I had planned a simple 300g consistency test on the Gene Café CBR-301.
The idea was to roast three batches back-to-back:
- Roast 112 — cold start
- Roast 113 — warm start
- Roast 114 — warm start
Same coffee. Same batch size. Same machine. Same broad target.
The goal was not to prove perfect repeatability. It was to see whether I could manage three full-capacity 300g roasts close enough to each other to be useful.
That was the plan.
Roast 112 had other ideas.
The coffee
This was a Central Kenya washed coffee.
- Origin: Central Kenya
- Altitude: 1,600–1,900 masl
- Varieties: SL28, SL34, Ruiru 11, Batian
- Process: Washed, double fermentation
- Moisture: around 10–11%
- Batch size: 300g
- Target temperature: 250°C
This was not a gentle, low-energy coffee. It was a high-altitude Kenyan coffee with plenty of structure, and it moved much faster than I expected.
What happened
The roast was meant to be the cold-start anchor for the test, but the early checkpoints came in fast.
- 200°C: 4:10
- 230°C: 6:43
- 245°C: 8:20
- First crack: clearly audible around 10:16
At 245°C by 8:20, the original longer roast plan was no longer sensible. I shortened the roast heavily and aimed to cool around two minutes after first crack.
Even with that adjustment, the roast went much further than intended.
Final result
- Input: 300g
- Output: 243.6g
- Weight loss: 18.8%
The beans expanded dramatically, smoke became heavy, the colour was uneven, and some beans were much darker than intended.
Then a chamber part came loose: CR73-005A, the blade cushion.
At that point, the decision was simple.
Stop the test.
Do not run Roast 113. Do not run Roast 114. Inspect the machine, reset, and treat Roast 112 as a failed first attempt rather than a usable consistency anchor.
Why I stopped
This is the main lesson from the roast.
The planned video was meant to be a three-roast consistency test, but after Roast 112 I no longer had a useful baseline.
I had:
- an overdeveloped roast
- uneven colour
- heavy smoke
- 18.8% weight loss
- a blade cushion detached from the chamber
That is not a reference roast.
That is a warning.
It does not mean 300g is impossible on the Gene Café CBR-301. I have roasted 300g before.
It does not mean the machine cannot roast larger batches.
It does not mean Kenyan coffee is the problem.
The narrower lesson is this:
This specific 300g Kenyan roast moved much faster than expected, and once the physical signs became questionable, the sensible decision was to stop the sequence.
What I’ll do next
Before attempting this again, I need to:
- replace and check the blade cushion
- inspect the chamber
- make sure the drum is rotating cleanly
- clean the chaff path and exhaust
- treat Roast 112 as a failed test roast, not a reference roast
If I return to this coffee, I will probably step back first.
That may mean testing it at 250g before trying 300g again.
Or, if I go straight back to 300g, I will use a much shorter planned endpoint.
Not 13 minutes plus.
Not even close.
Roast 112 taught me that this coffee can move fast, and the next plan needs to respect that.
Download the JSON roast file
For anyone interested in the specific Gene Café JSON profile from this roast, I have made the zipped JSON available here:
Download Roast 112 JSON ZIP:
Download the Roast 112 zipped JSON file
Please note: the Gene Café app did not start cleanly at the very beginning of this roast, so the JSON should be treated as useful but not perfect. It still captures the main shape of the roast and the later machine behaviour, but it is not a clean full-session reference file.
Final thought
This roast did not go where I expected.
The title I wanted was probably:
Can I repeat three 300g roasts on the Gene Café CBR-301?
The title I got was:
Gene Café CBR-301 300g Roast Test Went Wrong
And that is fine.
Because it may be more useful.
Roast 112 was not a success as a consistency anchor, but it was a success as a reminder:
- Watch the machine.
- Watch the coffee.
- Do not worship the plan.
- And when the roast gives you enough warning signs, stop.