Terms used throughout this documentation. Definitions are given as they apply inside TilauScope, which is not always the widest possible sense of the word.
!!! note This glossary grows with the documentation. Every chapter that introduces a term defines it here and links to it on first use.
The moment the green coffee is dropped into the drum. It is the origin of the roast clock: every phase duration and every milestone time is counted from it.
The lowest bean temperature of the roast, reached shortly after charge, when the beans stop absorbing heat faster than the machine supplies it and start climbing again. A rising RoR after TP is normal recovery, not a fault.
The first phase, from CHARGE to dry end. Free moisture leaves the bean. Little colour develops.
The end of the drying phase. In TilauScope it is a marked milestone, detected and proposed for confirmation, and one of the anchors the roast plan is built around.
The middle phase, from dry end to first crack. Sugars and amino acids react and produce most of the roast’s aromatic complexity. The phase where heat reductions are staged.
These reactions absorb heat rather than release it, so the burner is feeding them, not merely warming the bean. Cutting the heat here does not slow the reaction gently — it starves it, and the bean temperature keeps climbing on the drum’s own heat while the chemistry stalls. This is why the plan holds the heat through Maillard instead of stepping it down early, and why it is a phase to judge by the settings held rather than by the curve. See baked.
The audible cracking as bean structure fails under internal pressure. It marks the start of development and is the single most important milestone for reproducibility — which is why TilauScope learns its temperature from your own roasts.
The final phase, from first crack to DROP. Determines much of the cup’s balance and body. Measured relative to the whole roast as DTR.
The moment the beans leave the drum for the cooling tray. The end of the roast.
Roasting a second batch immediately after the first, while the machine is still hot. It needs different handling from a cold start, which is why TilauScope has a dedicated mode for it.
The probe reading closest to the beans themselves. Every temperature in this documentation is a bean temperature unless stated otherwise.
The probe reading of the air or the drum environment. On some machines — radiant ones in particular — ET can sit below BT, which is normal for that machine type and not a fault. A machine built without an air probe has no ET: once it is selected as your roaster, TilauScope stops showing ET and words its advice on bean temperature alone.
When several roasts are compared, the strip under the curves measuring each one against the roast holding the chart, from charge to each roast’s own drop. The reference is the flat line through the middle; a roast running warmer is above it, cooler below. It has a scale of its own, so a difference of a few degrees is visible instead of being lost against the temperature axis.
How fast bean temperature is climbing, in degrees per minute. The slope of the curve rather than its height. A roast is steered largely by keeping RoR inside a sensible band and falling smoothly.
The smooth curve a roast’s own bean temperature is fitted to — a parabola, a cubic or a logarithm — used as a yardstick by the profile analysis. Its slope is a rate of rise with every wobble removed, so the gap between it and the measured RoR shows where the roast sped up or slowed down. Being close to it describes an even curve, not a good cup.
The line Artisan draws ahead of the latest reading, showing where bean temperature is heading if it keeps its current trend. Artisan draws it only once the curve is long enough to extend.
What Artisan reads bean and environment temperature from: a roaster’s own link, a thermometer, a data logger. One meter is set at a time, in Devices.
An open thermocouple board, common on home roasters and conversions, that sends bean and environment temperature to Artisan over a USB cable. In Devices it is chosen instead of a meter, with the board channel each probe is wired to.
A source of readings beyond bean and environment temperature — an airflow extractor, an ambient probe, a crack counter, a roaster’s own burner and airflow levels. Each provides up to two readings, recorded with the roast and shown in the extra counters row when chosen. Set in Devices.
The share of total roast time spent in development, as a percentage. A common target sits somewhere between the high teens and low twenties, depending on the coffee and the intended cup.
How much the bean temperature climbed between first crack and the drop, in degrees. Where DTR measures development in time, the development rise measures it in heat: two roasts can share a DTR and still finish very differently if one gained twice as many degrees over that stretch.
The weight of green coffee going into the drum. It sets the whole scale of the roast: phase durations, how much heat the batch can absorb, and what counts as a comparable previous roast when TilauScope looks at your history.
A grouping of coffee varieties that behave alike in the drum, after the roasting research of Rob Hoos: Typica and Ethiopian varieties generally want a quicker roast, Bourbon ones a slower progression, and a hybrid is read through whichever parent is not the Timor Hybrid — a Catimor is paced like the Caturra in it. TilauScope reads the family from the variety on the coffee’s record and uses it to nudge the charge temperature.
It is a starting point, not a fact. Coffee cross-pollinates, nurseries mislabel, and varieties are often identified by eye, so a lot sold as one variety is rarely purely that one. TilauScope therefore keeps the nudge small, halves it for a blend, and drops it entirely for the many varieties it does not place — saying not known rather than guessing. It also stays silent when a record names varieties from two different families, such as a Bourbon and a Gesha: the two point opposite ways, and there is no honest way to pick one.
A roast meant to work both as filter and as espresso, given a development time between the two. Filter is given the shortest development, espresso the longest, and omni sits in the middle.
The weight the batch lost during roasting, as a percentage of its charge weight. It is two losses in one: the water the green coffee carried, which leaves whatever you do, plus the matter burnt off once the roast develops.
That is why the target TilauScope shows you is not a single figure per roast level. It starts from the lot’s own moisture content — a drier coffee has less to lose, so it should end lighter on the scale — and adds what the roast level and the development burn off on top. Two roasts at the same level, one held forty seconds longer after first crack, do not lose the same weight; the longer one loses about a point and a half more.
The target is a bearing, not a verdict. On a home-sized batch one point of weight loss is only a few grams — about what the chaff and the handling account for — so TilauScope only speaks up when the roast lands well clear of it.
The water still held in green coffee, as a percentage of its weight — how much water there is. It is the mass the roast has to heat and evaporate, so higher moisture means more power at the start and a coffee that resists heat for longer; it barely changes how long drying lasts. Distinct from water activity, which says how freely that water leaves.
How much the beans weigh for a given volume. Dense beans conduct heat inward more slowly and need sustained heat; light, low-density beans take heat faster and are easier to scorch. The plan acts on it only when it is clearly one or the other — an ordinary reading in the middle of the range is treated as no information, because a supplier’s figure cannot reliably separate two ordinary coffees. Measure it by water displacement rather than by filling a dry measure, which counts the air between the beans as if it were coffee.
The size of the green beans, given by the sieve they are graded on — a screen number in 64ths of an inch (17/18 = large, 15/16 = medium) or a grade name such as AA, Supremo, AB or Excelso. Peaberry (PB) is a single round bean where the cherry usually holds two flat ones. Smaller beans respond faster to each change of heat; size alone does not make a roast slower or faster.
A numeric scale for roast colour: the lower the number, the darker the roast. TilauScope uses it as the target you aim for, and — where a colour reader is fitted — as the measurement you compare against.
A cupping score out of 100 following the Specialty Coffee Association protocol, used as the standard reference for a coffee’s quality independent of any particular roast.
How much of the water in green coffee is free rather than bound, on a 0 to 1 scale — how readily it leaves, not how much of it there is. It predicts how well a bag will keep far better than moisture content alone, and in a roast plan it sets the airflow that clears the steam.
A network sensor’s last published value, kept by the message broker and handed to TilauScope the moment it starts listening. Without it a channel stays empty until the sensor next speaks of its own accord, which on a home automation network can take minutes.
The encryption a message broker can require on the link, so that readings and the password used to obtain them do not travel in clear text. The broker proves its identity with a certificate, which must come from a recognised authority: a certificate the broker issued to itself is refused, and the connection simply fails.
The idle time after which a quiet connection to the broker is checked. Short values notice a broker that has gone away sooner, at the cost of talking to it more often.
Asking a network sensor for a reading instead of waiting for one. Some sensors report only on their own schedule, far too slowly to follow a roast; polling requests a fresh value at a chosen interval. Only sensors on mains power can be polled — a battery sensor sleeps between its own reports and cannot be reached in between.
Power, in watts (W) or kilowatts (kW), is how much electricity a device draws at one instant. Energy, in kilowatt-hours (kWh), is power accumulated over time: 1 000 W for one hour is 1 kWh, 1 000 W for ten minutes about 0.167 kWh. A watt figure says how hard the machine is working now; a kWh figure says what a period cost. “W/h” is not a unit of either.
Where an energy figure comes from, stated beside it:
Artisan’s description of what draws energy on a roaster: up to four heaters or motors, each with its power, whether it runs on electricity or gas, and the slider that drives it. Without a meter, TilauScope estimates a roaster’s energy from the electric heater Artisan rates for it when neither its own machine database nor the Heater power set for the roaster gives a figure.
The energy of the roast, CHARGE to DROP, divided by the weight of green coffee charged. It lets batches of different sizes be compared. It is only given when the roast is fully covered by every expected source and the green weight is known.
How fast the environmental temperature is climbing — the machine’s own rate of rise, as opposed to the beans’. It moves within seconds of a burner change, while the beans take far longer to register it, so it is the earliest visible sign that a correction has taken effect. It says nothing about how the roast is going: only that the machine heard you.
