first profile experiments with the laptop

Roasters and roasting

Moderators: GreenBean, Gouezeri, bruceb, CakeBoy

first profile experiments with the laptop

Postby mnemonix » Mon Jan 26, 2004 1:28 am

This weekend I finally managed to combine all the various bits for the laptop roaster having finished building the thermocouple amp, so I was able to turn my mind back to the coffee roasting rather than the roaster itself.

It's early days, but so far I'm very pleased. Jim Schulman has posted extensively on alt.coffee about his profiling experiments with a variac and I decided to start by recreating one of his experiments: 3 roasts, the first typical of a popper with a steep rise in temperature at the start, flattening out towards the end, the second starting very gently and ramping up as the roast progresses (more like a hottop) and third an almost straight profile like a commercial roaster.

I tried the 'correct cupping procedure' but it went something like this: yuk... this isn't how i like my coffee, so 3 cups of filtered coffee with milk and sugar later things were tasting much better :wink:

More seriously though, the first (popper) profile showed all the deficiencies that lead me to this project in the first place, little scope to create an even light roast, far too much acidity and under developed origin flavour (requiring long rests to improve) as well as bitterness from the prolonged spell at high temps.

The second, concave - hottop roast improved on all counts and although the second and third profiles were very close, in my opinion the straight profile was the best on all counts, no acidic astringency, no bitterness, good origin flavour development and a well rounded, balanced cup.

One of the biggest surprises was how both the straight and hottop profiles allowed for good light roasts. With the popper profile colour seems to develop earlier in the roast but is usually either uneven requiring all roasts to go to a darker brown stage to even the beans out or just too acidic. With the new profiles, colour develops later in the roast (a feature of commercial roasters I've read), is far more even and controlable and I'd say might actually be lighter for the same degree of roast. I did a batch of Costa Rican and cooled them at an almost 'cinamon roast', they were very evenly tanned and without resting produced a smooth but floral cup. I couldn't have produced a roast that looked like this before and if I had I wouldn't have dared tasting it !

The features of the 'virtual roaster' - the software running on the laptop, are that I can do a manually controlled roast with realtime graphing, and also use any profile I created from a previous roast in an 'auto' mode so I can get the roaster to reproduce them without my input. It's also fairly easy to edit the saved profile data if I want to modify some aspect of it such as eliminating (or adding!) a temperature bump. I've attached a screenshot of it in operation doing a 'straightish' profile after a 300ºc preheat before dropping the beans in. The black shows the roast temperature (possibly reading a little high due to the thermocouple placement) while the red shows heater level as a percentage.

Sorry for the long post, I think in all I did around 12 roasts with differently tweaked profiles this weekend as well as reading a doctoral thesis on roasting (http://e-collection.ethbib.ethz.ch/cgi- ... s&nr=13620) so I'm somewhat saturated with data ! not to mention the best coffee I've managed to roast to date.

Chris
Attachments
profile.jpg
profile.jpg (154.63 KiB) Viewed 5212 times
mnemonix
 
Posts: 176
Joined: Fri Nov 14, 2003 11:03 pm
Location: Nottingham

Postby mnemonix » Mon Jan 26, 2004 1:57 am

a couple more profiles - the 'popper' style with an added short stabilization plateau at 230 and a gently 'concave' profile somewhat like a hottop as far as i can tell from the graphs I've seen (correct me if I'm wrong !). I hadn't written the subroutine to plot heater power when I did these and the straight faint grey profile is part of an unfinished graphical interface to input a profile so nothing to do with these roasts.

chris.
Attachments
2profiles.jpg
2profiles.jpg (141.49 KiB) Viewed 5210 times
mnemonix
 
Posts: 176
Joined: Fri Nov 14, 2003 11:03 pm
Location: Nottingham

Postby Raf » Mon Jan 26, 2004 7:34 am

Wow! That's impressive, Chris. Now you come over here and hook up my desktop to my Caffe Rosto! I want to do that too! ;)
This week I am eagerly anticipating the first god shots from my La Spaziale machine....

