• State
  • Locked Locked
  • Replies 1 reply
  • Subscribers 63 subscribers
  • Views 173 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.

LM5122: Noisy Gate Signal & Ringing – Overlapping Dead Time in LM5122 Boost (36V, 15A)

Arvind
Prodigy 175 points
Other Parts Discussed in Thread: LM5122, CSD18533KCS, CSD19505KCS

Hi everyone,

I'm working on a 36V 15A Boost converter using the TI LM5122 controller and running into gate drive ringing (M3 & M4) and overlap issues under high load conditions, Causing Heat and Losses, Hoping someone here has seen this before and can offer insight.

Specs:

  • Controller: LM5122ZPWPT

  • VIN: 22–30V, 27V TYP

  • VOUT: 36V / 15A

  • MOSFETs: CSD18533KCS

  • Gate driver directly from LM5122 (no external drivers)

  • Fsw : ~250KHz
  • Webench Design:             5758.WBDesign59.pdf

Problem Summary:

At higher loads (10A+), I’m observing severe gate ringing, dead time overlap and heating of Mosfet, where both MOSFETs are briefly on — leading to MOSFET (M4) failure. This behavior isn't present at lighter loads. Attaching the Schematic 

Original Sch :lm5122 Boost36V15A-1.rar

Modification done: Modifications lm5122 Boost36V15A-1.rar

What I've tried so far:

  • Removed MOSFET (M3) entirely and replaced with a fast Schottky diode (STPS30M60ST) — eliminated dead time issue and the failure of M4 but at a cost of power loss,

    along with VDS Snubber (10R 470PF) Circuit for M4, still the gate ringing persists for M4

  • Added gate resistors (10Ω–60Ω range) — little to no improvement.

  • Tried gate diodes (to speed turn-on) — even more gate ringing

  • Observed that the Miller plateau dips significantly at higher loads, may be self-turn-on?

  • Added Snubber (10R 470PF) for gate along with gate pull down resistor not much improvement.

Observations:

  • Strong oscillations during switching, especially on VGS.

  • Measured ~5–15 MHz ringing.

  • MOSFETs heat up quickly around 75degC at 10A for just a few minutes, Mosfets are mounted on a big heat sink box with air flow 27degC , the TAB is around 45degc

  • The Power loss calculations say 40-50degC, agreed that bad rise and fall times are the issue, how to make it better?

Oscilloscope captures: 

  1. The very beginning setup without any modifications: INITIAL VGS M4 1A.rar
  2. Dead Time : VGS of M3 and M4  : DEAD TIME.rar
  3. Current Stage  :  adding gate resistors, diode and snubber modifications, VGS of M4 at 1A and 10A load: VGS M4 1A.rarVGS M4 10A.rar

Questions:

  1. Any proven techniques for damping gate ringing without slowing switching too much?

  2. How to safely tune dead time in LM5122 to avoid cross-conduction?

  3. Better snubbers values help across gate or drain-source?

  4. Anyone used external gate drivers with LM5122 for stronger control?

  5. Better Gate resistor values? with proper calculations?
  6. Better Mosfets ? how about CSD19505KCS ?

Would appreciate schematics/layout feedback if someone has worked on high-current LM5122-based converters.

Also let me know if any other details required, I'm planning to make another layout. 

Thanks in advance!
Arvind

  • 0
    TI__Mastermind 30280 points

    Hi Arvind,

    Thanks for using the e2e forum and sending this detailed report.

    Based on the waveforms you attached, it is difficult to see how much cross conduction there is.
    You measure the high side gate with reference to GND, which shows the switch note voltage plus to gate drive voltage. To see the HO gate signal clearly, you would need to measure the HO Vgs differentially, or measure the switch note voltage in addition and subtract it from the HO signal.

    1. To reduce gate ringing, it is common to reduce gate resistance, instead of increasing it. Can you give the original gate resistance of R36/R37?
    2. It is not possible to change the dead time of LM5122 externally, but compared to other controllers like LM5123, this one already has comparably large dead-times.
    3. If you want to achieve a fast turn off, but slower turn on, you could try to implement a diode from gate to LO pin in parallel to the gate resistance.
    4. It is possible to use external gate drivers together with LM5122, but before changing the layout to use external drivers, I would suggest to have a look at the current layout first, to see if there are possible improvements with the LM5122 driver alone.
    5. It is common to increase gate resistance if you see ringing on the switch note voltage, as the drivers are too fast. However, if you see the ringing on the gate driver signal itself, higher gate resistance can even have a negative effect. Unfortunately, I am not aware of a calculation method to find the best suitable gate resistance.
    6. Unfortunately, I am not an expert for MOSFET parts. Right now the RMS current of the inductor can you up to 25A, which is a lot of stress for a single inductor. If two MOSFETs in parallel are used, the are able to share the current, which is helpful for thermal management.

    Could you share the layout of your design?
    Could you also share a waveform of the switch note, together with LO and HO signals to see the cross conduction more clearly?

    Thanks and best regards,
    Niklas