another_jim wrote:For roast profiles [...] there really isn't that much to it: keep your roasts around 10 to 15 minutes, running roughly equal segments from start to 300F (where the beans start to change color from greeenish to yellow-brown), 300F to first crack, and first crack to end of roast. You can do this on almost any roaster that gives you some control over the heat, without a lot of instrumentation, providing you are willing to roast in small batches and watch carefully.
...
I do about three minutes [at the end of the roast] for brewing roasts to let the caramels develop; for cupping I go 30 seconds to a minute faster. I'm at the point where I don't pay much attention to the time; but I'd have wasted about three years less time if I'd stuck to timing in this range when I was starting.

Quoted from this thread as this is deserving of its own. I've seen this topic rear its head repeatedly, as I'm sure you all have, however I haven't seen such a definitive answer offered from an authority before.

Actually, I did see this once before somewhere (sorry for lack of attribution, I don't recall where I found this image originally), but had assumed this advice was specific to espresso roasts, and possibly even to air roasters:




The first two of three phases make sense to me, and seems to be widely believed as The Way. However the final third stage, in particular for lightly roasted coffees, is still confusing to me. Obviously the above image does *not* apply to brewing roasts, as that profile would take the beans to 40F+ beyond the end of first crack if left to continue at that rate-of-rise (RoR) for 2+ minutes.

Can you clarify what the RoR looks like for that final third phase for brewed coffee? And are you able to maintain that RoR while also preventing a drop in environmental temperature, or is that not a concern?

TIA.

I was told that Willem Boot used to tell people that a light roast should take longer than a dark roast. I find that fast finishes for very light roasts are not a good idea, and a bit of stalling as the first winds down is called for.

My older profiles. like the one shown, are for espresso and for air roasters, where I used to go 4 1/2 minutes from the onset of the first crack to the end of the roast. I fiond these profiles on the flat side and ogo a bot faster. Since these were designed for use on Fuji ramp/soak controllers, they are based on the precision of those devices.

I'm not nearly as precise now with the M3 -- I don't have it PIDed, I've removed the TCs, and I dislike doing a lot of manual control. So for back to back roasts,
-- I drop in the next batch while cooling the previous one, with no heat or air, and leaving the little door where you drop in the beans open. (For the first roast I do the same, just preheat it to 200C, and the drum to 250C).
-- When the beans reach 155C (305F), I max the air and heat.
-- About 3 to 4 minutes later, the drun reaches 260C - 280C, 480F to 510F, and the beans get close to the first crack, to around 190C, and I back off the heat and air to the end of the roast, with a later back off and higher drum temps for high grown beans, and an earlier back off and lower drum temps for low growns.

The roasts taste better this way, since the basic profile is sound, and since paying close attention to all the sensory aspects of the roast to end it improves quality much more than tweaking a profile that's already in the ballpark.

If you have a PID, you have to pick (for instance) 425F for a segment marker, rather than 424F or 426F. But these variations are too small to be meaningful, so it's best to use roughly right, round numbers. Getting more precise will not improve the taste. Asking yourself if the beans look right and smell right, and ending the roast if they do, regardless of what the clock or thermometer says, will eventually get you better tasting coffee. You can train your eyes and nose to tell when the beans are right, you cannot train a timer or thermometer to do that.

In a cupping, the ground samples of each coffee have to be indistinguishable in roast color. A profile controlling PID on a sample roaster, roasting each bean to the same profile, will not accomplish this on the first go round, since the final roast color depends not just on the profile, but on the bean itself. In other words, whether you roast manually or automatically, duplicating roast colors will require several tries and a visual inspection of the beans. So the PID doing precisely duplicated roast profiles is just a convenience, not a fool proof way of either standardizing or optimizing roasts.

+1 on the PID comment. For how I roast, manual control still makes more sense (as much as I'd love an autopilot sometimes.)

Out of curiosity, whats your typical batch size? IIRC, the M3 has about 30% more wattage than the Hottop, but even so I suspect you're not able to do more than 7oz or so and follow such a profile.

I just roasted some Rwanda Gikongoro Nyarusiza. 6oz in a Hottop with extra fins in the drum, but otherwise stock. The idea was to roast for ~3 min from beginning of first crack to eject, aiming for a City+ roast. So I'd have 6 minutes to get to the onset of first crack. Looks like I got really lucky with a guess at a drop-in temp of 340F at the bean probe to get me there in almost exactly six minutes, but the drying phase was way overshot as a result. And the ET swung negative 10F towards the end to hold that 3 min and not overshoot the degree of roast. Beans looks nice and even, inside and out, but I'll be curious how it tastes tomorrow...

