BT RoR guidance for popcorn poppers and small fluid bed roasters? - Page 3

Discuss roast levels and profiles for espresso, equipment for roasting coffee.

#21: Post by Trjelenc »

GDM528 wrote:In case a drum roaster pilot is reading this thread: how important is RoR during the drying phase?
For drum roasters, I've been going off of the guidance that the minute details of the drying phase don't really matter as long as you're not doing anything to induce roasting defects onto the coffee, such as having drum too hot and scorching the beans. Allegedly no discernable flavor reactions occur prior to the color change to yellow. So the idea is to run the RoR at dry in a way that helps set up the rest of the roast. For example, if you're trying to go quicker through the middle phase, you might not want to start off with a slow momentum through to yellow.

minus1psi (original poster)

#22: Post by minus1psi (original poster) »

This may or may not be forward progress, but it is at least an incremental improvement in knowledge.

Following insights from @GDM528 and @Trjelenc I reviewed graphs of old roasts and came to the theory that the rapid drop in RoR always has an inflection point where the curve becomes more horizontal. Roasts with higher temperatures of that inflection point naturally had more vertical room to decline without flattening out.

I am pursuing the idea that the drying phase is less critical. Raising the temperature of the inflection point without worrying about the shape of the BT RoR curve pre-dry end is the current theory. I am trying to get the inflection point to happen right around the end of the drying phase. The hope is that at DE the inflection point of the BT RoR has been increased and has more room to decline gracefully.

I tried a couple of other approaches but the only way to bump the voltage just before DE (G/Y) was to have headroom at that moment. The voltage (effectively the heat) was increased when BT was at 300°F. For my typical roast DE happens closer to 320°F. It sort of worked. I chose to approach DE at 110V and increase to 120v at around 300°F. I do not love that the middle phase is so short.

On the graph below the RoR curve is less flat than in many previous roasts. This roast was allowed to go for a long time; I like for a portion of cold brew beans to go well into 2C and regard those roasts as excellent opportunities for this sort of experiment. 25% DTR would have been pretty close to the end of 1C.

minus1psi (original poster)

#23: Post by minus1psi (original poster) »

This is progress. I increased the sampling interval from 3s to 4s, reduced the drying phase voltage, and futzed with curve settings a little bit. This is looking more manageable. Maybe the software changes are cheating in some sense but the information on the graph feels like a good reflection of the mechanical changes during the roast. The target phase percentages were 40%/35%/25%. In this case the beans roasted through to 33%DTR based on color. Weight loss was 16.7%; in retrospect I could have pulled them earlier.

Current questions:
  • How low can the voltage (heat) go for the drying phase while maintaining a drying phase percentage of 40% or less
  • What factors influence the timing of the BT RoR inflection point -- how does the roast move it left or right on the x-axis
  • If the nuanced shape of the BT RoR curve under the drying phase doesn't make much difference, how much does the length of it matter -- essentially the length of the drying phase. The horizontal length of the RoR curve is a component of it's shape, right


#24: Post by GDM528 »

An interesting property of a fluid-bed roaster with aggressive airflow, is the consistency of the shape of the temperature readings. Empty chamber / full chamber, whatever - the curve is essentially the same. Chart below shows the temperature setpoints and the temperature readings in the chamber with and without a load of greens (thermocouple located in the spinning bean mass when loaded).

The RoR line is from the beans-in-chamber readings - but could've been from the setpoints and still look the same. I was able to prototype the RoR curve on a spreadsheet to get the initial profile, and it only took a couple tweaks to account for thermal inertia. Key to avoiding sudden drops in the RoR curve was making uniform adjustments to the temperature rise. In my case I could only make six adjustments, which was barely enough.

An essential component to my process was constant airflow. Changing both airflow and temperature is much more complex, so perhaps the constant-radius principle doesn't apply. Would've been handy if the Wearever had separate power cords for the fan and heater ;)

FWIW I don't use the RoR profile because it spends so much time below caramelization temperatures (320F-ish). I theorize that over-dries the greens before they hit the caramelization/Mailliard phase. As precedence, I cite drum roasters that are preheated prior to dropping in the greens, and even the Pro version of my roaster does a similar pre-heat. Lately I've been running drying phase down to about 15%.

minus1psi (original poster)

#25: Post by minus1psi (original poster) »

I am gaining on this thing slowly. The roast graph here uses the following parameters
  • sampling = 3s (default)
  • delta Span = 6s
  • smooth Curves = 6
  • smooth spikes = off
  • drop spikes = off
  • polyfit computation = off
The shape is much less concave. My roasting strategy was the following: introduce the beans at high voltage to get them moving. A few seconds later drop the voltage to close to a minimum. Watch the BT RoR curve LED; as soon as it peaked and began to decline, steadily increase the voltage (heat) until 1C. At 1C I let the popper coast until about 10% DTR, dropped the voltage until about 13% DTR, brought it back to the same voltage as the beginning of 1C and finished the roast there. Calling Dry End was arbitrary -- do not put any stock it the divisions of time pre-1C.

I'll have to see how the coffee tastes but the curve is improved :-). Next batch I will try a similar approach but increase the addition of heat less steadily -- which is to say increase it faster as soon as RoR peaks (RoR Turning Point?) and at some point reduce the rate of increase until 1C.

Thus far the biggest takeaway for my little set-up is to pay attention to the BT RoR Turning Point by watching the ∆BT LED and to apply heat once that number begins to decline. Due to the limited amount of adjustment available for heat, this necessitated going through the drying phase at a relatively low heat setting. Even at the low heat setting, bean temperature went above the boiling point of water soon about 30-35 seconds after the roast started.


