Level: Alchemist

Reading time: 15 minutes

I have been roasting the Belize Maya Mountain beans in 2lb batches in my home oven. So far I have tried two different profiles: 
1st - x/5.1/3.3 F/min @ 261F EOR (total time in oven: 22:45 mins) - oven set to 300F for entire duration.
2nd - x/17.6/11.5 F/min @ 261F EOR (total time in oven: 8:43 mins) - oven set to 375F for 6 minutes, then lowered to 350F
In my first roast, I noticed some heavy astringency and bitterness with very low fruit/acidity and a slightly lower chocolate flavor. After reading through your roast profile series, I noticed one paragraph that really stood out to me (link: http://chocolatealchemy.com/blog/2017/5/11/pnbipl43kdizhpho8hqlcwngi40kky):
"I’ll also make one more observation about roast 3.  It is pretty classic ‘long and low’.  The result totally tamed the bracing acidity.  True that.  But it also didn’t develop the chocolate and fruit (low development ramp) and the extended finishing phase heightened the astringency naturally present in the bean that was not noticeable when chocolate and fruit were present."
I found that this paragraph seemed to describe my first roast with the heightened astringency. I then decided that I was going to try and bring out the fruit in the second roast by going faster and hotter than before. Well... I sort of went overboard, though I definitely got some much stronger hints of fruit this time. While I'm pleased with the fruitier flavor, I'm still detecting some higher than I'd like presence of bitterness/astringency in the second roast as well. Any idea what I could do differently to lower the detection? 

I'm really glad to see you trying to use profiles.  The astringency is due to the fact that you are severely under roasting.

Unfortunately, the biggest issue here is that you are picking and choosing how you apply what I’ve written..  I’ve tried to lay out in multiple places that the profiles I discuss are ONLY good for drum roasting and that you can’t apply it oven roasting,  I’ll own that maybe I need to do a better job of that.  Without repeating myself, it is laid out in Ask the Alchemist #202  

Looking at your data though brings up another important reason that I wish to dive into.  It has to do with knowing what to expect when you roast and how to evaluate the data you are collecting.

What jumps out at me when I look at you profiles is that they are simply not possible.  I layout a basic oven roasting scheme (I’m not going to call it a profile) like this:

  • ·         375-400 5 minutes
  • ·         350 5 minutes
  • ·         325 5 minutes
  • ·         300 for 10-15 minutes or until done.

That scheme is 25-30 minutes and the oven isn't set to 300 F until the end yet you roasted at 300 F for barely 20 minutes.  Those beans just can’t be roasted.  There wasn’t enough heat for long enough.  It would be like reading a recipe for baking a pie at 350 F for an hour yet you decide to bake it at 300 for 40 minutes and expecting it to be done.

Good job on trying to ramp up the next one, but it is even worse in regards to being under roasted..  You started where I say to start (375-400) but stop after 10 minutes yet I say that you will need to go another 20 minutes or so.

My point here isn’t to make anyone feel bad.  My point is that you need to roast with some idea of what you are going to see and with a sense of expectation and if what you are seeing doesn’t meet your expectations, you need to find out why.  I want you to know and catch that if I'm calling for a 30 minute roast, you can't possibly be done in a fraction of that time.

In this case it has to do with you presumably adjusting the times and temperature of the oven based on what you think is the temperature of the beans. The key here is recognizing that since the roasting scheme was not working (since it was colder and shorter than it should be) then the data you were getting back was not accurate. And you should not feel bad.  This is a pretty classic error many non-scientists (and a few scientists) make, namely knowing how to properly take measurements and then evaluating those measurements to make sure they make sense.

And make no mistake, you have to be careful it doesn’t lead to bias in the data.  It’s all a matter of your approach. 

I have a remote thermometer I use that tells me how hot it is outside.  But I didn’t just put it outside and go my merry way.  I looked for a place where it would give me the kind of data I expect as the day gradually heats up AND roughly matches other sources.  Have a look.

air temp.jpg


Clearly I didn’t want it on my black grill where the metal absorbed heat and gave me very high bias (over 100 F on a 75 F day).  Likewise I didn’t want it in the grass exposed to the direct sun as it was inconsistent as the sun when behind clouds (see how the temperature bounce up and down).  The South porch was the right shape (it rose and then fell)  but trapped too much heat and didn’t cool off over night. Conversely the North side of the house, although out of direct sun, never quite warmed up to data from other sources.  I finally found a place under the eve of a greenhouse that was out of direct sun, on the south side and didn’t heat up or trap heat.  What I found was my little micro climate is pretty consistently a little hotter than the local airport reports (well, a lot hotter on August 2 when they reported 102 F)



The entire point of this is that your data HAS to be high biased. There is no other explanation.  Part of my intent in giving those roasting guidelines were so that you knew what to expect from any measurements you might take.

