Quote Originally Posted by joeri
at 100°C I have 11 kOhm; at 150°C I have 3 kOhm; at 260°C I have 500Ohm. I use a current source which drives a small current through the NTC, then I amplify the voltage over the NTC with a opamp and bring it into the analog pic input. So far, so good.

The oven is a samsung with 1500W. I added two heat resistors from a waffle machine (another 1000W), so total of 2,5KW.
This boosts quite fast. Also, I have a fan to equalize the temp inside the oven. The complete oven is packed in rockwool.

But my problem is the PID. Any experience on how to work with negative errors (overshoot)?

The calculate PID values are:
gain= 10
I= 0,04
D = 100

So it seems I have a PD controller. Due to the slow process, that might be ok. Or not?
It sounds like the rock wool is overkill, plus you want to cool things down pretty quickly once soldered. I crack the door of the toaster oven slightly for my high tech solution.
I really think the PID may be overkill. If it were me, I would take the current value at a given time and compare it to a desired constant value with ...

heat_loop:
for time = 1 to 10

If curr_temp_val < set_temp_val then
relay = 0; A lower number means the temp is too high.
else relay = 1 ; temp too low
endif
pause 1000
next time
goto get_next_heat_increment ; Put next temp constant into set_temp_val and start again.

Otherwise, the relay pulls in to increase temperature.You can check/regulate the temp this way until the next time increment is reached. At this point, update set_temp_val and compare. You can make the time increments every 10 seconds, or every second for that matter. A heavily insulated unit will require you to cut off power to the elements early, or you will have substantial overshoot. I'd say to use the rockwool for the hydroponic "tomatoes", and use less insulation or an automatic vent to the outside.