mesh basket and are taken outside for vigorous dechaffing by a vertical floor fan. The result is that they are COOL less than 3 minutes after the roast ends. My personal opinion is that rapid cooling is a generally good thing, as it allows you to accurately choose the roast level without the roast level advancing when the beans are supposed to be cooling down.

BradS wrote:
The shallower decreasing slope at the end of the roast was where I turned off the heat and left the fan running with the variac back at full (130VAC) to see if cooling could be done in situ with any degree of success. You can see - I got impatient and opened the roaster and dumped the beans! There is some more information on the roasting label below, mainly that I decreased the voltage from 110V to 105V to slow down the roast from 1st to 2nd... with fairly good results. I also seem to have a higher-than-normal offset on my temp reading as my temps seem about 40 degrees high compared to visual/audible cues.

It is hard for me to give any useful feedback regarding a roast I have not tasted, and being entirely honest there are a number of other HB regulars who probably could give you better tasting feedback on a roast than I could. I'm not really knowledgeable about SC roast parameters, and hopefully someone else can give you some meaningful feedback.

ken