When comparing roasting profiles across different machines and operators, there is room for interpretation. For example, what is SSFF/FSSF? What is a 6-3-3 profile when one person arbitrarily defines 300F as the end of dry and another person uses 320F as the EoD. Hopefully the expression format we adopt in this R&LT thread can bring more understanding between us and cause less confusion.

Neal (N3Roaster) made an insightful video clip explaining the Calibration of Roast Profiles between 2 Coffee Roasters. http://youtu.be/fS6eSKL8tlA When we use the same bean, we have several reference points (room temp @ time 0, time to 300F, time & temp at the First Crack). However, we don't have the time & temp to the second crack as the reference point for profile conversion, and therefore the absolute values of the RoR during development phase should be compared with caution in mind.

Rob Hoos uses a base line profile and change only one parameter when comparing various roasts from the same machine. Last weekend I employed his methodology and drove four back-to-back roasts of the Ethiopian. Since my previous two batches were (relatively) darker roast for SO espresso and I only got the typical blueberry smell in my batch #2 on day 3 when opening the container, but the fruity flavor didn't translate into the cup, I decided to try shorter and lighter roast. My batch #4 was the base line profile, #3 was same drop temperature yet shorter development time [Dev (-)], and #5 with the same development time but lower drop temperature [Drop Temp (-)] of the baseline. I also threw in a light roasted FSSF style #6 into the mix just for fun.

Here are the four profiles, and I'll update my notes next week.


Roasting Info:

Bean: Ethiopia Sun Dried Guji Hambela #3, SSFF Dev(-)
Roaster: KapoK 500 Sample Roaster
Charge Mass: 227.0g in, 193.0g out
Charge Temp: 311F
Dry/MAI/Development: 4:54/4:12/1:48
FC-start temp: 385.9F
Finish Temp: 408.0F (1C + 22.1F)
Overall Roast Time: 10:55
Weight Loss: 14.98%

***

Roasting Info:

Bean: Ethiopia Sun Dried Guji Hambela #4, SSFF Base Line
Roaster: KapoK 500 Sample Roaster
Charge Mass: 227.0g in, 192.2g out
Charge Temp: 307F
Dry/MAI/Development: 4:56/4:14/2:09
FC-start temp: 386.2F
Finish Temp: 408.0F (1C + 21.8F)
Overall Roast Time: 11:19
Weight Loss: 15.33%

***

Roasting Info:

Bean: Ethiopia Sun Dried Guji Hambela #5, SSFF DropTemp(-)
Roaster: KapoK 500 Sample Roaster
Charge Mass: 227.0g in, 192.5g out
Charge Temp: 308F
Dry/MAI/Development: 5:00/4:11/2:07
FC-start temp: 386.1F
Finish Temp: 408.0F (1C + 17.1F)
Overall Roast Time: 11:18
Weight Loss: 15.20%

***

Roasting Info:

Bean: Ethiopia Sun Dried Guji Hambela #6, FSSF Dev(-)
Roaster: KapoK 500 Sample Roaster
Charge Mass: 227.0g in, 194.3g out
Charge Temp: 352F
Dry/MAI/Development: 3:38/3:56/1:40
FC-start temp: 386.4F
Finish Temp: 399.9F (1C + 13.5F)
Overall Roast Time: 9:15
Weight Loss: 14.41%

___

2/15 updated:

My batches #3, #4 & #5 had very similar flavor profiles, and they were vastly different from my batch #6. I compared [#3, #4 & #5] as SO espresso, and [#5, #6] as drip (Aeropress). For espresso, I randomly prepared (A & B) in the same IMS filter baskets and one additional A packed in different filter basket (stock double, VST 18g, triple basket) and extracted at slightly different temperatures (199-203F) to introduce extraction variations. If B could be correctly identified from the two As, A & B were significantly different. OTOH, if both A & B prepared from the same IMS baskets and extracted at the same temperature were perceived more similar to each other, and the third cup of A extracted differently was identified as the most different cup, A & B were not significantly different. The drip was simply a non-blinded AB test.

#3 (truncated development time) was significantly different from #4 BaseLine in flavor intensity.
Despite the very short development time, there was no perceivable roasting defect (underdevelopment) in #3. However, the flavor intensity in #3 was significantly weaker than baseline. The gap between the two batches was huge in early days (day 3-6) and the difference gradually reduced with longer rest time; updosing of #3 also improved. #4 always had longer and slightly smoother aftertaste.

#5 (lower drop temperature) had slightly different flavor profiles from #4 Baseline
#5 made delicious SO espresso and wonderful drip. It had slightly more acidity and herbal notes than baseline, whiles the baseline had slightly darker notes and occasional hint of chocolate in it. However, sometimes the minor variations in extraction parameters caused greater difference in the cup.