On a slight tangent, that second profile is likewise just not possible because the ramp is just too steep.  You just can’t raise the temperature of cocoa beans 17 F/min during the development phase without smelling them scorching on the outside.  And for those that want to geek out on the numbers, you can calculate  how much energy is being pumped into the system and see there just isn’t enough energy in a standard home oven to heat up 2 lbs of cocoa beans that quickly.

 Do I expect you to know or do that?  Not really but hopefully you get to the point of noticing if it looks like your roasts are breaking the laws of physics.

So what can you do about it?  The main thing you have to understand is that thermometers are not magic. As you can see, they are only as good as their placement.  If you put one somewhere that is biased hot or cold, you will get hot or cold data.  That means don’t put it next to the cocoa beans where it will read the air temperature, or on the floor of the oven where it will see wild temperature swings as the oven cycles off and on.

You are going to be tempted to put your thermocouple (the business end of the thermometer) into the bean mass.  Unfortunately it isn’t that simple. Via email I found out that that is where the thermocouple was in the case above, the thought being that as long as the beans were touching the probe it would work fine.  But as you can see by the data, it doesn’t work.

It doesn’t work because it isn’t magic.  Physics and heat transfer get in the way. Just by putting the probe in the right area doesn’t assure you get a representative temperature. This (via awesome alchemist Paint art) is how the probe was basically set up.

probe in beans.jpg


I know it basically looks fine but what can (and did) happen is that that length of metal that the thermocouple bead is embedded in….ok, hold on.  I guess I need to explain the flavors of thermocouples.  Most people think of this when they hear thermocouple probe.

thermocouple probe.jpg


But really what you see there is only the metal sheath.  The temperature is actually measured at the welded junction of two wires and that is slipped inside a metal tube.


A small voltage is applied via those wires and a current flows.  The amount of current flowing is proportional to the temperature.  As the temperature changes, so does the current (because the resistance through the bead varies with temperature changes) and that current can be measured.

But that isn’t all of it.  I mentioned thermocouples come in flavors.  What I mean is that you can mount that bead different ways inside the metal sheath.  It can be touching (grounded), not touching (ungrounded) and exposed.

thermocouple junctions.jpg

Each has their advantages and disadvantages and it is up to you (with my help here) to pick which one is best for your purpose.  If I had to guess I would say the one used in this question was grounded.  Because the bead is touching the metal and metal conducts heat quickly these respond faster than the ungrounded option.  In the case here though that also means that heat can travel up the probe sheath from a hotter part of the oven and make that bead think it is hotter than it is right where there bead is.  The result is a high bias.

It is still possible for this to happen with the unground because the metal can heat up if you are exposing a large amount of it to high heat, but because the bead isn’t touching, the chances are less.  Ungrounded beads are my preferred probe style in my drum roaster since the slight lag time of the heat getting to the bead compensates for the slight heating up of the probe metal.

In theory you could use the exposed bead, but if it is not touching a cocoa bean then all it is reading is air temperature which can lead you to a very high bias.  And even if it is touch a bean, it is only one bean and isn’t going to be representative of all the beans.  I like exposed beads if I am measuring the air temperature of the oven.

Ok, back to where we were. The metal sheath can heat up giving your readings a high bias.  You can minimize that by making sure the length of probe is covered by beans and not touching the hot pan.

Better yet, don’t use a thermocouple at all.  Use an IR thermometer.  It works by taking an average of a patch of beans and gives you an automatic average.  Just be careful the cone of the reading isn’t hitting the pan or oven walls and do it immediately after stirring your beans.

IR cone.jpg

Even the IR thermometer won’t really let you profile roast in an oven but will give you a pretty good indication of your EOR temperature and that can be quite helpful.

I hope that gives some insights into temperature measurements, how to do them and what to look for when placing your thermometer probe.  Good luck and keep those questions coming in.



1 Comment