Ethiopia Sidama Dereje – A couple of disasters

Lots of work this weekend on the RoastGenie — I now have the sample camera working.  All the roast profiles will have small snaps of the roasted beans.  The camera is a standard Raspberry Pi camera with the focal length adjusted for close up shots.  The resolution is fantastic – probably does not come across all that well in the profile plots though. I will post more on that later.

Today I roasted three batches of Ethiopia Sidama Dereje which ended up being an object lesson in the hazards of software bugs.  A tiny error in code caused the heater to turn off right after first crack.  I suspect these will be _really_ bad but thought I give them a rest and a taste just for reference.


Although not on my profile plots just yet, I have set the RoastGenie up to calculate development ratio (time past 1st crack / total roasting time).  I’ve not studied this carefully yet but I’ve read a little that suggests numbers between 20 and 25% are ideal.  The roast above has a development ratio of 18%.  This is probably a little higher than it actually was though since I let the roast go a little while before I realized it was actually cooling — normally this cooling time would not count.

A second roast shows the same bug.  I stopped this one as soon as I realized the heater had cut out.  It has a development ratio of 11%.


And finally a proper roast:


This one started from a very well heated roast chamber (heated to ~500F for 10 mins or more then cooled just until inlet temp was at 200F).  First crack was very well defined and energetic.  Stopped this  one well before second crack.  Development ratio was a respectable 21%.

Yemen Mokha Ismaili – Musings on ROR (rate of rise)


Just roasted a 200g charge of Yemen Mokha Ismaili from Sweet Marias.  This bean has a subtle first and second crack (burned the hell out the first batch I roasted a few weeks ago because I was not paying close enough attention).  This was from a pre-heated roaster (chamber temp around 300F) and roasted a little ways into 2nd crack.  Before first crack, I ran the roaster at full power.  After 1st crack I backed the setpoint down to 510F resulting in about 80 to 85% power.  I calculated rate of rise for the (approximately) linear portion of the bean temp curve before 1st crack (16.9F/min) and after 1st crack (8.3F/min)  plots are below.




Going to have to wait a few days to sample this — I think SM recommends at least 3-4 days of rest for this one.

Some cursory research yields this information on development ratio (time from start of first crack to end of roast divided by total roast time).    Also found some information suggesting that ROR < 5 or 6 F/min is problematic.

A new control panel for the Roast Genie

The Roast Genie is on about the 5th generation of hardware.  After trying several different approaches to get a nice looking panel, I came up with this method using adhesive vinyl used for signs.


Details on how I did this are posted here.  I really have not seen many good ways to mount LCDs that covers up the ugly metal frame around the LCD.  I found some 3D printed bezels but nothing commercially available.  I was pleased with the result.

Guatemala Huehuetanango Cuilco


Another pass at perfecting this roast… After finding out how important pre-heating was, I did this roast with the chamber starting at about 220-250F.  I started the roaster empty set at about 515F and then let it cool down (so the heating element did not fry) until the chamber was somewhere around 250F.  Check out the post on effects of preheating for more information there.

Being impatient (and out of coffee) I actually tasted this one before the beans had cooled all the way down.  Results already MUCH better that the 11 Nov roast which was really terrible.


Effect of Pre-heating

So I found this to be astonishing…

On October 30th I did this roast no doubt one of the best roasts I have done recently.  The balance in this cup was perfect.  It produced a cup that just satisfies — you know when you get this just right.  It was that kind of roast.

On November 11th I did this roast.  Same bean, same temperature settings, same charge but a very different result.  Maybe not quite bad enough to throw out but I would have considered that if it was not the only batch I had for the week.  The dominant flavor in this roast was orange pith (NOT peel — pith,  not good at all).  You can tell the coffee is pretty good underneath but really dominated by this harsh/bitter taste.

Here is a chart comparing the bean mass temperature profile:



And a summary of the two roasts:


The good (really good) roast reached first crack about 1:18 sooner than the poor roast.  Overall roast times were about the same (right to first snaps of 2nd crack) but the good (really good) roast spent about 52 seconds longer between first and second crack.  The difference:  The good roast was the second in a series for the day — the roaster was substantially pre-heated.  From this test I have to say this makes a huge difference.  Does this slight difference in roast profile explain the huge difference in taste?

[EDIT] — OOPS the data in the table and plot above show the max bean mass temperature instead of the min bean mass temp.  This is important because of the large temperature swing that occurs as the roast chamber revolves.  Once during each revolution the themocouple is exposed causing the max reading to be about 35 to 40 degrees higher than the min temperature (when the beans cover the sensor).  I think the min value is closer.

The chart and table below is correct…

pre-heat(all 3)


Guatemala Huehuetenango Cuilco


This is a very short roast of the Huehuetenango that I have had such good results with lately.  With heat full-on until first crack and a setback to 510F, I was surprised at how quickly the bean moved to second crack (only about 3 minutes).  Some of the complexity seems to be missing from the cup.  After 2 days of rest, I really can not recommend this profile.

A tale of three roasts

This is an experiment to compare three different roast levels of Colombia Timana de Huila.  This first roast is the “calibration roast”.   Full power to first crack then roasted to the first snap of 2nd crack. In this roast, first crack started at 725 seconds and stopped 114 seconds later.  Second crack started at 355 seconds after first crack.



I stopped the second roast 295 seconds after first crack.  A full minute shorter than the first roast.  First crack started at 695 seconds and ended 111 seconds later.  I forgot to remove the outlet vent duct from the outlet until about 200 seconds into this roast.  You can really see the significant affect this has on the inlet temperature.  Outlet and Bean Mass temps are much less affected.



I stopped the third roast 235 seconds after first crack — a full two minutes less than the calibration roast.  I got distracted on this roast and, although I remembered to remove the outlet vent duct before I started the roast, I forgot to turn up the outlet setpoint to 550F at the beginning of the roast and I also forgot to set the post first crack temp up to 510F.  I really need to get the automatic roast profile setup completed for the RoastGenie.  It looks like I caught the initial setting error before it made any difference.  The post first crack setpoint mistake did result in the bean mass temperature dropping a bit. First crack on this roast lasted between 120s and 132s compared to about 111-114s for the previous roasts.