900g Full Flame test

I am going to have to try this to understand how your D settings affects the roast. Your PH, P and F settings are similar to mine for 1kg but your D is way lower. What kind of cup results are you getting?

1 Like

Extremely good roasts.

Let me explain what is going on.

You have three modes of transferring heat to any material.

Conduction, Convection and radiation (very minimal in coffee roasting so we are going to ignore it).

Conduction is generally more efficient in transferring the heat between objects where convection is better suited to cover the atmosphere around and object.

Drum speed and power modulates the conduction.

Drum Speed, Power and fan speed modulates convection.

Power is the common dominator in the Aillio for both major modes.

Air Flow modulates the amount of cooling or heating during Convection. Less Airflow leads to higher transfer during convection and more airflow leads to the reverse where you cool the beans and in addition removes any excess humidity in the chamber (Humidity is basically a battery for heat).

Drum Speed is a unique setting, where at slow speeds you have mainly conduction transfer with less convection as you up the speed it becomes less conduction and more convection to a point where beans start to stick togather and then it becomes more conduction while you are creating a tornado effect in the drum that is still increasing convection.

The secret to a good roast is that you have a good transfer of heat on the overall bean:

1- At the start of the roast where you have a large amount of heat stored as preheat transfered as much as possible and thus your require conduction. You might say that scorching might occur but scorching occurs when the point of low water content is dried throughly during the roast. In addition, You have a minimum drum speed for that to occur. From my testing it was around d3 for 900g batchs.

2- At the start of yellowing and mallard reactions, Most of the free water has already dissipated. Now you are trying to evenly develop the roast. That occurs with a mostly convection type of heat transfer and that’s when you start to up the drum speed evenly and steadily.

3- Just prior to FC, You require to have as much stability as possible in transfering your heat. Maximum point of conduction and convection and trying to smooth out the overall power requirements to keep the beans from running away to over medium roasts. A buddy of mine actually drops power after 1min after FC and lets the beans glide to 0 ror prior to dropping them.

My routine on 900g is basically start at d4, hit d5 prior to 150’c and increase the drum speed until d9 prior to FC.

Fan speed at F2 until yellowing hit f3 and f4 prior to FC and maybe F5 to stabilise the development.

All changes after that are at the PH and Power settings depending on the type of beans.


@otaibimn Thank you for one of the most well articulated writeup on this topic. It is worth a “pin” (but I don’t have a way to do that). I think you’ve put it very concisely in less than a page! I’m going to venture to guess you’re probably an engineer maybe in the thermo field? (I’m not one). :slight_smile:

Since you generally roast at 900g, let me ask this, would you follow the same principals for smaller batches but maybe start at a lower PH and all else is pretty much the same? I’m asking because sometimes I’ll roast 500g, but I generally roast 1kg (which I think your 900g principals would work well - I hope to try this weekend). I’m also asking because I wonder whether how the bean mass occupying the volume within the drum plays a role in this - the bean mass is a variable too IMHO.

Lastly, you you wrote:

I’m very curious what your buddy did there … esp I’m not sure what you mean by “1 min from roasting” (whether that is a typo or missing some words).


1 Like

Oops was meaning after FC. In addition, You are correct I am a simple mechanical engineer trying to understand my roaster.

As for smaller batchs split them into three:

1- 800-1000g: Standard roast.

2- 600-800g: I drop one degree in all settings from standard roast and 15-25’ preheat reduction.

3- 350-600g: I drop two degrees from standard roast in all settings and reduce 25-50’ preheat.

But always overlay the established roast and verify and modify.


This is a systematic method :slight_smile: I’ll have to save this and try this sometime. Thanks for sharing how you approached this.

1 Like

A few additional (and perhaps minor) insights:

The faceplate of any roaster emits radiated heat and therefore should not be ignored;–hence, preheating properly/adequately is important.
You can monitor the environmental & drum ambient temp. by using an air temp. sensor, which will ensure that the roaster is properly heated up each time, before dropping a charge.

If one desires to keep the variables that influence a roast to a minimum, it might be a good idea to keep drum speed @ D9 (highest speed). In this manner, one can focus on power and ventilation with the added bonus of having an even roast due to proper mixing of the beans and especially making sure that larger batches get “thrown up in the air” and lofted properly during roasting (please have a look @ Aillio’s diagram of how air flow traverses the roaster from the back into the Bullet). Good control of air flow becomes then easy and a great way to control the roast, especially since most drum roasters rely primarily on convective heat (with the Bullet not being an exception).

Preheating temps determine the momentum of the roast and are also batch dependent, which consequently results in a specific turn around point (which if kept equal for the same batch size), will ensure consistency. On the other hand, if one reduces the batch size, but aims at keeping the turn around point the same by reducing accordingly the preheat temperature, the ensuing roast profile will deliver a proportional/“faithful” flavour profile if the other variables are adjusted accordingly.

In conclusion: each bean and harvest is different and so is each Bullet (every roaster has its own quirks which make it unique, therefore get to know your roaster “thoroughly inside-out”), no fixed rules … just a few standard parameters, which if applied methodically set us up for great roasts; have fun and keep experimenting :wink:


I agree with @blacklabs . Maybe @jacob can pin this?

1 Like

In order to avoid having too much posts pinned, I marked the most popular answer as the solution so that it can show up in the Docs - Roast World Community. This should help with its discoverability.

I’m also open to renaming the topic if that helps with discoverability. I’m all ears with possible names @prehistoricperk @blacklabs


@derrxb Not sure if @otaibimn’s post explaining convection vs. conduction is an “answer” so to speak for this thread. It’s really a separate topic on this subject while the OP is about @krame’s full flame testing of the roaster. I don’t know if there is a way to create a new thread from a post in Discourse, (and of course my replying to you here is already another tangent :slight_smile: )

We need a “Philosophy of Bullet Roasting” topic, lol

1 Like

Drama Peek GIF

Was having P9 power reduction issues with firmware 591. Updated to 605 in hopes that this will solve issue thanks to community comments.
Thanks and happy roasting.


Ran a test of full flame.

After noticing what looked like a power fall off after the 2nd or 3rd minute, I added a wattage meter inline to the Bullet.

Setting: P9
Input: stable 118v
Usage: 1550+ watts (rounded up, would be a solid 1560w)

Power was stable through most of the roast, and just as the IBGT sensor got above 95Celsius, the watt usage went into the 1540’s. Correlation at this point, more than causation.