Several times I have read that guidance for gas roaster BT RoR curves does not apply to fluid bed roasters. So far I have not found specific guidance for small fluid bed roasters like the FreshRoast or Popper.

Does that guidance exist?

In case this is useful info...

I am using a 1400w Wearever™ Popcorn Pumper with a short glass chimney. Chaff blows up into the air and is swept up after roasting. Unless the roast moves significantly into second crack there is not a lot of smoke. Roasts happen in an unfinished basement at a relative constant temperature. A Phidget™ IR sensor is mounted above the popcorn popper and tracks the temperature at the top of the bean pile using Artisan. The popcorn popper is attached to a voltage transformer so that the voltage can be controlled. As far as I can tell increasing the voltage increases both heat and air flow; decreasing it reduces both.

A sketch of the set-up is below. The height of the IR sensor is shown as 25 inches above the bottom of the popcorn popper. That was the original plan but after numerous roasts and tweaks it is a bit closer.



The IR sensor is being used out of spec. The sensor itself becomes too hot. As it heats up the readings become less accurate. The readings are pretty much spot on at the beginning and off by something closer to 50°F at the end of the roast.

The inaccuracy doesn't bother much as the curves are useful anyway. Moving forward I plan to take a few manual readings at key moments and use Artisan's Transposer tool to reshape the BT and BTRoR curves to reflect actual temperatures. Recently I started pointing two fans at the sensor to cool it during the roast. This has significantly improved the repeatability of the readings and improved the accuracy a little.

Two roast curves are below. One shows a roast with no adjustment to voltage. That is Roast 61 -- Decaf Moka Java that is taken through second crack. My wife has developed a taste for cold brew and we are honing in on the best way to roast for her decaf cold brew.

The second roast below is Roast 57 -- Brazilian Santo Antonio. The voltage was reduced about 90 seconds into the roast hoping to slow the second phase.

Over the months I have been roasting I have tried several approaches to adjusting or not adjusting the voltage (heat+air) to the system. My roasts are becoming more consistent, and I am having great fun. I would love to know if there is reliable well-grounded guidance for best practice in terms of RoR curves for small air roasters.

Many thanks.

The question was sorta touched on in another thread - this was the final(ish) post: Ikawa Home vs Bullet You can find time/temp points elsewhere in the thread.

TLDR: don't sweat RoR in a fluid bed roaster.

Peeps still do sweat it nonetheless, and for good reasons. However, I think a there's a significant issue with variability in how we taste that leads to reductive arguments to just do what you like - hard to argue with that.

I found it really (really) touchy to fine-tune a precision fluid-bed roaster to achieve a linear RoR curve, so I expect it will be even more challenging with the OP's setup - but experimenting can be fun and delicious!

minus1psi wrote:Several times I have read that guidance for gas roaster BT RoR curves does not apply to fluid bed roasters. So far I have not found specific guidance for small fluid bed roasters like the FreshRoast or Popper.

Does that guidance exist?

To answer your question, I don't know.

What I did want to do is to make the point for people reading along, because people constantly miss this point, that you can't just latch onto whatever reading you happen to get from your roaster and just apply temp or ROR readings from another roaster to it. Setting aside the nature of the probes, the probe positioning matters a great deal. For a, say, 15kg drum roaster, a 5cm long bean temp probe in the bottom of the face plate of a roaster will be immersed in a very large amount of green coffee, which probably has very good contact with the probe tip. This is the paradigm information that most of the BT ROR writing on the internet is about. If you have a 100gm popper with a standard temp probe, then, depending on the positioning, that probe may barely have any contact with the beans at all. The air may hurl them around so much that the probe barely contacts the beans. So it shoudn't be any surprise at all that if you blast a very sensitive probe in an air roaster with 300C air, it instantly heats up a great deal and you get a massively high derivative/ROR (I'll use the lets-not-call-a-spade-a-spade "ROR" expresion) at the very beginning and then it'll be super flat later on. It's not an apples and apples comparison. Particularly, I might add, w/r/t the Ikawa, which doesn't even have a probe in the bean mass at all.

Your IR temp sensor I have no idea about, but it wouldn't surprise me if it's more useful than a k-type for an air roaster.

Thank you GDM528 and Luca.

The IR sensor has its own set of issues. One was already mentioned. As it heats up it under-reports the temperature until it shuts off completely. This is not a fault of the sensor, I am subjecting it to heat well above its documented maximum environmental temperature. Another issue is alignment. Seen from the top, sez me the view of the IR sensor wants to include beans and only beans. If the angle of the IR sensor relative to the bed of beans is off a little bit, meaning not perfectly perpendicular *and* perfectly centered *and* the right height it will see the hot inside of the popper chamber , the plastic body of the popper, or both. Confounding this issue is that the only practical way to empty the popper at the end of the roast is to pick it up and dump it. (Maybe some sort of fancy vacuum I do not have would work without moving the whole thing). It sounds like that is not a problem - put on gloves dump the popper and move on. But the alignment on the rig is finicky and has to be re-assessed after every 100g batch. Not a huge deal, I rigged a plumb bob for the purpose but another place for variations to creep in. It is really the angle of the sensor relative to the popcorn popper that worries me. I have leveled both surfaces as best I can; neither is perfectly level. This seems like it could be fixed with a perfectly square jig -- 80/20 aluminum comes to mind, but that is another $100+ USD and still has the issue of aligning the sensor perfectly squarely with the jig.

