Roasting Notes from the Oven Based Coffee Roaster

My mark 3 roaster is the bastard offspring of a rotisserie oven and a convection roaster. If you got here directly somehow, you can read the main page about the oven roaster or the highly pictorial build page.

Uncontrolled Roast

This is a roast without any control. It was simply set to 250C on a PID controller and when the roast looked done I tried to cool it. Unfortunately, I hadn't planned that step properly and it ran on too long, resulting in a horribly over-done roast, as you can see.
3/1/09, 150g Colombian (but quite old - I forget exactly what it is). The ambient temperature around the roaster was about 0C (maybe a bit below, actually), no heating profile, just PID control set to 250C and beans in from cold.
The temperature ramp was roughly 50C @ 65 seconds, 100C @ 100 seconds, 150C @ 145 seconds, 200C @ 205 seconds, 220C @ 235 seconds, 240C @ 270 seconds 250C by 300 seconds (5 minutes). This is plotted on teh graph.
First crack started quite slowly, but was at more than 1 per second by 7 minutes and 10 seconds.
At 8 minutes, the PID was reporting that it was running the elements at 20% to 30% power (though it moved around quite a lot) suggesting this arrangement has plenty of reserve for bigger batches.
At about 9 minutes 15 seconds, the cracking suddenly took off. This must have been first running straight into second. At 9 minutes 30 seconds I cut the heat, but it took a little while longer (another 30 seconds?) to get the cooling started. Cooling was just me opening the door and blowing an air gun in the front of the oven, so not highly effective.
The result is the massively over-done beans shown. Very dark, very oily. Probably not nice to drink, but I'll give it a go. To get to this in under ten minutes is also too fast, in my opinion. However, it proves that there's enough heat available for this size batch.

First Roast

4/5/09, 200g of the same Colombian (so now even older and I still don't remember exactly what it is).
Much slower than uncontrolled roast above, this was a 12 minute roast to 240C, actually following a profile that gave me a darkish roast in the popper roaster. Here, it was well into first crack, but still cracking when it got to the cool step, so the roast develops rather slower than in the popper. I think it wants a bit longer at the same peak temperature.
This photo is a bit fuzzy, but the bean cross-section shows decently uniform colour (though my camera's colour rendition is a bit flaky with so little to choose from - the colour in the main picture is more representative): .

Second Roast

5/5/09, 150g of the same Colombian.
Smaller batch, it progressed faster than the first roast. Odd cracks at 7m10s, 8m40s, building from about 9 minutes and running at 1 per second from about 9 minutes 24 seconds. That stopped by about 10m50s, then second crack started at 11m55s and was at one per second by 12m15s.
I had set the final soak to 5 minutes, with the intention of stopping the roast when I thought it had gone far enough. I did that at 12m25s by switching off the elements.
You can't really tell from the photos, because this one is taken with a flash (I did the roast too late in the day for a decent daylight picture), but compared to the first roast the beans are somewhat darker, and a hint of gloss without actually showing oil. Ideally, I'd probably have cut the roast a little earlier - say 12m15s.
After 24 hours resting a very few of the beans (one in 20?) was showing an odd spot of oil. Next roast, I'll try 12m15s.

Third Roast

9/5/09, 150g of the same Colombian. No photo because it was too dark again, and I don't think the photo with flash is particularly helpful.
Program modified as above (ie, roast to 12m15s). Also, this has a slightly wider proportional band on the controller - 20 degrees for all stops, rather than 15 which was used for the above. The most marked change on the data is that it seemed to ramp up faster in the initial heating stage. I'm not sure why that would be. This graph now has the percentage power output line (purple line, constrained to the 0-100 range).
First crack was well under way at 9m30s, and almost over by 10m35s. Crack started up again at about 11m50s, well under way (more than one per second) by 12m05s, and heat stopped at 12m15s, but cracking kept going for several seconds. So it developed a bit faster than the second roast (getting to key steps about 10 seconds sooner).
The roast seemed to be basically identical to the above, maybe slightly fewer beans were spotting with oil at 24 hours.

