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.

LMR12015 failing after various run times

Other Parts Discussed in Thread: LMR12015

Hi

I am using the LMR12015 in a standard step down design produced by Web bench. Have selected own capacitors, inductor and diode. These still meet the part design parameters. Operating current 120mA, input voltage 12V, output voltage 7.7V.

We are having a number of failures where the design is blowing/burning up in the corner region of pins 8,9 &10.

With this devices onboard protection what could cause the device to fail in this way? I am trying to understand if this is a placement/solder issue or a design issue.

Any help would be appriciated.

Thanks

Luke

 

 

 

 

 

 

  • Please send us your detailed schematic with part numbers/values and also the board layout in pictures.

    Thanks,
    Akshay

  • Hi

    I have attached a word file with circuit and BOM information. It also has a picture of the PCB layout.

    Luke

    8053.LMR12015 circuit details for TI.docx

  • Your component placement on the layout is good. But please use copper pours for Vin, Vout and Gnd nodes. Sending power through vias can always be trouble some because you are adding parasitic inductance in the path. The parasitic inductance will cause spurious spiking and could harm the part. Also please remove relief connects from component pads. Your SW pad is broken in to many small traces if you look closely. You should have one pour covering all the components so as to reduce parasitics.

    You could probe the Vin pins on the IC with open bandwidth on the scope and look at the noise on that pin. I am attaching a presentation on EMI and layout techniques also. You could also add an input filter if you think your input is not well behaved at all. This will help control some noise upfront. Method to calculate the filter is explained the presentation that I have attached.

    Regards,
    Akshay

    Switching Power Supply Design_ EMI.ppt
  • Hi Akshay

    Thank you very much for the information supplied. I already have a filter on the input path. Its on the top layer and thus the vias coming through.

    You mentioned removing the relief connections from the component pads. Is this with reference to the vias under the IC? These are thermal relief vias. Should they be moved to the side of the device. WEBBench suggests using 8 vias, are these in the wrong place?

    Do you have any advice on what could cause the device to burnout in the way we have seen? What element on the die is located in the area of pins 8,9 &10? Maybe this could help identify the failure mode.

    Regards

    Luke

  • Hello,

    If you have filters on the top layer then you should have more than just two vias to send the input down to the bottom layer. Also the vias add line inductance, therefore you must have some damping post filter. An additional electrolytic cap after the filter on the bottom should help. Please go through the filter selection guide in the presentation I sent across. 

    By relief connects I mean the the way the copper polygon is poured on the nets. Relief connect is a way of having reduced connection to the component pads to facilitate soldering. Direct connect is essentially covering the entire pad in copper. Please check the image I have attached. Your vias are in the right place. They should be right below the IC.

    The pins being destroyed are the input pins of the IC. If there is too much stress on them then that can happen. For this device a max voltage rating on the input is 20V. If there are any spikes on your input because of parasitics, then the input can hit 20V or more and cause damage.

    Regards,
    Akshay

  • Hi
    I am still working on tying to solve this failure problem. I have conducted tests to ensure there are no transient spikes on the input pins, and have relaid the PCB with better solder past and via arrangement. The part is still failing.
    The failure is apparant at power with the circuit drawing more current than expected. I can make it fail with power cycling the circuit. There is no extra load connected to the output other than the 750R load resister.
    Could there be a case that the 750R load is too large, ie my mimimum load current is too low.?
    I have noticed that on a failed unit it sat in frequency foldback for a while (20-30sec) before increasing frequency and shorting the input to GND as it failed.

    Thanks
    Luke