My analogy is that, my batches #4 & #5 were within the "strike zone". Specific bean needs corresponding roast plan, and by careful execution of the first two thirds of the roast (Dry & MAI phases), there will be broader strike zone in the development phase. There can be multiple hits (drop temperatures & times) within this strike zone to yield a generally good roast.

#6 (FSSF with very short development) was underdeveloped.
My batch #6 was dropped when the first crack was rolling. This was an uneven roast and there were more under-ripen tipping beans. Although both #3 & #6 both had very short development time, only #6 was "under"developed. The long Dry phase in #3 makes a difference.

I personally prefer the FSSF style for high grown washed beans, but the relatively short dry phase in the FSSF style roast may not be sufficient for this Sun Dried Ethiopian. In #6, I can taste the familiar flavor spectrum as in my other washed-Ethiopian FSSF-style roasts, and IMO huge delta (steeper declining RoR during the Dry-MAI phases) in #6 resulted in fuller base notes and smoother body.

However, two major defects (eccentric acidity and bitter-astringency) killed the cup. I attributed those major defects because the taste in the cup couldn't be improved by altering brewing parameters. Comparing to batch #5 which produced layered acidity, the acidity in batch #6 was stronger and more in the sour-end. In my understanding, the bitter-astringent aftertaste is usually said to be due to incomplete Drying, and this is the taste of chlorogenic acids. I won't claim myself knowledgeable for all of the chemical reactions in the incomplete drying prior to the Maillard rxn or the roasting theories about sufficient heat to penetrate the core, and therefore will simply stop here for now.

I did a significantly lengthened drying phase and an early drop with high air. THIS IS THE FIRST DP ETHIOPIA I'VE EVER ROASTED THAT TASTES LIKE IT SHOULD. This is a watershed moment for me.

Thank you all.

graph? What were your times in each phase?

jalpert wrote:I did a significantly lengthened drying phase and an early drop with high air. THIS IS THE FIRST DP ETHIOPIA I'VE EVER ROASTED THAT TASTES LIKE IT SHOULD. This is a watershed moment for me.

Thank you all.

Sweet!



(see what i did there...)

johnny4lsu wrote:graph? What were your times in each phase?

I had to travel for work today, will load the graph up tomorrow or Tuesday.

If I recall it was around 5/3.5/1.5, but not totally sure

jalpert wrote:I did a significantly lengthened drying phase and an early drop with high air. THIS IS THE FIRST DP ETHIOPIA I'VE EVER ROASTED THAT TASTES LIKE IT SHOULD. This is a watershed moment for me.

Thank you all.

Congrats you nailed it!

cimarronEric wrote:Sweet!



(see what i did there...)

Hi Eric,

I know it sounds like splitting the hair, but on the tedious technical side, when entering the first crack, how hard you hit the brake? What are your routine roast level and the development time for a DP Ethiopian?

In my batch #5 (post #61) the average RoR in the development phase was 8.1F/min on my BT readouts. However, in the first minute of the first crack I didn't hit the brake very hard and gave it some time to roll. My impression was that it needs some momentum to roll in the first half of the first crack. When I think of your earlier suggestion, I want to try stretching the development phase (probably into Rao's 20% window) whiles keep drop temperature as low as possible without stalling.

Still have some DP Ethiopian to play with...

Here's my roast of the GCBC DP Ethiopia:



Generally fairly slow drying period with a flattening ET, dropped at the end of first crack (still some snaps in the cooling tray). Don't pay attention to the ROR spike at the end of the roast, it is an artifact of Artisan's interpolation algorithm and not real.

Problems I typically have with DPs:

- Too-fast roasts cause severe underdevelopment
- Too-slow, or too-deep roasts cause dead flavor profiles

So I tried to solve this by doing a longer drying phase to promote even heating, but a short development period to promote retention of aromatics and acidity.

However, after the Dec/Jan R&L, I am now a very firm believer that the graph only shows half the story. Everyone knows you're supposed to dry slowly and use a short development period for DP Ethiopias, and it never worked for me like that. What I've since learned is that air flow is so, so important. I used very high air flow during the second half of MAI and through the crack, in order to promote development. But the high fan setting sucks cool air into the drum and torpedoes the ROR. So I did something I almost never do - I increased the gas pressure as the roast went on to keep the roast moving along.

The combination of an elongated drying period that brought ET and BT closer together than normal by 300F, high air flow/increased gas pressure later on, and short development time appear to have been the ticket.

jalpert wrote:Here's my roast of the GCBC DP Ethiopia:

<image>
... Don't pay attention to the ROR spike at the end of the roast, it is an artifact of Artisan's interpolation algorithm and not real.

...

The alog's do the same for mine, only in reverse. I end my roasts at around 4-5* ROR. The alog's when saved in graphs always make my profiles appear to crash into negative territory even when I don't crash below zero.

I think when Artisan v0.9.8 was released that started happening for me too (delta "blip" at end of roast that appears after pressing OFF). Don't remember it happening on the older versions.