Roasting basics

This is basics about roasting coffee.

My name is Therese. I started this roasting guide in 2016 because I got a new roaster: the Bullet R1 (my 4’th coffee roaster). It’s a 1 kilo roaster with many setting possibilities and data output. Since it was new to the marked there were no one to ask. Here in Denmark we were a bunch who got it. So we have been exploring and learning together. This made me collect info on this site. Along the way input has come from Bullet owners around the world (from the Facebook group “Aillio Bullet R1 User Group”).

But of couse much of it can be used on other roasters too.

Just remember; nobody has the ultimate truth about roasting … there are many approaches … and it depends on the type of roaster .. and the beans … and what taste you want in the coffee. So I can’t tell you exactly what to do. You have to explore for your self. Enjoy !



Scope of the roasting process
Batch size
Preheat temperature
Bean characteristics
Bean temperature (BT)
ROR curve
Turning point (TP)
First Crack (FC)
Development time (DT) & Roast degree
When to stop the roast
Shape of ROR curve
Weight loss
Scorching and other roasting defects
Fine-tuning the roast profile
Taste is king

When you got hold of these basics → you can move on to Roast Profiles


Scope of the roasting process

When you roast coffee you want to get the best taste developed AND avoid bad tastes.

To achieve this the main things to focus on when roasting:
1) The heating up process: how the beans are heated up
2) Determine when to stop the roast

beans-smoke-sh21. Heating up
It takes time for the heat to travel from the surface to the core of the bean.

If the heat is too high during the roast you risk the outer tipping-in-beansof the bean get roasted quicker than the inner – the coffee will taste both burnt AND underdeveloped (like hay and cucumber).

If the heat is too low you will not get enough development of the taste. The aroma is flat. This roasting problem is called baked (taste like old bread).

Too little heat will give a late First Crack – which gives a weaker sound = more difficult to hear.

First Crack (FC)
When roasting coffee at some point, the beans start to crack; they expand and make a noice like breaking a tree branch (like popcorn, listen here). When the first cracking is over, its silent for a while, and then starts a second crack (SC).

To me a late FC is like after 11 minutes. And a early FC is before 7 minutes. But there are different oppinions on this. You got to explore what suits your beans and your taste.

When I roast coffee, my aim is to get a First Crack (FC) start around 6-9 minutes but thats for a light roast of high grown beans. For darker roasts and espresso, I found 10-11 minutes works well.

A step deeper: instead of the total time from start to FC … you can look only at the time from Yellow point to FC. I know a roaster who focus on this instead of the whole time to FC. He like that phase to be 3 minutes.


2. Stopping the roast
roaster-door-dump-sh2During the roast chemical compounds are formed and broken down all the time. So the point where you stop the roast is very important for the taste.

Coffee people typical have a strong preference for roast degree: either light or medium or dark – or burned-black dark as hell. There is no common definition of what exactly is medium or dark roast.

How do you determine when to stop the roast ? There is no ultimate technique, because the development of the taste i multifactorial. Some look at the bean temperature. Others go by the color of the beans or the smell of the roast.

I use the time from FC start and adjust it acording to the bean temperature rise (quicker rise -> shorter time). Listen more in episode 5.

survey-when-to-stop-roast-27decI asked a coffee roaster forum on Facebook what they go by. Here is what they answered – I only gave them possibility to make one tick: which is the most important ? Many commented that they look at several things.


I make a podcast about coffee roasting. The first episodes are about the different approaches to stop the roast.

Batch size

Each roaster has a max bean load capacity. Like a Probat P12 is 12 kilo and the Bullet roaster is 1 kilo.

For the Bullet a minimum is also specified from the producer: 350 grams. This is to make sure that the Bean Probe give a good reading. Of cause you can still roast less, see Roasting small batches.

But a roaster typical has an optimum load that works the best. Several longtime Bullet owners reports they find it is around 600-750 gram. While with 1 kilo the beans reacts slower to changes and you need to use max-power; P9 and have no headspace to do more . With a 300 grams batch it responds too quickly (“fly around”, Albino calls it).

But you have to try out for your self. Bob Werby found that it is also affected by the surrounding humidity.