Another bad view is if the sensor sees 100% of the beans plus some other stuff. Maybe that would be more consistent in some way but it does not intuitively seems like a good design. I guess one advantage is that it would raise the IR sensor higher, reducing its environmental temperature. Maybe I'll try it... eventually.

Almost forgot to mention this. The column of beans is a few centimeters thick. The IR sensor is only measuring the top layer. The beans are moving up and down in the popper but not nearly as quickly as the IR sensor is reading or as quickly as Artisan is recording the BT. If I were an engineer this would bother me. Ok it bothers me anyway. Which beans are being read, beans that have been on the top for a bit or beans that just got there?

In the end I still find the information very useful. I love the development phase LCD and generally gauge repeatability by weight loss and density. I like being able to at least compare time, percentage, and proportions of phases. Artisan's area-under-the-curve (AUC) function is really interesting but so far I have been unable to get results consistent enough to rely on it. I am sure the inconsistency is due to operator issues and not Artisan.

I do not want to abandon the popper. I am getting perfectly enjoyable coffee from it. I am incrementally learning about roasting and about Artisan. At some point I expect to invest in another roaster but that could be a year or more (and 100s of roasts) down the line.

I went full-nerd on my fluid-bed (Ikawa) roaster and pointed a thermal video camera into the chamber as it was roasting a batch of greens:
Color stills - Ikawa Home thermal performance
Video clips - Ikawa Home thermal performance

The testing helped me feel more settled on just planting a low-mass thermocouple directly in the spinning bean mass, described here: Ikawa Home thermal performance

As luca points out, it's a freaking maelstrom inside a fluid-bed roaster, so a temp probe is largely just reading the cooling effect of the beans on the airflow. As borne out from the thermal camera measurements, the surface temperature of the beans is effectively the same as the air flowing around them. So, seeing the heat (IR) or feeling the heat (thermocouple) end up with similar readings.

Location matters, however. When I took thermal readings of my FreshRoast (RIP), moving the thermocouple just a centimeter in any direction would produce a radically different reading, like 20C, the difference between a medium or a dark roast. I wasn't able to use the temperature readings to 'dial in' a roast a-priori. I just had to pick a probe position, stick with it, and over time learn how to correlate the temperature readings to my observations of the finished roast level. And that correlation only applied to my specific hardware setup.

Nonetheless, I pay close attention to the Artisan (and otherwise) profiles posted on HB, regardless of the roaster, because there is often some tidbit in those postings that will inform my roasting profiles at a strategic level, which is more useful IMO than simply copying time/temperature points.

I think the second best measurement for a fluid bed roaster might be 3 or more probes, one below the beans so you know the air temperature, one after the beans so you can get an idea of how much energy the beans picked up and maybe one in the top 20 % of the bean column to see if trying for bean temp is viable. When I roasted with a popper, back in the early 80's modified it by putting a switch on the heater so I could control the heat and used a kitchen thermometer stuck through the top of the roaster so the tip sat in the beans. Somewhere I had read what a profile should look like and tried to follow those rules. I remember being happy with the coffee. I think the most important thing is to instrument the roast in a way that lets you see the differences between two roasts so you can start understanding how those differences effect the outcome. And while the optical thermometer might be best when it works, if it doesn't work perfectly 100% of the time, it might end up being a lot worse for learning than a probe that reads consistently.

Milligan wrote:The Ikawa seems to be fast to react to bring a temp up and overshoot but slow to try to fix a high temp issue.
...
Steady is key.

I second that motion! Have you tried strictly inlet temperature control? That will deliver a fixed heat/temperature profile regardless of whatever's in the chamber. Steady, repeatable results.

GDM528 wrote:I went full-nerd on my fluid-bed (Ikawa) roaster and pointed a thermal video camera into the chamber as it was roasting a batch of greens:
Color stills - Ikawa Home thermal performance
Video clips - Ikawa Home thermal performance

GDM528 wrote:...I wasn't able to use the temperature readings to 'dial in' a roast a-priori. I just had to pick a probe position, stick with it, and over time learn how to correlate the temperature readings to my observations of the finished roast level. And that correlation only applied to my specific hardware setup.

That describes my situation nicely. By early in the roast the IR sensor is not returning a true temperature. The primary goal is to get consistent results when the same controls are applied. Adding fans to cool down the IR sensor is definitely helping.

I am still getting a handle on the controls available to the set-up. This is a 'duh' type insight; the size of the batch affects the roast. Eventually it dawned on me that the beans have thermal inertia. My question, not yet fully explored, is whether, to a point, increasing the mass of beans actually decreases the roast time due to the beans holding and sharing heat. If that turns out not the case there is still an element of control from the batch size that affects the roast. A secondary question for my roaster is whether increased thermal mass of the beans evens out roast color as long as the beans move sufficiently.

GDM528 wrote: Nonetheless, I pay close attention to the Artisan (and otherwise) profiles posted on HB, regardless of the roaster, because there is often some tidbit in those postings that will inform my roasting profiles at a strategic level, which is more useful IMO than simply copying time/temperature points.

This.