In the months since I published my post on sample roasters, I’ve heard from countless Roest users who shared my experience—and perhaps a few frustrations—and who hoped I’d be willing to shed a little more insight into best practices for getting optimal results from the roaster.

Every coffee that comes through my lab is roasted first on the Roest; millions of dollars per year of purchasing decisions are made, in no small part, due to the quality of the roasts I am able to produce using the cube-shaped machine—as well as the speed, cleanliness and ease of its workflow.

In an effort to demystify the roaster and make it easier for a new user to achieve quality sample roasts quickly, I’ve put together this best-practices and quick-start guide.

I think it’s helpful to understand the physics of a roaster and how it works before we try to manipulate it.

In traditional drum roasters, heat is applied to coffee through two means: contact between the surface of the coffee and the hot drum (conductive heat transfer) and contact of the coffee with hot air inside the drum (convective transfer). Air passing through the drum is sucked from a motor positioned in the exhaust stack rather than pushed or blown. In many drum roasters, air from the roasting environment is first passed over the flame, heating it before it passes through the coffee-filled drum (the Diedrich IR series and Primo roasters are popular exceptions to this—but there are others).

In these roasters, increasing the speed of the exhaust fan will evacuate air from inside the drum at a greater rate (as well as smoke and chaff) as well as increase the velocity that ambient air is drawn from the roasting room and passes over the flame; this means that higher air settings will, in effect, lower the temperature of the air coming into the drum unless the flame is also increased (which will increase the surface temperature of the drum). While roasters may be trying to take advantage of an ostensible increase in convective air transfer by increasing the rate of airflow, any effect, if it exists, would be extremely temporary; once built-up heat passes through the coffee and is evacuated from the roasting chamber, the effect is that the air temperature decreases. The best way to increase convective transfer is by changing the speed of the drum to optimize the loft of the coffee and the time it spends tumbling in the hot air of the roaster rather than touching the surface of the drum (you can also adjust your batch size and thus the ratio between the volume of coffee and volume of air in the drum).

Single-pass air roasters operate in a different manner—with a fan pushing hot air through or around coffee, often to loft and mix the pile. Most often, this one fan also serves as the exhaust. In this style of roaster, heat transfers into the coffee nearly purely through convection. This means that the temperature of the air (inlet temperature as well as the temperature of the air in the roasting chamber) will impact the rate of heat transfer, as will the velocity of the hot air through the system (think of convection ovens: if you bake scones in a convection oven, you’ll require a lower bake temperature setting). Higher airspeeds have a greater evaporative effect and potentially have the ability to transfer more heat. How much heat air can hold at a given moment as well as its efficiency at transferring heat through convection is a function of density and air pressure, and as a result, air roasters are sensitive to elevation as well as changes in drum pressurization. Finally, the specific design of where air enters the coffee pile, how and where it’s exhausted, and how the coffee is mixed/agitated will impact these dynamics as well.

In the Roest, however, we have two fans: one at the inlet, blowing air into the drum like a normal single pass air roaster, and an exhaust fan above the drum which both sucks air from the drum and forces it through the exhaust stack. The differential in pressure created between the two fans has a dramatic impact on heat transfer, air humidity in the drum, evaporation, coffee expansion and the dynamics of first crack, and therefore is an unseen variable that we must control to achieve replicable, usable results.

There are areas of conduction in the machine—some minor, such as the agitation paddles, and some potentially major, such as the inlet screen—and while we must contend with them for the best results, we’re primarily concerned with convection.

In the Roest, we have the ability to control a few parts of the system directly and independently:

Batch size is the amount of coffee you load into the machine for a roast and impacts both the ratio of air to coffee in the drum (volume) as well as the amount of energy required to roast. In the Roest, it also impacts how well the coffee is mixed (due to the size and geometry of the drum) and how/what the probes read depending whether they’re primarily contacting air or coffee. Batch sizes below ~125g will render the BT probe of declining reliability at 50 RPM or higher, with 100g being a usable minimum; batch sizes higher than ~150g will bury the exhaust and air temperature sensors, making them less reliable. Greater batches sizes will, in theory, require greater application of power and thus inlet temperatures to achieve similar roast times.

Inlet temperature and power are two methods you can use to control the roast; you can either adjust power up or down (like on a traditional roaster) or use inlet set points to have the roaster adjust the power via a PID to achieve that inlet setting. Either way, inlet and power will impact each other: using higher power will result in higher inlet settings; setting a higher inlet setting will require the machine to deliver higher power (Because of the variability in ambient temperature in my roasting space and my desire for consistency, I use inlet-based profiles). Power can be adjusted either programmatically or in real-time during a roast; this means you can establish a power profile and then intervene using the encoder.

