I'm scratching my head about this one.

I set up Artisan with the same TC4 type of themocouple Arduino system I use on our Huky to an Artisan 6 fluid bed roaster. Ambient temperatures were correct but end of dry occurred at a bean probe way over 300F and in general temperature readings were 30 deg F higher than expected. Ambient temperature readings were correct. For example, we dropped beans in second crack, fairly early, but the BT read 479F. The bean probe was the same type K TC he had been using and his drop temps for the same time roast was 450F. However the ET probe (which sits in the mouth of the air flow into the roast chamber) was 1 cm closer to the wall.

I switched to Celsius for a batch, trying to mimic my HUKY profile, and dropped before the end of first crack, BT read 218C. (???!!!)



As I write this, I think what happened is: The ET was fairly hotter than the ET probe was reading. The BT reading in such a roaster is offset upward by the fact that the heat transfer is convection. That is, the BT probe reads hotter than the true BT and if the roaster sets ET say 40 deg F hotter than expected (due to inaccurate information) one may reach the landmarks at a higher temperature.

My air roaster reads lower ETs and higher BTs than my drum roaster for identical profiles and taste. Not sure exactly why except that the the positions of the probes are not readily interchangeable. In a drum roaster you are readuing the ETY at the drum surface, the BT inside the beand mas. In a fluid bed, you read the ET where the air enters a the bottm, and the BT where the air leaves, at the top. Moreover, there are several different air roaster designs -- pure fluid bed, rotating fluid beds and bean fountains -- they all have their own thermometry issues.

This is why I think a lot of the profile advice is BS: roasting gear is too individual for the kinds of one size fits all rules people imagine they need.

I've been struggling with replacing the basic thermometer that came with my Artisan 6 with a data logging probe for use with a Phidget 1048 and Artisan software.

I'm learning that placement of the probe is tricky. In order to be able to do small sample roasts of 200g or so, the probe needs to be as close to the bottom of the conical roast chamber as possible. Too low and it will get into the path of the hot air and you will get a hybrid reading of BT and ET.

Here is a small experimental roast to illustrate the issues:

First, I calibrated the probes. The TC I used read 212* in boiling water:



The probe on the All Clad thermometer that came with the roaster read 209*:



I placed them side by side on the probe bracket:



At rest they read 3* apart, in line with their calibration numbers:



Beans added:



At 1 minute the All Clad thermometer took the lead:



Same at 2 minutes:



Same at 7 minutes. I noted that dry end occurred when All Clad probe was 302* and Artisan TC was much lower.



BTW, here is my TC placement for ET. The one on the left came with the machine, the other I added for logging purposes:



and at 8:10, this is what they read on their respective devices. That's a 43* swing. Hence the issue with TCs and placement!



By 11:00 the All Clad and Artisan probe evened out.



And this was the final profile.



Not my best work, but this was not about roasting, but about data logging.

The beans look OK for an Ethiopian natural. I'll try them later.

So what was learned by this? Maybe nothing new. TC placement is critical. A few millimeters one way or the other results in significant differences in readings. The key is to find a place that reads similar to what others get for Dry and 1st crack and then just learn your particular roaster and compare your results from batch to batch instead of results others get on their respective roasters. Again, nothing new.

As an aside: Fluid bed machines are a different animal than drum machines. It takes way more air to loft the beans in the beginning of the roast when they are small and moisture laden than it does at the end when they are larger and lighter. This necessitates continually turning down the air flow so the beans don't fly out the top. When the air is turned down, it results in an increase in ET because air is held below the bean mass for a longer time. When ET goes up, so does BT.

So turning down the air equates to increasing ET and an increase in delta BT. This works out well in practice by helping maintain a nice decreasing RoR curve, but it wreaks havoc on Delta ET curves.

As I'm trying to get my head around roasting (about to start my first experiments) inwas reading this topic and wondered...

Can it be explained by the fact that the mass of beans per volume of air is different between various types of roasters?
Probes differ, so comparing those will not say a lot, the location of the probe will have an impact but what are you trying to measure? Air temp or bean temp? Assuming it's bean temp, but you are likely measuring a mix of both depending on the contact area between the probe and air/beans.
I apologize for any ignorance at this stage...just a thought.

