Measuring your MET on Nano 7 and the impact of batch size

[Sam.law]

Update 27 Oct 2019 : Apologies all. It looks like the original thermocouple sensor that I used was on its way out, hence leading to incorrect temperature reading and observation....

Hi all,

Sorry for the lack of update...life's got in the way.

Just want to share a finding and possibly get some input/confirmation from y'all to check that it's not all just in my head...

After many years of roasting on the Quest M3, one thing I've learned is that monitoring the MET (Maximal Environmental Temperature) can be quite crucial. Obviously the Nano 7 is quite a different system, but the concept still applies : A high MET may mean that some of the beans are experiencing high temperature locally and may pick up some unwanted roast defects. I had some questions about the probe placement of the Nano 7, so I got curious and wanted to know how the MET tracks on the Nano 7 (compared to the usual BT probe) .

About the impact of MET on roast quality, Jim Schulman did a much better job in explaining that than me:
https://www.home-barista.com/home-roast ... ml#p538330

So in one of the weekends, I snaked a bare k-thermocouple into the vent of the Nano 7 and did a couple of roasts as per usual (Steve you may recognize the meter ).

Probe location.jpg (104.59 KiB) Viewed 6830 times

Logging setup.jpg (93.32 KiB) Viewed 6830 times

So by doing this, I get the 'MET' reading through the k-thermocouple, as well as the 'BT' as read by the usual BT probe on Nano 7. I extract the roast logs and overlay them in Artisan.

These are my results below (I've got more roasts but these two are representative enough). So, again, the same coffee, same profile (stock Firestarter), but roasted at different batch size.

Bensa 50g.jpg (94.05 KiB) Viewed 6830 times

Bensa 100g.jpg (95.78 KiB) Viewed 6830 times

Mainly, note that the spread between the 'MET' and 'BT' between 50g load and 100g load.

In restrospect, it is not unexpected that you needed hotter air(and more energy) to roast more coffee. What was surprising to me was that in some of the roasts, the MET got up as high as 270 deg C for the higher 100g load. 270deg C may have been in what some considered to be the 'scorching' territory. As an additional observation, I had some experimental roasts cupped blind in one of our cupping sessions back in Aug (basically a couple of us coffee nerds in Melbourne getting together). For the cupping, I have a coffee roasted with the same profile and same drop point, but one was loaded at 50g and the other at 100g. The 50g-roast came out tasting more superior than the 100g-batch roast on the cupping table. So this MET observation seems to support it somewhat. In later roasts with larger batch size, I tried to compensate by roasting to a lower roast level number, but that did not seem to work (tasted too light and coffee didn't taste as good, but that's once off experiment).

If you're nerdy enough and happen to have a logging thermocouple sitting around, I would be curious to see what your MET looks like.

I shared this with Chris and he also gave me another useful tips. He mentioned that not all vents put out heated air at the same temperature, so you might want to verify the vent which you're measuring is the hottest spot. To do that, you need to start the roaster empty (no beans) and go into the 'secret' timer mode (it's in the instruction booklet). I haven't got around to repeat this with Chris' suggestion, but that's in the plan.

Alternatively, if you just place the probe at the vent which looked most discoloured, my gut feeling is that there's where the hottest air is coming through (I may be wrong though).

So if you're chasing clean light roasts (and are using Firestarter profile), especially for filter, may be give this a shot by halving the batch size (50-60g) and see what you think. I found that stock Firestarter profile, 50g load, roast level 2.2 seems to give me the best results so far (compared to my own profiles).

If what we observed is valid, another implication is that when we're designing a roast profile, the ramp after first crack shouldn't be too steep for larger batch size, otherwise you may get into this too-high MET scenario. This also can be evidence for why it's ok to have almost flat/zero ROR during development phase (the MET may still be going up, but not picked up by the BT probe)

p/s: changed the title to better reflect the batch size is apparently what is affecting the MET

[Angela]

Sam, a few of observations.
1. The roasts I've been getting are pretty even by colour. And colour is a good indication of roast temperature. Of course there will be hotspots but the beans are flowing so they will move in and out of any hot areas - averaging over time. Even coloured beans suggest even temperatures.
2. Your probe most likely gets driven by the beans away from where you place it, unless those tails are very rigid; probably to the side wall. High energy transfer readily available there, on the aluminium wall; much better than the thermal conductivity of air or a coffee bean.
3. We have no real idea what the Nano 7 probe measures. Is it air temperature; bean surface temperature; bean core temperature; some of everything; or just an analogue indication of how the roast is progressing? I've come to regard it as the latter.
4. FC happens at 210C and above for me, as indicated by the probe, but don't beans in drum roasters FC at 196C or thereabouts? That is a real good clue to what the probe is not measuring.
5. Chris has stated that the Nano 7 won't go above 240C - measuring whatever it measures.

