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.

LMZ35003: Possible Causes of Blown LMZ35003

Part Number: LMZ35003

Hi,

I am using the LMZ35003 in a design that operates off 15-50v from lipo batteries to control 12v LED lights. I have built around 25 units to test with and out of those I have seen 4 failures where the LMZ35003 is blown. In these cases the rest of the circuit remains functional after the LMZ35003 is removed.

The LMZ35003 failed in the same way for each of the blown parts. The VIN pin, INH/UVLO pin, and a path to the AGND pad burned up. You can see the damage to two of the LMZ35003 packages in the image below.


The LMZ35003 is configured for a 12v output with an input range of 15-50v in accordance with the example circuit provided in the datasheet.



In all cases where the LMZ35003 blew, it happened when power was connected and the load current draw was < 1A. Since the failures occurred when power was applied, I suspected the inrush current to be the issue. I have measured the inrush current when connecting to a 12S lipo battery on a few modules I have on hand, and it stayed within safe limits. None of the modules I have tested with have blown from connecting the battery yet.

My product is used in a high vibration environment, so my suspicion at this point is that the failures are caused by a broken solder joint. I believe that one of the supporting passive components is susceptible to a broken solder joint and that causes a cascading failure that burns up the LMZ35003. My question is whether this theory sounds likely, and anyone knows which component failing open would cause the LMZ35003 to burn up in the way that it has.

Since all of the failed LMZ35003's have the same damage to the VIN and INH/UVLO pins, I am lead to believe that the broken solder joint is on R11 (schematic in attachments). I believe that if this resistor opens, the INH/UVLO pin will see a voltage too high too handle, then the LMZ35003 will fail. Do you agree with this assumption?

The second component I suspect is C14. If that capacitor is opened, then the LMZ35003 would no longer be configured for slow start, and the inrush current would be much higher than normal.

Since the parts are expensive, I do not want to arrive at the solution through trial and error. Any help you can give me in finding out why these failures have happened would be greatly appreciated.

  • Matthew,

    The images did not attach properly. Please attach the image and the schematic and I'll take a look.

    My first thought is EOS. If VIN is 15-50V, and this device is rated for 50V max, there isn't much margin to avoid potential damage if the 50V max has a transient spike exceeding 50V. Check VIN on an oscilloscope during power-on and during transient conditions to make sure the voltage isn't higher than expected.

    For your other theories I'd need to see the schematic to conclude one way or the other but likely yes, since the INH/UVLO is only rated to 5V, those resistors are suspect if you are suspecting broken solder joints.

    -Sam

  • Sam,

    Thank you for the reply. I have re-attached the photos.

    I measured the input voltage as you requested and I believe you are correct about EOS being the cause of the failures I have seen. Below is a capture of the input voltage when connecting a fully chgarged 12S LiPo battery to my board. For this measurement, I put a 1 ohm resistor in series with the input, connected CH1 to the PCB side of the resistor, and connected CH2 to the battery side of the resistor. The math waveform that is plotted tracks the current draw of the PCB by subtracting CH1 from CH2.

    As you can see, the input voltage on the battery side rises to 65v before falling back down to the 49.2v of the battery. I do not fully understand what causes this, but 65v is the absolute maximum voltage allowed by the LMZ35003 for surge voltages so I expect that any higher of a surge would cause damage to the IC.

    Do you have any recommendations of how to prevent a surge like the one I have captured from happening?

  • I was able to reproduce the failure.

    I connected the battery voltage directly to the input of the PCB, and this time the LMZ35003 blew. I was still monitoring the voltage, so I was able to capture the waveforms below showing the input voltage during the failure.

    All of the solder joints on the PCB were checked before I began my testing, so I no longer suspect broken solder joints. This failure is a clear indicator that the failures are caused by EOS during the connection of the fully charged battery to the PCB.

    Going forward, how can I prevent these voltage surges from occuring? My first thought is to add some circuitry to limit the input surge closer to 50v but I am not sure at the moment what that circuit would look like.

  • Matthew,

    This overshoot is caused by the inductance of the connection from the battery to the IC. High inrush current stores some magnetic energy in the parasitic inductance of that loop which continues to push the current to the IC even after VIN reaches 50V. You can reduce this overshoot by adding more input capacitance. The datasheet recommends 10uF ceramic or 22uF non-ceramic. I think you should increase both of your 2.2uF capacitors to 10uF, or 10uF ceramic and 22uF electrolytic. The ESR in the electrolytic will dampen this overshoot.

    Otherwise you can add a soft-start inrush current limiting circuit on the input.

    -Sam

  • Sam,

    Thank you for the advice. I will try this out and report back with the results.

  • Matthew,

    I should also mention that you should be able to test this without the IC attached. This will give you worst-case results and will prevent any unnecessary blowups during testing.

    -Sam

  • Sam,

    Thanks again for the advise. That is exactly how I plan to test the design once the parts arrive. It's good to know that that will be the worst case scenario.

  • Sam,

    I have tested the circuit by placing only the front end components (D2, C13, C22, R10, R13) and measuring the voltage seen at VIN when the battery is connected. In the plots below, CH1 is always measuring VIN, and CH2 is measuring VCC.

    The first test I did was the original circuit to confirm what I was seeing before. This measurement is taken with no additional components in the circuit than what is listen above.

     

    The next test I did was to replace C13 and C22 with 4.7uF and 10uF respectively. 4.7uF was the largest electrolytic cap I was able to place in my design without changing the layout.

    I noted that these results show far worse max voltages at VIN than I saw when testing with the 1 ohm series resistor to measure the inrush current. Because of that, I added the resistor back in and got these results.

    The measurement below was taken with 1 ohm in series with VCC and all of the original components placed.

    Then I performed the same measurement on the circuit with the new C13 and C22 values.

    From these results, I believe that the solution to my problem is adding a 1 Ohm 1W reistor in series with VCC as well as replacing C13 and C22 with 4.7uF and 10uF respectively. What I am unsure about at this point, is the effect of a 1 ohm resistor in series with VCC on the voltage regulation of the LM35003. I know to account for WC current draw of the system and how it affects the series resistor, but I do not know if there are other affects on the regulation that prohibit me from using the series resistor.

  • Matthew,

    I'm not clear on where the 1 ohm resistor is placed. In series with D2? Where is VCC going?

    -Sam

  • Sam,

    As shown in the image below, the 1 ohm resistor is placed in series with D2 (as you guessed). VCC is connected to the input voltage of the system. VCC will be 24-49.2v in normal use.

  • Matthew,

    Makes sense. 49.2V battery on a 50V device does not give a lot of margin. Using a higher rated device may be a better way to go vs these necessary precautions we're discussing.

    The 1 ohm will reduce efficiency and will reduce the voltage seen on VIN by 1*IIN. This is an issue if your VIN approaches your VOUT voltage. And you may also see some ripple if you have a pulsed load (IIN changing over 1 ohm will give you a changing VIN) but I don't expect this to do much against proper operation.

    You could also try 0.5 ohms or 0.1 ohms. They may work just as well.

    But my final recommendation is to use a device with more VIN margin.

    -Sam

  • Sam,

    Thank you for your support and all of the advice you have given.

    I will look for a part with a higher input voltage rating for my next revision of the layout. Right now, to mitigate the overvoltage spike on the units I have on hand, I will add the series resistor. For the series resistor value, I will use the lowest value that results in a safe maximum input voltage.

  • Matthew,

    Sounds good!

    -Sam