LTC4020 MOSFET blow-up issue

[Tinytux]

Hello designers,

I just registered and have taken a quick look if my problem is discussed already. Maybe I missed it, but this is mine:

The circuit is the LTC4020 with Si7252 MOSFETs and extra 5V supply (like the application on page 39 of the datasheet, just Li-ion instead of lead-acid mode).

Even if the 4020 is disabled via the SHDN input, every now and then the MOSFETs are killed together with the 4020. I've never seen this during normal operation, only at switching on - even without load.

Now I started measuring the gate-voltages without MOSFETS and saw, that sometimes they go up to 5V, sometimes not (I've got two similar circuits to compare at the same time - they behave randomly).
My guess is, that when some or all 4 MOSFETs are driven the supply gets shorted via the MOSFETs and the charge of the electrolytic capacitors is enough to kill the components.

Has anyone ever had this issue and maybe a solution?
Is the 4020 sensitive to repeated on-off cycles?
Is there anything I haven't considered?

Thank you so much for support, Manitux

 

[ahmed saad]

Please upload schematic & Datasheet or link to Datasheet to make it easy for others to help you

 

[Tinytux]

Sorry. This is the link to the datasheet:
http://cds.linear.com/docs/en/datasheet/4020fc.pdf

I'll make an extract of the specific schematics and put it here too.

 

[bowman1710]

What are the specifications for your design?? Do you have the exact setup as is page 39? What is your switching frequency, what are you using for your input voltage/supplies etc?

 

[Tinytux]

For now I will not continue with the actual design. I have ordered the genuine Demo Circuit 2044A and will do a redesign on this platform. Appearently this design does a correct and reliable startup.

Anyhow, my initial intension for this thread was to hear about experience with the errant drive of the MOSFETs and short circuits...
Thanks for the patience so far, I will report if there is better luck with the new setup.

Tinytux

 

[bowman1710]

Yes it is best to get a demo board in and mod it to your needs, I have had a lot of issues with power supplies when I have tried to make some in the past. All I have learnt it that is way more efficient using a demo board and modifying it to your needs. You are less likely to fall foul of tracking issues, plus any fault finding tends to be because of design rather then layout.

 

[treez]

IT is BEING USED AS CURRENT SOURCE? (sorry about caps)
Make sure you have plummed in the output overvoltage protection in properly, as current sources will blow the output otherwise.

 

[FvM]

Anyhow, my initial intension for this thread was to hear about experience with the errant drive of the MOSFETs and short circuits...

I wonder how you achieved the results indicated in post #1? The device uses bootstrap drivers for the high side, its operation is not so easy to debug. Seeing the low side transistors switching isn't a problem as such, shoot-though should be blocked by the basic driver hardware.

Guessing about possible problem causes, I would put inappropriate layout at first place. In so far it's a good idea to start with a verified demo board.

I designed a power supply around a similar but slightly different (36V V LTC3789) buck-boost switcher and it worked flawlessly.

 

[treez]

I would get it working on the simulator (free ltspice) first, then post your sim on here and people will really know what you are doing then.