Author Topic: Interesting failure of buck converter  (Read 7009 times)

0 Members and 2 Guests are viewing this topic.

Offline mikeselectricstuffTopic starter

  • Super Contributor
  • ***
  • Posts: 14117
  • Country: gb
    • Mike's Electric Stuff
Interesting failure of buck converter
« on: May 24, 2019, 03:11:37 pm »
I have a prototype board using a ST ST1S14PHR buck converter, converting 48V to 13.5V at 1.25A max output current.
I'm seeing an odd failure mode - after it's been running hot at high loads for a few 10s of secs, instead of the ST1S14 going into thermal shutdown, the diode fails to a near-short.
At lower loads ( ~1A) the thermal protection works OK, the chip cycling on and off.
I've confirmed that it is the effect of heat ( from the chip - the diode produces minimal heat itself), as putting a soldering iron on its ground pin induces the same failure quite quickly.

The diode is an STPS5L60U, rated 60V 5A. (Datasheet recommends STPS3L60, the 3A version)

I know Shottkys get quite leaky at higher temps, so my current theory is the 60v rating isn't enough when it gets hot. I've ordered some 100v diodes to try as I can't find any to hand.


 
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 
The following users thanked this post: InductorbackEMF

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13523
  • Country: gb
Re: Interesting failure of buck converter
« Reply #1 on: May 24, 2019, 03:41:49 pm »
Failure of the diode appears to be a challenge to other designers as well.

See here....

https://electronics.stackexchange.com/questions/86791/cause-of-diode-failure-in-step-down-switching-mode-power-supply

May be nothing to do with your failure mode but some interesting comments. High voltage ringing looked to be one possibility and with high currents and any inductance high frequency, high voltage ringing could potentially occur ? An RC snubber does look like a good idea if this is suspected.

Another suggestion mirrors yours. That of thermal runaway due to external heat causing increased leakage.

Hope you solve the mystery.

Fraser
« Last Edit: May 24, 2019, 03:45:35 pm by Fraser »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13523
  • Country: gb
Re: Interesting failure of buck converter
« Reply #2 on: May 24, 2019, 03:50:58 pm »
The last comment on that referenced thread looks pretty convincing......
 To quote it here .......

“It will be over-voltage spikes that are causing the Schottky shorts (many years experience in SMPSU). If you look at the waveform with a higher bandwidth scope you will seen them. 100 MHz not really good enough, suggest 500 minimum to see all. Use snubber as suggested or higher voltage rating.

I realise this is very late, but may help others, who find the thread as I did.”

End quote.

Hope this helps

Fraser
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Interesting failure of buck converter
« Reply #3 on: May 24, 2019, 04:13:21 pm »
Shouldn't be much overshoot on turn-on though, at least not unless the bypassing is terrible?

That diode is rated for avalanche, so it's very suspicious that it would fail due to overvoltage without also heating up.  It's probably not rated for much avalanche, so maybe it fails pretty suddenly when subjected to a large enough surge?

The other way diodes can fail suddenly, is large peak currents.

And dV/dt, but I don't really know what that does for diodes (and MOSFETs).  It's hard to even generate that much rate, though.  (Easier with GaN nowadays.)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline thinkfat

  • Supporter
  • ****
  • Posts: 2161
  • Country: de
  • This is just a hobby I spend too much time on.
    • Matthias' Hackerstübchen
Re: Interesting failure of buck converter
« Reply #4 on: May 24, 2019, 04:56:54 pm »
I would expect the diode to generate some heat. At the given conversion ratio, there should be a significant ripple current which the diode has to bear.

Gesendet von meinem Nokia 6.1 mit Tapatalk

Everybody likes gadgets. Until they try to make them.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: Interesting failure of buck converter
« Reply #5 on: May 24, 2019, 07:13:01 pm »
This is just an idea but at high currents, the parasitic PN junction in parallel with the schottky junction can become forward biased enough to store charge and create reverse recovery.  A short is created during the reverse recovery which can lead to reliability problems.

Maybe make a measurement of the high speed diode current waveform to see what is going on?
 

Offline Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 2766
  • Country: ca
Re: Interesting failure of buck converter
« Reply #6 on: May 25, 2019, 09:52:12 am »
Mike,
I suspect that  this is just over voltage on the diode.


There will be transients on the switch node. This might be very short, but enough to damage the diode.

While you are waiting for the higher voltage diodes, try testing the circuit with a lower input voltage, say 36V

The higher current diode, 5A versus 3A, may have higher junction capacitance which could increase the transient on the switch node.

The transients on the switch node will change with temperature.

60V diode in a 48V (+/- _ %) application isn't enough margin.

Regards,
Jay_Diddy_B
 

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: Interesting failure of buck converter
« Reply #7 on: May 25, 2019, 10:13:31 am »
Schottky diodes have significant junction capacitance - much higher than the equivalently rated PN diode - and it's typically high-Q, so like the commenter reposted by Fraser, my vote is on ringing as the most likely cause of death here, too.

