Coffee | One Roast at a Time

Gene Café CBR-301 | Sample roast (≈50 g)

Why this roast exists

This roast wasn’t about flavour discovery or proving whether Geisha is “worth it”.

It exists to answer a simpler, more practical question:

Can the Gene Café CBR-301 roast a true sample-size batch in a controlled, meaningful way?

Professional sample roasters typically work with 30–60 g. Home roasters rarely test this range, and when they do, results are often dismissed as unreliable or “not real roasting”.

This roast was an attempt to test that assumption rather than repeat it.

Coffee & setup

  • Coffee: Costa Rica Geisha
  • Process: Fermented & washed
  • Batch size: 52.5 g in → 45.1 g out
  • Weight loss: ~14.1%
  • Set temperature: 250 °C
  • Stirring: Level 2
  • Fan: System-controlled only (no manual changes)

This roast intentionally used the same baseline settings as full-size roasts, with no compensations or tricks applied for the smaller batch.

What happened during the roast

At this very low mass, the roast progressed quickly, but not erratically.

The familiar system behaviour seen at 250 g and 300 g still appeared, just shifted in timing and temperature:

  • At ~4:20, the roaster triggered its over-temperature protection logic
  • The fan automatically increased from level 2 → 3
  • After temperatures stabilised, the fan returned to level 2 later in the roast

Crucially, this was not a failure or interruption, it was the same predictable airflow inflection already observed across many larger roasts.

The system behaved consistently, not defensively.

First crack & decision-making

First crack was clearly audible, with the app flag pressed at ~10:20.

Rather than treating this as a cue to chase development, it was used only as confirmation that the roast had crossed a structural threshold.

Cooling was initiated roughly 50–60 seconds later, based on:

  • aroma
  • bean colour
  • rate of progression
  • and prior weight-loss experience

The roast ended cleanly at ~11:10 total time.

Outcome & structure

Despite the small batch size, the roast landed exactly where intended:

  • Even colour
  • No scorching or tipping
  • No stall or runaway
  • No surface oil
  • No obvious under-development

At ~14.1% weight loss, this roast sits comfortably in the lower–mid development band that has repeatedly proven drinkable and balanced across other coffees.

The key point here is not the percentage itself, but that weight loss remained a reliable validation signal even at sample size.

What this roast proves

This roast demonstrates that:

  • The Gene Café’s thermal and airflow logic is batch-size agnostic
  • Over-temperature protection behaves predictably, not destructively
  • Time, aroma, colour, and weight loss remain valid anchors at ~50 g
  • Small batches can be roasted coherently, not just “heated until done”

In short:

The machine did not need special handling to produce a structurally sound result at sample scale.

That matters.

What this roast is not

This roast is not:

  • a flavour benchmark for Geisha
  • an argument for or against expensive green coffee
  • a replacement for professional sample roasting equipment
  • something to optimise or repeat endlessly

It is a system experiment, and it answered the question it was designed to ask.

Where this fits in the project

  • Logged as Sample Roast 1 (separate from the main roast sequence)
  • Treated as Notes-only
  • Not eligible for Lazy Mode
  • Not Guide material, yet

Its value is observational, not prescriptive.

Closing reflection

If earlier roasts explored the limits of capacity, airflow, and development at full batch sizes, this one explored the opposite edge of the system.

And at that edge, the Gene Café didn’t fall apart.

It behaved like the same machine, just faster.

That consistency, more than any tasting note, is the real result of this roast.