Introduction

This note captures what changed after replacing the heater in the Gene Café CBR-301, and how the system was recalibrated using real roasts.

It is not a theory piece.

It is based on:

  • Roast 70–79
  • Controlled use of a single variable (TRT)
  • Weight loss as the validation output

The goal is simple:

understand how the machine behaves now, not how it behaved before


What Triggered Recalibration

The first roasts after the repair appeared to confirm a problem.

Early results (Brazil)

  • 13:05 → 12.5% weight loss
  • 12:35 (warm) → 13.7% weight loss

These are materially lower than expected.

At this point, it looked like the machine had become underpowered.

That interpretation was incomplete.


What the Data Actually Showed

The recalibration sequence moved from Brazil to Honduras and then to a washed Colombian (Regional Jess).

Honduras (Roasts 72–75)

  • 15.5% → 16.5% → 14.6% → 15.7%

These roasts demonstrated something important:

the machine is still capable of reaching full development

This ruled out a simple “loss of power” explanation.


The Clean Signal, Regional Jess

The most useful data came from repeated warm roasts on a single washed Colombian.

Key roasts

  • 12:55 → 11.8%
  • 13:20 → 16.0%
  • 13:30 → ~16.5–17%

This is a very strong response curve.

A small change in TRT produced a large and predictable change in weight loss.


The Core Insight

The heater replacement did not break the system.

It changed how clearly the system expresses the coffee.

Before repair

  • Behaviour was more buffered
  • Different coffees produced similar outcomes

After repair

  • Behaviour is more transparent
  • Coffee characteristics show up more clearly

What Actually Changed

The relationship between TRT and weight loss is still valid.

However:

it is now more sensitive to both time and coffee type

This shows up in two ways:

  1. Small changes in TRT have a larger effect
  2. Different coffees separate more clearly in behaviour

The Fan Behaviour

The change in fan step timing was real and visible in the data.

Before repair

  • Fan step ~4 minutes
  • Around 200–210°C
  • Useful as a phase marker

After repair

  • Fan step occurs much later
  • At higher temperatures
  • Sometimes near the end of the roast

The correct interpretation is:

fan behaviour reflects system heat, it does not control it

It is now a diagnostic signal, not an anchor.


The Working Model After Recalibration

The roasting approach has not changed.

TRT remains the only control variable

Fan, temperature adjustments, and drum speed are not used as active controls.

The system is:

  • set TRT
  • run the roast
  • validate with weight loss

New Anchors

Warm, washed coffees (Honduras, Regional Jess)

  • 13:20 → ~16% weight loss (centre)
  • 13:30 → ~16.5–17% (upper bound)

This is the primary working range.


Cold, washed coffees

Derived from observed offsets:

  • ~13:40–13:50 → ~15.5–16.5%

Naturals (Brazil)

Naturals now behave differently to washed coffees.

  • Lower weight loss at the same TRT
  • Require slightly longer roast times

Approximate adjustment:

  • +10–20 seconds vs washed

Decaf

Decaf remains a separate system.

  • Faster early behaviour
  • Lower overall weight loss

Typical working range:

  • ~12:30–12:50 (warm)

What This Means

The system is now:

  • stable
  • repeatable
  • more sensitive

The key change is not capability.

It is visibility.

the machine now shows more of what the coffee is doing


Final Position

The recalibration confirms three things:

  1. TRT remains a valid and sufficient control variable
  2. Weight loss remains the most reliable validation signal
  3. Coffee type must now be accounted for more explicitly

Summary

TRT-only roasting remains valid

The machine is not weaker

The system is more responsive

Calibration required ~+20 seconds for washed coffees

Fan behaviour is no longer a control signal