Who is here? 1 guest(s)
 Print Thread
Am I baking or am I just 'Green' at roasting?
My roaster set-up: Pop-Lite air popper with Chicago John's funnel hack to ensure good bean mixing. Separate power to heat and air- air is always full power (not moderated speed). Temp sensor is a Pt RTD that comes through the wall of the funnel and sits in the center of the bean mass. This also means the tip of the probe is in the center of the funnel axis- i.e., in line with the air flow, but covered by rising beans. Temp/ heat control is via a lab process temp controller with PID that can follow any temp profile I set to within the accuracy of the RTD. Yet, regardless of the roast profile I use, I get flavorless roasts. Frankly, some of my best roasts were before I hacked the popper and just ran those out-of-control, try-your-luck, run outside to cool the beans roasts. I haven't seen a difference between pre-heating and just starting from ambient temp- it's only a 90-100 gram charge and this machine can get things hot FAST if I choose.

I've tried Colombian Supremo's, since that is usually my 'daily' choice before I started roasting, and I've tried an Ethiopian sampler since I really enjoy a good Yirgecheffe. All came out boring, although I did get one roast of Yirg to really blast me with citrus and floral 48 hours after roasting, but then faded to boredom the next day.
Bought the beans from Coffee Bean Corral and moisture loss is usually about 13%

I brew with an Aeropress. Attached is my latest roast of an Ethiopian Queen City Harrar Natural. RoR is calculated at 1 minute intervals using (Temp(n)- Temp(n-1))/(Time(n)-Time(n-1)) FC starts at 435F at 7:30.

Help? How important is fan speed control? Right now I am set up with the fan always running at full. Roast is always nicely even, just boring. First cracks tend to be spread out- am I not roasting long enough, even though these Ethiopians call for light roasts? FC with this roaster starts at 430F or thereabouts (depending upon the bean) - Would like to show you Colombian roasts, but those were done during popper transformation so not a lot of control. I've roasted these for much longer and still no joy.

Do I need to pull the T-probe closer to the wall of the roasting chamber?

I can take some photos if needed.

Monarda attached the following file:
ethiopian_queen_city_harrar_4-14-19.7z [88.73kB / 699 Downloads]
According to Scott Rao, your ROR (Yellow line in your graph) shows a pronounced flick and crash. In other words, you baked your coffee.
JitterzZ attached the following image:
Thanks- I know my graph is not well marked- haven't figured out how to annotate in LibreOffice Calc. The 'flick and crash' that you see is actually the T-drop from when I turn off the heat to stop the roast. The 'flick' is simply the app's (LibreOffice Calc) algorithm for drawing a smooth line through the points. So, that line should really go straight to the next point and then drop- that is where I turned off the heat and let the fan cool ("quench") the roast. For this bean, I read that the best flavor came from ending the roast just before or right at the end of first crack.

I've roasted beans longer than this (constantly decreasing or 'flat' RoR through 470+ degrees), and I still get unremarkable coffee. Starting to think my taste buds are dead! I slowed down the drying time and browning time to ensure complete loss of water so that Maillard could start- am I still not drying enough? Conversely, should I heat through this phase faster? Taking too long through Maillard? Or, am I reading the temp in the wrong place- maybe move the probe tip closer to the wall and out of the actual air flow?

I read the Rao blogs (and will read your links to see if I missed one), and made sure my profile would properly control RoR and avoid the crash. I am starting to wonder if I am not reading the temperature I think I am reading, i.e what seems like a perfect probe tip placement is perfectly wrong. Or, there is something about running with full airflow throughout the roast that is impacting heat soak for the beans(??)

Thanks for your thoughts- my first thought was also a RoR issue- hence I added it to my graph and made a point to set my profile to avoid it. There's something else going on and I know there are 'secrets' that Rao doesn't publish- that's how he gets his consulting fees. Fair enough- I am hoping someone in this group will have an idea of what else is happening here.

Thanks for your thoughts, keep them coming- you are always quick to try to help!
I record my temps manually. I calculate RoR point-to-point, and one of the articles from OilSlick calls RoR a moving average. So, one thing I need to do is create a 'moving average' instead of (Current temp - temp at last minute)/ Current time - last time (=1 minute). How are others calculating RoR?

If a moving average, then it is ((current temp + temp(n) + temp (n-1) +...+ temp (0))/current time (0) in minutes, yes?
RoR could be thought of a moving average of period 1, although that really doesn't make any sense to me to think of it that way.

RoR is the slope (first derivative) of the temperature of the temperature curve. If you look at the attached image, then think of the blue line as the bean temp curve, and the other lines as the RoR at the point where each straight line touches the blue line. For example, the green straight line indicates the RoR at x=1. The slope at that point is 7, which indicates how steep the blue curve is at that instant. The slope at x=4 is 19; much steeper than x=1.

Now, during a roast we can't actually determine the slope because we don't know what the bean temp curve will look like in the future. So, one approximation to the slope is looking at how much the temperature has increased since the last time we noted temperature (your first formula). For manual readings - I used to note a measurement every 30 seconds - that works fine. OTOH, if you are using a sensor that might not work so well - for example, my current setup takes about 30 measurements per second. Just looking at the current temp and the temp 1/30 of a second ago is probably going to look very noisy. We can try to get rid of ("average out") some of that noise by using some type of moving average that takes into account not just the temperature immediately prior, but the pattern of a few (or maybe a lot) of past measurements. (There are many types of moving averages, all with different formulas.)

