Roasts 84 to 91 ended up being a very important block.
Not because every roast was perfect.
And not because this suddenly turned the Gene Café CBR-301 into a machine with no uncertainty.
What this block gave me was something more useful than that.
It gave me confidence that, even with the heater currently fitted to this machine, I now understand the way this version of the roaster behaves well enough to roast consistently and make sensible decisions.
That matters.
Because by this point, my working understanding is that the heater currently in the machine is probably the wrong voltage, and that Gene Café are likely to replace it. I’m saying that carefully on purpose. This is my working understanding, not a final engineering statement. But practically, it means I’ve been roasting through a phase where the machine has not behaved like the one I had before the failure.
And that is really the story here.
Not panic.
Not crisis.
Calibration.
Two personalities already, maybe a third still to come
One of the things I’ve said repeatedly in this project is that you need to understand your own machine.
That has become very literal for me.
This Gene Café has effectively shown me at least two personalities already:
- the machine before failure
- the machine after repair
And if the heater is replaced again soon, there may well be a third.
But I do not think that makes the machine useless.
It just means the reference point can change.
That is the part I think people miss.
The process can still be stable.
The coffee can still be good.
The machine can still be consistent.
What changes is the need to understand the version of the machine actually sitting in front of you.
That is where Roasts 84 to 91 were so useful.
Roasts 84 and 85 — the difference between cold and warm stopped being theoretical
This block really started to settle down when I stopped treating warm state as background context and started treating it as part of the roast plan.
By this stage, the pattern had become clear enough that I later described it very simply: in this recent test block, cold starts clustered around 14.2 to 14.5% weight loss, while warm starts on the same coffee, with similar roast time and target temperature, clustered around 15.8 to 16.1%. That was the same rough setup producing meaningfully different results because the starting condition was different.
That is what Roasts 84 and 85 really helped make obvious.
They were not interesting because one was “right” and the other was “wrong.”
They were useful because they showed, very clearly, that the thermal state of this machine changes the outcome in a way that has to be accounted for.
Once I accepted that, the roasting got calmer.
Roasts 86 and 87 — the process still worked
The next useful thing in this run was that the logic kept holding even when the coffee changed.
That mattered to me.
Because it would have been easy to tell myself that the machine had become awkward enough that nothing reliable could be built on top of it. But that is not actually what the roasts showed.
What they showed was that I could still:
- repeat a standard process
- choose a roast direction
- validate the result afterwards
- and learn something useful from the outcome
That may not sound dramatic, but it is probably the most important thing a home roaster can establish.
Not that every roast is identical.
But that the process still makes sense.
Roasts 88 to 90 — one coffee, one evening, clear decisions
Roasts 88, 89, and 90 were especially useful because they were practical.
This was not abstract testing for the sake of it.
It was one coffee, one evening, with different intended outcomes:
- one leaning more toward pour-over
- two leaning more toward espresso
That created a very honest test.
Roast 88 was the cold-start roast.
Roasts 89 and 90 were warm-start roasts.
And that split matters, because by this stage I was no longer pretending those were interchangeable.
Roast 88 landed as a good, balanced roast, but slightly fuller than I had originally pictured for a clearly separate filter batch.
Roast 89 then showed what the warm machine was capable of when left to carry more momentum.
Roast 90 was the most useful of the three from a calibration point of view, because it was the correction roast. The first warm espresso roast ran a little high, the next one was shortened, and the result moved back in the direction I wanted.
That was important.
It showed that the current TRT model was not magic, but it was useful.
That is all I really need from a working method.
Not perfection.
Just useful control.
Roast 91 — cold-start Colombian for espresso, exactly the sort of result I wanted
Roast 91 felt like a quiet confirmation that this current heater may not be the heater I want long term, but it is now a heater I understand.
This was a 250g cold-start roast of Colombia Supremo (Washed). It started at 20°C, finished at roughly 13:21, and produced 210.7g out from 250g in, which is 15.7% weight loss.
That is exactly the kind of result I wanted for a cold-start Colombian with espresso intent.
It was not a flashy roast.
It was not experimental.
It was not trying to prove anything dramatic.
It was simply a roast built around a stable process, with a clear target in mind, and it landed in a very believable espresso-friendly zone. The app log shows the same basic pattern I’ve been seeing on this heater: a normal climb through the roast, a move into the upper band, a fan jump to 3 around the hotter phase, and then a return back to 2 later on before cooling.
That sort of roast gives me confidence.
Not because it proves I have solved the machine forever.
But because it proves that this current version of the machine can be used well.
What this block really taught me
The biggest lesson from Roasts 84 to 91 is not that the heater issue no longer matters.
It does matter.
If Gene Café replace the heater and the machine shifts back toward its earlier behaviour, I will need to recalibrate again.
But the more important lesson is that none of this is actually a disaster if the process is right.
That is the bit I care about most now.
Because the CBR-301 is still capable of producing consistent coffee when:
- I understand the current state of the machine
- I keep the setup stable
- I understand the beans as well as I can
- I choose my roast level from what I actually want in the cup
- and I use TRT plus weight loss to validate where the roast really landed
That is the method.
Not because TRT tells me everything.
And not because weight loss tells me everything.
But because together they give me a very practical, repeatable way to work.
In the recent control video, I described the model I’m working with as a simpler one: hot air, rotating glass chamber, target temperature treated more like a ceiling, thermal state built into the plan, total roast time set deliberately, weight loss measured afterwards, and the cup used as final judgement.
That is exactly what this roast block reinforced.
Where I am now
So where I’ve landed after Roasts 84 to 91 is fairly simple.
I still believe the heater currently fitted is probably not the correct long-term part for this machine.
I still expect that to change.
But I also now believe something else with more confidence than before:
I understand this heater in this roaster.
And that is enough to keep roasting well while the next part of the story plays out.
Because the real goal here was never to pretend the machine was flawless.
It was to understand it honestly.
One roast at a time.