#26: Post by pivoc »

minus1psi how did the coffee taste?
Any progress with the profile? I was on this track as well, starting with low heat and increase it later on. Found it hard to stretch the middle phase, so I gave up that strategy for a while. Now I am trying to add heat in the beginning and decrease it during the roast.

minus1psi (original poster)

#27: Post by minus1psi (original poster) »


Thank you for the note. Please share more of your experience.
pivoc did the coffee taste?
The coffee was good. I am not reporting tasting notes of tiramisu and pomegranate, just reporting good tasting coffee. To be honest that is pretty much my goal => coffee that tastes good to my palate. Nuanced flavor notes are still a ways away for me.

There are some unsatisfying details that became clear using the approach in my previous post.
  • Consistent results are hard to come by, possibly due to the large number of adjustments. It is much easier* to consistently nail a weight loss with a maximum variance of .5% if the number of adjustments is minimized. (*easier as in 'occasionally possible'!)
  • In my set up the heat is exhausted out of the roasting chamber almost instantaneously. The beans themselves have thermal mass and no doubt affect the roast, but it seems that the overall environment offers less influence than in drum roasters or any roaster that slows the escape of heat from the roasting chamber. Worrying about spontaneous changes in bean temperature has begun to feel kind of silly. The beans are going to go through an exothermic phase but a second or two later my roasting chamber has exhausted that heat and the BT has changed. Nor can I anticipate that phase to the second at this time. I am still trying to figure out how much I should be concerned about those type of BT changes.
Currently I am pursuing ways to minimize adjustments and increase repeatability. Although this is sort of thing tends to change as my roasting evolves, at the current time I am judging consistency using end-of-roast BT and weight loss % of the cooled beans.

This is the most recent procedure. Every detail is subject to change
  • Charge the beans (chamber temperature around 200°F-215°F, voltage ~115v)
  • Increase the voltage/heat/airflow for a second or two to get the beans moving (increase voltage to > 120v)
  • Drop the voltage to an evolving low setting (varies, 110v +/- 1-5v) for part of the drying phase
  • Increase voltage to an evolving middle level (115v or 116v) at the BT RoR Turning Point as judged by watching the change in the BT ROR LED on Artisan. By the theory of the moment this adjustment starts a couple of seconds before BT ROR TP and finishes a couple of seconds after.
  • Coast through the rest of the roast at that setting
  • Drop the beans when they hit a target temperature that may or may not correlate to a DTR percentage. In any case it does not correlate tightly enough to rely on DTR... yet.
Trying to nail consistent weight loss using DTR is challenging with my roaster. Maybe it is challenging with everyone's roaster :-) With barely 100 roasts under my belt, I am striving for techniques that improve consistency. At the moment that means observing the ROR curve but not relying on it mid-roast. Post-roast the ROR curve is more of a focus.

minus1psi (original poster)

#28: Post by minus1psi (original poster) »

Here is some data from three roasts this afternoon. The voltage was adjusted multiple times midroast trying to shape the ROR curve with unsatisfying results. The data show why I am relying more on drop temperature than DTR -- not to say that that doing so makes sense for any other roaster. I am getting very consistent weight loss by roasting to a consistent temperature; DTR and total roast time varied quite a bit.
           Roast             #113           #114            #115                     
 Roast Time (m:ss)           7:03           6:36            6:42  poorly correlated  
 Temp. @ Drop (°F)        368.1°F        368.2°F         368.5°F  tightly correlated 
      Wt. Loss (%)         -15.6%         -15.4%          -15.6%  tightly correlated 
           DTR (%)          38.3%          31.1%           40.3%  poorly correlated  
      Drying Phase   1:51 (26.2%)   1:51 (28.0%)    1:39 (24.6%)                     
    Maillard Phase   2:30 (35.5%)   2:42 (40.9%)    2:21 (35.1%)                     
 Development Phase   2:42 (38.3%)   2:03 (31.1%)    2:42 (40.3%)                     
I would love to use DTR as a guiding metric for my roasts but am having a terrible time getting that to work out.


#29: Post by Trjelenc »

minus1psi wrote: I would love to use DTR as a guiding metric for my roasts but am having a terrible time getting that to work out.
I wouldn't lose sleep over it. DTR is a meaningless metric.

minus1psi (original poster)

#30: Post by minus1psi (original poster) replying to Trjelenc »

There are two aspects of the metric that challenge me. The first is simply identifying the beginning of first crack; my hearing is okay but not great. The audio environment around my roaster is cluttered with sounds from the popper itself and from two fans. I am listening for the first triple pop, as in "pop-pop-pop" not "pop-pop-pause-pop" or "pop-pause-pop-pause-pop". That means audible-to-me triple pop; I am pretty sure there are subtle pops going on that I do not hear, perhaps some triples. Also because my batches are in the 100g range, if I hear quite a few double pops before hearing a triple pop there is a concern that 1C began prior to being marked. On some occasions I never hear a solid triple pop, especially if the heat is lower vs higher.

The second thing is a little harder to describe. As I understand it DTR is the ratio of Development Time to (Drying Time+Maillard Time+Development Time). As such it only factors time. It does not factor changes in heat or air flow. For the sake of discussion imagine two roasts both reaching first crack at exactly the same time and BT. Next imagine that starting from 1C, one of the roasts has the heat decreased and the other has the heat increased. If both are dropped with a 20% DTR, sez me, you've got two different roasts but they have identical DTR metrics. Generally if one roast has the heat (or air flow) treated differently than the other but they have the same total time, they have the same DTR but seem to me to be different roasts.

The metric may be useful to me as a data point but not as a be-all-end-all guiding metric. Maybe if I can standardize all other roast manipulations such that they are dependably/reliably repeatable, DTR can have a significant influence on when to drop the roast. For now BT is working better for my set-up.