Now that the finest minds of Hottop heating have assembled in one spot can anyone tell me what Michael won't. How is it that my heating element is sometimes glowing and sometimes dark? This happens while I am watching the watts on the kill a watt meter hold steady at 735. If there is really binary control of the element why don't the watts fluctuate just as they do when the elements shut down when segment temps are reached. Without knowing this how can I really tighten up on the program if the machine has a mind of its own? What are parameters that cycle the element and where are the watts going if it really is binary?

I have to say, personally, that I can't really answer that. My electronics skils are long-term-hobbyist only, and I don't have an electrical engineering background.

The heating element itself appears to have a pretty large thermal inertia, meaning that it takes a while to respond with more heat after power is applied to it, even at full power. And conversely some time to cool down when power is no-longer applied. The red glow comes when the element itself is red hot, it's not necessarily a reflection of the applied power at that instant.

After quite a lot of roasts, I find the "P" a little frustrating. Intuitively, it feels to me that if I wanted to slow the rate of temperature rise of a particular part of the roast, I would like to force lower power. If I wanted the rise to be faster, I would force higher power. On the face of it, this ought to correspond to lengthening the segment time / lowering the segment target temperature, and shortening the segment time / raising the segment target temperature. But, there's always the constraints that reaching the target temperature will turn off the element completely, and segments are limited in both number and maximum duration. This seems to result in most people's profiles making use of really high segment target temperatures to ensure the element never goes off. Roast progress is then largely determined by how fast the hottop will heat at full pelt with the given bean mass, bean type and the ambient temperature. It doesn't feel quite right IMHO. A good PID algorithm, subject to the limits of the maximum possible rate of temperature rise, ought to be able to feather the heater so that it never needs to go off, that required segment temperatures are reached with a reasonable degree of accuracy, and low overshoot, and potentially, with some regard to what is supposed to happen in the next segment as well. This is why I started to look at different control methods. Not that I necessarily claim to be an expert on this, it's just something I'd like to look into. Even if that ends up being to run it like the "B" model.

A bean mass probe, and possibly a better environment probe, seem to be the first steps to evaluating what's really happening in the hottop at the bean level, with tweaks to the segment temperatures and times to influence this, in an iterative process over several roasts. I think this will give a clearer picture of what effect the program and the heating element are having at any given time, than will the power consumption or color.

Jon

Yep, the P is frustrating indeed. Programming seems to be a race against time, as opposed to managing heat AND time control. The B is much better, just limited in its programming.

All the P really needs are in-roast over rides. Of course the other option is to be able to forecast the power flow to the heating element so you can stage it better across time periods. ha!

Go manual, young man.