Marcelnl wrote:As I'm trying to get my head around roasting (about to start my first experiments) i was reading this topic and wondered...

Can it be explained by the fact that the mass of beans per volume of air is different between various types of roasters?
Probes differ, so comparing those will not say a lot, the location of the probe will have an impact but what are you trying to measure? Air temp or bean temp? Assuming it's bean temp, but you are likely measuring a mix of both depending on the contact area between the probe and air/beans.
I apologize for any ignorance at this stage...just a thought.

Ideally, it would be great to be able to stick a probe in the center of the bean, like you do with a meat thermometer. Everything short of that is a compromise and a best guess.

My thinking: Air temp around the bean is very transitive; air is an insulator and does not hold heat well. Beans are more of a conductor and the temp of the beans themselves are more stable. So the TC should give a reading based on the temperature of the outer shell of the beans themselves that are in contact with it.

That is likely where my issue stems from. Not all beans are being heated to exactly the same degree at exactly the same time. As the bean mass circulates, sometimes hotter beans are in contact with the TC and sometimes cooler beans. Overall, everything averages out, but taken in ministeps this results in temperature swings.

Thanks for clarifying that, I think I see your point in bean temp not being uniform throughout the bean but that will be the case with every type of roaster (well it will vary given the speed of heating etc).

Air is an insulator yet I"m not so sure if it is during roasting as water will evaporate and steam has a large specific heat capacity...

To me it reads as if the measurement is also depending on what air setting/drum speed is used which is something you can probably do nothing about other than picking the best place for your probe which may be difficult enough. Sounds to me that averaging or damping the readout can be beneficial too yet for now I tend to agree with Jim's assessment

When I bought my SR500 and tried to place some thermocouples I ran into the sort of issues described by the OP. I never did find a placement for BT that matched the typical figures one finds in articles in books and on the web. Eventually, I gave up and just stuck the probe more or less where I thought it should go into the bean mass. In no time at all I ignored it and became a SOTP (seat of the pants) roaster. I roasted by sound, smell and appearance with great success, at least according to everyone who tasted my coffee.

With the acquisition of a Quest M3 the first thing I noticed upon installing EricS probes was that the temperatures not only did not match the popular notions of correct temperatures for roast milestones and types of roast, they also did not match very well with the reports of other M3 owners using similar set-ups. Alas, it is way harder with a drum roaster, despite the window, trier and exhaust port, than it is with an SR500 to be SOTP. So, I continued to use the probes and eventually discovered the temperatures displayed that match in the milestones I was seeking. We're now very happy with the roasts and our ability to control them.

I suppose an alternative, should one want to be able to compare roast stages to "book learnin'", might be to use offsetting feature (calibration) of the interface between the thermoprobes and the software (Artisan in my case), so that the numbers roughly match the commonly accepted figures. If I were to do this I think I'd probably use the 1C point, which is way off (~180 C) on my quest, but which is fairly consistent from roast to roast using many different greens.

In summary, I suspect it would be much easier to put the BT and ET probes in some 'reasonable' position where the readings are fairly consistent, then calibrate and damp the readings to match accepted milestones, than it would be to move the probes about to achieve the desired display temperature.

My 2-cent's worth.

Almico wrote:I've been struggling with replacing the basic thermometer that came with my Artisan 6 with a data logging probe for use with a Phidget 1048 and Artisan software.

I'm learning that placement of the probe is tricky. In order to be able to do small sample roasts of 200g or so, the probe needs to be as close to the bottom of the conical roast chamber as possible. Too low and it will get into the path of the hot air and you will get a hybrid reading of BT and ET.

Here is a small experimental roast to illustrate the issues:

First, I calibrated the probes. The TC I used read 212* in boiling water:
.
.
.

Your Phidget us using a type K thermocouple, you thermometer is more than likely a thermal resistor, two very different probes and the later is much slower to respond and inherently inaccurate. I would say the +/- on that thermometer is also quite high over its range so you are really comparing apples to watermelons with that .