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.

LM5104: To work as a switch, not with PWM

Part Number: LM5104

Hi!

I'm working on a prototype with LM5104, basically the datasheet recomended circuit for a halfbridge system, with some adds. It's working perfectly with a PWM input signal, but I need to modify this to work just as a switch. I have an DC inductive load that need to be turned on and off with a 3.3-5v signal, it will remain turned on for hours. I noticed that the bootstrap capacitor will discharge just a few ms after the input signal is high, making the halfbridge output voltage to drop a few volts.

The load is 30VDC, the mosfets can drive it nicely, it must be turned on from the high side and I need to be sure that its grounded when off.

  • Hi Rodrigo,

    Welcome to e2e! Thanks for asking about LM5104. Im an apps engineer with this device and will help you out!

    The charge required for initial turn-on of the highside must be supplied by bootstrap cap. After that Cgs acts like an open circuit and if there is an Rgs then current will discharge out of Cboot per ohms law to the point where HB UVLO is reached. There is also bootstrap leakage current that will discharge HB.

    One option is a floating bias supply to keep the bootstrap from discharging during this time. In order to have 100% duty cycle on the highside we will need to generate a small separate floating bias rail for HB-HS capacitor, Cboot. Using an unregulated transformer driver like SN6505 you can drive its primary from a separate supply and rectify the output. You may have to add a zener diode on the output to keep the HB output with in limit.

    You also have the possibility of using a charge pump to keep the bootstrap charged. Section 3 from www.ti.com/.../slva444.pdf highlights this method.

    What topology do you have? if a buck converter, you might be able to get away with connecting the bus voltage VIN to HB through a resistor sized to allow an average current to flow thats larger than the worst case Cboot discharge current.
    Also, I was thinking, for a buck, if you reference HS pin to GND pin you can use a PMOS for the highside and turn it always on by inverting the input signal. If HO is OFF then the FET is ON because Vs-Vg>Vth. If HO is ON then the FET is OFF because Vg=Vs=VIN.

    Does this help with your question? please let me know if you have any more questions.
    Thanks,
  • Hi Jeff, thanks for this very explanatory answer!

    I tried to use an isolated 15v supply in parallel with boostrap capacitor, but the LM got unstable and burned, I don't know exactly why. Attached below the schematics I'm using,

    VIN - 30v

    VCC - 15v

    Input - 3.3-5v signal

    My load is connected to OUT and ground, when the input signal goes high, I expect that my OUT terminal keep with 30V from VIN.

  • Hi Rodrigo,

    Thanks for the update, sorry to hear about this burn issue. I have not heard of this issue before, it should work since both VCC=iso supply. The iso supply will be 15V and since theres a diode drop from VCC then the iso supply will have about 0.7V to charge with.

    Do you have D1 in parallel with the internal bootstrap diode to reduce internal conduction/recovery losses?
    Since your circuit works fine with PWM and burns with the iso supply on HB-HS, can you show me the iso supply circuit that you are using?
    Can I see the HB,HS,HO WRT to GND waveform during the failure?
    Can you tell if only the highside or lowside was damaged by measuring HO/LO to GND? is it a VDD-GND shorted too?
    also, Can you tell me the purpose of C3 and R7? is this to shape the gate signal? or for noise filtering?

    Thanks,
  • Jeff,

    D1 is backup diode that a coworker suggested to use. I noticed that in my application it isn't a must have part.

    The power from VIN comes from a huge battery bank, also there's a very simple voltage regulator connected to the battery to power the VCC terminal. I tried to supply HB-HS with a 15v bench power supply, completely isolated from the system.

    We already put away the damaged one, but I can replicate the failure, with this waveforms log. Do you have any recomendation for this test?

    C3 and R7 are just noise filters.
  • Hi Rodrigo,

    Thanks for the quick update.
    Just a note for the input filter, since the input voltage signal = the output voltage signal this parallel RC circuit transfer function does not act like a filter unless the input is driven as a current source.

    To bench test these waveforms, make sure the voltage scale is the same for all signals and probe has short GND connection measured as close to the IC pins as possible. Show a few cycles of the issue, if there are any fast/large spikes zoom in to see whats going on. Hopefully this will give us a clue as to whats going on and why the IC is burning.

    Thanks,
  • Attached some pictures of what happend. After a few cycles with that oscilation, it stopped and the HS went to something arround 5 volts. After that, changed the osciloscope probes to HO and LO

    After this oscilation stopped I got this:

  • Hi Rodrigo,

    I am confused with the names of the waveforms, let me explain. HB voltage (pink waveform) should be HS voltage plus 15V or 45V. It seems to only be 5V in your waveforms. HS voltage should be 30V right? HO (blue waveform) appears to be about 30V. Also for the LO/HO waveform, it looks to be less than 1V, is HO following HI? the HO in the top pic is much different from the HO in the bottom pic according to the time/div.

    I would like to see what these signals look like when for the working system when you are PWMing at a high enough kHz to keep the bootstrap charged. (similar setup to the second pic just with the same V/div)

    are you doing anything to the circuit when the oscillations begin? after how long does it take before you see oscillations?
    is there a load?
    is there a current sense or temp sense in your circuit/FETs?

    Thanks,
  • I guess that the waveforms were weird because the chip already died.

    I'm attaching a picture of my last LM5104, 5V 1khz PWM input. Now I'm feeding all the circuit with only 15 volts.

  • Hi Rodrigo,

    Everything looks good for these waveforms. Can you retest with HB-HS floating supply?

    Thanks,
  • Hi Rodrigo,

    do you have an update with this issue?
    I will think it is resolved otherwise.

    Thank,