[volt-nuts] monitoring LTZ1000 chip temperature

[ed breya]

If I understand what you're trying to do, I think you may be able to 
run the experiment backwards to more easily figure it out. I'd 
recommend focusing on junction to case characteristics first - the 
case to ambient will have a lot of variables.

With the case attached to an "infinite" heat sink near ambient, or 
even regulated separately, or at least monitored, and the thermal 
regulation loop in action, you can watch what happens when the loop 
is disabled and the heating power is shut off. The thermal time 
constant should become evident. I believe it should be mostly a 
single pole, linear response.

You can set up a normal external circuit to run the device, but add 
provisions to override the loop and utilize Q2 for temperature 
sensing on demand, then digitize the thermal response curve.

If better accuracy is needed, you can use the old trick of measuring 
Q2 at two different collector currents, say in at least an eight to 
one ratio - an old-school current DAC is easily set up for this task, 
and can be directly digitally driven and synchronized with the 
measurements. The cycle/sample rate would have to be much faster than 
the expected thermal bandwidth, and the average power dissipation of 
Q2 due to measurement current would need to be included in the analysis.

This whole process - let loop stabilize, switch to measurement mode, 
switch to normal mode - could be repeated indefinitely, and the 
results averaged to get more precision.

Ed