Steve's Ultimate Profiles 100g - Constantly Declining Fan Profile (Fan Cal 0.91)

[Mark]

I haven't tried it yet ... need a few more days ... but I roasted the same bean straight after using MD4 at 5.9. See screenshot.
Colour wise both are dark ... but perhaps the Power at 1.4 at touch darker.

Cheers Mark

[mr. bean]

I just had an aeropress with the Colombian roasted last night. Can't say much about aroma as we've got oil based paint drying in the house : \
I'm not good with correlating what I taste with official labels, but I'd say on the positive it was bright and fruity, but on the negative it was probably underdeveloped and perhaps slightly vegetal (if I know what that descriptor is referring to!).
I tweaked the numbers and experimented with some Ethiopian Limu (will post logs later), and I'm hoping the third try will be good!

ps. My experience with the MD2 wasn't great. Tasted kind of dirty compared with the other profiles I cupped it against.

[Steve]

If users do not understand the how and why of the "Power" profiling, probably best to stick to using automatic profiles they are familiar with.

This is from the original power profile post....

"This is where I am at....with the caveat that this is for ONE particular green coffee, 2020 washed Caturra from Costa Rica, 1400 - 1700 MASL. Any hope of getting this approach to work with other coffees will likely require several test roasts with careful observation of how they play out then with an understanding of how and why power profiles work make careful adjustments to the calculations to get the best possible result.

These settings will not be directly transferable to anyone else's roaster or location environment even using the same green coffee but likely a good starting point."



Additionally, Colom Excelso or Volcon with a 5 min First Crack and 1:30 DEV time i have no doubt will be very underdeveloped. Both are average commercial grade suited to Med - Dark roasts for espresso / milk drinks.

The Costa Rica Caturra is a good quality specialty green from 2020 and roasted with filter brewing in mind. Easily the best developed roast ive had from the KL, sweet, clean and focused with no baked roast defects.

In conclusion, I am abandoning all automatic PID controlled profiles and will only be using POWER profiles tailored to individual beans from now on. Explanations in my previous post stand and I dont really have anything more to add.

Cheers

[Howard W]

Thank you so much Steve, I've only tried tinkering with the latest POWER profile a few times but already noticed an obvious boost in clarity / flavour separation. A major breakthrough!

[kaffelogic]

Interesting zone 3 doesn’t work, and that was zone I tried and failed, when I should have used zone 2. I wonder if there is something different with the zone 3 coding, and should have worked for me?

Yes, there is something different with zone 3 coding. This was not an intentional difference and my apologies. It has now been fixed, see https://kaffelogic.com/community/viewto ... =271#p1735

[Geronimo]

Interesting zone 3 doesn’t work, and that was zone I tried and failed, when I should have used zone 2. I wonder if there is something different with the zone 3 coding, and should have worked for me?

Yes, there is something different with zone 3 coding. This was not an intentional difference and my apologies. It has now been fixed, see https://kaffelogic.com/community/viewto ... =271#p1735

Many thanks.

[Simon]

Hi All,

Any thoughts on what to tweak if I'm getting "temperature rise too slow" errors a few seconds into the roast on MD3 and MD4? This is in NZ with 230V. Increase the input power slightly? Or is there an early temperature vs time test that it's failing?

Cheers
Simon

[kaffelogic]

Any thoughts on what to tweak if I'm getting "temperature rise too slow" errors a few seconds into the roast

There is an engineer-level profile setting called 'preheat check gradient time'. It defaults to 30 secs, but is scaled according to preheat power. I'm thinking that the scaling algorithm could be improved. To that effect, if you could send me a couple of logs of roasts where the "too slow" error occurred that would be useful.

Your quick fix, i.e. workaround, would be to increase 'preheat check gradient time' to 45 secs or 1 min.

The criteria for the "Too slow" error is that the roast has not reached at least half of the profile ROR by the time 'preheat check gradient time' is reached.

[Simon]

Thanks Chris. Adjusting the preheat check gradient time from 0:30 to 0:45 did the trick. log 296 is one of the aborted roasts and 299 is a completed roast. A bit of a ROR dip at first crack, but the proof of the pudding will be in the drinking.

[kaffelogic]

Hi Steve,
Because your roast profile curve starts off a little concave it creates a steeper slope than it would otherwise have at around 15 to 30 secs and this sets the bar particularly high for the heating speed test. So extending the time frame out to 45 secs makes perfect sense.

2020-12-24_17h02_41.png (12.27 KiB) Viewed 13116 times

There is an algorithmic adjustment that operates to further shorten the time frame when you use more than default preheat power. I plan to modify that adjustment so it no longer shortens that time, however that is a very minor firmware tweak and will not really affect anything except make it less likely people will encounter the need to change preheat check gradient time. It will eventually make its way into the firmware, but you have achieved the same result by increasing preheat check gradient time in your profile.

Chris