La Spaziale S1, Vibiemme Domobar (retd), Mazzer Mini Electronic, Behmor 1600 230V
User avatar
Raf
Founder Member
 
Posts: 1706
Joined: Sat Aug 23, 2003 9:48 am
Location: Brussels, Belgium

Postby Ian » Mon Jan 26, 2004 11:18 pm

Chris, this is the most interesting project I've seen in a long time, well done -I'm impressed. This setup makes the I-Roast look like an expensive toy.

A few quick questions for you:
a)What is the roast load?
b)Where is the TC placement?
c)What is the green two digit number under the time? I'm guessing it's the roast log no.

Ian
User avatar
Ian
 
Posts: 423
Joined: Thu Jan 01, 2004 11:07 pm
Location: Hampshire,England

Postby mnemonix » Tue Jan 27, 2004 2:45 am

Ian, the short answer is: a standard popper full. I'm using 80g as a comfortable benchmark. The thermocouple is in the roast chamber/bean mass just out of the direct blast of hot air from the vents. The number you refer to is actually a relative indication of heater power.

The long rambling answer follows but isn't necessary !

I've been using 80g as a comfortable benchmark but could go to 100g+ without having to stir the beans at the start of the roast. I'm running the fan with a 24V transformer. In the next couple of weeks I intend to add some fan control to the software, not so much to roast more beans but to be able to reduce the power later in the roast to compensate for the loss of bean mass and more importantly to avoid/investigate the possibility of 'aroma stripping'. The dissertation I referenced previously touched on air/bean ratio and suggested to me that the minimum amount of airflow necessary to fluidize the beans was preferable to create a microclimate in which the roast could develop without removing the volatiles produced. As this project develops I may try and build a larger capacity roaster and I'm looking at various components at the moment, but initially it's the nature of the roast itself and the control mechanisms and possible interfaces that are driving this. Orders from friends and my own consumption are becoming an increasingly important factor though !

As far as 'absolute' temperatures go I may be way out, due to the difficulty of getting accurate bean mass temps in this 'chaotic' environment and simply because my calibration of the thermocouple amp output may be off. Actually, I'm not placing too much importance on this at the moment, it's the profile itself that's important, the shape - the rate of change of temperature during the roast that's interesting, and it's me, using my sense of smell and sight that's deciding when certain milestones have been reached. Having said that, solid data is useful too, so a calibration session is in order as well as introducing a thermocouple into the heater chamber below the roast chamber to better understand what's going on, though I'm reluctant to do too much that's wholly specific to the thermal characteristics of a specific popper.

One interesting observation, I have noticed that at both first and second crack there is a sudden and marked variance in temperature readings in the roast chamber.

Ultimately, what I'm not aiming to do here is simply emulate a PID - setting rigid target temps or temp gradients smacks of trying to predict the future, I'd like to find a more organic approach whereby both the software and the human roaster can respond to the specifics of a roast and its variables during the course of the roast. Computer Aided Roasting rather than Computer Controlled Roasting ! The data that I'm collecting at the moment is not only educating me but ultimately, might also inform the roaster, helping to develop a set of 'rules of thumb' for obtaining types of profile whilst leaving some of the specifics of each roast to be determined during roasting.

The number you refer to is an indication of the heater power at any given time, specifically it's the percentage of the timing loop (ie: the duty cycle) that the heater is 'on' for, in the cases illustrated it's frozen at the moment the roast went into cooling. It's graphed in red on the first (but more recent) profile I uploaded. 92 roasts ?!?!? not yet ! I'm identifying each stored profile with a descriptive sentence typed in rather than a number at the moment but it needs to be added to the interface.

I'd be interested to hear what results in the cup your profiling with the variac have produced. Being quite a visual thinker I'm finding the graphing very useful both during and after the roast for understanding the effect of profiling, but I'm also spending far more time smelling and observing the roast now I don't have to manually intervene so much.

( Neku, it's certainly 'do-able' :wink: )

Chris.
mnemonix
 
Posts: 176
Joined: Fri Nov 14, 2003 11:03 pm
Location: Nottingham


Return to Roasting - Equipment and Techniques

Who is online

Users browsing this forum: No registered users and 28 guests

cron