These two roasts were planned as a useful comparison pair.
Same coffee.
Same machine.
Same batch size.
Same target temperature.
Different start state.
The coffee was REDBER Colombia Swiss Water Decaf, roasted on the Gene Café CBR-301 with the restored heater setup I have been recalibrating around.
The aim was simple enough:
Can I move this decaf closer to a sweet, balanced Fellow Aiden filter roast?
I was not trying to roast it dark. I was not trying to push it into espresso territory. I wanted a clean, drinkable, rounded decaf that would work well in the Aiden.
The target was roughly:
14.2–14.8% weight loss
That felt like a sensible step up from my earlier decaf reference, Roast 103, which landed at 13.1% and looked safe but probably low-edge for the result I wanted.
What actually happened was more interesting.
Both roasts stayed light.
Not visually light necessarily — decaf can be awkward that way — but light by weight loss.
Roast 107 landed at 13.5%.
Roast 108 landed at 13.6%.
That was not the result I expected.
And that is why the pair is useful.
Roast 107 — cold start
Coffee: REDBER Colombia SWP Decaf
Batch: 250 g
Output: 216.2 g
Weight loss: 13.5%
Start temperature: 21°C
Target temperature: 250°C
Cooling / endpoint: 11:50
Crack markers: none recorded
This was the cold-start leg of the pair.
The plan was to extend beyond the earlier Roast 103 decaf reference, but not push too hard. During the roast, it looked as if it might be moving slowly early on, then catching up.
The JSON later showed a slightly different picture.
The key timings were:
Marker Time
200°C 3:58
230°C 6:00
240°C 7:05
245°C 7:37
249°C 8:05
250°C 8:13
End 11:50
The important correction is that, live, I thought 245°C had arrived earlier than it actually had.
The JSON showed first 245°C at 7:37.
That matters because the roast did not spend quite as long in the upper band as it felt like during the roast.
Time after first 245°C was about:
4 minutes 13 seconds
Time after first 250°C was about:
3 minutes 37 seconds
That sounds like a decent amount of upper-band time, but the final weight loss was still only 13.5%.
So this was controlled, but still lower than the intended Aiden target.
Roast 108 — warm start
The second roast was the warm-start comparison.
Coffee: REDBER Colombia SWP Decaf
Batch: 250 g
Output: 215.9 g
Weight loss: 13.6%
Start temperature: about 61–62°C
Target temperature: 250°C
Cooling / endpoint: 11:01
Crack markers: none recorded
This one moved much faster.
The key timings were:
Marker Time
200°C 2:55
230°C 4:48 / 4:49
240°C 5:54
245°C 6:26
249°C 6:56
250°C 7:04
End 11:01
At the time, this looked like a roast with real warm-start momentum.
And it was.
Compared with Roast 107, it reached the main milestones roughly a minute earlier.
But because I also stopped it earlier, the final weight loss barely moved.
Roast 107 was 13.5%.
Roast 108 was 13.6%.
That is the lesson.
A faster warm-start curve does not automatically mean a more developed final roast if the total roast time is shortened enough.
Cold vs warm comparison
Detail Roast 107 Roast 108
Start state Cold Warm
Start temperature 21°C 61–62°C
Input 250 g 250 g
Output 216.2 g 215.9 g
Weight loss 13.5% 13.6%
200°C 3:58 2:55
230°C 6:00 4:48
245°C 7:37 6:26
250°C 8:13 7:04
Endpoint 11:50 11:01
This is the part I find most useful.
Roast 108 was clearly faster through the machine-temperature milestones.
But the roast ended 49 seconds earlier.
The extra warm-start momentum mostly cancelled out against the shorter total roast time.
The result was almost the same weight loss.
What happened
My working interpretation is this:
This specific Swiss Water decaf did not shed weight as aggressively as I expected.
That matters because I had been carrying lessons from recent Guatemala and Nicaragua roasts, where the upper temperature band seemed to push weight loss quite hard.
With those coffees, staying in the 245–250°C region too long could move the roast further than expected.
This decaf behaved differently.
Even with Roast 108 reaching 245°C at 6:26 and 250°C at 7:04, it still only reached 13.6% weight loss by 11:01.
So the warning is not:
Decaf always needs to be roasted shorter.
The better warning is:
Decaf can look and behave differently, but the actual result still needs to be checked with weight loss and the cup.
That is less tidy. But it is more useful.
First crack and decaf
There were no crack markers recorded in either roast.
That is not unusual for this project.
On the Gene Café CBR-301, first crack can be difficult to use consistently. Some coffees are clear. Others are quiet, scattered, or hidden under machine noise.
Decaf adds another layer of uncertainty because it can look darker earlier and may not behave like a regular washed or natural coffee.
For these two roasts, I would not use first crack as the main control point.
The useful anchors were:
start state
total roast time
time to 200°C
time to 230°C
time to 245°C
time to 250°C
final weight loss
eventual cup result
Crack, if heard, would have been context only.
What I expected
The original plan was to move the decaf toward a more rounded Fellow Aiden result.
I expected the cold roast to land somewhere closer to the mid-14% range.
I expected the warm roast, because it started at around 62°C and moved quickly, to land at least slightly higher again.
That did not happen.
Instead, both roasts stayed close together:
13.5% and 13.6%
That does not make them failed roasts.
It makes them good reference roasts.
They tell me the next step.
What I would do next
For this specific coffee, on this current machine state, I would now extend the timing.
For a future cold-start REDBER Colombia SWP Decaf roast, I would likely try:
12:15–12:25
For a future warm-start roast around 60–65°C, I would likely try:
11:20–11:35
That is not a universal decaf rule.
It is just the next sensible step for this coffee, this machine, and this Aiden/filter target.
The cup still decides.
If these two roasts drink beautifully, I may not need to push much further. But based on weight loss alone, they look like lower-edge decaf references rather than the final target.
What this means for the Range Finder idea
This pair is useful for the future Range Finder logic.
It shows why the tool should not simply say:
warm start equals much shorter roast
That is sometimes true.
But it depends on the coffee.
It also shows why decaf needs careful treatment.
A Swiss Water decaf may look darker, crack less clearly, and move quickly through machine-temperature milestones, but still not land at the intended weight loss.
So the Range Finder needs to keep doing what this whole project is trying to do:
compare like with like
separate cold and warm starts
use weight loss as validation
avoid pretending the graph is bean temperature
avoid treating one coffee’s behaviour as universal
let the cup result close the loop
This pair helps the model, even though it did not hit the intended target.
Actually, that is probably why it helps.
Working conclusion
Roasts 107 and 108 were not the optimised Aiden decaf endpoint I was looking for.
They were both controlled.
They were both safe.
They were both probably on the lighter side for the intended result.
The warm-start roast moved much faster through the temperature milestones, but because it was stopped earlier, it ended almost exactly where the cold roast did by weight loss.
That is the useful lesson:
Faster milestones do not automatically mean a more developed roast. Total roast time and weight loss still have to be read together.
For this REDBER Colombia Swiss Water Decaf, my next test would be slightly longer.
Not dramatically longer.
Just enough to see whether the cup gains sweetness, body, and balance in the Fellow Aiden.
For now, Roasts 107 and 108 become a clean paired reference:
cold start: 11:50, 13.5%
warm start: 11:01, 13.6%
A tidy reminder that coffee keeps refusing to behave like a spreadsheet.
Which is annoying.
And also exactly why this is interesting.