Solution Finder Get Tech Support

LNK304 Blew up

Posted by: empower on 06/19/2012

We are using LNK304 as our low voltage supply. The input voltage is 347-480V. We have been testing this for a while and it has been working great. We recently install our system to one of the location and we have 34 units fail out of 100 units with the same symptom. The LNK304 blew up the top of the chip opened up. The Mosfet is shortage and some blew up too. The fly back of the primary inductor went to zero. Not sure what cause this blow up. Please help. Attach is schematic. The rectifier DC is about 678V.

Thank you.


Submitted by PI - Traveler on 06/22/2012

empower - 


Can you attach your schematic as a PDF?  The resolution is too low on the image you attached and I can't read it.


What kind of differences are there between where you were using the product before and the new location? 


Just out of curiosity, what is the basis of your username?



-The Traveler

Submitted by empower on 06/22/2012

Hi The Traveler,


Please see attached file of schematic in pdf. The differrent between the 2 loactions is the first no problem location is the car dealer which we insatlled about 40 units and second problem location in the mall which we installed100units. Both locations have 480V line. Username is company name.

The failure mechanism you're describing is usually associated with BVDSS of the MOSFET getting exceeded.  With the design working fine in one location but not in anothe, my hunch is one of both of the following: electrically, something is different about this new location and/or the design you've created is borderline in some way.  


For the location, it's not uncommon to see differences in the line voltage transients between two very similar locations.  Power line quality can be a huge issue in some areas.  Where the power supply is physically installed can also have a big effect on the line voltage transients experienced by your power supply. 


From your PDF, I can't see anything in the circuit before RECT+.  In a three phase 480V use scenario, you will want to make sure you're doing everything possible to protect your power supply from differential and common mode line voltage transients.  Depending on installation location, you might have to design your power supply to be capable of dealing with 6kV line transients.  IEEE has some nice documentation and reports about line voltage transients.




-The Traveler

Looking through your schematic, I spotted an issue that might be the source of the problem you're having.  I noticed that in your clamp circuitry, you're using a 1n4007 diode.  Our StackFET documentation recommends and UF4007 for this application.  The reason for this is that if the 1n4007 hasn't finished turning "off" when the LNK304 turns on, there will be no primary inductance to slow down the current rise through the MOSFET...the circuit becomes a short circuit.  


This diode choice combined with possible differences between the physical locations (and possibly inadequate transient suppression) is my best guess as to the failures you're seeing.  If you insist on using a 1n4007, make sure it's a GP version (glass passivated) so that the reverse recovery time period of the diode is actually tested by the manufacturer.  The non GP versions often don't even have a reverse recovery time spec.  This is why we don't recommend them for use in clamp circuitry components.



-The Traveler

Submitted by empower on 06/25/2012

Hi Trveler,


Thank you very much. We will change 1N4007 and test it out. We have a part that we use DIODE ULT FAST 1200V 2A SMB (STTH212U). Is this OK to rpelace 1N4007? Please let us know. Thank you again for all your help.



Submitted by PI - Traveler on 06/25/2012

In reply to by PI-Surak

We recommend UF4007 in our StackFET documentation.  UF4007 has a reverse recovery time of approimately 75ns.  An equivalent diode with a reverse recovery time around this value would like be just fine.


Besides the design itself, I would also encourage looking for differences in the two sites you're using this power supply.  If the existing design with the improper clamp diode was working fine (despite being incorrect) and suddenly the new site is experiencing lots of failures.....well, it tells me something is really different about these two sites.  


A good experiment, if possible, would be to use a scope with a high voltage differential probe configured to trigger when it expereiences voltages much higher than the typical line voltages.  You could configure the scope to save the waveform for each trigger or just do an overlay capture where it stores multiple captures over and over again onto the same plot.  This might give you some insight into any differences that might exist between the two sites in regards to line voltage transients.  Again, from past projects, I can tell you from experience that the line voltage transients experienced by power supplied installed close to the utility drop experience significant quantities of high-voltage line transients.  6kV line voltage transients are fairly common when located at the utility panel and this is just for 115/230 Vac setups.


