WEBENCH® Tools/LM5117: M2 mos on the shunt resistor much hotter than M1 mos

Hello

Can you send us your schematic and a snapshot of your PCB layout.

Thanks

Hello, this is the screen shot:

and this is the schematic. It is exactly the WEBENCH output:

There is an EMI filter section, the LM5117 section and an putput measure and switch section.

Hello

Can you put a scope probe on the low side FET gate and check to be sure that

it is getting a good drive voltage.

Also, be sure that the FET is well soldered down to the PCB.

Since the PCB is the heat sink for the FETs, you may need more

copper area to the FET connections; the source and drains.

You may also need more vias to connect the top and bottom planes

to help dissipate the heat.

In this design the low side FET will be on for a longer time than the high side, so it will dissipate 

more power.

Thanks

Hello and thank you for your support!

Please let me know your opinion on the following photos

Francesco

Every board has two 12V 10A power supply  (A and B) and I have found difference between the two channels. Tha should not to be corrrelated to thermal limits

This is channel A, ch1 is the hi side gate and ch2 is the low side gate. No load condition

This is channel A, ch1 is the hi side gate and ch2 is the low side gate. With 7A load

This is channel B, ch1 is the hi side gate and ch2 is the low side gate. No load condition

This is channel  B, ch1 is the hi side gate and ch2 is the low side gate.  With 7A load condition

Hello

It looks like you have some bad spikes in the waveforms,

This could be from noise in the current sense signal or from noise on VCC or Vin.

Can you take a scope shot of the VCC pin.

C66 and C68 should be grounded near the device.

Also, you may want to replace R73 with a finite value, say 100Ohm, to provide filtering.

Thanks

Hello,

Vin is generated by 3 lead acid battery in series, so it is quite clean. What about an RC snubber network on the inductor?

Hello,

we have checked there are some spikes on VCC (LM5117 pin 16).

C66 and C68 are near the LM5117 grounded with a via, I wil try some more connection to GND

Best regards

Francesco

"Also, you may want to replace R73 with a finite value, say 100Ohm, to provide filtering."

This was very helpfull. All the spikes disappeared!

Thank you!

Francesco

Hello

If you have no further questions, I will close this post.

Thanks