John Platô tells:
“When roasting let’s say 10Kg. There is so much mass that a declining RoR just happens automatically. It also takes a lot more input to make small changes to the curves. Whereas with the Bullet, you actually have a lot more instant control over the curve due to the lower mass and lower temperature inertia.

fil-29-11-2016-16-43-56With large batches, your charge temp and power during the first 4 minutes of the roast will basically dictate how the rest of the curve is going to look like. So in some ways, it is easier to visualize how your roast will go when roasting large batches. However, for example if you notice your RoR is too low at 3 minutes, you are pretty much screwed for the rest of the roast as it would require a LOT more energy to get things back on track.”

Preheat temperature

How the beans are heated not only depends on the power-setting during the roast, but also of how warm the roaster is before the beans are added/charged.

Roasters talk about the “charge temperature” = reading of a bean probe inside the drum. But the amount of heat the roaster can give to the beans do not only depend of a temperature reading, but also of how long the roaster have been preheating. The mass of the roaster itself absorbs heat. When the inside is first reaching 200°C, the outer parts of the roaster  still lack in heating up. So if you drop in beans right away the mass of the roaster will take some of the heat -> your beans will heat up slower than if the roaster had been preheating for 1 hour.

beans-smoke-sh2On the Bullet R1 roaster you choose a temperature for the preheating before you start up the roaster. If you watch it heat up, you will see after first reaching the chosen preheat temp it stay there for a long time fluctuating → that is to heat up the whole mass of the roaster.

Choice of preheat temp
Choose the preheat temperature according to batchsize (grams) and depending on when you want to reach First Crack (FC). And then consider that some beans easier get burned: lower density and naturals, like brazil. Whereas high-density and washed (like Kenyans) can tolerate higher heat. Also see Bean Characteristics below.

Too low preheat temp will give a too late FC time. You cannot catch up during the roast by using high power (like P9).

Too high preheat temp will give a too soon FC time / scorch the beans = roast the outer of the bean quicker than the inner. A FC start before 7 minutes is too soon.

I prefer a FC start around 6-9 minutes (Therese speaking) for light roasts – and 10-11 minutes for espresso and darker roasts. I know others who are content at 11-12 minutes. Note: Late FC gives a weaker crack.


12 batches, all 500 grams – roasted by Steffen, starting on P8 and then declining.

The graph shows FC start around 9 min with 170°C and around 8 min with 180°C. Clearly other things are affecting too, like type of bean.


Our experience with the Bullet

400 grams batch: At least 170°C (in combination with P7-6 and Fan2-3) or 175°C, to reach FC around 7-9 minutes.

500 grams batch: 170-180°C. Steffen gets FC start around 9 min with 170°C and around 8 min with 180°C with power starting at P8 and then declining.

7-800 grams batch: 190°C

1000 grams batch: 195-200°C. Jens gets FC around 9 min with preheating at 195°C and P9.

Later on we went on a roasting course and here Michael suggested more heat to begin with. He used preheat 200°C for 700 grams batch. This is for hard beans/high grown = they can tolerate such high heat.

Notice: the first batch needs more heat because the roaster itself isn’t fully heated.

The initial push of heat is like a push of a boat in water → it helps getting the boat going. The speed of the heat transfer into the bean depends on the temperature gradient from the outside to the core.

Preheat time

Its not enough that the Drum temperature reads the set-temperature – because it takes time to fully heat up the roaster. Thats why it takes a while before the Bullets says “Charge”.

I like the preheat time to be at least 25 minutes. Steffen use 45 minutes of preheat.

Its a common experience in coffee roasting that the first batch is a bit slower than the following batches. Thats because the roaster isn’t fully heated to begin with.


First the preheat temperature is important – then the power during the roast has influence on how long it takes to reach FC.

Develop your own strategy for when to change the power setting depending on the bean temperature (BT). For instance; when the BT reach 150 C, then turn down to P5.

There is a general consensus to: Decrease power in the latter part of the roast.

See further under Roast Profiles.


Fan/ airflow

You can have airflow at one setting during the entire roast (like on F2). Or you can increase airflow when the beans are getting brown to blow out created smoke and chaff. If you airflow is too high to begin with you rist the beans will get roasted quicker on the outside than in the core.

We have seen different patterns in what happens to the bean temperature (BT) when increasing Fan:

At 400 grams the BT increases when increasing Fan but only until F4 (like applying heat with a hairdryer). From F5 it cooles the roast.

But at higher batch sizes it takes higher fan setting to result in cooling. Thomas reports that it is F9 with a 1,2 kg batch.

Note: this is what the bean probe registers.

Also, wether the filter in the chaff collector is clean or full of chaff and oils will make a difference to the airflow in the drum.