That said, a high junction capacitance will also allow a big pulse of current to flow at the instant of reverse biasing which looks an awful lot like reverse recovery in a PN diode. This capacitvely coupled current shouldn't hurt the diode, though, since it isn't using charge carriers to travel across the barrier junction, but I am not a semiconductor device physicist so take that with the proverbial grain of salt.

The idea floated by David Hess - the guard ring diode becoming forward biased at a high enough current - is another possible cause, but that one doesn't seem as likely to me unless the buck inductor is saturating and peak currents are way higher than the 5A diode rating.

 

Offline Ice-Tea

  • Super Contributor
  • ***
  • Posts: 3229
  • Country: be
    • Freelance Hardware Engineer
Re: Interesting failure of buck converter
« Reply #8 on: May 25, 2019, 10:50:03 am »
I would expect the diode to generate some heat. At the given conversion ratio, there should be a significant ripple current which the diode has to bear.

The diode is only there to bridge the dead time between top and bottom fet switching..

OP: I assume you checked the diode out-circuit? If not, you may be measuring the bottom FET tossing it.
 

Offline mikeselectricstuffTopic starter

  • Super Contributor
  • ***
  • Posts: 14117
  • Country: gb
    • Mike's Electric Stuff
Re: Interesting failure of buck converter
« Reply #9 on: May 28, 2019, 02:32:27 pm »
I replaced the diode with a V3FM12-M3/H , which is rated at 120v, and it will now sit with its cathode lead in a blob of molten sodler and carry on working fine!
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13523
  • Country: gb
Re: Interesting failure of buck converter
« Reply #10 on: May 28, 2019, 04:14:10 pm »
Excellent news  :-+ So you think thermal runaway is the issue ?

Fraser
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline mikeselectricstuffTopic starter

  • Super Contributor
  • ***
  • Posts: 14117
  • Country: gb
    • Mike's Electric Stuff
Re: Interesting failure of buck converter
« Reply #11 on: May 28, 2019, 05:18:16 pm »
Excellent news  :-+ So you think thermal runaway is the issue ?

Fraser
Not runaway - I'm pretty sure that it was the heat from the converter that was pushing the diode over the edge. 
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 2766
  • Country: ca
Re: Interesting failure of buck converter
« Reply #12 on: May 28, 2019, 10:34:49 pm »
Hi Mike,

I had a look at the datasheet:



That is not a lot of margin when operating from 48V supply.

I also think it is telling that the do not present a data when operating at 48V on the datasheet. They show scope shots operating into a short with 43V input and lots of data at 12, 24 and 32V.

They also show 50V input capacitors. I would want to see 20 or 30% margin.

I know that they do say 'Operating input voltage from 5.5 V to 48 V'

You might want to take a page out photonicinduction's playbook and turn one up till it pops. :D


Regards,

Jay_Diddy_B
« Last Edit: May 28, 2019, 10:37:29 pm by Jay_Diddy_B »
 

Offline mikeselectricstuffTopic starter

  • Super Contributor
  • ***
  • Posts: 14117
  • Country: gb
    • Mike's Electric Stuff
Re: Interesting failure of buck converter
« Reply #13 on: May 29, 2019, 09:10:47 am »
Unfortunately there aren't many small 48V converters like this at sensible prices - I'd seen that chip used in another big installation with several hundred nodes, so was fairly happy using it. Something I was concerned about, and did test for, was whether overvoltage transients caused by connecting & disconnecting the supply caused issues, but it seems fine.
 
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline thinkfat

  • Supporter
  • ****
  • Posts: 2161
  • Country: de
  • This is just a hobby I spend too much time on.
    • Matthias' Hackerstübchen
Re: Interesting failure of buck converter
« Reply #14 on: May 29, 2019, 01:23:45 pm »
I think I'm going to try this on one of my LMR14030 based buck converters. I'm guessing it'd show a similar failure if I drive the input voltage high enough. That's going to be interesting to watch.
Everybody likes gadgets. Until they try to make them.
 

Offline Ice-Tea

  • Super Contributor
  • ***
  • Posts: 3229
  • Country: be
    • Freelance Hardware Engineer
Re: Interesting failure of buck converter
« Reply #15 on: May 29, 2019, 01:24:54 pm »
I agree that that ain't much of a margin...

MPS has some:

Digi Link
 

Offline Neomys Sapiens

  • Super Contributor
  • ***
  • Posts: 3268
  • Country: de
Re: Interesting failure of buck converter
« Reply #16 on: May 30, 2019, 01:24:57 am »
What did you mean by 'prototype board'? Is it a demo board for the regulator or some other prototype where it does supply something? When you compare the actual layout with their recommendation regarding heat removal, how far apart is it? I have made the experience that higher power SMD parts can't be used if not provided with the correct thermal vias etc. So if that heat is coming from the regulator, which you say is running correctly, it has to go somewhere else instead of towards that diode.
 

