Hello,

I'm not sure, but it seems to me temperature cannot be changed during a transient simulation. Maybe this is related to the fact that it is an input parameter to the device models.

Paul

Hi Andrew and Paul
   Thanks for your response!
   i didn't understand what you mean the following

that said, it's not been done yet AFAIK.
~~~~~~~~~~~~~~~~~~~~~~~~~

best regard
microant

If the device supports self-heating, one could use a time-dependent voltage source to set the temperature from the simulation time.  The device temperature in Kelvin is the same as the "voltage" of the thermal terminal.  However, if you set the temperature in this manner, the device won't exhibit any *self*-heating; the voltage source is a perfect heat sink.

The advanced bipolar transistors (mextram, hicum, vbic) support self-heating, by which the power dissipated in the device makes it warmer, which changes parameters such as the saturation current.

I think some high voltage MOS models in commercial simulators also support self-heating.  And some implementations of BSIM3 have added it, as well.

But the point is, the model equations are already set up to change the parameters during the simulation, not just a dc temp or parameter sweep, but also a transient analysis.

Jaw,

  Some circuits are sensitive to thermal effects, for example, the input
of a comparator. If one device is on and one device is off, then the
power and the temperature of the devices are different. When the
inputs return to the balanced state, there is now a time-dependent
offset voltage due to the difference in the device temperatures. This
can degrade the comparator's sensitivity.

The issue is rela