Fourth Roast

11/5/09, same again - 150g of the same Colombian. No photo because it was too dark again.
This is the same program. The reason for including it is that it doesn't have the oscillating temperatures that the other roasts have had. I have no idea why this is - it is an identical program on identical version of controller software as roast 3 (immediately above).

back to the oven roaster build page
back to the main oven roaster page
back to the top roaster page
back to Ian's contents page
back to www.astounding.org.uk top contents page

To comment on anything (please do) email ian@astounding.org.uk

Uncontrolled Roast
	This is a roast without any control. It was simply set to 250C on a PID controller and when the roast looked done I tried to cool it. Unfortunately, I hadn't planned that step properly and it ran on too long, resulting in a horribly over-done roast, as you can see. 3/1/09, 150g Colombian (but quite old - I forget exactly what it is). The ambient temperature around the roaster was about 0C (maybe a bit below, actually), no heating profile, just PID control set to 250C and beans in from cold. The temperature ramp was roughly 50C @ 65 seconds, 100C @ 100 seconds, 150C @ 145 seconds, 200C @ 205 seconds, 220C @ 235 seconds, 240C @ 270 seconds 250C by 300 seconds (5 minutes). This is plotted on teh graph. First crack started quite slowly, but was at more than 1 per second by 7 minutes and 10 seconds. At 8 minutes, the PID was reporting that it was running the elements at 20% to 30% power (though it moved around quite a lot) suggesting this arrangement has plenty of reserve for bigger batches. At about 9 minutes 15 seconds, the cracking suddenly took off. This must have been first running straight into second. At 9 minutes 30 seconds I cut the heat, but it took a little while longer (another 30 seconds?) to get the cooling started. Cooling was just me opening the door and blowing an air gun in the front of the oven, so not highly effective. The result is the massively over-done beans shown. Very dark, very oily. Probably not nice to drink, but I'll give it a go. To get to this in under ten minutes is also too fast, in my opinion. However, it proves that there's enough heat available for this size batch.
First Roast
	4/5/09, 200g of the same Colombian (so now even older and I still don't remember exactly what it is). Much slower than uncontrolled roast above, this was a 12 minute roast to 240C, actually following a profile that gave me a darkish roast in the popper roaster. Here, it was well into first crack, but still cracking when it got to the cool step, so the roast develops rather slower than in the popper. I think it wants a bit longer at the same peak temperature. This photo is a bit fuzzy, but the bean cross-section shows decently uniform colour (though my camera's colour rendition is a bit flaky with so little to choose from - the colour in the main picture is more representative): .
Second Roast
	5/5/09, 150g of the same Colombian. Smaller batch, it progressed faster than the first roast. Odd cracks at 7m10s, 8m40s, building from about 9 minutes and running at 1 per second from about 9 minutes 24 seconds. That stopped by about 10m50s, then second crack started at 11m55s and was at one per second by 12m15s. I had set the final soak to 5 minutes, with the intention of stopping the roast when I thought it had gone far enough. I did that at 12m25s by switching off the elements. You can't really tell from the photos, because this one is taken with a flash (I did the roast too late in the day for a decent daylight picture), but compared to the first roast the beans are somewhat darker, and a hint of gloss without actually showing oil. Ideally, I'd probably have cut the roast a little earlier - say 12m15s. After 24 hours resting a very few of the beans (one in 20?) was showing an odd spot of oil. Next roast, I'll try 12m15s.
Third Roast
	9/5/09, 150g of the same Colombian. No photo because it was too dark again, and I don't think the photo with flash is particularly helpful. Program modified as above (ie, roast to 12m15s). Also, this has a slightly wider proportional band on the controller - 20 degrees for all stops, rather than 15 which was used for the above. The most marked change on the data is that it seemed to ramp up faster in the initial heating stage. I'm not sure why that would be. This graph now has the percentage power output line (purple line, constrained to the 0-100 range). First crack was well under way at 9m30s, and almost over by 10m35s. Crack started up again at about 11m50s, well under way (more than one per second) by 12m05s, and heat stopped at 12m15s, but cracking kept going for several seconds. So it developed a bit faster than the second roast (getting to key steps about 10 seconds sooner). The roast seemed to be basically identical to the above, maybe slightly fewer beans were spotting with oil at 24 hours.
Fourth Roast
	11/5/09, same again - 150g of the same Colombian. No photo because it was too dark again. This is the same program. The reason for including it is that it doesn't have the oscillating temperatures that the other roasts have had. I have no idea why this is - it is an identical program on identical version of controller software as roast 3 (immediately above).