Smoothing might be good, but not smoothing helps with resolution.
Once a small bean got stuck between the drum and the door, and I saw the change in temps very quickly. Another time the bean chute plug was dislodged a little and I noticed the temp changes very quickly.
If there is an over zealous smoothing feature added to the roast curve it may prevent detecting other anomalies.
I think the issue is the IBTS is not accurately reflecting the gradual increase in bean temp. There is no way a constant application of heat to the beans would cause IBT to rise in a stepped fashion resulting in the ROR swings. Not sure if this is algorithmic or what.
Having scanned my whole roast log, the issue only started to happen 3 months after I started using the Bullet (unless I was roasting smaller samples, where it was always the case) and currently it is significantly most pronounced in the first roast of the day, smoothing out from the 2nd onward. So I’m not sure if it’s an issue with the sensor.
I don’t know if our issue is the same, but we had our mainboard changed a couple of months ago from one that we have been using since 2020. The previous board was a workhorse and was easily handling 1.05KGs with enough oomph for us to play around with Power, Fan and Drum for roast profiling.
The new powerboard has a CE certification sticker on the side, a new cover but is comparatively less powerful than our old one. Preheating to 310C took about 14 mins for the IBTS to show 310 and beyond. The same happens about 3/4 minutes later at about 18mins. Compounding the issue is the IBTS that fluctuates wildly, so much so that the readings reach a certain temperature, say 160C and then it goes back sometimes by about 5C or so and then back up to the previous level, a few times in each roast.
Had a long exchange of emails with Matt from Aillio, sent profiles suggested and at the end, was told that either the Bullet be sent to Taiwan for calibration or I had to reduce my load to 850gms or so. I am doing that but the wild fluctuation has not gone away. In fact, I am now using my Bullet, which I used to talk so highly about like a Gene Cafe at P9 throughout. Any decrease in power and the roast targets are not met. It has basically been a tragic fall from grace from one of the most capable and intelligent roaster to a dumb one with no use for all the controls.
Unfortunately for the R1, changing the power board will cause it to not be calibrated, as the calibration requires the specific coil, board and drum to be present.
In production, every Bullet is running on a calibrated power supply which we use to tweak the power, so each Bullet match. The R2 is different as it can auto calibrate itself without the use of external hardware.
We are developing a piece of software which will allow you to calibrate new induction boards at home, but it requires you to get hold of a power meter.
I will have Matt reach out to you about this option.
About the IBTS fluctuations, can you try running D9 and check if this helps?
Thank you! Was feeling a little helpless and this is definitely a really good bit of news for us. We’ve had a terrific run with the R1 V2 so far.
Anyway, tried about five roasts on D9 and the IBTS fluctuation was a lot more less. It still happened but was manageable.
I’ll chime in, but it will mostly be ramblings of a mad man out of which someone smarter than me could come to some sort of conclusion. This did start happening towards the end of 2023 on my machine. My only conclusion so far is that the Bullet does not preheat properly and has an unstable temperature before the second roast. Now, I also think that this is happening to many owners of the Bullet, but it’s very close to a non-issue, as bab says above, so nobody reports it. I’ll put myself in that category. I can’t see the history of changes in various firmwares, maybe because it is not a click away and I’m too lazy to dig for it, but it seems to me that this started happening after the introduction of F1 during preheat, which was somewhere between 601-605. That small change has also impacted my approach to roasting the first batch because the machine started behaving differently. I did ask about that and got a very good reply explaining what exactly happens. Here’s the link.
What I also noticed is that soaking exacerbates this behaviour. If you hit it with P9 right out of the gate, it should be a lot less, which might not be true, but it is what I think after analyzing all my roasts. As noted by magneticfede the smaller the batch, the bigger the fluctuation on first roast. But, as it turns out, people who roast 1kg batches have this too, so I’d go to previous heat instability due to the different preheat routine. Also supporting this theory is the fact that it smooths out towards the end when everything gets hot enough, as pointed out above by bab, who also explains how the I-RoR is calculated and reflects the swings of the I-Temp curve. I don’t know if there’s a solution for this, but I’m pretty certain that there are no failed parts in your machine that need changing.
Well, that’s my 2, or maybe more, cents. I don’t know if it makes any sense, but now that I’ve typed everything, I’ll just click on the blue button. 
I use D9 only, so that doesn’t explain the IBTS fluctuations for me.