DRV8262: Efficiency vs. Switching Frequency not as expected

Tool/software:

Hello,

We used the DRV8262 in our design to power TEC elements. We connected two 12V TECs in series to a full bridge so we could use the full 24V supply voltage. In this configuration, we achieved a maximum current of 4.5A per full bridge. We calculated the PWR loss using the formulas from datasheet 7.1.1 and then measured it with the EVAL kit. We used resistive loads and tested the PWR loss at different frequencies.

The measured values ​​are unfortunately much higher than we calculated.

We removed the filter coils when comparing the curves in our calculation, so we have no idea why our measured losses could be so much higher. We would have expected the losses to differ more significantly at different frequencies. At the end, at 100%, it makes sense, since switching losses no longer play a role. But in between, the curves are not as expected.

Test setup:

measured with digital multimeter at the input and both outputs, current and voltage respectively.

R Load ~ 5.3R ~4,5A @24V / 100% PWM

As an example, the measured values ​​of the 200kHz PWM measurement with 200kHz PWM calc

It looks like we have an offset of about 4W that we cannot explain.