My measuring equipment is a TO-220 packaged LM35, using 9V battery as "floating power supply". The output has 100ohms series resistor to avoid the problem of the short circuit, and 2 banana plugs. It can connect directly to the DVM, and the output voltage is 10mV/K. The only backwards, that it can work in the positive temperature range.
Sajti
Sajti
You have a specific device in mind? There are power transistors that have an additional diode on the die for the purpose of bias stabilization with temperature. Implies that the Vbe of that diode is a measure of the junction temperature, or at least the die temperature.
Jan
Take for example a LTP consisted of a pair of 2SA1015 TO92 devices, physically coupled for better temp matching. Dissipation is about 180mW per device, absolute maximum is 400mW. However, due to non-standard mounting, Rth junction-to-ambient given in the datasheet is not valid anymore.
What are the options to check if the actual Tj is on the safe side? I used the approach explained in the OP and got Tj=70C. Is this trustworthy?
YOU-SHOULD-HAVE-MENTIONED-THIS-BEFORE 🙄
And not only on post #23 🙄
You sent us in a wild chase imagining POWER transistors 🙄
I suggest:
Heat worrying you comes from DC; signal does not have much influence, since pair current total is presumed constant and thermal time constant,even for such small devices, will probably be longer than the lowest frequency applied, so they will integrate temperature over time.
In that case, apply no signal, let it thermally stabilize for a few minutes and measure Vbe.
Previously calibrate one such transistor in the 0C to 100C range as suggested earlier.
I just gave an example. But the same consideration applies also to medium and high power devices. As long as they have class A operation, the "worst case dissipation" is with signal absent.
Makes sense. Although requires some cooking 🙂
On the other hand, calibration has to be performed with the target Vce and Iq. And this would further rise the junction temperature beyond 100C of the surrounding boiling watter.
Last edited:
@bonsai
from your experience, do you have some insight how much can Rth junction-to-case deviate from the figure specified in the datasheet (+- tolerance)? That is from batch to batch and among devices from different factories?
from your experience, do you have some insight how much can Rth junction-to-case deviate from the figure specified in the datasheet (+- tolerance)? That is from batch to batch and among devices from different factories?
It is hard to define. I try to kill one IRF530 about 15 years ago. I found that it was able to handle 20% more power than I expected, from the dissipation, and the case temperature. So the specified Rth j-c is the absolute worst case.
Sajti