Drum speed or RPM will determine how the coffee is mixed as well as lofted in the drum/roasting environment. In other words, it impacts the efficiency of heat transfer in the roaster. Lower drum speeds will result in more conduction and contact with the bottom of the drum; inlet air enters the roaster through the bottom right of the drum behind a steel grate which absorbs considerable heat. Using lower RPM will deliver more accurate BT readings, but requires lower inlet temperatures to prevent damage to the coffee from this conduction, and will also result in longer roasts due to less convective heat transfer. Higher RPMs will result in greater lofting and mixing of the coffee, which can also produce defects (tipping, specifically, and internal scorching), will accelerate the roast, result in worse BT contact at lower batch sizes (100g), and at larger batch sizes (200g) can toss coffee into the exhaust system at the top of the drum. In general, I use 55 RPM for 100-150g batches.

Exhaust fan speed, which can be set programmatically or in real-time during a roast, dictates how fast the exhaust impeller spins. This fan extracts air from the roasting drum and pushes it out of the exhaust in the back of the unit. If the exhaust is low relative to the speed of the inlet fan, it can create a positive-pressure environment in the drum by creating back pressure; this can be favorable for transferring heat into the center of the coffee and promoting even development, but also will not dissipate smoke at crack, which can, in theory, result in coffee that tastes roasty; it will also make development rapid and reduce the margin for error for a targeted development level.

Inlet fan speed can only be set via the service menu and controls the speed of the fan that pushes inlet air into the drum. The primary engineering function of this fan is to keep the heating element cool; at higher elevations, because the coefficient of heat transfer is lower, you may need higher-than-standard heater fan speeds to appropriately dissipate heat from the heating element and metal inlet plate. The inlet fan speed will directly influence airspeed in the system as well as pressure inside the drum through its interaction with the exhaust fan. The inlet fan speed self-regulates with power, apparently—which means that it interacts with exhaust to create variation in drum pressure through the roast.

The effect of each of these levers—and their relationship to each other—is what makes Roest simultaneously compelling and perplexing.

A few quick notes before we begin:

• BATCH SIZES: Profiles and guidance for one batch size do not necessarily apply to another batch size. In other words, it’s not possible to extrapolate directly between 100g and 200g profiles. Not only do these different batch sizes occupy different volumes of the drum and therefore impact the ratio between air and coffee, but they will also change the way that probes are embedded either in the coffee pile or air; further, differences in power will lead to differences in drum pressure. It’s critical to understand the limitations to using the batch size you’d like to use, and select a suitable profile accordingly. I primarily roast 100g batches—so you’ll find most of what I say applicable to that batch size.
• EXHAUST VENTING: The exhaust in the Roest is powerful; because of this, the specific stack or venting configuration seems to have minimal impact on drum pressure relative to an exhaust setting. Members of the unofficial Roest discord have tested configurations with multiple 90-degree turns as well as vertical stacks going up or running down to the floor without observing a difference inside the drum using a manometer. An obstructed venting system or a booster, however, will impact drum pressure and air speed. Because the exhaust on the machine is sufficiently powerful—and because it would impact the pressure and airspeed inside the drum—I bypass the AC Infinity Cloudline booster I use when roasting with Kaffelogic or Ikawa roasters.
• WARM-UP AND BBP: The best way to achieve replicable roasts is through implementation of a sound warm-up profile (at least 10 minutes of an air temperature profile at your max roasting temp or above your average temp) and the use of a between batch protocol. The goal is to heat the metal surfaces of the roaster sufficiently so that they begin the roast day with enough energy from the first roast. In general, observation of the drum temperature and ensuring it reaches the same starting temperature each batch will yield consistent results. For my purposes, this is a starting drum temperature 290ºF/143ºC for 100g batches (I do hope that Roest will add drum temp BBP in a future firmware release). For reference, here is my warm-up profile (which I load as the first profile slot in my roaster) and my BBP. If you don’t have a machine capable of an automated BBP, you can use one of the five profile slots on your roaster as a BBP and program an air roast with the same shape of values to end at your charge temperature.
• USING IT/BT OR POWER/BT PROFILES: when using the same profile, coffees dried in cherry or with high water activity will run slower but require less post-crack development, while traditionally washed coffees will run faster. Coffees with lower water activity or lower densities will crack later. You can adjust your charge temperature accordingly or simply move your first few (for naturals) or last few endpoints (for lower water activity coffees) a few degrees higher to account for their differences in heat absorption/resistance.
• CONDUCTIVE TRANSFER?? IN AN AIR ROASTER?? While the company states there is no conductive transfer in the machine, in practice, I haven’t found that to be quite true. Using higher inlet temperatures—particularly with lower inlet fan speed, lower exhaust, lower RPM and larger batch sizes—will result in the metal inlet plate inside the drum getting quite hot. If you are experiencing tipping after reducing roast speeds and experimenting with different exhaust settings, this may be the cause and can typically be alleviated by reducing your inlet temperatures or changing the inlet fan speed.
• 110V OR 220V: There do seem to be differences in the way the profiles run on 110v machines (N. America) versus 220v machines (Europe). This is not supposed to be the case; however, in all of the testing I’ve seen, it is. No matter what you do, you will need to modify shared profiles for your machine as a result; use a reference roast curve from the profile originator as well as targets for weight loss and color to guide your modifications.

