I really like the ability to have RoasTime automatically mark Yellowing on the Recipe program.
I would like to request a new feature. For me, I find First Crack Start occurs almost always around 205C ITBS. Could the programmers also insert the ability to automatically mark First Crack Start on the Recipe program just the way Yellowing can be done now?
I agree with @pnrenton. This would be a nice feature to have for recipes.
The first crack will normally be at the same temperature everytime I run a specific bean. (If I get a new batch of the same green beans, I can check to make sure that First Crack is in the same temperature range and adjust the Recipe if needed)
FC identification is a great feature that other sample roasters have. I don’t believe an update was sent to me, is this feature automatically active?
To add on to this request – the ability to run another step based on FC or Yellowing + another variable (time since FC, for instance) would be pretty slick…
Is there a “Time Since FC” trigger somewhere? I’m roasting now, and clicking around in one of my recipes and can’t find it. But yeah, if it’s already there somewhere, consider me newly enlightened!
Also along these lines, a trigger based on DV% would be something I’d put to good use.
This is exactly what I’m talking about @billc. And, quite honestly, with every added feature like this, I think we get closer to what I consider “holy grail” territory for roasting software. If I could tell the software something like…
My total weight is 800g
My pre-heat temperature is 295C
My turning point is 1:05
My yellowing temperature is 165C and I want to hit that at 4:30
My FC is 200C, and I want to hit that temperature at 8:00
My drop temp should be 210C or less, and I want 15% DV
…then the software itself could fiddle with everything in between the various milestones I’ve set. My recipes are generally just tweaks to P/F/D that help me get from each milestone to the next, and the milestones themselves are the parts that are important to me.
Well yeah I suppose. But the volume of training data would be microscopic compared to an LLM AI. The bulk of the processing involved would really be computational, and the training data could start with the first couple of seasoning roasts on a new machine, and could be ongoing with each roast, as long as the input from the human doing the roasting is considered trustworthy.
I used to have a HotTop roaster with Artisan that has the capability of doing some of what you are talking about.
If I recall correctly.
You can design a roast profile in Artisan Designer and then turn over the control of the HotTop to Artisan and it will watch the temperature and using PID control algorithms, it will change the heat (and possibly air) settings to match the profile curve.
When you turn the control of the HotTop over to Artisan, all of the safety features of the HotTop (deadman switches) are overridden. The HotTop also has an automatic drop and cool feature. When the HotTop reaches the drop temp that you set, The door opens and drops the beans and starts the cooling tray with stirrer.
The biggest downside to the HotTop for me was that it was only capable of roasting a little over 1/2 pound batches.
I really liked using Artisan with the HotTop.
Supposedly, you can use Artisan with the Bullet, but hardly anyone is doing that.
Artisan is not using AI to get there. We aren’t telling it that we are using a specific bean with specific characteristics and asking ti to develop a roast that Brad Pitt would like.
Maybe I should give Artisan another shot. Tried (not terribly hard) to get it going with my Bullet a couple years ago, but gave up kind of quickly when I realized I wasn’t in a learning mood…