Here is the cleaning routine for some of the Bullet owners:


Update May 2017: Aillio has now released a new bigger chaff filter.



We are still exploring. The general recommendation says that larger batch needs higher drum speed.

Many of us use d8 and dont change it during the roast. In most cases use at least d6 (unless your own experimentation for your specific bean/environmental temp/charge size says otherwise).

Bob Werby and Kaffe Thomas have looked into how different drum speeds affect on the roast. It seems that higher drum speed does transfer heat quicker into the beans – like d8 instead of d4.

Bob Werby ran some tests. With a Kenya bean; First Crack started slightly later at d4 than at d8, but only like 10 seconds. With an Etipean Bob could not hear First Crach at d4, where as it was clear at d8.

John-Paul McCarthy use lower drum speed in the beginning of the roast to reduce risk of “tipping” of the beans. He use d5 until yellowing point where he increase drum speed one step each minute until FC. He roast batch size 7-800 grams.

Rounds per minute
Here Klaus has registered what the drum speed setting on the Bullet corresponds to in rounds per minute (rpm) – he counted it manually: putting a white mark on the drum and then counting the rounds within a minute.

Mixing versus sticking
The point in having a turning drum is to get the beans mixed to get an even heat transfer to the beans.

If the drums speed is very low → the beans will get burnt on the side where they touch the drum

If the drum speed is very high → the beans will stick to the drum wall due to centrifugal forces (like clothes in a washing machine during spinning) and will beans will get burnt

Drum rotation diagram

Bobby Chow found an engineer article that presents a formula for the optimum drum speed depending on drum size:

Drum speed optimum

The inner radius of the Bullet drum is like 8,5 cm. Using this formula the maximal rpm before beans do not fall off the drum wall is: 102,5 rpm.

Fil 22-06-2017 10.57.04

Take 25% off this and you will end up with 76,9 rpm … which is the level of d8 on the Bullet 😉


Temperature readings affected by drum speed
Lower drum speed (like d4 instead of d8) seem to give higher BT readings. Note: this has to do with the temperature probe and not the actual temperature of the beans.

Kaffe-Thomas tested what changes in the drum speed during the roast did to the BT reading and ROR. P and F settings were fixed. He did it twice during same roast.

Lowering drum speed from D8 to D4 increase BT ROR by approximately 5 degrees.

Increasing from D4 to D8 lowers BT ROR approximately 5 degrees

At 4:30 (ROR stabilized nicely) going from D8 to D4 and at 5:30 from D4 to D8. Then again from 7:30 D8 to D4 and at 8:00 D4 to D8. Same pattern twice. Lower drum speed equals higher ROR reading.

Higher BT readings could be because the beans dont cover the bean probe as good at d4 than at d8 .. or that the heat accumulates more on the surface because the heat transfer into the beans is slower. Just guesses.

Bean characteristics

The heat needed to get nice development depends on the characteristics of the beans.

Larger beans needs more heat. Harder beans (high density) needs more heat. On the other hand a small bean with low density should have lower heat to avoid getting burnt.

preheat-ifht-bean2The moisture level in green beans is normally 10-12%. If the moisture content is low the beans get burnt much easier – so reduced heat is necessary.

Joe Marrocco have written a blogpost about bean density

Bean temperature probe

Most roaster use a temperature probe placed in the middle of the bean pile in the roasting chamber to follow the temperature in the beans during the roast. It’s called BT = Bean Temperature.

It is just not that simple to get an exact temperature reading of the beans. Keep in mind that the BT reading first of all has to do with the probe – it is not exactly what is going on in the beans.

The probe is typical a thermocouple or NTC. It differs how quickly they measure (= responsiveness). Remember the temperature during roasting changes quickly and steady-state for temperature takes time. On the other hand metal do absorb heat quicker than the cellulose chunk beans – so changes in airflow temperature is picked up quicker by the metal probe. See my test how airflow affect th BT reading in my roaster.

Rob Hoos writes: “The thermocouple is affected by sheath thickness, probe placement relative to the roaster design, probe placement relative to the bean pile, the bean mass, the airflow filtering through the bean mass, air velocity and burner adjustments.” in the article “Our World Thru A Keyhole” from RoastMagazine sep/oct 2017

The bean temperature (BT) measurements are not comparable from roaster to roaster. In one roaster FC starts at 200°C, in another roaster it can be 170° or 207°C.

You just have to get to know your own roaster.