[Sam.law]

Hey Angela,

Thanks for sharing your observation!

I agree, the roast comes out of the Nano 7 have always looked pretty even, perhaps even a touch more even than what I get out of the Quest M3. That's the advantage of fluid bed roaster I suppose. But we're chasing after marginal improvement here (so I am looking beyond beans appearance). In my understanding, all else equal (beans roasted to the same development level), beans roasted at lower MET will generally taste cleaner and have less of the roasty flavour components. When I say 'scorching', it doesn't literally mean the beans are scorched at all (apologies as it wasn't clear). But there may be more pyrolytic flavour molecules generated (associated with roastiness). It's probably not so much of an issue if the end-goal is a darker roast where the flavours from the darker roast spectrum are expected. But if you want to develop lighter roast more without introducing the roasty flavours, having a lower MET (less hotspot/temp gradient) will likely be helpful.

You have a valid concern about the probe and it was my concern too. The probe does seem to be semi-rigid. You may have noticed in the pic I had a springwire (made-shift from a portafilter retainer spring) holding the probe in place. I did inspect the position of the probe before and after the roast, it seems to stay intact. So I think it's fair to assume the probe does stay where it is during the roast.

No BT probe is a true measurement of bean temperature in that sense. It is always a weighted average between BT and ET. The ratio/influence of ET over BT depends on the actual roaster design/airflow/how the probe is setup. On a drum roaster, even a well placed BT will still be influenced by ET to a certain degree. On the Nano, I reckon the BT is influenced by ET to a greater extent than in a drum roaster (hence higher 1C temp).

Of course, these are all to some extent just speculations and were based on my interpretation of what I read over the years.

At 100-120g load, the BT probe is in contact mostly with bean mass, so it's a truer BT reading in that sense. At 50-60g load, the beans definitely hits the probe less frequently(hence less like BT and more biased towards ET). This is probably where a transparent roast chimney will be helpful for observation (wink at Chris ) I thought of removing the lid during roast to observe the beans, but that will probably affect the dynamic of the airflow & temp (observer effect).

5. Chris has stated that the Nano 7 won't go above 240C - measuring whatever it measures.

That's actually the point I was trying to demonstrate: There definitely exists a temperature gradient within the roaster, especially when you have beans in the chamber. The more beans you have, the greater the temp gradient it seems (because the probe only measures the temp after the heat has been absorbed by the beans). If you have too great of a temp gradient, it may affect the roast quality.

[Angela]

Sam, if you've done a lot of drum roasting I can see your need for parallels. The problem is there probably isn't that much that is comparable. The beans rotate bottom to top to bottom with a period of a few seconds all beans will go through any hotspot fairly uniformly. I suspect if you scrutinize temperatures in all parts of a drum roster there will be considerable variation too. I found taking my machine out side and just roasting with the lid off was a good way to see what was going on. It didn't influence my roast as far as I can remember. Just be ready for a face full of chaff!

If you are worried about individual molecules in beans being over-heated remember the definition for temperature is measure of the average molecular kinetic energy in a system. Thus by definition some molecules will have more energy than others; it's the nature of the Universe and there is not much you can do about that.

I tend to take a pragmatic view; change things you can and don't bother too much about the things you can't.

I'm sure, as you say, there is a temperature gradient through the chamber but I'll have to live with that.

You're playing with batch sizes - good - keep us posted. Though isn't 90g a recommended machine minimum?

The big difference drum vs hot-air is air flow. That hi-speed air must be force evaporating volatiles from the beans that probably do not get driven off during the drum roasting process. That is a debit point for air roasters and will likely prevent Chris becoming a millionaire. It might be the increased air flow in the small batch that is giving you the results you see. Perhaps some of the nastier volatiles are more readily driven off in the small batches.

Now what you as a drum roaster can really help with is the temperature difference (hot-air vs drum) for the observation points - bean colour change, First Crack and Second Crack. Is the temperature difference between machines the same across the range?

As an afterthought; I saw a comment from Ben about Ninja working well with Yirgacheffe at 2.6 so gave it a try. It wasn't perfect but better than a lot of the stuff I've been getting. My mistake was take too much from drum roasters who said Ethiopians like a slow shallow rise, a similar length of drying to Maillard and a short time after FC. I can tell you now the drum roasting gospels did not work for me and my Yirgacheffe. That Ninja hump seems important.

