• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 1 answer
  • Subscribers 63 subscribers
  • Views 181 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

LV5144: LV5144 design assistance

Anastasios
Prodigy 30 points
Part Number: LV5144

Tool/software:

Hello,

I have designed a 60V nominal (max Vin may be a bit higher, hence why I chose LV5144 and 100V mosfets) to 18V @ 8A buck converter, based on the schematics for the LM5146EVM and the webench suggestions.
The switching frequency was set to 200kHz in order to reduce switching losses and DCM mode was selected to improve efficiency.

Idle current on the prototype PCB was very high (~76mA @ 60V in), the IC and both mosfets were getting very hot, but it worked up to 8A output.
There is some ringing on the SW node (~15V peak), which I assume may cause some crossconduction?

I have tried various modifications in order to reduce the idle current (and subsequently the dissipated heat), but I can't get anything to work properly, so I would appreciate your assistance.

Some things I've tried:

  1. Modifying the compensation network to get a flatter response after the crossover frequency. No significant improvement.
  2. Replacing the inductor with 33uH and the compensation components according to the quickstart calculator recommendations. Seemed less stable and produced audible buzzing sound above 1A current.
  3. Supplying an external 8V supply to Vcc and increasing the Bootstrap resistor (up to 10R). Reduced the idle current to ~40mA, but seems like some protection kicks in above 6A, there is audible intermittent "clicking" and the pulses at the SW node are very short for this duration (I assume it's an overcurrent protection, tried increasing Rilim, but didn't seem to work).
  4. Supplying an external 8V supply to Vcc and using a 1nF/3R3 snubber on the SW node seemed to provide the best results. The idle current dropped to 20mA, essentially no ringing and the converter worked up to full output current. But of course this setup had fairly poor efficiency (2-8% less than the previous option).


Am I missing something critical? 

I would appreciate any feedback.

  • 0
    TI__Guru 74422 points

    Hello,

    Please send a completed quickstart calculator for this design https://www.ti.com/tool/LM5146DESIGN-CALC, so we can check stability. Make sure to enter the derated value for ceramic Cout with voltage.

    The schematic looks okay. High light-load current may be caused by excessive inductor core loss. What's the inductor P/N?

    Regards,

    Tim

  • 0 in reply to Timothy Hegarty
    Prodigy 30 points

    Hello Tim, 

    Thank you for the quick response.

    The inductor is a Bourns PQ2617BHA-220K. As far as I know, these drum core inductors do indeed have quite high core losses at idle, but I estimated it would be less than 0.5W (sadly the datasheet is not very detailed).
    However, this supply will be working at 4-8A most of the time, so based on calculators it should be more efficient due to small DCR. Also, if the high idle current is due to inductor core losses why does it drop to 20mA when using the aggressive snubber option?

    The forum won't allow me to upload excel or zip files, so I uploaded the quickstart, as well as the full schematic and pcb gerbers here (I suppose they might be helpful as well).
    If you prefer I can send you the Altium files via email or something.

    Please note that for most tests the charger section was severed and the load was connected directly to +18V.
    I was initially worried that the additional capacitors at the output might adversely affect stability of the LV5144, but there was no significant difference from severing the +18V track to the charger section

  • 0 in reply to Anastasios
    TI__Guru 74422 points

    Hello, 

    You can drag the file into the window here and it will attach. I can't access sharepoint.

    Check your switching waveforms, especially the low-side gate voltage. Adding a snubber slows down the SW voltage, so this is helping something...

    --

    Tim

  • 0 in reply to Timothy Hegarty
    Prodigy 30 points

    Hello,

    I keep getting "file type is not allowed" for the .xlsm extension.
    I have renamed it to .xls for the upload.

    LM5146-LV5144_quickstart_calculator_revB1_09-05-2024.xls

    Also attached are the SW & low side gate waveforms (with the original PCB configuration as per the schematic/quickstart, without snubber and mods. IC powered from internal regulator).
    The corresponding idle current @ 60V is 83mA (if I switch SYNCIN to FPWM mode the idle current is even higher, ~110mA)

    1. +1 in reply to Anastasios
      TI__Guru 74422 points

      These waveforms look reasonable. Note that the BSC034N10 FET is quite capacitive, leading to high switching loss. Use a device with lower Qg, Qoss, Qrr, etc., particularly as the load current is only 8A.

      , 

      You could install the BSC146N10 on the low side as a test to see if that helps.

    2. 0 in reply to Timothy Hegarty
      Prodigy 30 points

      It was the low side mosfet indeed, I didn't know it can have so severe adverse effects!
      I based the fet choice on the quickstart calculator, which apparently is not accurate under some conditions. 

      Using BSC146N10 for both fets, the circuit works much better. Idle current at 60V is now 5-10mA, minimal overshoot/ringing and pretty good efficiency up to 10A (around 94-96% with ext Vcc, more or less lines up with the quickstart predicted efficiency as well in this case).
      I might try one or two fet options out of curiosity mostly, but in any case the issue is solved.

      Thank you very much Tim!

    3. 0 in reply to Anastasios
      TI__Guru 74422 points

      Great, thank you.