-The Traveler

Submitted by empower on 06/28/2012



We replace the diode with DIODE ULT FAST 1200V 2A SMB (STTH212U) which has 75 ns recovery time. We tested and after 5 hrs we had a the same failure. The M1 mosfet blew up and the LNK304 chip also blew up. So we got the new system and put thermo couple on the MOSFET M1 and the temperature is 105 C before it blew up. LNK304 chip's temperature is 120C.  Do you think this cause the problem. I did think it will run that hot or may be something esle cause M1 and LNK304 to run hot. The part for M1 is in the attached file and diode.

Submitted by PI - Traveler on 06/29/2012

empower - 


At this point, I'm going to need some additional detailed information about your design.


If you could please provide the following documentation:

Your complete schematic (including the input EMI filter, MOVs, Xcaps, etc)

Your PCB layout (PDF preferred)

Your PI Expert design files

Your transformer build diagram

Your transformer leakage inductance measurements at  50kHz, 100kHz, 500kHz and 1Mhz


Additionally, the following information would also prove helpful:

Your LNK304 Drain-to-Source voltage waveforms directly before failure

Your Stack-FET Drain-to-Source (IC Source) voltage waveforms directly before failure


Other questions:

How did you arrive at using a 130V zener in your clamp circuitry?  

Is the 130V zener running hot?

Are there any other components that are running at elevated temepratures?

If you run this power supply from a controlled, quality AC source (lab bench supply), do these failures occur?

What is the nature of the installation of this power supply?  Is it near high-power machinery?  Is it at the end of a long over-head power-line run?  Is it installed in a location that is much hotter or colder than your original installation?  


My next guess is that your zener clamp is chosen at way too low of a voltage or your transformer wasn't designed correctly and has a lot of additional leakage inductance. This could very easily overheat the clamp components which would then fail as a short.  With the clamp circuitry failing closed (Short), the BVDSS of of M1 and LNK304 wouldn't be exceeded.  However, once the LNK304 tries to switch, there would be a direct short from Bulk+ to SRC and M1 and LNK304 would easily get destroyed by the large surge current.  


If you can please provide the requested information, I think we can nail down this problem rather quickly.


-The Traveler

Submitted by PI - Traveler on 07/10/2012

empower -


Victor forwarded me the information you sent him since we were already working on this issue together.  


Looking at the information you sent him, I have a few more thoughts/questions/suggestions:

  • The kind of failure you're describing is typically caused by either Vds of the device exceeding the rated BVDSS, transformer saturation or failure of the clamp circuitry.
    • You've mentioned that the MOSFET and PI device are running very hot.  Have you checked the temperature of the transformer?  High core temperatures will cause the transformer to saturate.  Also, what is the temperature of your clamp circuitry components?
    • If the transformer is starting to saturate, you'll see a very tell-tale sign when you look at the drain CURRENT waveform.  You should have a nice linear ramp.  As a transformer starts to saturate, the peak of the ramp will start to rise exponentionally.  I've attached a quick sketch of what this looks like.  It's a very classic sign of transformer saturation and will cause catastrophic failure of the IC.
    • If your clamp circuitry is running hot, it's possible DL3 could be failing closed.  If this happens, the primary winding of the transformer is bypassed and there is no inductance to slow down the current ramp when the MOSFET turns on.  This will cause catastrophic failure of the IC and likely the StackFET.
  • Looking at your schematic, I'm also noticing that it looks like you're using the feedback winding for your PFC.  This isn't recommended.  The power draw from external components will affect the voltage seen by the LNK304 IC and could affect the regulation of your outputs.  If you need an additional output, use an additional winding.
  • I also noticed some problems with your transformer documation:
    • You are specifying 5mH @ 1kHz.  You need to measure the primary inductance at LNK304's switching frequency 66kHz).
    • You arn't asking your transformer vendor to measure your primary leakage inductance.  Why not? If you have excessive leakage inductance, your clamp circuitry will run hot (possibly causing failure) and you'll suffer from efficiency issues.
    • Where is the winding diagram?  This is a VERY important part of your transformer documentation.



-The Traveler

Submitted by Kanwar on 01/06/2017

In reply to by PI - Traveler

The arrangement of stackfet is flawed, the Gate to Source terminal should also have a clamping resistor across it around 22k in parallel with zener in order to facilitate fast turn off of upper mosfet. Otherwise the mismatch in turn off causes the VDS to rise instantly and the chip goes boom !