TPS54360: Output shows short pulses when Input Voltage switches off.

Part Number: TPS54360

I have a design with 3 TPS54360 on one Board: 12V, 5V and 3,3V

The Input Voltage is about 38VDC. When I switch off the Input Voltage it ramps down for about 300ms to a Voltage of around 4V.

Then the 3.3V Regulator shuts off the Output.

Approximately 100ms later the Regulator starts again for around 2ms (up to 3V) and another 500ms later it generates another spike!

The behaviour decreases when I lower the Input Voltage. At around 15V the problem does not persist any more.

You can imagine that a Processor module tends to ruin certain memory cells when confronted with such short pulses.

On the oscilloscope it seems, that the Input Voltage drops below 4.3V and recovers again after the TPS54360 switches off. Then the IC starts over again.

Because of a rectifying input I have quite high capacitance on the input (3x470µF Caps). Do you think they recover

How can I improve this "hysteresis"?

BR Markus

  • The EN Pin is not connected...

  • Hello Markus,

    Could you please provide a circuit schematic for review? I would like to take a look at your schematic and make sure all is okay. Also, it would be very beneficial/helpful if you can please provide waveforms of the phenomena.

    Thanks,
    Alejandro

  • Hi, you can see the VIN Voltage in cyan and the 3,3V in yellow. I have replaced the TPS54360 chip yesterday, and the issue is much better now. But still sometimes as you can see in the waveform. I have to say that this board has already been running for months (not switching OFF/ON). And I have 4 similar boards (where not running) where this problem is not that bad. Maybe it is some aging process?
    We got caught on this because the Processor module was "bricked".
    BR Markus

  • Now I have the worst case reproduced. It is with a little daugtherboard using an AVR Chip. You can see that the power drawn from the caps is sufficient to start over 3 times! The VIN is regenerating vividly over the hyteresis...

  • Hi Markus,

    Thank you for sharing these waveforms. They are very helpful.

    I find this behavior to be very strange. Let's try the following tests:

    1. I would recommend to not populate input cap C7 & C8. Please share results. A speculation is that the device is triggering ON because the input caps still have plenty of storage to which the device thinks it can power back on. Kind of like your speculation on the first post in regards to the hysteresis.
    2. Can you please also include the SW node on the waveforms. I would like to see the behavior.
    3. Also, we can adjust UVLO by using an external voltage divider on the EN Terminal pin. Equation 4 and Equation 5 calculate the resistance value necessary.

    Thanks,
    Alejandro

  • Hi, even with only one Electrolytic capacitor, the problem still resides. It is just the time of regeneration of the VIN Voltage which is then reduced.

    In the pics you can see the SW Pin on channel 3. The second will be zoomed in a little bit.

  • When I totally remove all the Electrolytic Capacitance, there is no Spike any more. But I need to work with AC voltage on the input which makes a rectification with smoothing caps mandatory. For my test I use 45V DC power from a benchtop powersupply.

  • Markus,

    Understood, now let's try option 3 - adjusting the UVLO thresholds with a voltage divider. The device is disabled when the VIN terminal voltage falls below ~4V or when the EN terminal voltage is below 1.2 V. The EN terminal has an internal pull-up current source that enables operation of TPS when the EN terminal is left floating (as it is in your circuitry). We might have to increase the threshold to not cause a faulty trigger.

    You mentioned that the device has input voltage of 38V, correct? Let's set the new UVLO thresholds. For this example test, the device should turn on and start switching once VIN increases above 9V and after regulation, it should continue to do so until VIN falls below 7.25V. Therefore, place
    515kΩ between VIN and EN and 73.5kΩ between EN and GND.

    If you would like to change the threshold values, please use Equation 4 & 5 in datasheet.

    Please share results.

    Thanks,
    Alejandro

  • Hi, yes that is working ! If I set VStart to 9.18V and Vstop to 7.58V I calculate R1 470k and R2 at 66k. That would give me nice resistor values that I already have on the assembly machine in this project. Do you agree on these values?

  • That's great news to hear Markus. I'm glad we were able to resolve this. And yes, the following calculated values are correct.

    Kind regards,
    Alejandro

  • Hi, I have one more question.. We all know that the chip market is crazy right now. But do you think that the lead time of 35 to 50 weeks of many chips is realistic? Or is there a tiny little chance that we can see some chips coming earlier?

    It seems to me, that many of the bigger companies have been carefully hoarding some chips because distribution spread fear from January on...
    Or is the production really on hold at TI?

    Other than that, could you recommend any chip to replace the TPS54360 which is ans will be better available the next months? I know you would need to have more Info... don't worry I will check on my own. Just a little hint which one I could check would be really great!

    Thank you very much,

    Markus

  • Hi Markus,

    Due to the shortage it is difficult to confirm. I would recommend to go the Step Down Buck overview page (link) and type in your specific requirements for a list of potential ICs.

    Kind regards,
    Alejandro