There seems to be a growing group of drum roaster users here, so I'd like to share what I'm doing and hopefully get some input on how to use this information-

After reading Jim's exploration of the M3, I decided to wire up my 2kg Has Garanti. I've been relying exclusively on the stock bean temperature read out, and while I have gotten reliable results based on that I was intrigued by the method of roasting by keeping a consistent MET reading.

Here is a photo of how my roaster is wired - There are four temp readouts: one stock bean t/c [Bean], one new t/c in the drum [ET] (out of the beans and out of the airflow), one t/c between the drum and the casing [MET] and one in the path of the exhaust [Exhaust].



The first roast was just an experiment - starting at a much higher drop in temp than I usually use for a 320g load (157C rather than 130C). I kept the gas at ~66% (10kPa) and got the shortest roast I've ever seen on this machine.

[the temps are in degrees C and the air and gas are in %]



In the 2nd and 3rd roasts (also using 320g) I tried out something new, dropping in the beans in a very hot drum to pull the temp down (rather than stabilizing the drum temp first) then increasing the gas pressure to to achieve a certain 'turn around' temp. It worked much better than expected, and saved me a lot of time between roasts (~2mins rather than ~10 mins)



After these three roasts my big question is "how can all this extra information help me?"

The 'MET' temp shows the most sensitivity to gas and air changes, with the 'ET' and 'Exhaust' behind it. However, in these roasts I based the air and gas changes and drop timing on the bean temp and crack sounds, and was able (in the 2nd and 3rd roasts) to replicate previous roast profiles very closely.

This begs the question: why collect this extra information? In what situation could it possibly help me?

I was working with a familiar bean; could the other temp info help me when working with an unfamiliar bean? My roaster is in an environment with large temp/humidity swings (seasonal); could this info be helpful in adjusting to those changes?

Any thoughts are appreciated,

Regards, Henry

Nice work.

I would give three reasons for having all this data:

• If your environmental probes respond faster than bean temperature, they are easier to use for control. This is especially true if you change weights.
• The shape and upper limit of the environmental profile has an independent influence on the coffee's taste.
• A problem in the heat or airflow might show up more quickly, since it will spoil the relation between the various readouts and controls. Changes in overall roast speed may be due to change since gas pressure or weather, so may not be as diagnostic as, for instance, the ET not responding as quickly as normal to heat input changes

Nice and informative posts! Please keep them coming!

On the mini500, I use the stock TC for perhaps ET or bean mass; I still have to contact the vendor or take apart the roaster to see the exact location of the TC tip, but it is on the upswing location of the drum. Because I roast the same 300g all the time, the readings and profiles are reproducible so I am leaving it alone for now.

From my rough understanding, the heat transfer in drum roasting consists of conduction and convection, and in case of perforated drum, infrared. The exhaust AND ET/bean mass temperature can be manipulated by the damper. I use the exhaust temp gauge as a measurement of the damper's opening and therefore convection.

The TC in the drum or bean mass measures more or less the conductive heat.... For example, the same TC temp reading can be achieved by opening the damper and increasing propane, or by closing the damper and decreasing the propane; however the convection will be different, and therefore different flavor profile. I guess this has something to do with heat transfer carried by water called "vapor front", contributing to the different flavor profile with the same temperature reading.

Several observations I have for past three weeks....

To avoid tipping, I turn off the propane for about one minute after bean drop. Tipping was more significant if propane was not turned off, and with dry processed and pre-blended espresso blend. I can see the drop in temperature to reach "turning point", then turn back the propane.

The drum speed is roughly 75rpm, tipping occurs if it is 60rpm.

To preheat, I open the damper all the way, so the roaster can reach a heat equilibrium, so the exhaust metal tube does not act like a heat sink and affect the exhaust temp drastically once it is adjusted during roasting.

Another extra reason (apart from Jim's nice ones). Sometimes a good roast comes out of luck (because of different beans, etc). The information collected will help me remember how to redo a roast the exact same way. So if I have a way to record automatically a roast and then one day I hit jackpot, I'll know how it happened. What I am trying to say is that you are the judge and the tools are the map, instead of the map being the judge (like following rules blind). Once you have several jackpots then maybe there is a trend for your roaster, your liking, beans, etc.

Cheers

hbuchtel wrote:Here is a photo of how my roaster is wired - There are four temp readouts: one stock bean t/c [Bean], one new t/c in the drum [ET] (out of the beans and out of the airflow), one t/c between the drum and the casing [MET] and one in the path of the exhaust [Exhaust].

Nice roaster! Nice size
Seems the range of batch sizes you might do with that roaster will need very different MET adjustment. 320 grams will get a lot of drum contact conduction where a large batch will get much more bean to bean transfer. What size batches have you tried and how differently do the beans react to the MET. You can fine tune the MET TC. to improve the stability of the readings. As I remember Jim did several tunings and ended up with his MET TC pretty close to the drum to avoid other effects.
You'll find that all those reading you have will become very interesting once you have more readings,roasts and cups to compare.

