SquidMan wrote:Visually I can't really tell the difference

Bingo! That was the goal: make them look the same but taste different. Can't say I have any good reasons for doing that, but it does highlight the IKAWA's precision and repeatability.

Speaking of taste different... did they?

I have very mixed feelings about Klatch's Brazil Diamond Reserve. Unusually wide range of bean sizes, roasts very unevenly, and virtually inaudible first cracks.

With the caveat that I don't know this specific coffee but some confidence in who is selling it, Brazils are often naturals with less sorting. So I wouldn't come to quick conclusions about bean variations, other than culling obvious defects. Also Brazils are often low density coffees that benefit from a gentle roast with a more gradual application of initial heat. And since you would be emphasizing chocolates, I would take this close to start of second crack or at least in the darker medium range, dark enough for good caramelization.

Added: Researching this, you may have more bean variation because it's a blend. Klatch roasts this medium, but they usually roast lighter than my preference. I would try both ways and let us know.

https://www.klatchcoffee.com/products/b ... nd-reserve

Taste wise with the matching set points, I found curve 9 to be more acidic, curve 7 for me did not exhibit much acidity but more dark chocolate, "roasty" flavors, I got the best results when using a lever machine as a straight shot.

Made a few updates to the spreadsheet:
1) Hopefully fixed the formatting bugs that caused time readouts to show 60 seconds, e.g. now shows 2:00 rather than 1:60
2) Added a 'labelmaker' at the top that creates the label I use to record the roast profiles: Curvature StartTemp StopTemp DropTime
3) Scrunched and reformatted a bit to improve the screenshots

Link to the file (now 'V2') is here, as before, please make your own local copy.
https://docs.google.com/spreadsheets/d/ ... JNMpU/copy

I had also attempted to refine the 'rise' calculation: the temperature rise post first-crack - but that went sideways, which is elaborated below:

Working off comments in this thread: IKAWA inlet profile calculator and IKAWA inlet profile calculator And from another thread: Guatemala Xinabajul Aurelio Del Cerro - Sweet Maria's I embarked on using the spreadsheet to compare a relatively flat profile with a steep profile. I tweaked the numbers so each roast would have the same post-FC development time, but would straddle the suggested post-FC temperature rise to cross the 'dark chocolate threshold'. The greens I planned to use (Klatch Brazil Diamond Reserve) seem kinda prone to burning/tipping, so I also tried a more gradual starting ramp for one of the two runs - unfortunately that breaks a core rule for design of experiments, so sorting through the results is much more complicated.

Screen shot of the calculator:



So, one batch should have a post-FC rise of 5C/9F and the other 15C/27F. Theoretically I should detect enhanced dark chocolate notes in the 15C/27F post-FC rise run...

Despite one of the runs dropping 10C/18F hotter, both batches of roasted beans were visually identical, as in, like, totally the same-looking - was not expecting that. When ground, the hotter finish roast did look a tad darker, and the grind felt smoother when WDT'ing. Shot flow resistance was identical, and I use a long soak low-pressure preinfusion before pulling a 1:1.7 ratio. Could not detect a difference in dark chocolate notes. But there was a distractingly distinct difference in 'brightness': the hotter finish was more complex - but also more astringent/sour/acidic for my "2.5 Wave" coffee tastes - again, not expecting that.

Besides stopping at a higher temperature, that same profile spends a full minute less in the Mailliard phase. Apparently that's more significant than the finish temperature - or at least it is when things go wrong. I've now learned to stick with higher starting temperatures.

I logged the temperature data in the chamber, to see how well it lined up with the spreadsheet estimates. Color-coded per the respective profiles, solid lines are BT and the dashed lines are the inlet setpoints.



This might explain my tasting results: my goal of a 10C/18F post-FC difference turned out to be only a 5C/9F difference. The cooler roast finished hotter and the hotter roast finished cooler - a worst-case scenario for figuring out how to correct for this in the spreadsheet - so no quick fix here. For the time being I would advise any users to not put too much stock in the rise calculation, especially when comparing significantly different profiles.

Good news is the BT RoRs are still nicely linear. Linear enough to encourage me to read arguably too much into the nuances of the lines. In particular, note how the slope changes over the course of the blue BT RoR line: there's a 2x change in slope on either side of where the roast hit first-crack temperature. My personal theory is the specific heat capacity of beans changes before/after first crack - just as drgary suggested in post #11 of this thread - good call, sir!

Knowing what the curves should look like before the roast can significantly enhance the ability to analyze the results post roast. This is easy to accomplish on the IKAWA, but seems like it might be more challenging for a drum roaster, unless it's fully and tightly automated.