GDM528 wrote:Do RoR curves make sense for the Ikawa? I get their purpose in the context of a drum roaster, which has significantly higher thermal mass than the beans. The Ikawa on the other hand, has negligible thermal 'mass', and the aggressive convective heat transfer in the Ikawa causes the beans to linearly track the applied temperature profile 1:1.

This is from one of my roasting runs:

image

Thanks so much for looking at that! Super interesting. And quite helpful, I think.

Much of the RoR curve (blue) is largely flat, because the bean temp (green) is simply tracking the chamber temperature. I've read that the goal is a steadily declining RoR, which implies the beans are perpetually catching up with the chamber temperature - but I hit that milestone just two minutes in. I suppose I could tweak the Ikawa's temperature setpoints to mimic a drum roaster curve, but would that make the resulting roast taste better?

What this shows is that it's impossible to get a true "bean temp." You can only get the "environmental temp." This is what we experienced with the Fresh Roast when we tried sticking a probe in the bean mass. You have so much air pushing through the beans in fluid bed roasters that it's fairly impossible to get a reading on much more than the air.

If you stick a piece of chicken in the oven, a temp reading in the air around the chicken doesn't tell you much about how done the middle is. If you stuck a probe in the middle of the chicken it would show the chicken heating up slowly. The rate of rise would naturally be declining, and if you wanted a longer "development time" you'd set the oven temp lower, closer to what you want your final chicken temp to be. This would give you chicken that is more moist, but less browning would occur since that happens over 350. (I just happen to be cooking chicken at the moment).

I do think Rao would say you absolutely want a declining rate of rise in any roaster, thus avoiding the baked flavor. You should get a lot of that naturally if you really can control the temp of the chamber well. A declining profile would keep it in the development stage longer. However, beans don't act in a nice way like chicken. There are stages where it takes heat better, where steam happens, and you have this tough exterior on them. The flick, the crash, make things tricky, I guess.

I think there are enough visual and audible clues to get us most of the way there though. And I'm definitely encouraged by how you show its ability to really be accurate in the temp. But maybe this last bit shows me I don't need a temp probe to really dial in a bean.

Thanks again! Super nice.

GDM528 wrote:Do RoR curves make sense for the Ikawa?

Oops, my first "thanks" comments I put up in the quote instead, by accident. So again:

Thanks for charting out the RoR stuff. It was super interesting and I think quite helpful. Glad to have people working on this!

I've observed how some of Ikawa's profiles for washed beans have a "spike" that goes really hot, right at the beginning of the roast. I wondered if that spike would overshoot and scorch the beans. So, I set up a run using their generic (non-Ikawa) recipe for filter washed to a medium+ roast level:



Those clever peeps at Ikawa knew how to avoid the overshoot by quickly pulling back from the peak of the spike, essentially exploiting the ringing effect to slingshot the temperature back down - nicely done, Ikawa. Still a pretty freaky profile tho.

I strictly roast for espresso, so I left the chamber empty for the filter-centric roast. To see the effect on bean-mass temperature, I switched to the recipe for Ikawa's Blue Ayarza Dark+++ roast:



Same deal with the spike, and it seems to have the effect of hovering the bean-mass temperature just barely in the browning range for about a minute before slowly ramping up the heat. Fan speed was elevated during the spike such that all the beans were moving very fast and elevated off the bottom of the chamber. The resulting roast is nice and dark with a very subtle sheen. Heard one or two second cracks just before the drop.

There's a point in the roast where the beans are getting hit with an air temperature very close to the first-crack threshold. From what I've read about natural-processed beans, the spike could scorch their more delicate exterior. Fortunately, Ikawa doesn't use high-temp spikes for their generic natural-process recipes, but I did find one of their recipes for their Dumerso (natural-process) light+++ that looks kinda scary to me - I'd keep the batch size small for that one, so the beans don't get pinned to the bottom of the chamber.

It's not obvious to me why the spike is even necessary, as I've found other ways to extend the hang-time for browning without significantly overshooting. Perhaps Ikawa is just showing off their crazy thermal agility?

GDM528 wrote:I've observed how some of Ikawa's profiles for washed beans have a "spike" that goes really hot, right at the beginning of the roast. I wondered if that spike would overshoot and scorch the beans.
Those clever peeps at Ikawa knew how to avoid the overshoot by quickly pulling back from the peak of the spike, essentially exploiting the ringing effect to slingshot the temperature back down - nicely done, Ikawa.

