Heat resistant microphone to attach to the Gene Cafe? - Page 3

Discuss roast levels and profiles for espresso, equipment for roasting coffee.
boren (original poster)
Posts: 1117
Joined: 14 years ago

#21: Post by boren (original poster) »

@Quidgybo - the cardboard addition looks very simple. I'll try that. I already subdued the noise of the beans hitting the glass by covering the inside of the drum with BBQ grill mesh, but muffling the mechanical noise should further help.

@Marcelnl - the left side of your diagram (marked in red below) looks much "quieter" than the right. Does it represent pre-first crack phase, or am I misreading the diagram?


Marcelnl
Posts: 3837
Joined: 10 years ago

#22: Post by Marcelnl »

boren wrote: @Marcelnl - the left side of your diagram (marked in red below) looks much "quieter" than the right. Does it represent pre-first crack phase, or am I misreading the diagram?

image
the lower parts (two) represent idle noise, the higher parts are the numbers for noise when loaded, FC is not any different from baseline noise pre FC.... I'm sure it can work but not with this transducer. Perhaps Phidget has something in their inventory that is better suited (a 3D accelerometer f.e.)
LMWDP #483

tompoland
Posts: 269
Joined: 3 years ago

#23: Post by tompoland »

I bought a cheap $29 personal microphone/speaker from ebay and put a plastic tube over the microphone part to insulate it from the heat. It goes into the Trier just before first crack. Works a treat. It's lasted over 100 roasts so far.

A little obsessed.

Marcelnl
Posts: 3837
Joined: 10 years ago

#24: Post by Marcelnl »

great idea, that is in a Bullit I assume?
Not sure if the Genecafe has a trier...but the principle may be adaptable.
LMWDP #483

jfife
Supporter ♡
Posts: 55
Joined: 4 years ago

#25: Post by jfife »

It is no surprise that the noise of first crack has higher frequencies in it than the background noises before first crack. A while back I used some audio equipment and looked at the frequency spectrum in a Mill City roaster before first crack, and in a 5-second sample during first crack.
Attached are graphs of amplitude (Y) versus audio frequency (X) showing increased volume of high frequencies above 16khz during first crack.

One could filter out low frequencies with simple RC filters, and when noise above say 12K increases one could ring a first crack alarm. Perhaps as an Artisan alarm?

I can forward the Audacity spectrum files to whoever is interested.



See the frequencies above 16K are louder during FC.

Marcelnl
Posts: 3837
Joined: 10 years ago

#26: Post by Marcelnl »

interesting and indeed sortof expected...you'd need to baseline a bunch of beans to establish a safe threshold for how many dB the volume at those freq increases before making final filters though...think that it's doable but the underlying issue stays...how to get a microphone (that is heat resistant) IN the Genecafe that has no tryer and as far as I can tell it only measures (and controls) ET.

I'm not sure if my assumption that it simply sets the final ET and heats until it is reached or combines duration with a preset ET, assuming BT gets to the point where it correlates with the roast levels in the manual https://www.topcoffee.net/media/coffee/ ... Manual.pdf
If so I'd probably tinker together my own PID and BT probe and use that with the tip from Rickpatbrown.
LMWDP #483

jfife
Supporter ♡
Posts: 55
Joined: 4 years ago

#27: Post by jfife »

When I did this sound recording the microphone was several inches away from the hot drum. For a general FC alarm useful on any machine, It wouldn't need to be real close because low frequency noise and rumble can be filtered out electrically. Good idea about a background bean-baseline to set the threshold. There shouldn't be a lot of 14-20khz background ambient noise in the roast room unless someone is rattling silverware... so any uptick of noise in the 14k-16khz range might be accurate enough to trigger a first crack alarm.

We use our ears to detect several instances of cracks and pops ( I use 3 for example) before calling FC. If Artisan used popups, or speaking alarms a person could see if 3 detections had occurred and distinguish FC from an incidental noise. A sort of persistence filter.
I'm not sure how one might signal Artisan that it's time to turn on an alarm from an outside source. I see Phidgets.com has devices which can detect a voltage or event and send a signal through its USB connection. Any ideas out there? I know roasterdave was interested in an idea similar to this a while back.

Marcelnl
Posts: 3837
Joined: 10 years ago

#28: Post by Marcelnl »

makes sense! with some fancy computer based filtering it would probably work reliably!

Now for the Artisan whizz kids to translate the idea in something real!
LMWDP #483

jannus
Posts: 88
Joined: 10 years ago

#29: Post by jannus »

There was this post some time ago on the same subject..

RoastLearner for Artisan: train Artisan to listen to your roast

I've never tried it but it seems very practical? Either way it speaks to artisan as well!

jfife
Supporter ♡
Posts: 55
Joined: 4 years ago

#30: Post by jfife »

Wow! This is all put together and it works. I like the visual display showing cracks as a spike on the Artisan plot.