Automated first crack detection and analysis

Discuss roast levels and profiles for espresso, equipment for roasting coffee.
popeye

#1: Post by popeye »

I'm slowly working on a program to detect and record first crack data to see if we can extract usable info (rate and start/stop info) for correlation with cup quality. I know that this isn't the first data we want to pull from a roast but I've already got ET, BT, Drum temp, gas pressure, and some other info. I'm really curious about whether we can delay/hasten the onset of first crack for a particular coffee, and what effect that will have on the cup.

I googled and found this paper: http://asa.scitation.org/doi/full/10.1121/1.4874355 but nobody's done this yet as far as I can tell.

Anybody hear about someone developing this? Please don't tell me cropster did it five years ago and i just haven't heard.
Spencer Weber

9Sbeans

#2: Post by 9Sbeans »

I remember reading this article before. I was thinking of FFT of the sound spectrum at the time and there were several threads. :lol:

Acoustics of Coffee Roasting: Machine Learning Edition

User avatar
millcityroasters

#3: Post by millcityroasters »

popeye wrote: I'm really curious about whether we can delay/hasten the onset of first crack for a particular coffee, and what effect that will have on the cup.
We developed a first crack detector for deaf users a couple of years ago and I think Rob Hoos wrote an entire book (Modulating the Flavor Profile of Coffee) about the effects of delaying or hastening first crack (among other things, but I'm pretty sure he invented it :) ). I think we've even got an archived live stream on YouTube somewhere with him talking about his book. (Search "Minneroasta")

User avatar
drgary
Team HB

#4: Post by drgary » replying to millcityroasters »

Steve, you've been hiding it from us? On those really quiet coffees I could use some help.
Gary
LMWDP#308

What I WOULD do for a good cup of coffee!

btreichel

#5: Post by btreichel »

Somebody did this a while back, posted about it either here or homeroaster.

User avatar
drgary
Team HB

#6: Post by drgary »

Here's a thread where we wandered around the concept.

Apps that Display First Crack Visually?

But Steve says he actually has something he can supply.
Gary
LMWDP#308

What I WOULD do for a good cup of coffee!

User avatar
millcityroasters

#7: Post by millcityroasters » replying to drgary »

My genius son Nick figured it out, but we've got a long list of projects we need to complete before crack detection becomes a commercially feasible, stand alone package.

I'd suggest working the tryer more. Most competent roasts of good quality coffee present a sweet scent in the tryer after 1st crack indicating Maillard reaction in the center of the bean and the approach of complete development. That scent will dissipate within seconds and will be followed (perhaps overwhelmed) by a sharp scent associated with the cellulose off-gassing and crossing the threshold of roast flavor. I drop my cupping roasts of an unfamiliar coffee about 15 seconds/+ 2.5F after the start of of that sharp scent. If the coffee cups underdeveloped, I repeat the roast twice dropping at further 15 second and 30 second intervals. Whether by luck or design, I usually get a drinkable coffee out of at least one of them.

I'm open to other interpretation, but Joe Marocco opines the sweet scent is associated with Furfural and the sharp scent with Acetic Acid.

popeye (original poster)

#8: Post by popeye (original poster) »

millcityroasters wrote: I think Rob Hoos wrote an entire book (Modulating the Flavor Profile of Coffee) about the effects of delaying or hastening first crack (among other things, but I'm pretty sure he invented it)

well sounds like i've got a little reading to do.
Spencer Weber

popeye (original poster)

#9: Post by popeye (original poster) »

9Sbeans wrote:I remember reading this article before. I was thinking of FFT of the sound spectrum at the time and there were several threads. :lol:

Acoustics of Coffee Roasting: Machine Learning Edition
Wow thanks! Very impressive work by Allen (luma). i'll have to install the code and see how it works. Machine learning is a little more advanced than i was thinking :D
Spencer Weber

User avatar
drgary
Team HB

#10: Post by drgary »

millcityroasters wrote:I'd suggest working the tryer more. Most competent roasts of good quality coffee present a sweet scent in the tryer after 1st crack indicating Maillard reaction in the center of the bean and the approach of complete development. That scent will dissipate within seconds and will be followed (perhaps overwhelmed) by a sharp scent associated with the cellulose off-gassing and crossing the threshold of roast flavor. I drop my cupping roasts of an unfamiliar coffee about 15 seconds/+ 2.5F after the start of of that sharp scent. If the coffee cups underdeveloped, I repeat the roast twice dropping at further 15 second and 30 second intervals. Whether by luck or design, I usually get a drinkable coffee out of at least one of them.

I'm open to other interpretation, but Joe Marocco opines the sweet scent is associated with Furfural and the sharp scent with Acetic Acid.
Thanks, Steve. I'm thinking that a way to train this method is to roast the same coffee using the trier first to sample every 30 seconds and cupping the samples. Then repeat the roast and smell the trier on the way to dropping at the previously preferred level of development.
Gary
LMWDP#308

What I WOULD do for a good cup of coffee!