Level: Alchemist

Reading time: 15 minutes

I have recently had the chance to experiment with a very powerful 10 tray combo/convection oven. I was really excited about all the features but the results I have got so far have been disastrous! So I would be really grateful for your thoughts on resolving this mystery.
I usually follow your "basic oven roasting scheme" in my home oven with great success and satisfaction (I have even applied it to a wok roast with great results!). I then tried to apply that to the convection oven. Pre-heated to 350F and programmed to 10 min@350F, 5 min@325F, 5 min@300F and 10 min@275F as a start. 600 gram of beans spread in a single layer on a single perforated pan so plenty air all around them. 
After about 5 minutes already the beans started popping in the air like popcorn, measuring 350F on my IR thermometer! And after 12 minutes they were all burned and dead. I know you should look at and smell the beans and not blindly just pre program a profile (it was a first test). I also know that this oven has powerful air circulation and much faster recovery times than a home oven. But 350F is 350F. Why do the beans reach oven temperature so much faster in the convection oven than in my home oven? Is it really just due to these mentioned factors? I have done a couple of roasts now with lower temperature settings and more beans on one pan (about 1 kg making 2 layers). But every time the beans tend to reach oven temperature really fast. 
I have since been reading thoroughly through your roasting profile articles (which are nothing less than brilliant!) and I know they apply only to drums. BUT I simply won't accept not being able the make profiles for a convection oven. :) I know many people use them and they have their advantages in areas like steam injection for better bacterial kill and the option to roast nuts and dehydrate things as well. 
So if I were to develop a profile for a convection oven that wasn't a "long and low" profile but looked more like yours in time efficiency and flavor development, using the 4 phases you describe, how would you approach it considering the above issues? Would you maybe set the oven temperature curve to just follow the desired bean temperature curve (of course when you know your beans and how they roast best) because of it's efficiency in heating up the beans fast? Or would you put the beans in glass dishes like in a home oven to slow down the heating and stir every 5 minutes? Or would you simply give in to making a long and low roast instead? In these ovens you do at least have a large window for great visibility of the beans and you can open the door and smell them without much temperature drop, so it does offer some control.
Even though I know you are not fond of these ovens, I think your detailed thoughts on how to approach them would be highly valuable to the people who use them and a relevant addition to your roast profile articles. A passable profile template must be possible to develop somehow so you don't have to waste a ton of beans every time you get a new origin...

I really wanted to try and cut that question down, but there is so much good stuff in there I have just left it and I am going to do my best to answer it all.

I need to get something out of the way up front and I will take full responsibility for not making it clear previously.  It is true I am not fond of oven roasting. The key is that is has virtually nothing at all to do with the results.  You are 100% correct that there is some amazing chocolate out there that is made from cocoa beans roasted in a convection or even conventional oven.  I never said otherwise.

I am not fond of oven roasting because for two reasons:

1)         It is basically impossible to share profiles.

2)         It is basically impossible to change the profiles systematically and get a predictable result.

I am sure there are exceptions to this, but so far my observations of makers who successfully use ovens to roast are that:

1)         They take many roasts (dozens in some cases) to dial in a good one.  I’ve often heard about using a full 150 lbs bag to get a good chocolate, in increments of a few pounds at a time.

2)         They only use beans that work for the standard in house ‘profile’ they use.

Number 1 occurs because oven roasting seems to thwart your best efforts to subtly adjust the actual temperature profiles of beans.

I’ve spoken to more than one maker that has turned down great beans because they could not get them to taste good with their house profile.  And in this case I mean oven settings when I say profile.  Over time they work out how much to load into the ovens, how often to stir and to what surface temperature (or just time) makes good chocolate and use those settings over and over, rejecting beans that don’t work with that profile. Many thought they were just bad beans "because my profile is good". 

Let’s circle back to this “basic oven roasting scheme”.  This is a variation of why I don’t like oven roasting.  Those setting work pretty well for a home oven with 2 lbs of beans.  Most home ovens have a range of BTU output and I have defined the weight.  They problem you ran into is that you tried to apply it to an oven with a radically higher BTU output, is convection (which increases efficiency some 40%) and you only used 600 grams.

My point is you should not be surprised it failed.  Think of it this way.  I tell you how to get to a certain speed in an old inefficient 1968 Ford Pinto that has 89 hp.  I say load 4 adults in the car, and press on the pedal until the tachometer reads 3000 rpm and you should reach 60 rpm in about 1 minute. Instead you load just yourself into a new 435 hp Mustang GT and put it at 3000 rpm. Are you the least bit surprised you hit 60 mph in under 10 seconds?  No, I bet you are not.