Regardless, Rao has an always declining slope is an objective, i.e., as the roast progresses, he wants the rate of increase in the bean temp to be declining and never rising. However, you have to keep in mind his assumptions: washed high-grown specialty greens (no naturals, Robusta, or low-grown Brazils ...) roasted light (dropped well before second crack, and before the end of first crack in a lot of cases), in relatively small Probat-style drum roasters, no airflow adjustments throughout the roast, no drum speed adjustments during the roast, solid drums, no air recirculation, etc. How far one's situation differs from his wheelhouse may lead to different results.

Hope this helps. Sorry about the long-winded explanation, but I used to teach stats.

Source for the image: https://www.intma...ciples.php
baldheadracing attached the following image:

Edited by baldheadracing on 04/20/2019 10:44 PM
Wish I could be of more help, but that's about the extent of my knowledge on this. Really, it would be me writing the same posts if I were in your shoes. The more experienced members here will hopefully weigh-in so your not left with what little I've been able to add. ThumbsUp
Edit: baldheadracing beat me to it.
Edited by JitterzZ on 04/20/2019 10:32 PM
Thanks, JiterzZ, I've been engrossed in Oil Slick's blogs, so you were a big help.

Thank you so much, baldheadracing- not too wordy. I was trying to find someone's blog who explained which type of moving average Rao uses (or what apps like Artisan use). As for my manual T recording- I do indeed record every 30 seconds, but I still found that calculating every minute made for a smoother curve to tell me the trend. I was wondering if I needed to average over a group of adjacent points. I'll look more closely at your post tomorrow and maybe try some different moving average calcs on previous roasts to see if they show me anything. As you've noted, I may also not be using the 'right' beans for Rao's formula. Or, my T-probe may be positioned in a sub-optimal location. Could be too much influence of hot air temp and in fact, the beans are not that hot? (Although Temp behaviors, FCs, etc., are consistent from roast to roast.)

Maybe I need to buy bigger batches of the same beans (instead of just 1lb or the .5 lb samplers) and just keep roasting until it finally tastes the way I expect!

Still, if anybody sees something I might try, please chime in!
Here is the moving average formula used by Artisan: https://artisan-r....html#more

One thing isn't now clear to me: Are you calculating RoR for use during the roast (which I assumed), or afterwards? In the former case, you have no knowledge of the future; in the latter case you know what will happen after a particular data point, and can take advantage of that ... for example, you could make a new bean temp chart where each point is the average of the point 30 sec before, the original point, and the point 30 seconds after - and then calculate RoR as usual using the new data points.
I am indeed calculating RoR after the fact, BUT, my T-controller will do such a good job of producing a T-profile that I set up a profile ahead of time and calculated what should be RoR at each minute using (current T- Current-1T)/(Time(n) - Time (n-1)). The Temp controller then reproduced the profile that I programmed as shown by the recorded actual temps and calculated RoR. Unless my RoR calcs are insufficient, I am thinking more and more that the Temp probe is not properly measuring bean temp and is too heavily influenced by the temperature of the air flowing through them. i.e. instead of being indicative of the beans absorbing heat, it is indicating the temperature of air flowing through and around the beans, if that mental picture makes sense.

The tip of the probe protrudes to the center of the cone-shape chamber. I wonder if I want to pull it back so that the tip is closer to the wall and measuring the temp of the bean mass after its last exposure to the heat, while it is absorbing heat from the outside of the bean to the inside. (i.e., the beans are blown upward by the hot air, and fall to the side once they reach the top of the bean pile, where they slide down the sides for their next turn in the direct heat. I am currently measuring temp in the bean mass that is riding up the heated air. Maybe I should be measuring 'after' they've been exposed...)

Anyone have advice on probe location in an air roaster that might enlighten me about my problem?

Off to more Easter gatherings. Thanks for all help!
Very interesting reading for a freshy like me. I thought i'd done a decent job at my latest roast - but having a read through this, it seems i've got the dreaded flick and crash. I was focussed on timings for Dry End, F/C, Drop and was relatively happy with the 'composition' of the roast.

Was also focussed on keeping my RoR between 15-20 for the dry phase, 7-12 for the browning phase, but didn't have a plan for the final phase. Any thoughts/suggestions?

This is in a bread machine/heat gun corretto style roaster.

RoR is desirable to be continuously declining, not having horizontal level sections, as above.
And, guess what... this is much simpler to do than it sounds... all you have to do is to maintain a heat as constant as possible.
Check recent discussions about "well tempered roaster" for more insight in this subject.
Thanks for that, have read through it and makes complete sense!

Now I have the urge to roast again, even though I have more than enough coffee for the next couple of weeks!
test roast 45%; 40%; 15%
popper!!an etiopía coffees!
What are these percents, power or fan levels ?
If power, why that dramatic drop on final ?
hi!!!drying , maillard and development.
popper!!an etiopía coffees!
I put a power between 75% or 80% and after drying I lower the fan a little or not, it depends on the ROR so that it does not stagnate.
popper!!an etiopía coffees!


optopusi_dama wrote:

hi!!!drying , maillard and development.

Ah, got it, good ratio !
Jump to Forum: