Roast Planner

A roast plan is three durations laid end to end: drying + Maillard + development = the total roast. Cultivar is the one input that sets a time you cannot negotiate — when first crack lands. The seed's genetics fix how much sugar and acid it carries and how heat gets into it, so each cultivar browns and builds its aromatics on its own schedule. A Bourbon rushed to a 7:30 first crack tastes thin and papery no matter how well you handle the rest; a Gesha dragged to 9:00 goes savory and loses the florals you paid for.

That is why the FC anchor below feeds Time Ledger line 1 and everything else in the plan is a correction of seconds around it. Treat the cultivar on the bag as your best hypothesis, not a fact — labels are wrong often enough that the "After the Roast" section exists to correct them.

The drum has one heat setting and one clock — you cannot give 55% of the mass a different schedule than the other 45%. Times are additive, so a weighted average of two first-crack anchors is another valid time on the same clock. An average of two temperatures means nothing: first crack is a property of the chemistry, not a dial you set.

If one component is 60% or more of the batch, roast the whole batch as that coffee and let the rest ride along. If no component dominates, weight each anchor by its share. And if components disagree by more than about a minute at first crack, split-roast and blend after — one compromise profile serves neither half.

Processing changed the chemistry that browning starts from. Naturals and dark honeys keep far more free fructose, and fructose browns faster than sucrose — the same profile lands visibly darker. That is why natural and honey lots take 15–30 seconds less development and want a slightly earlier drop, and why processing never touches the first-crack anchor. Change development and drop, hold everything else, and the experiment stays readable.

Roast level is the one input that is purely your decision — and development is a bell curve, not a slider. Shorter development pushes acidity up, longer pulls it down; but too short reads sharp and astringent with no fruit behind it, and too long goes flat and dull. The pleasant region is in the middle and narrower than the acidity curve suggests, which is why the fix is always a 15-second step with a cupping in between, never a big swing.

Moisture, density and screen size change how much energy the plan costs, not its shape. Wet greens (over 11.5%) stall as they near 100 °C while water boils off — expect the flat spot and don't panic-add heat. Dense, high-grown seed tolerates a hard early push; soft, low-grown seed gives you defects instead of speed. Very large beans scorch outside while the core lags — the safe way to go faster is more airflow at a lower inlet temperature, never more burner.

None of these auto-adjust the ledger, and that is deliberate: they inform the ±0:15 judgment in Net correction and your heat/fan choices, nothing more. Don't have lab numbers? Leave moisture and density blank — record only evidence you actually have.

A bean that is still gaining heat, but gaining it more slowly, is moving through its reactions in order. Two shapes break that. The crash: RoR falls off a cliff just after first crack, when the exotherm ends and an oversized heat cut lands at the same moment. The bake: RoR flattens toward zero, the bean sits at temperature without progressing, and the cup goes flat, papery and sweetness-free — nothing after the drop rescues a bake.

So the rule is: decline steadily from the post-turning-point peak, through small frequent heat cuts rather than a few big ones, and stay ahead of the machine's roughly 60-second lag — what you change now shows up a minute from now, so cut before the exotherm, gently. If a phase would need a rate outside the proven bands to hit its milestone, the plan is not reachable — fix the ledger, not the roaster.