Here are a few tips and practices that have helped me get the most of out of Roest roasts.

1. If you are at sea level or close to it, you may find benefits reducing your inlet fan RPM speed and exhaust speed.

The inlet fan setting is accessible via the service menu and is by default set to 3400 RPM which is a safe setting for most conditions and effective at cooling the heating element to prevent it from burning out prematurely. However, this is entirely dependent on your use—not all profiles are equally demanding of the heating element, nor are all batch sizes. Further, roasting at sea level or close to it allows for greater dissipation of heat from the heating element at lower fan speeds, allowing us to manipulate this variable safely (I have put in over 800 roasts using an inlet fan speed of 3100 RPM and have yet to experience a heating element failure). 

It’s still unclear to me why this simple change seems to have a massive impact on roast quality, but for me, it did: I saw less tipping and better flavors in the cup. Where before, at 3400 RPM, coffees would taste baked or “brown” even if the roast log looked okay, with 3100 RPM the coffees were juicy with longer finishes and more intense aromatics.

I believe that because air is pushed away from the heating element with less velocity with lower burner fan settings that the machine doesn’t need to deliver as much power to the burner to achieve an equivalent inlet setting. This is, therefore, a more efficient way to roast—and also likely means that the metal grate above the inlet fan stays cooler through the roast, which would account for the lower amount of roast defects I’ve experienced.

Reducing inlet fan settings also allows us to use lower exhaust fan settings to achieve the same drum pressure (e.g. 45% at 3100 RPM instead of 75% at 3400 RPM on my machine) and thus have greater ability to manipulate the pressure instead the roaster, thereby impacting heat transfer.

2. Use a manometer to determine your drum pressure

To determine the pressure inside the drum, I use this manometer (disclosure: I am an Amazon affiliate and will receive a commission if you purchase through that link) with the metal barb running through a drilled-out silicone plug which I place into the trier port and observe the pressure differential in pascals during a roast. Others use a manual manometer, which offer a bit more of a smoothing effect and are thus easier to read.

Adjust the exhaust fan setting during a roast, observe and record the resulting values.

You’ll notice that there’s some oscillation that occurs during a roast as the roaster draws more power—and in turn, this will lead to a change in drum pressure. That’s fine. The thresholds that were most important for me to note were: (1) when the drum begins to read positive pressure during the early stages of the roast (which for me is around 2-5 pascals at 45% exhaust), and (2) when the drum reads consistently negative at crack (for me, 55% exhaust resulting in -10 pascals). 

While your approach may differ, I have found the best results by roasting with positive-to-neutral pressure until crack, and then increasing exhaust to vent smoke and chaff as well as evacuate excess hot air (abating the risk of unintentional added development). Using this approach, I’m able to increase inlet temperatures significantly relative to higher exhaust settings without risk of scorching or tipping. I also find that development is more even. Observing the difference in these roasts between air temp and exhaust temp curves, I believe this is because heat is transferred to the coffee earlier in the roast, allowing for better and more efficient development.

3. It’s a good idea to roast manually using a power profile until you first achieve acceptable results to understand the shape of the roast curves.

It’s not really possible to extrapolate perfectly between different roasting systems; what you’ve learned roasting on a Loring or Diedrich or Arc will not perfectly apply to a Roest. The only real way to know what is a “normal” shape of an Inlet curve, or the acceptable highs and lows, or what happens with different power applications—is to roast manually. I do encourage you to roast manually—using the same coffee, if possible—recording your final roast data (weight loss at a minimum and ideally color) and tasting every batch. I would use the findings you have from your study of the drum pressure and try to achieve a minimum possible exhaust setting—but your mileage may vary.

Once you’ve found an approach that works well using your power profile, try to understand why it works and build your new profile from the data and trends you observe.

4. Always record your roast’s final weight loss. And ideally color. And know as much as possible about the green you’re roasting.

Roasting is about trends and consistency. Disconnected from data about the starting state of the coffee (moisture, for example, or process) as well as its end state (how much weight it lost, or how a change to pressure or exhaust or post-crack development affected its final color), you can’t really make any assessments about the quality of your profile. Track everything. Take good notes. And always be cupping. Observing the whole bean color and expansion of your coffee can be a good way to understand how well heat is penetrating the coffee in a given profile (versus just darkening the outside).

