Post Go back to editing



I would like to know how to make the most accurate measurement possible of the junction temperature of an LTM8025 buck power supply module.  Here's why...

I am using the LTM8025 development board configured for 5V output.  My input voltage is 30V, output current is 1.5A.  I measure the following at the module:

Vin = 29.94 V

Iin = 288 mA

Vout = 4.988 V

Iout = 1.500 A

Calculating from above measurements ...

Pin = 29.94 V * 0.288 A = 8.623 W

Pout = 4.988 V * 1.5 A = 7.482 W

Ploss = 8.623 W - 7.482 W = 1.141 W

eff = 7.482 W / 8.623 W = 86.77% 

The measured efficiency matches closely the efficiency shown in the LTM8025 data sheet for this input voltage, output voltage, and output current.  So far, so good.

A graph in the data sheet (pg 7) shows that the module junction temperature rise (at my specific input and output conditions) should be 20 C.   The graph title doesn't specify if the junction temperature rise is above ambient, but that is my assumption.  The data sheet states that the temperature rise graphs were developed using a board of (58 sq-cm) which I assume is the LTM8025 development board.  The development board measures 7.6 x 7.6 cm = 57.8 sq-cm.

However, when I attempt to estimate the junction temperature based on the module case top temperature and the thermal impedance listed in the data sheet, I get a wildly different result.  Here are my measurements and calculations...

The datasheet (pg 2) lists the thermal impedance of the module from junction to case top as 42.7 C/W.  On the development board, I measure a temperature rise of 37.0 C - 21.2 C = 15.8 C from ambient to the top of the module (see attached thermal images).  From calculations above, the module is dissipating 1.141 W.  1.141 W * 42.7 C/W = 48.7 C rise from case top to junction.  Adding in the measured rise of 15.8 C from ambient to case top brings the temperature rise from ambient to junction to 48.7 C + 15.8 C = 64.5 C.  This is obviously nowhere near the 20 C as shown in the data sheet graphs for these conditions.

Because of the above discrepancy, I'd like to more directly / accurately measure the junction temperature of the LTM8025 module.  What is the best method available to accomplish this?