Context
These four roasts were part of a focused attempt to understand what the controls on the Gene Café CBR-301 actually do in practice.
The goal was not to optimise flavour, but to isolate behaviour:
- drum speed (minimum vs maximum)
- mid-roast temperature adjustment
- and the effect of starting the roast cold vs warm
All roasts were:
- 250 g batch size
- target temperature 250°C baseline
- total roast time ~13:00
- same coffee (Honduras, washed)
- same setup and cooling approach
What was done
- Roast 80 (cold) — drum speed set to minimum for full roast
- Roast 82 (cold) — drum speed set to maximum for full roast
- Roast 81 (warm) — baseline behaviour (no clean mid-roast intervention)
- Roast 83 (warm) — temperature reduced to 240°C at 06:00, returned to 250°C at 08:00
What was observed
Across all four roasts:
- Changing drum speed altered how quickly the roast progressed, with higher drum speed consistently reaching milestones later
- Mid-roast temperature adjustment also altered timing, delaying progression and shifting later-stage behaviour
- In both cases, the machine’s internal behaviour (including fan transitions) shifted alongside those changes
However:
- Final weight loss remained effectively unchanged in both tests
- Drum speed: 14.3% → 14.5%
- Temperature step: 15.8% → 16.0%
In contrast:
- The difference between cold and warm starts resulted in a much larger shift
- ~14.4% (cold average) → ~15.9% (warm average)
- Warm starts reached all key milestones significantly earlier
Interpretation
The controls on the CBR-301 do have an effect.
They:
- change the timing of the roast
- shift when key events occur
- influence how the machine behaves during the roast
But within this test:
- those changes did not materially alter the final outcome when total roast time was held constant
The machine responded to input, but still converged toward a similar end point.
What appears to matter more
Within this dataset, the most significant variable was:
- starting thermal state (cold vs warm)
Compared to:
- drum speed adjustments
- mid-roast temperature changes
The starting condition had a substantially larger impact on both:
- roast progression
- final development (weight loss)
Where this leaves things
At this point, the working understanding is:
- user inputs influence the roast
- but that influence appears bounded within the system
- and outcome remains strongly anchored to time and starting state
Further testing is needed, particularly around:
- fan behaviour as an isolated variable
- whether longer or more aggressive interventions change outcomes
- and how these effects translate into the cup
Notes
This is not a conclusion about “control” in general.
It is a record of what happened under these specific conditions.
As with all roasts in this project, this will either hold — or be revised — as more evidence builds.