Thanks for the replies guys, I'm feeling inspired again

farmroast wrote:What size batches have you tried and how differently do the beans react to the MET.

Good question - with the new t/c setup I've only tried 320g, but tomorrow I'll try 640g and 960g roasts.

farmroast wrote:You can fine tune the MET TC. to improve the stability of the readings. As I remember Jim did several tunings and ended up with his MET TC pretty close to the drum to avoid other effects.

The t/c is placed at around 3 o'clock, between the drum and the inner wall (but not touching either). It is 'upstream' of the flame: ie the drum turns clockwise, pulling tendrils of flame with it and away from the t/c. I don't know if that is an ideal location, but I'll keep it there for the next couple roasts just for consistency.

Arpi wrote:The information collected will help me remember how to redo a roast the exact same way.

Good call. I just wish it didn't take so much effort to record all those numbers!

chang00 wrote:The exhaust AND ET/bean mass temperature can be manipulated by the damper. I use the exhaust temp gauge as a measurement of the damper's opening and therefore convection.

On this model of roaster the drum and cooling bin run off the same fan, the speed of which cannot be adjusted. The damper offers a range between drum+cooling bin and drum only. Seeing how sensitive MET is to airflow changes, I think I now have a way of quantifying and using the murky range between the two extremes.

chang00 wrote:For example, the same TC temp reading can be achieved by opening the damper and increasing propane, or by closing the damper and decreasing the propane; however the convection will be different, and therefore different flavor profile.

I'm really looking forward to learning more about this!

another_jim wrote:The shape and upper limit of the environmental profile has an independent influence on the coffee's taste.

Jim, I'm interested in trying to profile by keeping the ET steady, as you wrote about here. Based on my description of the t/c placements and the graphs I uploaded, do you think I should focus on my 'ET' probe or 'MET' probe?

Regards, Henry

As I said, I don't feel comfortable (yet) trying to control the BT exactly. Instead I have three MET profiles, one each for fast, medium, and slow roasts, which are idiot proof repeatable. The medium roast is a flat line at 250C. The fast roast is a flat line at 270C, then dropping to 250C at first crack. The slow one drops the MET temperature to 200C until the beans are at 150-160, then raises it back to 250 by the 1st crack. For a 150 gram load, this gives me 11, 13, and 15 minute roasts respectively for first pops of the second roasts. Here are some examples:

Fast


Medium


Slow (This roast is 5C lighter, and would have taken another 45 seconds or so to first pops)



As always, YMMV -- these levels will depend on the exact sensor placement. My sensor placement is such that the fast profile at 270C to 250C will just avoid ashiness in a Brazil or other low grown.

I haven't cupped and pulled enough shots to make any claims for these profiles other that they are competent roasts that don't screw up the beans.

Ed's point about the size of the load, and the amount of bean to drum versus bean to bean heating is hugely important (I never payed much attention when I was air roasting). I am doing roaster simulations which arenlt quite ready for prime time yet (the red estimator line on BT and parameters in the graphs above), but they do indicate that the effective MET the beans see drops as the load increases when the drum temperature is kept steady. A full drum might allow one to get away with higher drum temperatures.

Hi.

I tried switching to keeping a constant ET. I placed the PID probe in the bottom chamber close to the heating elements (but not inside the drum). Previously, the PID sensed the bean temperature. Then I used a data logger to record bean temperature. This is the graph I got

beans: Burundi Bwayi lot #5
load: 200 grams
fan: 5
power setting: PIDed constant ET 600F
Red line = bean temperature
1C starts at 390F 7:34
2C starts at 10:24 END
No power or fan adjustments made





I think the roast went a little fast. So maybe next time I'll try a setting of 550F. The beans showed no defects. I think that once the difference between ET and BT gets smaller, then the graph should flatten naturally at the end.

Thanks for the ideas

The 2nd started early on this roast, since the splits on the beans are still light. They go dark when the oils start smoking.

I'm guessing the PID has an easier time with controlling this temperature.

I have a very good reason for monitoring any version of the ET that will tell you whether you are approaching the ash zone. I've just cupped the last of about a dozen professional roasts, some from the best roasters in the country, and every single one of them from a winter area is ashy (the ones from Hawaii are fine). I asked around about this. In the winter, the ventilation air on the roaster tends to be at a lower temperature than usual. The roasters are running fast profiles, and are trying to maintain the same roast timing. As a result, they are exceeding the upper limit of drum temperatures and are ruining the coffee (at least for me).

There's no point in finishing your roast in exactly 12.5 minutes if it tastes like it's been dipped in coals.