Offline mikeselectricstuffTopic starter

  • Super Contributor
  • ***
  • Posts: 14117
  • Country: gb
    • Mike's Electric Stuff
Re: Interesting failure of buck converter
« Reply #17 on: May 30, 2019, 08:39:53 am »
What did you mean by 'prototype board'?
It's the pre-production protoype PCB, so as good a thermal layout as possible with the mechanical constraints I have.
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: Interesting failure of buck converter
« Reply #18 on: May 30, 2019, 09:43:46 am »
Unfortunately there aren't many small 48V converters like this at sensible prices

You get what you pay for ;) I agree that 50V is way too slim margin for 48V supply, especially when transients can be there. Better add some 1$ to BOM for better regulator and sleep well ;) In your case I would use http://www.ti.com/product/LMR16030
 

Offline mikeselectricstuffTopic starter

  • Super Contributor
  • ***
  • Posts: 14117
  • Country: gb
    • Mike's Electric Stuff
Re: Interesting failure of buck converter
« Reply #19 on: May 30, 2019, 10:09:58 am »
Unfortunately there aren't many small 48V converters like this at sensible prices

You get what you pay for ;) I agree that 50V is way too slim margin for 48V supply, especially when transients can be there. Better add some 1$ to BOM for better regulator and sleep well ;) In your case I would use http://www.ti.com/product/LMR16030
Efficiency of the TI part looks way lower - about 82% vs. 91% for the ST, based on nearest graphs in the datasheet - ST has a sync. rectifier fet across the diode.
For this application I'm very constrained for space to get rid of heat.
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: Interesting failure of buck converter
« Reply #20 on: May 30, 2019, 10:21:28 am »
Efficiency of the TI part looks way lower - about 82% vs. 91% for the ST, based on nearest graphs in the datasheet - ST has a sync. rectifier fet across the diode.
For this application I'm very constrained for space to get rid of heat.

Synchronous switcher then, definitely. http://www.ti.com/product/LM76002 May blow your BOM, but efficiency is way better than ST part.
 

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: Interesting failure of buck converter
« Reply #21 on: May 30, 2019, 10:23:01 am »
Well, honestly you ought to know better than to try to run a 50-55V rated part on 48V, and all just to shave less than a pound off the BOM... Given that the most common failure mode of a buck converter from overvoltage is a shorted switch, you are definitely playing with fire here, and perhaps not entirely figuratively, either.

Personally, I'd consider the MPS MP9846 or the Microchip (nee Micrel) MIC28514 just from a quick sort of buck regulator parts on Mouser that can handle at least 60V input and 3A output current.
 

Offline mikeselectricstuffTopic starter

  • Super Contributor
  • ***
  • Posts: 14117
  • Country: gb
    • Mike's Electric Stuff
Re: Interesting failure of buck converter
« Reply #22 on: May 30, 2019, 11:10:12 am »
BOM cost isn't a huge deal, it's more that I'd already seen this part used in a reliable 48v application
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline Ice-Tea

  • Super Contributor
  • ***
  • Posts: 3229
  • Country: be
    • Freelance Hardware Engineer
Re: Interesting failure of buck converter
« Reply #23 on: May 30, 2019, 11:45:56 am »
Unfortunately there aren't many small 48V converters like this at sensible prices

You get what you pay for ;) I agree that 50V is way too slim margin for 48V supply, especially when transients can be there. Better add some 1$ to BOM for better regulator and sleep well ;) In your case I would use http://www.ti.com/product/LMR16030
Efficiency of the TI part looks way lower - about 82% vs. 91% for the ST, based on nearest graphs in the datasheet - ST has a sync. rectifier fet across the diode.
For this application I'm very constrained for space to get rid of heat.

MP4559 might still be worth a look.
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13523
  • Country: gb
Re: Interesting failure of buck converter
« Reply #24 on: May 30, 2019, 12:23:42 pm »
This thread is really interesting as it discusses component ratings. I come from a background where components used in mission critical applications were heavily over engineered. Even when a component had a rating of X Volts, it was a pessimistic value and the components would often have better capability, it was just not guaranteed. We often tested components to destruction and were impressed with the true headroom on a component rather than that indicated by the data sheet. Whilst it is true that it would be unwise to deliberately run a component beyond its maximum ratings in an important role, it might not be such an issue to run it at or near that rating if the manufacturer has already calculated a decent margin of safety to cover issues like transients. If transients are considered a risk them maybe a TVS can be used in the supply to the chip ? In important applications, some decent supply filtering and transient suppression is wise anyway.

If I were in Mike’s position I would sacrifice a few of the chips in a torture test to see at what voltage they fail and the failure mode. Such is useful information anyway as a non safe failure mode would either steer me away from a component or cause me to employ additional safety measures that might otherwise have not been required.

When there is little choice of components for a task, it is sometimes worth a little experimentation to see just what can be used safely without breaking the bank or having to completely change a design. This is, after all, what OEM’s do when prototyping, often to test reliability, but also to see how cheaply a design can be built and still meet the needs of the customer. Those who get it wrong do cause issues with premature component failures though.

Fraser
« Last Edit: May 30, 2019, 03:48:40 pm by Fraser »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf