Some lessons after using a fluid bed roaster for two weeks - Page 2

Discuss roast levels and profiles for espresso, equipment for roasting coffee.
lins05 (original poster)

#11: Post by lins05 (original poster) »

GDM528 wrote:
I wonder if one of the reasons behind the declining-RoR profiles, is to create a roasting gradient within the bean, resulting in a blend of roast degrees and a richer range of flavors. Thoughts?
There are a hundred ways to do coffee roasting, even Scott Rao himself can't say "follow the declining RoR rule or otherwise the coffee would taste terrible".

My 2c is: if we keep the burner/fan constant, then the declining RoR is a natural result as BT increases and converges to ET. So a smoothly declining RoR means very little adjustments to the burner/fan, and by Occam's Razor, it allows for the simplicity to reason about the whole process. The more frequent burner/fan adjustments, the more difficult it is to reason about it and reproduce a batch.

User avatar
Almico
Supporter ★

#12: Post by Almico »

lins05 wrote:There are a hundred ways to do coffee roasting, even Scott Rao himself can't say "follow the declining RoR rule or otherwise the coffee would taste terrible".
Not terrible, but it will taste better if you do. And will be much more repeatable.

Otaibimn

#13: Post by Otaibimn »

I tried to replicate Scotts Steady Declining Technique on a lot of roasts.

Once I threw it out of the window and just focused on phase durations I found a lot of my roasts got that nice "TING" of quality.
Unless you roast your own coffee, You still didn't reach 90% of the rabbit hole.

User avatar
drgary
Team HB

#14: Post by drgary »

@ Richard Green:

I'm really thinking about "useful" measurement of BT, not "accurate" measurement. My speculation in this case starts with looking at the OP's roast profile image and trying to understand whether variations in measured BT are less accurate because the probes are picking up a larger proportion of air temperature than occurs in my drum roaster. Your suggestion of monitoring exhaust temperature is very helpful and applies also to using drum roasters, especially when roasting a very small sample where BT is less accessible.

BTW, I don't think the steak analogy is close enough to what happens in a coffee bean which has much less mass and is building up pressure inside a plant cell matrix. Do we really need to know BT exactly to track something useful? What's happening on the surface of the bean that is contacting the probe is obviously different than what's happening in the internal pressure cooker. How much air versus bean is contacting the probe would mean interpreting BT differently. It comes down to getting more consistent results between how I'm operating my roaster based on the measurements available, sensory inputs during the roast, and expected versus tasted roast and bean properties in the cup afterwards. And then, with the imperfect tools we have, trying for good, consistent results.
Gary
LMWDP#308

What I WOULD do for a good cup of coffee!

User avatar
Peppersass
Supporter ❤

#15: Post by Peppersass »

I don't disagree, Gary. Relative or "proxy" measurements can certainly be used to effectively manage a roast. We all do that now.

But because such measurements don't directly tell us what's happening inside the bean, we have to use a trial-and-error process to find the right profile -- i.e., the one that cooks the interior to the degree we seek. It seems to me that if we had direct measurement of the interior of the bean and how far the heat has penetrated (like a heat map), we could at least reduce the number of trials necessary to find the right profile.

User avatar
drgary
Team HB

#16: Post by drgary »

My main inquiry, though, was whether there's an added inaccuracy in this fluid bed roaster either because there's more intermixed air or because of probe (mis)placement. Your idea of prioritizing exhaust air seems to be sound.

To the greater accuracy you envision, I believe internal versus external measurements have been calculated with microprobes. Rather than real-time measurement, a table could be created to help estimate what's happening in a bean mass. The estimate could be adjusted for roaster characteristics, etc. Until then, it's roast, view, taste and adjust.
Gary
LMWDP#308

What I WOULD do for a good cup of coffee!

User avatar
luca
Team HB

#17: Post by luca »

FWIW, you can chalk me up as another person that bumbled around trying the pretty ROR on a pure fluid bed air roaster and finding that it didn't work. As others have said, you can't just arbitrarily pick whatever temperature measurement the machine gives you and expect that you can declare that to be the BT and therefore a pretty ROR will result in a good roast. I'll say again, that one fairly objective experiment you can do to affirm the utility of temperature measurements is to exactly duplicate the temperature graph on two roasts, then taste those two roasts next to each other and see if they taste different. If they do, then the temp measurement you have clearly isn't giving you the whole picture.
LMWDP #034 | 2011: Q Exam, WBrC #3, Aus Cup Tasting #1 | Insta: @lucacoffeenotes
★ Helpful

mathof

#18: Post by mathof »

luca wrote: I'll say again, that one fairly objective experiment you can do to affirm the utility of temperature measurements is to exactly duplicate the temperature graph on two roasts, then taste those two roasts next to each other and see if they taste different. If they do, then the temp measurement you have clearly isn't giving you the whole picture.
I'm not sure I'm understanding that point. I have a tc in the swirling bean mass in my Ikawa Home roaster. Using Artisan's background function, I can see when the temperature curve of a new roast closely tracks the temperature curve from a previous roast of the same beans using the same inlet profile. The results taste the same too, or at least I've never noticed a difference; but I'm open to testing two "identical" roasts side-by-side, if that is what you are questioning.

User avatar
Almico
Supporter ★

#19: Post by Almico »

On my Artisan FB roaster I added a stubby 1/2" BT probe to the wall near the bottom of the hopper. The short length kept it out of the lofting air path so I was confident in getting a consistent, if not totally accurate, BT reading.

The main issue with FB for me was how easy it was to quickly change the roasting temperature up and down. Because of the high airflow it is easy to just turn a knob and "steer" the curves. I could practically write my name with the RoR curve. My coffee was pretty good, but frustratingly inconsistent. One roast would be nice and juicy, the next roast of the same coffee would be dry, even though the curves looked identical. One roast would be sweet as can be, the next not so much, even though the roast level measured the same agtron #s.

When I switched to drum roasting the improvement in my coffee was dramatic. In hind sight, it likely had little to do with drum vs air roasting. It had mainly to do with the consistency of the roast environment temperature. Adding a thick steel drum and heavy faceplate to the equation stabilized the environmental temperature to a great degree and thereby stabilized the "actual" bean temperature. I couldn't make swift, agile temperature changes if I wanted to.

From my experience, coffee beans do not like swings in roasting temperature. Most people focus on the word "declining" in Scott's recommendation of "steadily declining RoR". The key word is more likely "steadily".

Once I developed the technique of using a consistent roast plan on my FB, my coffee improved. Instead of steering the curves, I learned to predict heat settings in advance in order to hit the desired milestones and repeat those few adjustments over and over again. I stopped guessing and then making small corrections when I guessed wrong and my coffee improved. Roasting in a drum just makes this much easier because you can't make those corrections as quickly.


There might be another issue with electric fluid bed roasting, and that is how the heat level is determined and controlled. Is that knob (or software) controlling the current moving through the elements, or is it controlling a PID or thermostat? Is the heating element constantly on (to whatever degree), or is it pulsing on and off in order to keep the thermostat happy. On my Artisan I believe it was the latter, although I could never get a satisfying answer from the designer.
★ Helpful

User avatar
Peppersass
Supporter ❤

#20: Post by Peppersass »

Almico wrote:There might be another issue with electric fluid bed roasting, and that is how the heat level is determined and controlled. Is that knob (or software) controlling the current moving through the elements, or is it controlling a PID or thermostat? Is the heating element constantly on (to whatever degree), or is it pulsing on and off in order to keep the thermostat happy. On my Artisan I believe it was the latter, although I could never get a satisfying answer from the designer.
I think making more heat adjustments or having heat controlled (pulsed) by a PID in a fluid bed roaster isn't necessarily a bad thing as long as there's reasonable temperature stability and gradual change in the air contacting the beans. That probably depends a lot on the physical characteristics of the roaster -- i.e., how the air gets to the beans.

Below is a comparison of two light roasts of Ethiopia Duarte Bombe Natural got from Happy Mug in 2020 (in both cases, the beans were thawed after being frozen for over a year.) The first roast was 175g done in a Quest M3 with M3s drum and the second roast was 50g done in an Ikawa Pro V3. Now, both roasters being quite small, neither is representative of what might happen in a large gas drum roaster or large electric fluid bed roaster. But what's interesting to me is that 1) in both cases the heat was controlled by a PID, and 2) there was a significant difference in taste.

Here's the Quest M3/M3s roast:



I made minimal changes to heat. I bumped heat to max shortly after charge, then didn't touch it until about 90 seconds before 1C, when I lowered the heat. About 30 seconds before 1 C I saw that RoR was rising, so I lowered the heat again.The red MET curve measures heat outside the drum (i.e., closer to the heaters.) It shows the PID making some small adjustments here and there to keep the heat at the intended target. Undoubtedly, there are small changes from constant pulsing by the PID that you can't see. I only made two changes to airflow: an increase at peak RoR to keep the RoR curve from plunging and an increase to max at the end to blow off chaff (a little late on that one.)

Note that the orange ET curve and the purple Exhaust curve, the probes for which are much closer to the beans, are quite smooth, as is the BT curve, the probe for which is in the bean mass. RoR is declining steadily, but there does appear to be a flick at the end.

This roast came out just over the medium light side of the border between between medium-light and light. It tasted quite good, with hints of fruit, but there was some astringency at the finish.

Here's the Ikawa Pro V3 roast:



Here, again, you can see some bumps in the incoming air temperature the PID made to keep the exhaust temperature, which is measured just above the spinning bean mass, steady. Again, there's probably a lot of pulsing of the heater by the PID. As I've said before, the RoR curve looks like the roast is baking, but it isn't. No way I could get a steadily declining RoR in this roaster. I do find the RoR curve useful for marking 1C. There's almost always a distinct drop in RoR right when 1C begins, usually accompanied by a pop or two. In this roaster, waiting for a rolling 1C will result in much darker roasts than I'm seeking.

Probably the most surprising aspect of the curve is the time it takes to get to dry -- it's about half the time required in the Quest and other drum roasters. I always mark dry at 300F, so it's not necessarily when the actual color change to yellow takes place. I find that to be too subjective. However, 300F in exhaust temperature may or may not be the same as 300F in BT measure by a probe in the bean mass. Nevertheless, that's where I mark dry.

This roast came out just over the light side of the line between medium light and light. In fact, it was only one mark lighter on my RoastVision color meter than the Quest roast. Note that the development time was only 50 seconds, versus 1:27 for the Quest roast. Although the time to 1C is about the same in both roasters, time to dry and development time in the Ikawa are notably faster.

Another difference is the temperatures at 1C and drop. They're considerably lower in the Quest. I suspect that's just a difference in the probes and what they're measuring, though I've often been puzzled by how much lower the Quest temperatures are than temperatures I see in profiles from other roasters. Also note that the rise after 1C was 14 degrees in the Quest versus 9 degrees in the Ikawa. I'm sure this is part of the reason DT was longer in the Quest and the roast was slightly darker, though that much extra rise in temperature would have resulted in an even darker roast in the Ikawa.

Now, the big difference in these roasts was taste. The Ikawa roast was sweet, complex, and full of berry flavor. No astringency. It was one of the best roasts of a natural that I've tasted, right up there with any commercial roast I've bought. While the Quest roast wasn't bad, there was no comparison. Hard to say why. Maybe small electric drum roasters simply can't perform like large commercial drum roasters. Maybe the Ikawa's tiny batch size gives it an advantage. Who knows?
★ Helpful