Post Go back to editing

LTM4653 pulsed load

Afternoon Everyone,

I purchased a development board of the ltm4653 and used it during testing (DC2327A)

It was modified to output 12V at 15V with a switching frequency of roughly 1 MHz as per the datasheet.

Our operating input voltage range is 14V to 30V.

The output of the LTM 4653 is connected to a pulsed load with additional capacitance at the LTM4553's output to keep the bus stable.

Each time the pulse load activates the LTM4653 responds by supplying it'\s full 4A and the bus falls to 8V and then recovers.

After running the device for a somewhat extended period of operation the device now refuses to turn on at an input voltage lower than 25V.

This was a sudden failure or well seems to be a precursor to failure as the device is still operational but it's as if its UVLO circuitry or something has given in.

 The device did not get warm or display any other issues other than the output voltage dip.

The pulsed load condition occurs 10 times a second.

My question is whether this is expected? I will look into paralleling 2 modules in testing, however I'm concerned of experiencing the same failure condition..

Any help will be appreciated

Parents
  • Hi, so this behavior is definitely not expected. Whenever you have time to respond, could you let me know what exactly were the modifications made to the demo board? So we can check for anything that might be out of place from that point of view. Some other comments:

    • For your Vin/Vout, I'd recommend going with ~500kHz switching frequency. The problem could be related to compensation and loop stability. In which case it's possible that external compensation may have to be tuned. 
    • As for what was happening with the demoboard once the pulsed loads were introduced, it's possible that it's going into drop out with each pulse.
    • Going in and out of drop out can cause excessive rms current. You can check the IMONa pin to see if the current limit is getting activated transiently or it's being sustained. IMONa can also be grounded for maximum current. 
    • For the pulsed load itself, it's also possible that we need more MLCC at the input and at the Vd pin. This could help with preventing excessive depletion at the input cap during the transient. How much Cin do you have on the board right now? 
    • For now, I do not have any explanation on why the board now only turns on when Vin is above 25V. I'll look into this. 

    Thanks. 

Reply
  • Hi, so this behavior is definitely not expected. Whenever you have time to respond, could you let me know what exactly were the modifications made to the demo board? So we can check for anything that might be out of place from that point of view. Some other comments:

    • For your Vin/Vout, I'd recommend going with ~500kHz switching frequency. The problem could be related to compensation and loop stability. In which case it's possible that external compensation may have to be tuned. 
    • As for what was happening with the demoboard once the pulsed loads were introduced, it's possible that it's going into drop out with each pulse.
    • Going in and out of drop out can cause excessive rms current. You can check the IMONa pin to see if the current limit is getting activated transiently or it's being sustained. IMONa can also be grounded for maximum current. 
    • For the pulsed load itself, it's also possible that we need more MLCC at the input and at the Vd pin. This could help with preventing excessive depletion at the input cap during the transient. How much Cin do you have on the board right now? 
    • For now, I do not have any explanation on why the board now only turns on when Vin is above 25V. I'll look into this. 

    Thanks. 

Children
  • Hi  ,

    Glad to know it's not expected :)

    I double checked the modifications after the erratic behavior and they do align with the following:

    - From the user guide of the dev board, we made three alterations as follows:

    R4 was changed to 30k. R1 and R8 was changed to 121k and finally R13 was changed to 1 Megaohm to obtain a switching frequency of 1 MHz  according to the datasheet of the LTM4653:

    I started to investigate prior to its failure and this was my observation:

    The pulsed load would activate. The LTM4653 would immediately start supplying its max current and the output voltage would start to droop. The pulsed load would be removed and the LTm4653 would continue to supply the max current until the output capacitors were recharged. (This was monitored through Imon). No heating was noticed on the device.

    the capacitance at the input to the LTm4653 was not significant (maybe 10uF in addition to the dev boards capacitance). The capacitance at the output of the LTM 4653 was increased up till 5mF.

    As of not I am waiting on additional devboards. Once received I will change the switching frequency to 500kHz and increase the capacitance at the input of the LTM4653.

    I hope I answered all your questions, if not please let me know. 

    Thanks!