Also read about this on the Scott Rao blog: Your bean probe has been lying to you

The Bullet got a temperature probe to measure in the middle of the bean pile. You can feel it by sticking you hand inside the left side of the door window – its placed above the window.

ROR curve

ROR = rate of rise = temperature rise of the bean probe per minute
Calculated during the roast by computer software.

Roasting consultant Morten Münchow aim for: ROR-level around 10°C per minute in the middle of the roast and around 5°C from when First Crack starts.

I have heard of lower ROR levels than 5°C during the development phase, like 2 °C. But also higher; Mikkel Selmer from La Cabra go for 10°C for a light roast.

Roasting consultant Scott Rao suggests that the after the initial ROR peak (2-3 minutes into the roast) – the ROR curve should only be declining. Never rising or flattening. Note that this is only one strategy for roasting. I have tastes great coffee that didn’t follow this pattern.

Important: your ROR curve will also be affected by the type of probe you got. Rob Hoos published data in the RoastMagazine article where he compares three probes, the smallest is 1,6 mm and the thickest is 6,4 mm. The ROR difference at the end of the roast is like a couple of degrees Fahrenheit. See above under Bean Temperature.

So again – you need to gain your own experience with the equipment you got.

Turning point

“Turning point” (TP) is an artificial point on the ROR curve. It’s the low point shortly after the roast has started. Its typical around 1-1,5 minute in. Also called; bottom out.

The initial drop in Bean temperature (BT) is a mix of temperature in the bean probe before adding the beans and the temperature of the beans (room temp). So to beginn with the beans are cooling off the bean probe. As the beans get heated up at some point they make the BT probe rise again.

The time to Turning Point depends very much on the type bean probe. Again see the mentioned Rob Hoos article. He placed 3 different probes in the same roast; the thinnest probe was 1,6 mm and had TP at 47 seconds. The thickest was 6,4 mm and had TP at 2:46 minutes.

The temperature of TP says something about the speed of this initial “push” of the heating up process. So very dependent of the preheat temperature.


This graph is 12 batches, all 500 grams, roasted by Steffen. They were not that different in power and fan setting during the roast.

X-axis is the TP temperature and Y-axis is time for FC start.

This plot shows the tendency to higher TP temp gives a shorter time to FC start.


As the roasting proceeds the beans will go from green to yellow to brown – in various degrees.

You can note down when the beans are at yellow as a mark on the roast. It’s when the beans are no longer green and before they start getting light-brown. This point is also called Drying End (=DE)

If yellow comes too early the outer of the bean is getting roasted quicker than the inner. I don’t know what the limit is. It also dependens of the type of bean; high density beans transfer the heat better than softer beans.

If yellow starts around 3 minutes it is a fast roast -> that might be good for a light roast of a hard beans like Kenya. But if you want it slower then use less heat.

If yellow starts around 5 minutes it’s a slower roast -> might be just fine for a lower density beans, like brazil. But if you want a faster roast use more heat to begin with.

If the point of yellow/ the first sign of brown starts around 10 minutes I would think it was too slow. But taste for yourself.

Middle- or Maillard phase (MAI)
If you note the time for yellow (or mark it in the roasting software) you can get the lenght of the phase from yellow til first crack starts. It’s called the middle phase or Maillard phase.

Rob Hoos has written a book on Modulating the Flavour Profile of Coffee . He introduces talking about the importance of the stage from yellow to first crack – the midle phase between drying and 1st crack. He calls it the Maillard phase.

The more time spent in MAI the more you “increase the complexity of the sugar browning tones you are creating in addition to the weight/texture/mouth-feel”. Up to a certain point of course.

Also, it really depends on the coffee. But generalle speaking:

Less time in MAI: Lower complexity, but lighter body and greater clarity.
More time in Mai: More complexity, heavy body, lower clarity.

According to Rob Hoos.

Chris Kornman also looked into this. He finds that shorter time in MAI phase gives sweetness and acidity in the coffee. He writes about this and the effect of water activity of the green beans at:

First Crack (FC)

When roasting coffee at some point, the beans start to crack; they expand and make a noice like breaking a tree branch (like popcorn, listen here). This goes on for a while. When the first crack is over its silent for a while – and then starts a second crack (SC).

Its not that easy to determine exactly when First Crack starts. Read more here How to determine FC start . Episode 5 in my podcast Coffee Roasting Navigated is also about listening to the beans cracking.

Development time (DT)

DT = Time from FC starts to drop/end of the roast.

