It light of Rao's new book and the delicious results I'm getting following his 2nd commandment, may I suggest a 2nd derivative of BT LED?

Rao's 2nd commandment is to keep a steady decline of the first derivative of temperature (BT) during the roast.
A second derivative on screen would help, as a steady negative number would correlate to a certain ideal roast curve length.

You guys are dreaming The function (BT, in this case) is very noisy to begin with. A significant amount of digital smoothing, with the associated lag, is needed to make the first derivative, BT', usable. The amount of smoothing that would have to be applied before performing the numerical differentiation for BT'' would result in verrry long lags. I don't believe it is feasible.

Jim

Jim's spot on. In addition to noise, BT is not very precisely measured -- e.g. the chip I have only provides 0.25C of accuracy. With a 5 second sample interval, the minimum change will create a BT' change of 3 C/minute.

I threw together a patch to add two additional LCDs. For me, the numbers fluctuate wildly between -80 to +80. Not really useful, but if anyone else with an artisan checkout wants to see if they have better luck here's the commit.

Thanks Jim and Nicholas for answering this. Nothing to add.

Seeing video stabilization after effects being developed these days, I am certain the 2nd derivative can be done (as well as a smoother, more immediate 1st derivative). Knowing a algorithm that effectively works is clearly the hitch.

Everything is possible if you know the future. Ah. Sorry, I mean this in a technical sense. Artisan applies two different algorithm, one during the roast that is "weaker" as it does not know the future and suffers from a time-lag. The other one, applied on redraw after roasts takes the (by then known) future (at any point in time, of the profile) into account and thus achieves better results without any time lag. That is mathematics.

Problem with this forum is that it is a melting pot of people with very different background, knowledge, education. Some are professors in chemistry, physics or computer science, some are coffee professionals others are not. While this is quite inspiring and positive on the one side, it seem to induce a lot misunderstanding and misinterpretations. I fear we have to make the best out of this. But second derivation live on screen. Nope. Never.

Want to add that Nicholas is right on the point identifying the low temperature resolution of being the real issue here.

Gotcha.

MaKoMo wrote:Want to add that Nicholas is right on the point identifying the low temperature resolution of being the real issue here.

Even if the temperature resolution could somehow be made perfect, I believe that individual beans striking the probe might jostle the readings enough to spoil 2nd derivatives estimated in real time.

Jim

PS - I saw the horse move, honest