...
I do think PID control is a bit overkill for the IR toaster-oven approach. The temperature on-board must vary wildly from part to part depending on IR absorbance, so I am skeptical that fine-grained control of the heaters will make much of a difference....
The toaster oven that I have has a convection fan. PID is a little overkill for sure but I did not know that until I watched the PID loop run many many times. I still think it is better to have it because it frees my brain to be concerned about other things. In my early uses of the machine I always mounted a thermocouple directly to the board i was soldering. The solder would melt extremely close to the time at which the sensor showed the board was passing the melt temperature. However, it takes a lot of effort to attach a wire to each board.
Recently I did three boards (one at a time, three runs) in which I left a sacrificial blank PCB in the oven next to the board I was manufacturing. This did not go as well. There was a big offset (maybe 10C?) between when I saw the solder melt and when the thermocouple said it would. As you said, the absorption of the radiated energy is different between the boards.
I need to think of a better way to do this. I may consider using one large copper plate underneath the board to be produced and also under the board with the sensor on it. One problem with that is that any thru-hole components sticking below the board will have to be accounted for by drilling holes in the copper plate.
a work in progress for sure..
my oven is documented here, only about the 100th person to do this
http://mjkuwp94.tumblr.com/ReflowOven