It's not obvious to me why the spike is even necessary, as I've found other ways to extend the hang-time for browning without significantly overshooting. Perhaps Ikawa is just showing off their crazy thermal agility?

What is the "ringing effect"?

I have noticed that drum roaster profiles use higher drop temperatures for denser beans (to get some heat into the centre of the beans). I wonder if the spike profiles serve a similar purpose.

Sorry, that was sloppy terminology on my part. The 'ringing' I refer to, is how the system responds to a rapid shift in temperature:



When instantly jumping to a new temperature, the control system will oscillate every 20 seconds or so. Ikawa's profile appears to have set timing between the peak and valley temperatures to sorta line up accordingly - so they're not aggravating the tendency of the system to ring, but not completely avoiding it either. The result is a near-perfect triangle-wave thermal response - good luck getting a drum roaster to do that!

MNate wrote:What this shows is that it's impossible to get a true "bean temp." You can only get the "environmental temp." This is what we experienced with the Fresh Roast when we tried sticking a probe in the bean mass. You have so much air pushing through the beans in fluid bed roasters that it's fairly impossible to get a reading on much more than the air.

If you stick a piece of chicken in the oven, a temp reading in the air around the chicken doesn't tell you much about how done the middle is. If you stuck a probe in the middle of the chicken it would show the chicken heating up slowly. The rate of rise would naturally be declining, and if you wanted a longer "development time" you'd set the oven temp lower, closer to what you want your final chicken temp to be. This would give you chicken that is more moist, but less browning would occur since that happens over 350. (I just happen to be cooking chicken at the moment).

I do think Rao would say you absolutely want a declining rate of rise in any roaster, thus avoiding the baked flavor. You should get a lot of that naturally if you really can control the temp of the chamber well. A declining profile would keep it in the development stage longer. However, beans don't act in a nice way like chicken. There are stages where it takes heat better, where steam happens, and you have this tough exterior on them. The flick, the crash, make things tricky, I guess.

I'm a recovering Fresh Roast owner, and very happy with the well-engineered Ikawa. The Fresh Roast had a sloppy thermal control system that would fluctuate +/-10C during the roast, beans would get stuck in hotspots at the bottom of the chamber, and there was a huge thermal gradient from top to bottom of the bean mass. I don't miss it one bit.

About that chicken:
A more accurate analogy would use a chicken with only 10% water content - pretty darn desiccated. And the oven would have a convection fan so aggressive, the chicken would be ricocheting around inside. I'm sure the chicken you prepared is much more appetizing than what I just described!

I prefer thawing frozen food in a sous vide as an analogy, as the frozen food will influence the water temperature until its reached equilibrium. I think something similar happens when the temperature I'm measuring of the spinning/rolling bean mass reaches the same temperature as an empty chamber. At that point the beans are no longer pulling down the air temperature in their immediate vicinity.

The stepped response I posted earlier shows that each stage of the roast can be deconstructed, and each stage could then be dialed into however many seconds are considered optimum for whatever chemical transforms occur in each stage. That's in stark contrast to the declining-RoR approach that's been perfected for drum roasting. I may very well end up duplicating drum roasters, but I'd like to explore some of the other possibilities first. Red pill versus Blue pill...

GDM528 wrote:I'm a recovering Fresh Roast owner, and very happy with the well-engineered Ikawa. The Fresh Roast had a sloppy thermal control system that would fluctuate +/-10C during the roast, beans would get stuck in hotspots at the bottom of the chamber, and there was a huge thermal gradient from top to bottom of the bean mass. I don't miss it one bit.

Super helpful context, thanks. One of the reasons I chose to dive into roasting at the $1000 level instead of the $200 level was controllability and repeatability. Your comment, along with my initial experience with the Home, has definitely contributed to the validation my purchase.

It could be interesting to pick out the 10 lightest and 10 darkest bean from each batch and analyze them individually.
That way you would get the true range of uniformity.

Lightest/median/darkest

Remember that beans are also often different in how they roast: Colombian seems to take heat well whereas some Kenyan I roasted was crazily uneven (but stated good).

Roast level also means a lot as darker roasts will have more uniformity. I always find it hardest to roast light or medium-light.

It was in response to post no 16