LM5013: This is a question about LM5013 design.

In both samples in question, destruction occurred near pins 2 and 3 of the LM5013 CHIP. After that, a short occurred between PIN 1 and PIN 2 and 3.

Hi Kim,

The issue could be due to input voltage transient during hot plugin. Can you please measure the input voltage across the input capacitor C2 , switch node and output capacitor during start up?

Thanks,

Nitya

Hi Nitya,

I don't understand exactly what you mean. Can you explain it easily? 
Which part do you want to measure based on the reference number?

Thanks,

Kim

What reference do you want to measure?

The + side of C2 is a yellow waveform, and the + side of C6 is light blue. 
The first picture is when the output is normal, and the lower picture is when the LM5013 CHIP is destroyed.

Hi Kim,

In failed case, the input voltage is within the abs. max limit as per above waveform. This could be due to sampling limitation of the oscilloscope as well. I would like you to do couple of experiments. 

1. With existing schematic and layout, capture the input voltage (C2 capacitor) and GND with 10usec/20usec/100usec time scale.  

2. It is recommended to use the electrolytic capacitor at the input of the LM5013 if there is lengthy cable from battery to input of the device to dampen the LC oscillation. Try soldering 33uF/47uF electrolytic capacitor in parallel to C2 capacitor. 

Thanks,

Nitya

thank you
I will do as you said.

Hi Nitya,

I changed the C2 value from 2.2uF to 47uF and measured it.
We were able to obtain much more stable values.
In the attached picture, the first picture is with C2 set to 0.1uF, 
the second picture is with 2.2uF, and the third picture is with 47uF.

The time was measured at 500uS.

I will also upload the measurement at 10uS.

The first picture shows C2 set to 2.2uF and the second picture shows C2 set to 47uF.

In my opinion, there will be no problem if you select C2 as 47uF.

Hi Nitya,

Thank you for your help. Let's do some more tests.

Hi Nitya,

I made several samples and tested them by replacing C2 with 47uF. However, 
in two samples, PINs 2 and 3 of the LM5013 CHIP are broken as before, causing GROUND PIN and SHORT. 
I am attaching the photo captured at this time. Please review

The yellow waveform is the input voltage DC75V, and the light blue waveform is the output voltage 48V, 
but when a fault occurs, the highest value in the picture instantly rises to 72V.

Hi Kim,

This is strange. The input voltage is pretty much stable. However the output voltage is soft starting as expected for some 700usec and suddenly the output voltage overshoot is observed causing the output voltage to reach around 70V. Can you measure the switch node waveform as well along with input voltage and output voltage to debug further?

BTW, on how many boards you are seeing this issue and what is failure rate?

Thanks,

Nitya

Hi Nitya,

Which part of the switch node waveform are you talking about? 
Can you tell me the exact location? 
Four sample test quantities and defective quantities were tested, and two of them had the same defect.

Hi Kim,

Refer below image marked in red. This is the location of switch node of dc-dc converter.

The input power supply method is first connected to the test sample using a cable with a 2-contact mechanical switch connected 
in the middle of the POWER SUPPLY, and then turning the switch ON/OFF to supply power to the test sample.

Among the attached pictures, numbers 1 and 2 are the input voltage and measured values ​​of the switch node, 
and numbers 3 and 4 are the output voltage and measured values ​​of the switch node.

I am sending you the 4th picture again. This is a wrong picture

When supplying input voltage to a test sample, a picture of when the sample fails is captured and attached. 
The yellow waveform is the output voltage, and the light blue is the switch node waveform.

Please review