[Steve]

Interesting measurements Sam! Thanks and quite timely as I have been playing around with 65g and 80g batches and cupping them blind against the 100g batches. Nice probe!

As Sam mentioned this is only really relevant to getting super clean light roasts especially when one wants to highlight a really great to exceptional coffee more in line with popular "Scandinavian" or" Nordic" roasting styles, most of which seem to be using Loring and sampling on IKAWA = low ET hot air roasting.

A few of my 65g batches had much more distinct aromatics and flavour profiles, some were quite dry in the finish though which I associate with some underdevelopment related astringency = a quick deal breaker for me. 80g batch sizes seem to be a pretty good sweet spot. A trend in my tasting notes with 100g even though the beans look light and grind light, ground aroma is caramel and they cup with muted acidity and lots of different "browning" type flavours.

Will definitely have to give Firestarter a go with the lower batch size as I abandoned that at 100g because light roasts tasted underdeveloped and roasty.

Took a video of 80g roast just up to FCS but the colour changes dont really come through right.
I use external ventilation from my drum roaster with a jam funnel connected. The blower on the Nano is much more powerful than the pull from the exhaust fan. Have accidently left the exhaust fan off a few times while roasting and its enough to still push outside through the exhaust. I have not observed any difference in the end roast logs running with the drum exhaust, just the black tube and no black tube. I have observed about 10 x 80g roasts for colour changes using my current 8 - 9 min profiles.

Start of rapid expansion - turning pale green 145 - 150
Start of Yellow 158 - 160
Full Yellow 170
Orange 185
Start of Browning 192
FCS 210 - 212

Drum roaster
Start of Yellow 148
Full Yellow 157 - 160
Orange 170 - 175
Start of Browning 178 - 182
FCS 194 - 196

I will probably have a play around with a bare probe and hook it up to the Mastech.
Sorry can not get the picture around right way.

20191005_190903.jpg (1.97 MiB) Viewed 6762 times

[Steve]

Ok a few sneaky late roasts. I only have cheapo spare K type probes so my data is not as great and I have not bothered merging. But I think it still gave me a pretty good snap shot and it would be safe to assume that my 100g batches were likely hitting 270C or more.

All 3 roasts were 2019 washed Yirgacheffe.
Roast one and two = 13% weight loss, 12.1% DTR, 59sec + 5.2C
Roast three = 13.5% weight loss, 11.9% DTR, 59sec + 6.4C

First 2 roasts are the same using my current filter roast profile, mimicking my drum / RAO decline ROR style 80g. Roast 3 was stock Firestarter with 65g at 2.2.

Roast 1 - sheathed bare tip hanging just above the bean mass, gave a similar curve to the Nano probe just a few degrees lower.

AboveBM.jpg (4.18 MiB) Viewed 6753 times

YirgKLabovebeanmass.png (77.86 KiB) Viewed 6753 times

YirgAbovebeanmass.jpg (30.58 KiB) Viewed 6753 times

Roast 2 - wrapped probe around Nano probe sheath and tucked into what looked to be the vent above the darkest spot. Had a jumpy start with some interference but finished ok, MET hit 260 - 261.

Invent.jpg (4.53 MiB) Viewed 6753 times

YirgKLinvent.jpg (30.67 KiB) Viewed 6753 times

YirgKLinvent.png (88 KiB) Viewed 6753 times

Roast 3 - probe in same spot as roast 2, MET hit 265.

YirginventFirestarter.png (85.76 KiB) Viewed 6753 times

YirginventFirestarter.jpg (31.02 KiB) Viewed 6753 times

[Steve]

Some more testing and fodder for thought.

Played around with a few things for 50g batch because of heat too fast error on these smaller Guat beans.
Lower fan calibration = much higher MET at the end of roast 270 - 275 into and through first crack.
Returning fan calibration back to 0.91 and swapping out the Firestarter fan profile for the Classic also resulted in high MET 270 - 275 into and through first crack.
Figured out leaving it stock and 60g batch with much lower preheat power is best, 60g seems to behave much like 80 to 100g with regards to colour changes and temperature.

Cupping blind today last nights Firestarter Yirg against some of my earlier 80g roasts of the same was very easy to pick. Much sweeter, cleaner / snappy clarity and actually makes my other roasts with the stretched finish and low ROR taste baked. So compensating for the higher MET on larger batch sizes by slowing the ROR through the finish to end up at lower end temp might not be so great.

I started tweaking the finish on my own profiles so that the 2.2 / 219C finish point matched the Firestarter and have roasted a bunch of 60g Guat samples.