How much heat the machine’s own metal stores. A high-thermal-mass drum keeps delivering heat after the power is reduced, so reductions must be made earlier to land on time.
Heating by infrared radiation rather than by hot air. Radiant machines transfer heat to the beans very directly, which changes both how fast they respond and how their two probes relate to each other.
An alarm held back by another alarm. With a guard, the rule waits for the alarm it names to act before it becomes live; the guard can also be negative, meaning the rule applies only for as long as the alarm it names has not acted. A rule that starts on a guard measures its sensor threshold as a change from the instant the guard acted, not as an absolute temperature.
A control loop that holds a temperature at a target by continuously adjusting power. TilauScope uses one to bring the machine to its preheat target.
Running a candidate model beside the active controller using the same measurements, while giving it no authority over the heater. Its predictions can therefore be checked on the real machine before it is allowed to contribute even a bounded fallback setting.
The temperature a PID is aiming for.
Sailing past the target temperature because the machine’s stored heat keeps arriving after the power has been cut. The higher the thermal mass, the more of it there is, and the earlier the power has to come off to avoid it.
How much TilauScope’s preheat control has learned about heating to a given setpoint on your machine — shown as a four-step reading (Learning, Estimated, Tuned, Calibrated) next to the preheat countdown. It reflects real preheats recorded at that exact setpoint, not a fixed property of the machine, and it can read differently for holding power and for knowing when to ease off — one can be learned before the other.
The highest Experience reading: enough real preheats have been recorded at this exact setpoint that the holding power TilauScope applies is a settled result, not an estimate carried over from a nearby setpoint or from the machine’s general thermal behaviour.
Following a previously loaded background curve live, from CHARGE onward, moving the same controls at the same points instead of following the guided plan. On a machine with controls, it tracks bean temperature once available and falls back to elapsed time before that; on a machine without controls it only announces the moves to make. The burner reaction time tells the replay how far ahead to look when following the curve, so the effect lands when the curve says it should rather than after the machine’s own lag — see Preparing a roast.
How a bag of green coffee is sealed: vacuum, valve bag, sealed jar, open cloth bag, and so on. It decides whether the coffee drifts toward the room’s humidity or holds its own water activity — sealed methods hold it, an open bag drifts.
The moisture a bean would settle at if left in the current room conditions indefinitely. A rough guide for what an open bag is drifting toward, not a measurement of the bean itself.
The release of CO₂ trapped in a bean by roasting. A coffee brewed too soon, before it has degassed enough, pushes back against water unevenly and extracts unpredictably — which is why a coffee needs a few days’ rest before it brews at its best.
The share of the ground coffee’s mass that ends up dissolved in the cup, as a percentage. Too little tastes thin and sour; too much tastes bitter and harsh — dialling in means steering toward the range in between by adjusting grind, ratio or time.
Water carving a fast path through the coffee bed instead of passing through it evenly — part of the bed is over-extracted, part under-extracted, at the same time. A grind change alone does not fix it; it is a distribution or preparation problem.
An abrupt collapse of RoR, usually just after first crack, when the roast loses momentum. Left alone it leads to a baked cup.
The opposite of a crash: RoR turning back upward late in the roast, usually producing harsh, ashy notes.
A flat, hollow, bread-like cup, caused by spending too long in a phase with too little thermal momentum — most often a drying or Maillard phase that ran long, or one run on too little heat.
The second case is the harder one to see: because the drum keeps radiating, bean temperature and RoR can both look impeccable while the reaction underneath has already stalled. A smooth curve is not proof that the roast is being fed.
Drying driven so hard that the bean surface dries far ahead of its core, leaving the roast unevenly developed.
Burnt ends: the two tips of the bean turn dark, sometimes black, while the rest of the bean has the expected colour. The ends are the thinnest part of the bean, so they dry and brown first, and too much heat early in the roast carries them past the body. It gives burnt, ashy notes in the cup. It can be marked on the result form — see After the roast.
Dark, burnt patches on the flat side of the bean, where it lay against a surface that was too hot. Most often the charge was too hot for the batch, or the drum turned too slowly to keep the beans moving. Like tipping, it gives burnt, ashy notes, and it can be marked on the result form — see After the roast.
The safe state TilauScope commands when you hold the crossed-out flame button: every automation that was driving the heat is stopped, the burner goes to zero, and airflow and extraction open fully. The drum keeps turning and the beans stay in it — emptying the drum remains your gesture. See The roasting window.