I hope that solve the 'mystery' of why your high powered convection oven gave you poor results.

Temperature does not equal energy input.

That is all an oven setting is.  It is the tachometer reading, NOT your speed or acceleration.

Let’s talk about this:

“ BUT I simply won't accept not being able the make profiles for a convection oven”

 My first response to that that is the nice thing about science.  It doesn’t care if you believe or accept it.  No matter how much you won’t accept that perpetual motion machines are impossible won’t change the fact that they break the laws of physics.

In this case, we have a case of minor semantics.  I’ve NOT said you can’t have profiles in an oven.  You can.  But they are basically not highly adjustable.

In order for a profile to be adjustable you need to be able to make a change to your heat input and KNOW  that it will result in a predictable change in your profile every time.  In a drum roaster, I can turn down my gas flow 10% and know my ramp is going to likewise slow down about 10% give or take a little.  And that is true for any drum roaster.  Maybe it is 8% in this one or 12% that one, but it is predictable. 

That doesn’t happen in an oven for a couple reasons.

1)         You are only turning down the tachometer per se.

2)         There is too much inefficiency and chaos in the system.

The second one is the real killer to predictable profile.  In a drum roaster, since the beans are tumbling, the heat transfer is very efficient and predictable.  If you turn down how much power you are putting into the roaster, those changes are transmitted very quickly to the beans and all parts of the beans.  In the oven the beans are on a tray.  Even if you have a lot of space around the beans, there is still going to be poor heat transfer on the lower side of the beans and you are down right insulating the bottom of the beans from quick changes because they are touching the pan which has a large heat mass (relative to the beans).

And here is the physics part that foils you.  There are equations out there that in theory will allow you to model the heat transfer gradients from oven temperature to the beans.  If you dig into those differential equations you find out that you need at least 3 and more like 5 to model the system.  There is one for the top side of the beans where there is some convection, one for the bottom where there is little convection and the point contact where there is no convection but only conduction.  Plus you have to model how each affects the other.

The issue is that they don’t change in anything related to a linear fashion and they change depending on the surrounding conditions.  Basically it is a chaotic system and those kinds of systems are in practice impossible to solve and predict on the fly.  The best analogy I can come up with is the 3 body problem.


Tiny changes in initial conditions result is radically different orbits.  The same holds true with oven roasting.  Minor differences in weights,  moisture and even bean position can cause different seemingly random outcomes.  There are a few stable solutions for the 3 body problem


But many of those (or similar) start off looking stable and then fall apart.


Sound like your oven roasting experience?

Drum roasting is more like a pretty standard two body orbit.  There are tons of solutions, they are stable and changing conditions allow very predictable results.


Basically, the only semi-workable solution in an oven is the old ‘long and low’ because it slows the system down so much that the convection, conduction and radiation heat transfer equations start behaving about the same and for long enough for the roast to complete.  Unfortunately you have to give up control of the individual profile segments.

Otherwise you are left with changing a condition (like lowering the initial temperature from 350 F to 325 F) and hoping for the best since you have no way of knowing how the pretty minor change will affect the entire, complex system.

Yes, I am making this sound very complicated.  That is my point.  It is.  So complicated that to my way of understanding it isn't solvable.

A last example.  You have a complex stunt driving course.  You can run it pretty well at a fast speed in a nimble small sports car that can brake, accelerate and turn in a controlled fashion but it is impossible for a large 2 ton truck to do the same.  It has too mass to brake or accelerate quickly, and doesn’t have the turn radius for the tight turns. The only way for that vehicle to make it through the course is to slow way down and do it slow and steady.

So what are my suggestions?  I have few and I’ve given them.  My “basic roasting oven scheme” is one semi stable solution to this 3 body problem.  Once you get outside of the conditions I’ve outlined you are in a chaotic system and you have to find your own stable solutions based on your own conditions.  Most people opt for long and low, since there seem to be more stable solutions there but they don’t work for all beans.

There you go. Those are my detailed thoughts on oven roasting.  Please don’t waste time, energy and most importantly beans, searching for the universal solution because that way lies chaos and until you recognize that you are just trying to invent the perpetual motion machine.  It just does not exist and can’t exist.  If that is what you have, ok, I get that.  Work with what you have.  But it is up to you to find out what works for you.  I wish I had better news.