I have been playing with constant fan speed, as opposed to trying to maintain a constant bean turbulence
Here is a fan curve graph
10mf.png (38.16 KiB) Viewed 6830 times
I couldn't embed the video, so here's a link showing the bean turbulence. The video is at 10x speed. https://youtu.be/O69sIH0ueK0
I have made some adjustments in engineer mode too, more on that later. I feel I'm getting close to where I want to be with respect to the ROR curve. The ROR curves are magnified 10x and 30s smoothing.
10m.png (85.57 KiB) Viewed 6830 times
this is the profile before starting today's experimenting
Damian wrote: ↑Tue 17 Mar, 2020 12:24 am
Thinking about this, with a faster fan and the beans jumping around more, the temperature is likely less accurately BT, and representing more so ET.
This is true. The air in the element chamber is being heated fairly evenly so it enters the chamber at a fairly steady temperature. The beans, on the other hand, are subject to several sources of randomness including their own natural variation and fluidisation effects such as slow (20+ sec period) density waves. The more the beans influence the probe the more the ROR will swing up and down.
On the other hand, faster moving beans may be less susceptible to variation caused by density waves.
Does the smoother ROR curve give benefits in the cup?
Does the smoother ROR curve give benefits in the cup?
Yes. That is part of what I got from Scott Rao's roasting class and has been my focus with profiling.
However, the way we measure bean temperature is not ideal, particularly in our case where the bean density around the probe can vary with air speed. Plus the prob is very thin (sensitive), so the curve is quite noisy. So I find it harder to determine roasting crashes from noise.
In saying that, I cupped these two this morning, Ethiopian Limmu. The difference is noticeable, the left has some baked after tastes, the right is cleaner. Yet both taste equally developed.
Damian wrote: ↑Wed 18 Mar, 2020 3:13 pm
However, the way we measure bean temperature is not ideal, particularly in our case where the bean density around the probe can vary with air speed. Plus the prob is very thin (sensitive), so the curve is quite noisy. So I find it harder to determine roasting crashes from noise.
Theoretically statistical smoothing should eliminate noise as well as the thermal mass of a heavier probe, however things don't always align with theory. I think there is also a factor of a 100g batch having little thermal buffering so some of that noise might be intrinsic to the batch size. I agree with you that the noise is a problem for picking roast crashes. I don't have any solutions at present except to play around with different smoothing values.
In saying that, I cupped these two this morning, Ethiopian Limmu. The difference is noticeable, the left has some baked after tastes, the right is cleaner. Yet both taste equally developed.
I would say that confirms that you are on the right track and that the smoother ROR curve is not just an artifact of air speed and bean dynamics, but reflects smoother ROR at the bean surface.
I think there is also a factor of a 100g batch having little thermal buffering so some of that noise might be intrinsic to the batch size.
For sure. And you can see this when comparing 130g batches
But larger loads can also work against us when it comes to the beans changing exothermic to endothermic around first crack, it takes more work to prevent or reduce stalling.
Here is the profile if people want to try it, I would love it if people let me know how it goes for them, I've only tested with one bean at this stage. I use 110g of beans
Thanks Damian, I will have a play with it and report back.
I have changed to the Beta firmware 7.4.4.1092. for the power profiling so not sure how this will work with your zone 1 settings? it says on the log Firmware 7.3.12.1062
Or should I just reflash back to the old firmware?
log0088.klog
Damians10v1.kpro