TomC wrote: it's reading the coolness of those beans :

the reading is of the metal the beans are trying to effect/equalize. Poor contact/mass and extra thick/large probes slow this down. My bead tip TA is about 30 sec

farmroast wrote:the reading is of the metal the beans are trying to effect/equalize. Poor contact and extra thick/large probes slow this down. My bead tip TA is about 30 sec

The probe isn't trying to equalize with anything. It's already at the same temp as it's environment. I agree that thicker bean probes will affect how quickly that tip changes its ability to sense a change in temp. The only thing "equalizing" in the drum are the beans soaking up heat.

I know the discussion may seem to have drifted from the OP, but I don't think it has. To understand the profiles is to understand the errors inherent in their measurement, which are not insignificant.

I don't stand alone in my perspective. It's the relatively few people in coffee without science backgrounds that do. There is nothing unique about this cellulose and steel that a particular failed musician could possibly understand better than a successful engineer.

The purpose of the BT TC is to measure BT, however the TC can only measure itself. Because the TC has mass, it will not be immediately responsive to it's environment, which should be only BT. Ill placed, it also measures ET. Well placed, and it's good to the degree within any reasonable ROR.

Part of the problem with drum roasters is there is not one spot where only a thick bed of coffee and no heated air or radiating surfaces exist. Low and to the side of the door is pretty good. Take a look at pictures of your hero's roasters to see just how poorly their TC's are placed.

I'm not claiming it impossible to roast well without good thermometry, but I am claiming that it is an inaccurate reference when TP and 1C numbers vary from <45sec and =200C respectively.

Understanding these errors gives the charts from the world roasting competition profiles meaning that they otherwise lack. If you double TP time, you get the first moment where ROR is meaningful, and you can extrapolate from error of 1C temp the actual profile.

The winning roast is a 4/3/1.5, charged with high heat tapering off aiming for a relatively gentle 10C/min entrance into 1C. Given the error, it could be as high as 12C/min. It's not possible to know if the heat was cut at 1C or not, since the final temp reading is below actual 1C temp. Seeing the upturn at the end of the roast ET ROR, I judge heat was not entirely cut off at 1C.

My roaster doesn't have thermal mass to coast with, so imitating this profile and leaving the heat alone going into 1C resulted in a very tasty brew. Big fragrance, aroma, sweetness, and acidity. I'll be playing with this concept more.

Which few people in coffee without a science background would that be?

The winner Jacky of the WCE roasting competition is from Kaohsiung, Taiwan. Here is his FB page for his shop Oh! cafe (握咖啡). Taiwan has become more prominent in coffee roasting, as last year the 2nd place was from Taiwan, as well as 1st place, Fika Fika, in the Nordic Roaster Competition in 2013.

https://zh-tw.facebook.com/ohcafe.shop

Here is blog about his shop. The KapoK roaster can be seen in the background.

http://imsean.pixnet.net/blog/post/3940 ... 頭---握咖啡-oh！cafe-%5B外帶咖

chang00 wrote: Taiwan has become more prominent in coffee roasting

Taiwan has been a hotbed for specialty coffee for a while... With people like Simon Hsieh leading the scene.

Has anyone found out what bean they roasted?

Thanks for sharing the links, Farmroast... very interesting.

I did run across an interesting tweet on turning point from Neal Wilson, but this probably deserves it's own thread.

Neal Wilson wrote:Considering adding an option for thermal lag compensation to Typica. Wonder what people would think seeing effectively no turnaround point.

endlesscycles wrote:Late turning points (more than a minute) and low 1C numbers (below 200C) are clear evidence of poor placement.

This is what we surprisingly discovered with the first roast using our GW15A, few years ago. We were shocked looking at the TP in 3 minutes from charging. Also BT at First Crack were much lower than in our former roaster with our custom placed probes.

But the truth is more complex and more scientific. The probe is not as problematic as their temperature computing. In all profiles from WRC, Cropster PCs were connected directly to roasters via the Ethernet link. So all profiles show us temperatures measured by the Giesen hardware/software system. But these temperatures are not exact, not strict, not real.

Lately, we changed the fabric probe to a new double sensor PT100 in our GW15A. One sensor is connected to Giesen system, the second is connected to the Artisan via the Phidgets 1046 4-bridge. We knew about data manipulation, but cannot imagine the amout of the manipulation. One can name it "digital signal processing DSP". Just look at these comparisions. In both pictures BT measurements are shown.

A roast profile. Blue line --> strict measurements with Artisan/Phidgets 1046; Red line --> Giesen. Note the TP.


Auto mode, ET set to 220 Celsius. Red line --> strict measurments with Artisan/Phidgets 1046; Blue line --> Giesen.


Any conclusion? The temperature measurement is somehow distorted (delay, lower values, mean value shift).

We are seeing a very slow response time from the Giesen stock probe. A curious byproduct of this error type is a flattening of the curve. It's possible declining ROR curves read as mostly flat post turnaround. I definitely prefer an overly fast thermometer than one that is sluggish.