Home roasting, the basics

Roasting turns a hard, grassy seed into something you would want to brew. It is the one part of the chain most drinkers never touch, which is a shame, because doing it once teaches you more about flavor than a year of reading. This is a working map of what happens inside the bean, the landmarks you steer by, and what the common home setups do well and badly. It will not make you a roaster on its own, but it will let you read a roast and know what you changed.

Green beans: what you start with

Green coffee is the dried, processed seed of the coffee cherry: dense, pale, smelling of hay or peas. Two numbers matter before you apply heat. First is moisture content, normally 10 to 12 percent for properly dried, well-rested coffee; too wet and it roasts unevenly and risks mold, too dry and it scorches. Second is density, driven largely by growing altitude: high-grown SHB or SHG beans are harder, hold heat differently, and take more energy to roast through than soft lowland coffee.

Green carries the fingerprints of everything upstream. Washed coffee roasts cleaner and more predictably; naturals carry more sugar and can run hotter and faster near the end. Watch the lot for defects and quakers, the under-ripe beans that stay stubbornly pale and taste papery; you pick those out after the roast. Buy small, buy fresh-crop, store cool and dry. Green keeps for months, far longer than roasted coffee.

The three phases and the two cracks

A roast is one continuous climb in bean temperature, but it splits into three jobs.

Drying is the first stretch, roughly the first third of the roast. The bean sheds most of its free moisture and shifts from green to yellow. It is endothermic, soaking up heat, and smells of grass then bread dough. Rush it and the outside runs ahead of a still-wet core, which bakes in a flat, hollow taste later. The papery silverskin flakes off here as chaff.

The Maillard phase begins as the bean browns past yellow. The Maillard reaction is hundreds of reactions between amino acids and reducing sugars, building the savory, nutty, chocolatey compounds (melanoidins) that give coffee its body and color. The aroma turns toward toast and almond. Much of the perceived sweetness and complexity is set up here, so most modern light-to-medium profiles give it real time rather than blasting through.

Development is the final phase, beginning at first crack. Here caramelization accelerates, acids break down, CO2 builds, and the flavors you actually taste lock in. Too little and the cup is sour and grassy underneath; too much and it goes flat, ashy, roasty. Development is not a synonym for dark: a light roast can be fully developed, it just spent enough time after first crack to finish the inside.

You find the phase boundaries by ear. First crack is a sharp, popcorn-like cracking, trapped steam and CO2 splitting the bean as it expands, typically around 196 to 205 C (385 to 401 F) bean temperature. The exact number depends on your roaster’s probe placement, so treat it as a relative marker for your machine. Second crack comes later: a quieter, faster snapping (more like cereal in milk) as the structure fractures and oils migrate to the surface, around 224 to 230 C (435 to 446 F). It marks dark roast territory; push hard past it and you head for charcoal. Medium roasts live in the gap between the two cracks. For the full map, see roast-levels-explained.

Your nose tracks all of this. Bread dough through drying, toast and nuts through Maillard, a sweet caramel-and-cocoa note around first crack: that is the roast going well. A sharp, acrid, scorched smell before you expect second crack means surface scorching from heat that is too aggressive. Thin bluish smoke late on is normal; thick, rolling smoke and a flat burnt smell means too dark, or a chaff fire starting. Keep chaff cleared, never walk away, and have a way to dump and cool fast.

Development time ratio: the number that keeps you honest

Development time is the stretch from the first audible crack to drop. Expressed as a fraction of the total roast, it is the development time ratio (DTR), the single most useful number a beginning roaster can track.

DTR = (time from first crack to drop) / (total roast time).

A common, defensible starting window is 18 to 25 percent for filter-oriented light and medium roasts. A six minute total roast with about 70 to 90 seconds of post-crack development sits comfortably in that band. Run DTR too low and you bias toward bright, sour, underdeveloped cups; run it high and you flatten acidity and dull the origin. Log first crack time, drop time, and the resulting DTR every roast, alongside a tasting note 24 to 72 hours later once the coffee has rested and finished degassing. After ten roasts you will see your own patterns far more clearly than any borrowed recipe.

Popper vs drum vs pan

Three common ways in, in rough order of control.

Pan or wok on a stovetop is the cheapest and hardest. You constantly stir a thin layer over moderate heat for even contact, get uneven, scorch-prone results, and have no real temperature data. A fine way to feel what a roast does for almost no money; a poor way to make consistent coffee.

Air poppers (hot-air, fluid-bed style) roast small batches fast, 5 to 8 minutes, by suspending beans in hot air. They strip chaff well, are cheap to start with, and give bright, clean results, but they offer little control over airflow and heat, struggle on dense high-grown beans, and run short of the long Maillard development that bigger roasts allow. Great for learning the landmarks, limiting once you want to shape a profile. Roast outdoors or under a hood; they smoke.

Drum roasters, including small home units, tumble beans against a heated drum (conduction) with some convection. They give you the levers that matter: adjustable heat, airflow, and usually a bean-temperature probe so you can log curves and DTR. Batches are larger and more even, and you can hold a slow, deliberate development. They cost more and demand more attention, and that is the point: they are the first setup where repeatable, profile-driven roasting becomes realistic.

Next

Roast three small batches of one washed coffee, changing only development time, then cup them side by side after a two-day rest. That single experiment, mapped against roast-levels-explained and development-time, will teach you what DTR tastes like better than any chart.