Myfilter and 9-10min for Espresso worked out pretty close to the Firestarter for end MET temps. They all come into first crack start in the high 250s and slowly creep up to 260 - 262. As Sam mentioned JimS has had a lot to say over the years about MET especially approaching and through first crack. If the MET drops off during first crack you are going to stall out the development and get some pretty nasty flavours.

So this dance between batch size, MET, fan speed and finishing temp is a bit of a rabbit hole indeed!

Guat60gFirestarter.png (50.67 KiB) Viewed 6745 times

Guat60gFirestarter.jpg (33.59 KiB) Viewed 6745 times

Guat60gmyfilter.png (48.38 KiB) Viewed 6745 times

Guat60gMyfilter.jpg (33.02 KiB) Viewed 6745 times

Guat60g9-10min.png (70.23 KiB) Viewed 6745 times

Guat60g9-10min.jpg (33.73 KiB) Viewed 6745 times

[Sam.law]

Wow, thanks for sharing Steve! You're quick to pedal into action, and so quickly! That's a nice set of data and lots of food for thoughts (still to digest fully).

Interesting that you commented about the heat too fast error. I did get them occasionally but only if I am doing back to back roast with minimal rest. Ideal rest time is ~10 min or so. Perhaps our airflow setting are not exactly the same either, hence the observation. That or your beans are behaving really differently.

What was surprising to me was that your 60g roasts were still going into 260 C ish, where as mine was ~240-250 C. Maybe my hypothesis about batch-size and MET is busted. Or perhaps my probe wasn't located in the hottest spot. I will redo the measurement again this weekend with more thoughts to the measurement location. Unfornately that will have to wait until the weekend where I have more spare time.

To anyone else reading, I hope that I did not come across as trying to find fault with the Nano 7, or that I am roasting to number/curve. Quite the contrary, the Nano is doing a good job roasting. I just find that it perplexing that the lower batch-size roast tasted better for me. The MET seemed like a potentially good explanation (and it is a measurement most roasters would measure). If our observation is valid (ie MET measurement is useful), it could help us fasttrack to achieve even better roasts on the Nano 7, instead of modifying the roast profile blindly through trial and error, whereas if you had the MET value the direction to the roast profile modification would've been much clearer.

[Sam.law]

Had some fun playing with Nano 7's manual control mode today.

With the Nano 7 in manual control mode, I poked around in the roast chamber with my temperature probe to find the hottest spot for MET measurement (hot hot hot!). Turned out my suspicion was right, the darkest vent (stained by the roast residue) is indeed the hottest vent so I was on the right track.

Further secure my probe with some Kapton tape, the probe should be sturdy enough for at least a couple runs.

Then another thought crossed my mind,why not turn on the fan without heat and see how the beans move in the roast chamber? The Firestarter profile has an initial fan speed of 14.7k rpm, so that's what I set it to. Loaded beans (washed yirgacheffe) at 50g, 75g, 100g and 120g.

Thought it would be fun to convert the video into gif and put them side-by-side for your viewing pleasure.

50g (left) & 75g (right).gif (3.76 MiB) Viewed 6712 times

100g (left) & 120g (right).gif (8.7 MiB) Viewed 6712 times

I am not sure how should an ideal bean-flow should look like but my uneducated guess is the 75g seems like a nice roll. The 100g was maybe overloaded slightly and needed a bump in fanspeed?

After the experiment above , I roasted the beans used as 2 x 60g roasts, Firestarter level 2.2 & 2.3. The MET I was seeing at the drop was only ~228 deg C (did not log that one though). Interesting that it's quite different from Steve's 60g roasts (he was seeing closer to 260C).

If anyone's curious on how to access the manual control mode, here's the instruction from Chris:

Put the roaster into timer mode by pressing profile and menu buttons together.

Select heater power manual control by pressing the minus button. Then hold down the + button to increase power to say 100 (meaning 1000 Watts). Element will not actually come on yet.

Press profile button to cancel power manual control. Press + for fan manual control. Then hold down the + button to increase fan speed to say 150 (meaning 15000 RPM).

Once the fan speed reaches a threshold the heater will come on. The roaster will stay on in a steady way now while you poke the thermocouple probe into each air vent one at a time, hopefully without burning your fingers.

[Steve]

Excellent work Sam.

It could be that the position of my probe was too far in / closer to element so picking up on more of an "inlet temp".

The gif looks to be similar with what I have seen in batch sizes 50, 60 , 80, 100 and 120g.

I might try fixing the probe as you have at some stage just to see if we get closer readings. (Looks up kapton tape)