Is it really that important to finish drying or reach first crack with certain short time frame?

The dry-processed beans can use shorter drying time, but not always. The wet-processed coffee can take longer. This makes some intuitive sense, not in the absolute moisture content, but the shape of the beans. Dry processed beans generally have a wider central crack; wet-processed beans have tighter central crack. The heat can penetrate the beans more easily with a wider central crack.

The bean density and size also contribute to the differences. The typical porous Brazil or Kona requires so little heat, as opposed to Kenya AA. High charge temperature for Brazil or Kona will inevitably tip the beans.

The SCAA cupping roast standard may be useful to detect defect, but at such quick and light roast levels, the coffee really is not enjoyable.

chang00 wrote:High charge temperature for Brazil or Kona will inevitably tip the beans.

Even if you have a very gentle rate of rise- say a 15min+ roast? I would have guessed the moisture in the beans would buffer the high temps through the end of the drying phase.

chang00 wrote: The typical porous Brazil or Kona requires so little heat, as opposed to Kenya AA. High charge temperature for Brazil or Kona will inevitably tip the beans.

I don't think this is quite accurate. Tipping occurs after the beans dry, and is based on too high drum temperatures (above 475F to 500F for low density beans on most drums), rather than air temperatures. I charge the drum as soon as I empty the previous roast, so the air is at 425F to 450F, and the drum is anywhere from 475F to 525F. But the heat is turned off, and all this residual heat is absorbed by the evaporating water which comes steaming out of the drop in port. No tipping occurs. It even occured to me I could do this on the first roast just as easily as follow up ones, so now I use the same "end of roast" drop in for the sessions first roast too.

It's the same idea as in bread baking, where you start with a very hot oven while the dough is still moist, and drop the the temperature as the bread browns

Later in the roast, as I get to the first crack, I have to make sure I max out the drum temperautre at an appropriate level. Too high and the low grown beans tip; too low and woody roast flavors out balance the caramel and chocolate ones. I push 490 to 500F even with Brazils, and get up to 525F for Centrals and East Africans; although a cool drum, "american oak" roast does work great for some coffees.

I probably did not make my observation clear. Generally tipping is caused by too high of conduction heat, in my case the hot drum. The ET can correlate with the drum temperature, if the damper is at the same position.

Because the drum is a rotating heavy piece of metal, it is rather difficult to measure the temperature with a thermocouple. So I have been using an IR Fluke thermometer to just measure the drum temperature, to avoid tipping, regardless of ET. Jim, like you have with the M3, I turn off the gas after charging the beans, so the beans absorb heat from the drum until turning point, to avoid tipping.

In my unmodified HotTop, the drums temperature cannot be measured reliably, and the heating element cannot be turned off after charging, so I only use the build in ET readout to estimate drum temperature.

Guess similar to grilling, if the grill is hot, I get a nice grill mark; the moisture does not carry the heat away rapidly enough. If the grill is cold, the food will still be cooked, but tastes more steamed.

Yeah, that's right. I wanted to stress that unlike air roasters, on drums the ET and drum temp can get too low as well as too high.

When I got the M3, I automatically tried to keep the drum and ET as low as possible, and the airflow as high as possible, like in an air or convection roaster. This helped initially, but by overdoing it, I lost some of the complex roast flavors that marks the strength of drum roasters. You want enough heat to get some dark caramels, chocolates and distillates, while still keeping the roast light and fast enough to keep the acidity and fruit.

another_jim wrote:I lost some of the complex roast flavors that marks the strength of drum roasters.

I took some pics of the inside of some beans from last weekend. To my surprise, some interior areas looked darker than the exterior surface. It was as if they were a little "burnt" in the inside (but no burnt flavor). And I think a little burn (darker) can give that complex flavor that differentiates drum from air roasters. These roast profiles were in the 15+ minutes (drum) meanwhile that an air roast could usually end before 10 minutes. So the difference in roast time could cause the complexity. Maybe the interior composition of the bean is different than the surface and cooks differently. I could put the pics of the interior beans tomorrow if anyone is interested.

Cheers