Roasting
Roast review and compare
A roast's numbers in one read, and two or more roasts compared phase by phase.
After a roast you want to know how it went: when first crack came, how long development ran, whether the rate of rise crashed or flicked, how closely the beans followed the profile and how hard the heater worked. The roast analysis reads all of that from the recorded roast in one go, and judges the numbers against what the profile, the roast level or your own earlier roasts of the profile lead you to expect. Comparing roasts puts two or more side by side, each number with its difference from a reference roast.
In the dashboard
- One roast: History → the roast. The page shows the marks, the phases, development, weight loss, AUC, the RoR slope from first crack to drop (statistics) and how its numbers sit against the expected ranges.
- Several roasts: in History tick two or more roasts and press Compare. The Compare roasts page draws their curves on one chart (Align on CHARGE, turning point, dry end, first crack or drop), with a Reference to measure against and a Consistency band. It shows the first ten roasts you picked.
- On paper: Compare report on the Compare page, or Reports → Compare two roasts: the reference as A, the other roast as B, both curves on one chart and every number as A, B and B − A (reports).
What a roast read gives
All times are from CHARGE, temperatures in °C, RoR in °C/min.
| Part | What it says |
|---|---|
| Marks | CHARGE, turning point, dry end, first crack and its end, second crack, DROP: the time, BT, ET, RoR, heater and fan at each |
| Totals | BT at CHARGE, roast time, development time and DTR, weight loss, the highest BT, AUC |
| Phases | drying, Maillard and development: how long, their share of the roast, the BT gained, the mean RoR |
| RoR | the early peak, RoR at each minute, the lowest RoR late in the roast, RoR at DROP, and the slope from first crack to DROP |
| Against the profile | BT − profile at 0:30, 1:00, 1:30 and 2:00, the largest overshoot in the first two minutes, how closely BT tracked from the turning point to DROP (RMS, bias, largest gap), first crack and DROP against the profile's targets |
| Heater and fan | the heater in the first 30 s, its mean and highest, time at 90 % or more and at 100 %, how busy it was, heater and fan over the last minute, BT against the setpoint |
| Expected | weight loss, roast time, first crack, development, drop BT and ground colour, each within, below or above its range (roast guidance) |
| Notes | a plain line for each number outside its range; "Little heater left late" when the heater averaged 90 % or more over the last minute; BT 5 °C or more above the profile in the first two minutes |
A roast compared with others adds about 25 of these numbers in rows, each with its difference from the reference, and from when each roast's BT stays within a set distance of the reference's (0.5 °C unless you change it): how closely it followed.
From the command line
roast review latest
roast review '#12' --vs '#11'
roast compare '#10' '#11' '#12' --ref '#11' --within 1 --curve
roast report compare '#12' '#13'A roast is its id, latest, or #N, its batch number (add --roaster when two roasters use
the same number). roast compare takes 2 to 8 roasts; --curve adds BT, RoR, heater and fan
every 30 s. Add --json for the whole answer. For a bean with fewer than three roasts on the
profile, roast review also says what to change next time (a new bean).
Limits
The analysis reads; it changes nothing. The dashboard still works out its own phases, AUC and
compare figures, so it can differ from roast review in small ways: its AUC follows Settings →
Roasting → Area under the curve.