L6491EVAL has excessive switching loss

Richard Tarbell · ‎2016-12-27

Posted on December 27, 2016 at 21:18

Project: Make one full bridge (non-isolated) rectifier, using two L6491EVAL boards as the gate drivers (the switching transistors mounted to them are FCH190N65F).

Initial testing approach: test the full bridge rectifier circuit as a DC-DC converter (just to verify functionality) by applying 10VDC to the input, vary the PWM duty cycle to each gate driver board, and then measure the output voltage (it should increase as the duty cycle increases). Just for safety/sanity, my 10VDC input source is current limited to 1 amp, and there is no load on the output of the full bridge.

Problem: With Vin=10VDC, and starting at a PWM of 0%, I get an output of 10VDC (expected), and almost no losses. However, as I increase the PWM, my current draw on the source supply increases pretty steeply:

PWM % Input supply current: Vout

5 0.06 A 11.0

10 0.20 12.3

15 0.41 13.85

20 0.58 14.8

25 Supply current limit at 1.0A

--> It seems that my current draw is WAY too high for ''normal'' operation. I suspect that Any ideas as to what I should be looking for? The L6491EVAL boards do have a dead-time potentiometer, but adjusting these seem to do nothing. Are there any other adjustments I can make, to reduce the power losses?

fabbri.fabio · ‎2017-01-13

Posted on January 13, 2017 at 11:30

Dear Richard,

we miss some important info: at which frequency and VCC voltage (L6491 supply) are you driving the application?

In the eval board you have to select and mount also the turn-on and turn-off resistors based on the selected MOS and application speed/characteristics. Which values are you using?

Based on your description seems that probably turn-off/on resistors are not sized correctly. If it is the case, the induced turn-on phenomena is occurring or, in an extreme case, dead time is not sufficient.

You can start having a look to a brief description of induced turn-on phenomena in AN2738 page 35 (chapter 9).

Just to speed up the analysis, very often turn-off resistors (R2 and R4) shall have a smaller value (often from 1/10 to 1/3) compared to the turn-on resistors (R3 and R5)

Monitoring with an oscilloscope MOS gates and OUT node you can tune the values to fit your application and check if the induced turn-on phenomena is present. Also monitoring which component(s) is dissipating such energy (which component is too hot) can help you to locate the issue.

Best regards

Fabio

Richard Tarbell · ‎2017-01-18

Posted on January 18, 2017 at 16:02

Thank you for your response Fabio! I am currently using the L6491s at a switching freq of 300kHz - I suspect that this may be too fast. I will try 50kHz in the next day or two, to see if that reduces this excess current consumption. If not, I will change the turn off/turn on resistors to see if that makes any difference.

--> Just a side question: if I were to use one of the IPM modules (with the transistors and gate drivers built in to the same module) instead of the L6491 eval boards, does this mean that I would NOT have to adjust the turn off/turn on times, since this would be managed by the IPM module itself?

fabbri.fabio · ‎2017-01-20

Posted on January 20, 2017 at 18:09

On IPM, turn-off/on currents have been selected by the IPM designers to fit the most variety of applications. These values cannot be changed by the user. IPM is much easier to use, but you cannot control all design parameters as gate driving speed and/or MOS characteristics.

With IPM probably you cannot maximize system performance by tailoring all system parameters to your application needs and characteristics.

By the way, 300kHz is a pretty high frequency and, if this MOSFET is suited for this requirements, you should use gate resistors with low value, probably below 10 Ohm.