RoasTime Question: Any roadmap item on 1) a ROR curve for IBTS temperature; 2) a toggle-able pane between the curve pane to show the recorded data in a list mode?

I am recording the IBTS temp by hand and check the ROR situation over time, and i think the captioned features might be helpful?

IBTS RoR is a big project. I don’t think we have an ETA at this time but I would also like to have it!

I am not sure what you mean about showing recorded data in a list mode? You mean like a spreadsheet?

yes like a spreadsheet, presumably every 30 second or so to indicate the recorded temp

why is this? Seems like it should just be a question of switching the source for the RoR calculation…

We have done this but suddenly we begin to see the real dynamics of the hot and cold beans moving around in the drum…and this is not so useful for many people.

Hello Jacob,
Have you tried using a windowing technique? That is probably the best way to smooth out the data from the sensor.
When I export the data from a run and import it into a spread sheet, the RoR curve seems pretty decent…

I assume you’ve already tried a sliding window average, as that is the most straightforward solution, it’s extremely easy to implement, and it’s very computationally cheap. If that doesn’t get what you want, maybe something more sophisticated, like a basic low-pass convolution or even Kalman filtering, would work better.

If you don’t have the computational resources to do it within the Bullet’s controller, maybe you can use raw IR temperature data which you’re already sending to RoasTime, have the PC do the work, then ship the RoR results back on a 1-for-1 basis. If you use the Bullet without RT, you can fall back on a less robust windowed averaging algorithm or all the way back to the old BT probe and algorithm. Or you could keep it as a feature of RT, only and have the Bullet always display the old algorithm.

1 Like

From what @Jacob is saying here, along with a couple of “clues” I’ve picked up from other posts on the topic, one possible area of complexity is that (I’m guessing) the “raw” IBTS measurement at any given point is actually itself derived, and not the clean, instant output number we see in data exports, on our displays, or in RT. Combining what I remember reading with the comment above around “the real dynamics of hot and cold beans moving around in the drum” makes me more confident in that conclusion.

With that in mind, @wm1 and @celticcupcoffee, you two seem to have better ideas than I do for mathematical techniques to come up with a reasonable curve here. That’s a project I admit I gave up on but may pick back up with some research into what you’re talking about. Do you think your transformations/algorithms could be applied rapidly enough on top of whatever is happening to calculate IBTS in whatever computational headroom the Bullet has, or is there another direction that might be a winner here?