5. A profile with efficient heat transfer will develop exceptionally quickly

Most of my roasts are between 0:35 and 0:50 of development. Depending on the coffee, a difference of 5 seconds might mean the difference between an excellent cupping roast and a cup that has roasty character. If you have been following recommendation #4, you’ll begin to notice patterns: coffees dried in cherry tend to require less post-crack development; coffees that have higher water activity do as well.

6. First crack doesn’t necessarily mean anything—but it’s a helpful way to understand how well the heat penetrated the coffee

If your coffee doesn’t crack much, that can result in an okay cup—the chemical changes necessary to make coffee taste like coffee don’t require an auditory crack. How vigorously crack occurs depends on moisture as well as the internal vapor pressure within coffee’s cellular matrix. If you’re able to drive heat into the seed efficiently, you are likely to get a more vigorous crack; this is generally a sign that development will be easy. EDIT: For reference, I do use first crack detection. My threshold is set to 390ºF and I have it set to start counting upon the third crack. Depending on how aggressive and vigorous crack is, I will adjust my development accordingly — less vigorous, longer development; more vigorous, less. In general I get best results with my profile when the machine hears between 12 and ~22 cracks in a 100g batch.

7. If you’re hitting a wall, it’s OK to change your approach to improve your data

This can mean that you increase your batch size, for example, if you’re trying to rely on BT data. It also might mean that you decide to change courses and approach your roasts a different way.

After months of using IT-BT profiles with 100g samples and being unhappy with how often I needed to re-roast to achieve good cupping roasts (20% or more of the time), I decided to switch to using Inlet-Time based profiles for my sample roasting work. If I had the luxury of larger samples, I might simply change to roasting 120g or 150g batches and stick with IT-BT because of the relative advantages of those profiles; but because I am often working with Pre-ship samples where sample material is limited, being able to roast 100g reliably—without re-roasts—is imperative. By switching to an Inlet profile, I bypassed my reliance on the BT probe’s poor data quality with 100g batches and was able to achieve replicable, high-quality sample roasts, reducing the number of necessary re-roasts and increasing the quality of my insights into the samples that came across my desk.

I have shared that inlet profile to the Profile Library as “CF Inlet”— or you can access it directly here.

8. If you’re at elevation, you’ll need to experiment.

Conductive heat transfer will be lower at elevation, so you may benefit from higher inlet settings and higher drum speeds to boost convective heat transfer relative to sea level. Because heat doesn’t dissipate as readily, you may benefit from default or higher inlet fan speeds to avoid burning out your heating element. And because heat doesn’t dissipate readily, your exhaust temperature and air temperature curves may be crucial to understanding how to optimize heat transfer in your system.

All in all: roasting with air at elevation requires changes in your approach.

Roasting on a Kaffelogic in Addis, I was able to slightly increase fan speed to achieve an equal roast. Using an Ikawa, I had to increase fan speed and decrease exhaust temps to achieve an equal roast. This makes sense; exhaust in the Ikawa is a proxy for the exit air temperature after the coffee has absorbed heat from it. If the coefficient of convective heat transfer is lower, coffee would absorb less heat from the air; this means that excess heat could build up in the system causing uneven development and roast defects. I’ve heard from friends that higher RPMs are indeed useful in the Roest as well as higher inlet settings, but haven’t yet had the opportunity to experiment first-hand.

9. Test your profile across multiple styles of coffee; you may need different profiles for different types.

It’s always been my dream to have a single profile to use universally for all coffees across my desk. With the inlet profile I’ve built, I’m comfortable using it as a first look to assess almost every coffee I receive. This won’t be the case with every profile—IT/BT profiles, for example, will require more aggressive heat at the start for naturals and high water activity coffees—but understanding the correlations between a green coffee’s physical characteristics and how it responds to a roast will inform your development of discrete profiles to handles the categories of coffee you create. As a heuristic, I would encourage you to examine processing style, screen size and water activity as categories of coffees requiring profile modifications or slightly different approaches.

The time and effort required to learn and master a new roasting system—particularly one as sophisticated as the Roest—can be frustrating and time-consuming (and make you question your skill as a roaster). But stick with it: I now find that the roast quality from my Roest L100 Plus is superior to any sample roaster I’ve ever used, bar none.

In weeks when I roast 40 samples, I might only flag two or three for a re-roast, saving me time, money and rejections. The time we put in upfront yields returns in time saved when it matters: when the sample is across your desk, when there’s limited material, when there’s a contract or purchase on the line.

I hope that with the guidance I’ve provided you’ll have success building a profile that is suitable for your needs and workflow.

Questions? Drop them below.