Hi all,
I built my own hobby controller for T12 soldering iron tips. As you probably know, these tips have heater and thermocouple in series, so you generate some PWM signal, stop it for few milliseconds, sense TC voltage and reenable PWM again. Easy. My circuit works as expected. Nice stable TC reading with stdev < 0.5C. Except I'm getting very strange negative thermocouple voltage jumps that seems perfectly proportional to the mean heater current flowing just before turning power off for sensing the TC voltage.
My circuit is attached.

Example of measured dependence attached too.

At first I thought I messed up my board, so I rewired for kelvin sensing TC, but as expected no change there. I also tried extending the power-off settling time before TC sensing. It has no effect on TC value when no power applied before, but the negative jumps vs current are lower. It really seems that negative signal comes from the T12 tip. Whatever it is, it has time constant of at least 50ms (judging from scope signal shape). That is obviously too much since my PID has sample rate of 100ms. Any ideas? Is this normal behavior of T12 tips? Note it is original Hakko. I cannot imagine how could any PID controller work with this behavior as it leads to oscillations.
Best regards,
Stanislav