Author Topic: Buck converter IC fail Vin shorted to ground  (Read 306 times)

0 Members and 1 Guest are viewing this topic.

Offline balor

  • Newbie
  • Posts: 2
  • Country: fr
Buck converter IC fail Vin shorted to ground
« on: May 22, 2021, 10:35:04 am »
Hi,
I am designing a buck to get 3.3 V (for a µC) from a battery (14 lithium cell in series 35-58 V).
In my first test using a bench power supply my design seem to work fine. I get a stable 3.3V output.
the graph below was taken with a LED as load and an additional 100µF electrolytic capacitor for decoupling and with 60 V on the input.


However when a try to connect it to the battery ( 56 V) instead of the power supply the LMR16006 fail and I have a short circuit between GND VIN and CB (bootstrap).

To me the main difference is the time it take for the voltage to go from 0 to 60 V. When I turn on my bench power supply it take around 20 ms to reach 60 V even with no load. To try and recreate the problem a installed a switch on the output of the bench power supply. That way when I close the switch the the voltage goes fro 0 to 60 instantly (1 µs with no load). This setup successfully recreated the failure of the LMR16006Y.

Unfortunately don’t have tension graph when the IC fail and I only have one IC left so I am waiting on for my order to arrive to do more destructive testing.

While I waiting for the component If anyone as an idea of what is going one I am all hears.

My current hypothesis are:

1. something with  inrush current but :
       I don’t know why inrush would damage the IC
       I don’t have that much capacitance on the input of the IC (2.2µF)  and at this voltage it should drop to 0.5µF which is lower than the value expected in my design (maybe this could the source of my problem)

2. something with the boot strap since it’s one of the pin that is shorted to ground. I am no sure what exactly but  the  since it create a higher tension than Vin I am suspicious.

I have also realized that the minimum on time of the IC (Ton) is to high. Thus I “shouldn't” have my 3.3V at the output but since I have 3.3V at the output I don’t think that’s the issue here.

I tried to recreate the circuit using LTspice to try and see some odd behavior but I think I am not good enough with spice. I also have access to Pspice if you think I could do something with that.

PS: I have made a lot of measurement that I did not include here because I don’t know which one are useful.
 

Offline Siwastaja

  • Super Contributor
  • ***
  • Posts: 3895
  • Country: fi
Re: Buck converter IC fail Vin shorted to ground
« Reply #1 on: May 22, 2021, 12:03:06 pm »
Overvoltage transient peak due to ceramic input capacitor ringing with the lead inductances. The difference to lab supply indeed is in rise time. Add some 100uF elcap to the input and I bet the problem is fixed.

See https://www.analog.com/media/en/technical-documentation/application-notes/an88f.pdf

Also LMR16006Y is only rated to 40V input and you are planning to run at 58V. How did you think this is a good idea? You are already at the edge of failing, not much peak is needed. And ceramic input cap easily generates peak of 2xVin. Semiconductors usually can't handle overvoltages well but fail short.

Use a part with at least 60V recommended operating voltage or 100V absolute maximum rating. Lots of suitable parts available. And add the snubber.
« Last Edit: May 22, 2021, 12:11:05 pm by Siwastaja »
 

Offline uer166

  • Frequent Contributor
  • **
  • Posts: 298
  • Country: us
Re: Buck converter IC fail Vin shorted to ground
« Reply #2 on: May 22, 2021, 03:34:28 pm »
I used that part in 48V regulated VCC with success, but even that is pushing it. At 56V input, I'm surprised it works at all, even with the power supply ramp. Like the guy above said, with ceramic caps at the input and no TVS or electrolytics, it's not surprising it fails when you slam the battery on.

At 56V you might be in the realm of a discrete FET+controller solutions, although there are a couple monolithic chips that do it.
 

Offline balor

  • Newbie
  • Posts: 2
  • Country: fr
Re: Buck converter IC fail Vin shorted to ground
« Reply #3 on: May 23, 2021, 07:13:55 pm »
Thanks a lot for your answer i'll try this week.

For the choicxe of the LMR16006Y i cheked and i found 2 data sheet. One with the 40 V  the LMR16006 Q1 (https://www.ti.com/lit/ds/symlink/lmr16006y-q1.pdf) and one with th 60 V limit (https://www.ti.com/lit/ds/symlink/lmr16006.pdf).

I will check monday but I am pretty sure I ordered the 60 V one. Thanks for pointing out the ambiguity I will update my documentation to remove it.

On the the topic of the IC choice the 60V datasheet has 65V as absolute maximum rating. When I choose It I was hesitant because that not a lot of margin but I fought I could get away with it. Do you think I was wrong in my choice ?
« Last Edit: May 23, 2021, 08:28:18 pm by balor »
 

Offline Siwastaja

  • Super Contributor
  • ***
  • Posts: 3895
  • Country: fi
Re: Buck converter IC fail Vin shorted to ground
« Reply #4 on: May 24, 2021, 08:03:09 am »
Even the 60V version has absolute maximum rating of 65V, which is exactly the same for both parts! It's well possible the two parts handle peak voltages similarly.

So it really seems to me the headline "recommended" value is too careless for LMR16006, and very conservative for LMR16006Y-Q1. The latter is AEC-Q100 qualified so it's well possible the part is basically the same but they have lowered the recommended voltage to create extra margin. For the 60-V rated part, it falls in to the classic category of manufacturer not being very helpful and you having to understand the concept of derating, i.e., not going anywhere near "absolute maximum". If abs.max. is 65V, 55V could be a good practical maximum, and even then, it's your job to make sure no transients ever exceed this.

58V is just pushing it too far, even if you were able to add perfect snubber completely removing any overshoot.

So tldr, get a part rated at 80V recommended / 100V absolute maximum at very least; AND don't forget to add the snubber, any cheap elcap is fine. If you use elcap, measure its temperature rise, it shouldn't heat up, if it does your ceramic is too small and leaving too much ripple for the elcap to handle. So leave all the hard work for the ceramic, and let the elcap work as a snubber, then it will last forever despite being rated for say 5000 hours.
« Last Edit: May 24, 2021, 08:09:20 am by Siwastaja »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf