LM5116 Keeps Failing

Hi

I'm not sure this is the case you are seeing.  First possibility is shoot-through, because LM5116's adaptive dead-time monitors HO-SW and LO-GND voltages, dead-time cannot be guaranteed if there is a gate resistor.  Check your dead-time.

Another possibility is dv/dt induced turn-on, it is possible because BSC340N08 has only 9.1nC (10V) gate charge while FDB2552 had 51nC (10V).

Regards,

Eric

Hi Eric,

Thanks for the reply.

Regarding the shoot-through, I have looked at the FET waveforms and I don't see any indication that there is any cross-conduction.  If it did exist I would expect it to manifest as failure of the FET's.  This isn't what I'm seeing.  I'm seeing failure of the LM5116 itself.  The FET's survive most instances.

While I agree that the lower gate charge FET might be more susceptible to dv/dt turn-on I haven't seen any indication that either FET is on when it should be off (or off when it should be on).  Again I would expect this to result in failure of the FET rather than the LM5116.

I suspected there might be an excessive 'glitch' on the CS pin.  The absolute maximum rating on this pin is -3V per the data sheet.  I originally saw some ringing on this line in the neighborhood of -2V (much larger than the actual sense signal).  I cleaned up some of the routing and reduced the ringing so this doesn't seem so likely any more.

Any other suggestions are certainly welcome.

Hi

Would you let me know which pin is damaged ?

Regards,

Eric 

I can't say for sure.  I am having a destructive physical analysis (DPA) performed.  Hopefully that will shed some light.

I do know that the internal reference appears to be damaged.  I'm suspecting that the gate drive current is overloading it.  I have reduced the switching frequency (based on the LM5116_quickstart spreadsheet I found at the TI site) to reduce the average gate current.  Initial results are promising but not yet conclusive.

Hi Wayne, hi Eric

did you find out something with your LM5116? I am experiencing a very similar problem: when switching loads: the controller, and the upper FET fail. I am using the IPP139N08N3-G FET. The design parameters are as per the quickstart excel. The input voltage is 36V, the output voltage 12V. Maximum power is 250W.

Similarly as in Wayne's case I did not see any sign of shoot-through. 

Any suggestions are more than welcome.

Thank you,

Martin

Hi Martin

Please check your MOSFET temperature. Please check your gate driver waveforms.

Need more details to investigate.

Regards,

Eric

Hi,

looks like I have the same problem here. I use the same components as the webbench specify.

Vin is 75V; Vout is 27V / 2A max.. I can't see any shoot-thru or overtemperatures at the FETS.

After Failure it also seems that the internal reference is damaged. It seems that the reference voltage breaks down on load. 

Is it recommended to use a boost resistor? This should reduce VCC current.

Was it helpful to reduce the switching frequency?

Should I left VCCX open as it is shown in the webbench or should I connect the pin to GND as the datasheet recommends?

Any other suggestions?

Thank you!

Regards,

Stefan

In my case reducing the frequency seems to have corrected the problem.  I used idfferent FETs than the Webench specified because they already existed in our system.  When I looked at the gate charge I realized I was violating the current capability of the internal regulator.  I considered providing an external voltage supply to VCCX which would have reduced the current demand from the internal regulator but by the time I discovered my error I had artwork in hand.  It was much simpler to reduce the switching frequency.  I have not experienced a similar failure since doing so.

On a different note I have not been terribly happy with the current limit performance.  It seems that if I load the supply near or slightly above the current limit it works fine.  A short on the output, however, results in a failure similar to what I experienced with the higher switching frequency.  As long as you are not exceeding the current limit set point you probably aren't seeing this issue.  I just thought I'd mention it...

Regards,

Wayne

I also ended up connecting VCCX to GND.

I forgot to add that in my previous response.

Regards,

Wayne

Hi, 

in my case the cure was to add a gate resistor (6.8 Ohm) to the upper FET. There was shoothrough, indeed. I was just unable to see it with my scope. The current limit is also working well since then.

Thanks a lot

Martin

Hi,

with "very low gate-charge"-FETS (<20nC) and a 2.2 Ohm gate resistor at the upper FET the DCDC works. But I don't know whether the gate resistor is required or not.

Thank you!

Regards,

Stefan