Here's a test that shows the effects of the roaster warming up:

Ran the same profile four times, with the first run starting with a 'cold' machine that hadn't been run for at least 12 hours.



My first observation is running the machine just once is enough to warm it up, and it's remarkably consistent thereafter, like, really-really consistent. Wow. From now on, I'll be doing an empty-chamber warm-up run before loading in the greens.

The other thing I'm noting is a slow drift in chamber temperature, about 1.5 degrees/minute at first-crack temperatures. It's steeper (2.5C/min) at the 200C setpoint, but that could be because it was preceded with a pretty large initial temperature jump.

The drift effect is probably not such a big deal at lower temperatures but could be significant when trying to dial in exactly when first-crack/second-crack is supposed to happen if it's not accounted for. The drift will also flummox any rules of thumb for estimate the offset between the input setpoints and the chamber temperature. I used to think the gap was 40C, but now I'd say it's 40-30C at drying/browning temps, and 45-35C for FC/SC temps.

Another thermal drift and offset measurement; this time with beans in the chamber:

Instant initial jump to emulate dropping beans into a preheated chamber, followed by a straight ramp to first crack, brief pullback to accommodate the beans going exothermic, and finally a slow ramp to a City roast level:



Also showing results with an empty chamber, which illustrates how the beans are largely following the shape of the empty-chamber curve after about 2-3 minutes. I theorize the slopes line up once the beans are finished drying out.

The ramp over the first 5 minutes is faster than the beans can follow, which causes the temperature offset to increase over time, but starts tracking 1:1 for the final 3 minutes. The good news is the temperature offset stays pretty close to a steady 40C over the phase of the roast where it matters the most.

Here are short clips of the beans as they're heating up.

The camera software is kinda dodgy, hence the pulsing pixeliciousness in the clips. Also switched to greyscale to show details a bit more clearly. For you IR camera nerds: Seek Thermal CompactXR with a detection range of 7-14 microns - ain't a lick of visible photons in these images. The recognizable landmarks in the image (dosing funnel, magnets, screw holes, aluminum frame) pop out due to their IR emissivity. The beans are actually well suited for thermal imaging compared to all the metal surfaces in the Ikawa.

The 50g clip shows how hot that dome at the bottom of the chamber gets - which is covered when the dose is increased to 100g. So, for larger batches there may be some 'accelerated roasting' for the beans that contact the dome, which might be why the bean temp is a bit higher in the center of the 100g batch size.

50g batch: 100g batch:

Want the hot-air exhaust of the Ikawa to go somewhere other than straight up? Copper tubing to the rescue. I got these at a local hardware store:



Both are 1 inch (Ikawa is a British company after all); one is a 'cup x cup elbow' and the other is a 'fitting x cup street elbow'. They nest together into an articulable bend and drops neatly into the exhaust port of the Ikawa:



This allows me to redirect the airflow laterally, such that I can park the Ikawa anywhere near my range hood and point the airflow right to the intake. This could also be used to direct airflow out a window.

No discernable impact on airflow during either the roasting or ejection phases - but I wouldn't be surprised if Ikawa disapproved. After successfully roasting with the glass cover completely off (for posts #64 and #65) I'm impressed with the robustness of their design. Clearly an insult to the Ikawa's industrial design ethics, but they should have included an airflow deflector in the first place, IMHO.

Warning: it gets litigiously hot, the first clue being the reddish hue the copper picks up after the first few uses. The copper is also very good at holding and transferring heat, so, no touch!

PSA for Ikawa Home owners: warm your machine up before roasting.

As I am sure many Ikawa owners do, I was roasting consecutive batches of greens - such is life with a small-batch roaster. Same bag of greens, same batch size, same roast program, same, same, same. Normally I run them back-to-back, but this time I got distracted for about fifteen minutes before starting the next batch. First crack timing was delayed about ten seconds from the prior runs - in the range of 10-15% lower development time. Hardly a disaster, but got me wondering if some of the variation that others observe in first-crack timing might be related to the thermal inertia of the machine...

From an earlier post in this thread, temperature readings in an empty chamber for consecutive runs, starting from cold:



The Ikawa Home is only aware of inlet temperature and is unaware of what's actually going on in the chamber. Sometimes that's really useful and other times it's a liability. My setup workflow includes running the machine empty before loading the first batch of greens. Now I'll be doing the same if the machine sits idle for more than five minutes or so, depending on how OCD I'm feeling that day.

ET-controlled Pro models should first-order correct for warmup effects. I imagine there's a longer delay before the bean-drop signal when starting from cold. A Pro user could confirm this by making consecutive empty-chamber runs (starting from cold) and comparing the curve overlap.

Does the Ikawa care what greens you put in? Not so much.

Ran an experiment with four different greens, all sourced from Ikawa:

Mexico
Olam Extended Fermentation project, Huatusco / Veracruz
Varieties: Columbia, Costa Rica, Sarchimor, Gesha, Oro Azteca, Marsellesa
Washed process
1,000 - 1,500 masl

Zambia
Northern Coffee Corporation, AA Isanya Kateshi
Varieties: Catimor 129, Java
Washed process
1,300 - 1,630 masl

Ethiopia
Dumerso Mill 750-800 Farms, Yirgacheffe
Varieties: Indigenous heirlooms
Natural process
1,900 - 2,100 masl

Ethiopia
Buie Bora, Yirgacheffe
Varieties: Indigenous heirlooms
Washed process
1,984 masl

Each green was weight-matched to a 60g batch size. Underdosing allows the greens to fully elevate as the roast proceeds, spinning and corkscrewing while fully saturated by the heated airflow. Fan was set to a fixed 80% (whatever that means) for the entire duration of the roast, which greatly simplifies understanding the heat transfer.

Roaster was fully warmed up, and I recorded empty-chamber thermal readings before and after the runs to check for any thermal drift. There was absolutely no drift - magic Ikawa moment.

Roasting program was a linearly-declining RoR profile calculated using a spreadsheet per an algorithm posted here: IKAWA Home - profiles Roast level was City+, just shy of Full-City.

The chart is kinda busy (sorry-not-sorry) but viewing it all together is more instructive.
Red = Ikawa Home setpoints - could also be setpoints for a Pro, set to inlet control.
Green/Yellow = empty chamber temperature data - bask in the awesomeness of their sameness.
Brown/can't-tell-doesn't-matter = temperature readings in the spinning bean mass of each batch. Mexico is brown, the rest identical.
Various wigglies = RoR of each run. Generally dance around each other, but coalesce around the 5min mark... why??
Orange staircase = Target RoR from setpoints. Ikawa nails it.



Zambia hit first-crack at 5:50 and lost 15% weight
Both Ethiopias hit first-crack at 6:00. Buie Bora (washed) lost 14% and Dumerso (natural) lost 15%
Mexico hit first-crack at 6:05 and lost only 12% weight - kinda odd... dry greens?
IMHO, I consider audible first crack to be an unreliable indication of the roast progressing into the development stage. Rather, I look for clues in the thermal data, and in this case there's a curious quieting of the RoR curves as the beans cross the 200C mark, which might signal a shift in chemistry.

All greens hit 223C just as they 'dropped' and all hit first-crack inside a 15-second window. The only observable deviation was from Mexico, which might be related to their lower-than-expected weight loss. Despite that, I'm inclined to say that the source greens don't affect the temperature profile. Program a profile, and the greens - all greens/any greens - will follow it. That should make sense, given the temperature feedback loop is closed prior to the air entering the chamber, combined with the very high airflow/batch-size ratio of the Ikawa. The Ikawa p0wns greens - experimentally confirmed.