This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM3409HV: How to protect the IC?

Part Number: LM3409HV
Other Parts Discussed in Thread: LM3409

Hello,

I have a board with 48 LM3409HV driven with a picoZed board from Avnet. When I receive these SOM there is already a program in the FPGA.
As soon as I connect these 2 boards (power management board and picozed) I always have a current sense resistor burning and an IC exploding  (1 out of 48).
I think this is because of the program loaded in the FPGA. Here is the schematic of a LED controller:

What could happen? Even if the shunt FET (T3) is always on and enable the IC should maintain the a good current value (about 0.5Amp).
The Iadj voltage comes from a DAC081 that should be at 0 V at startup. I was thinking at a too high frequency on the LVDS line, is this possible?
Rsns is 0.25 Ohm, 2W.

The solution is to clear the boot flash on the picoZed and reprogram with the right program but technicians sometime forget it.

Thanks, Serge.

  • Hello Serge,

    Do you know if the LM3409 is getting damaged or is it just R2 (0.25 ohm sense resistor)?

    Were you able to take waveforms of what happened when it got damaged? It would be important to look at the SW node (Connection between T2 (top side fet) and D2 (the catch diode). Were the LEDs turned on at all before it got damaged? I would also like to have seen the output voltage. If the shunt fet turns off very quickly it is possible to get large voltage spikes on the puts due to parasitics. You can clamp any overshoots on the output with a diode from the output to the input. You may want to put pull down resistors on gate fets to prevent transients from turning them on. You also have some inputs that are potentially left floating when the micro isn’t on or is in reset mode.

    Here is an update of your schematic with some ideas about protecting the part.  We need to catch what happened.  I am guessing that maybe the part was damaged and that pulled the SW node to ground permanently  and then it blew up the Rsns.  That is only a guess.  

    Let me know what you find.  I hope that this was helpful.

    -Francis Houde

  • Thank you for your answer Francis.

    I'm almost sure the problem happens with Avnet's software in their picozed boards. I have a lot of these materials working without any problem as long as my program is loaded in the FPGA.
    The IC is totally exploded with the die pad spotted on the PCB ground plane. I suppose the sense resistor burns first but I'm not sure since I cannot reproduce the problem on bench because of the cost of these boards.

    I think your solution with a diode to avoid spikes is a good idea, I believe I can use the same as D2.

    Iadj voltage comes from a DAC081 which is at 0 volt on startup. Enable comes from picozed with a 74AVCH1T45 and pull-down resistors in between to avoid floating voltages.
    I don't think the LVDS receiver can produce a logic '1' when inputs are not connected.

    I cannot reproduce the problem but I'll try to check what is coming from the picozed as soon I have a new one coming from Avnet. This is the best way to know what are the conditions at first.
    Thank you to give me the way to solve this problem.
    I asked this question at first because I thought this was a known problem.

    I'll be back with more informations soon.

    Serge.

  • Hello Serge,

    You can use the same diode as the catch diode, but that is over kill.  I would go with a Schottky that is 200mA or more rating and have a voltage rating of what ever the max operating voltage plus some margin.  Our part is 75V, so I would go with 75 or even 100V Schottky to be safe.     

    Let me know how it goes.  

    -Francis HOude

  • Hi,

    I always prefer to maintain the BOM list as short as possible so I'll use the same if there is enough room on the board, I'll check that.

    Thank you for now Francis, I'll come back as soon as I have more information on signals coming from the processor board.

    Serge.