A lot is happening to the taste from the point of First Crack start. Therefor it’s more meaningfull to focus on how long the beans was roasted from FC start – than the total roasting time.

Episode 5 in the podcast Coffee Roasting Navigated is using Development time to decide when to stop the roast. Listen here:

Roast degree

The longer DT → the darker a roast.

But the temperature rise during DT also makes a difference. If temperature rise is low, the darkening of the beans will be slower.

At some point Second Crack (SC) will beginn. This sounds weaker than First Crack.

I know many roasters who stop right at the first sounds of second crack – its called “on the verge of second crack”, so before SC is really going. This roast degree is called Full City. Read about how Peter van Wijk is roasting.

Very dark roasted coffee is roasted into Second Crack – several minutes. Here the oils will come out to the surface of the beans. This roast degree you find at Starbucks and in Italy.

On the other hand, very light roasts is stoppede before First Crack has ended.

BUT notice – there is no common definition of when its a light roast or medium or dark. Here in Scandinavia what we would call a medium roast is a light roast in USA.

Fil 12-12-2016 10.14.41


Stop the roast

During the roast chemical compounds are formed and broken down all the time. So it’s very important for the taste when the roast is stopped.

You have to search to find the optimum for the bean at hand. Ten seconds earlier the coffee can taste underdeveloped. Ten seconds later the coffee can taste flat or burned. While at the right timing the coffee has a big aroma and is nice balanced in acidity/sweetness/bitterness. Its different from bean to bean where this optimum lies. And a coffee can have several optimums – at different roast degrees.

But not only the time from FC onset determines the taste. The temperature rise does as well (ROR). A quick rise (like more than 7°C pr min) gives more acidity and clarity. A lower rise (2-5°) gives more body. Zero rise gives a flat taste (called baked).

How do you determine when to stop the roast ?
There is no ultimate technique, because the development of the taste is multifactorial. I use the time from FC start combined with the rise in temperature. Some look at the bean temperature. Others go by the smell or color of the bean.

survey-when-to-stop-roast-27decI asked in a coffee roaster forum on Facebook what they go by. Here is what they answered – I only gave them possibility to make one tick: which is the most important ? Many commented that they look at several things.

I have started a podcast about coffee roasting. The first episodes are about the different approaches to stop the roast.

Shape of the ROR curve

It’s a general principle that after the initial drop and rise of the ROR curve (whitin the first 4 minutes) – the ROR curve should be declining.

Dont overfocus on getting a nice declining ROR curve. So many other things are more important.

But avoid the ROR curve going beyond zero (=BT is falling instead of rising) – and avoid ROR dramatically rising in the last part of the roast.

A rule of thumb is aiming at a ROR of 10-15 degrees per minute after yellowing  and around 5 degrees per minute going into First Crack.

In order to get a low ROR after First Crack onset – you need to turn the heat down a while before. With bigger batches, like 900 grams, Steffen turns the heat down to P1 1,5 minute before FC start – or else ROR will rise to much after FC start.

The roasting consultant Scott Rao have propose some theories about a constant declining ROR and a development time ratio of 20-25%. Read his elaboration of this on his blog

I don’t use this rule. I have great tasting coffees at 10% DTR.

See graphs and read about working on the ROR curve in


When you stop the roast fast cooling is necessary.

Aim: Get below 40°C in 4 minutes.


Weight loss

If you weigh your roast before and after roasting, you can calculate the weight loss in percentage. This says something about how far the roast went: the darker roast the greater loss-percentage.

A light roast is around 11-13%. A real dark roast is 20-22%.

tipping-in-beansScorching and tipping

These are roasting defects caused by too much heat.




Here is a brazil bean with tipping. The ends has cracked open and are darker in color.

This is to much heat, too quickly.

Try lower preheat temperature. Aim at later FC start.




Read more:

and see pictures

Fine-tuning the roast profile

The taste is very much affected by small changes in the roast profile. There are a lot of advice out there on how to fine-tuning the roast profile.

But here you have to watch out … A rule of thumb that works on one type of roaster doesnt work on another … and there is differences in taste preferences … and there is theories out there that works in one context but not another. Beans are different. Roasters are different. Taste preferences are different.

You have to try out for your self.

Read experiences on the next page

Taste is king

The only thing that determines the success of a roast is how it tastes.

For every roast make notes of the taste. This is the key to improve your roasting skills.

Read more under Tasted Coffee. And under Cupping.



Discuss your roasting with the Bullet

The unofficial Bullet forum

Groups on Facebook like