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.

UCC28700: UCC28780

Part Number: UCC28700
Other Parts Discussed in Thread: UCC28780, LMG3410R070

Ulrich

What I observe on the EVM ( I got it delivered today) is that they used 3x .33 uF with 2.2M bleeder.

I have not run the calc yet, but why 3x .33uF?

calculations I did on my needs force me to accept "dc derating to 40%" but why so when there is no DC across it?

#2 issue is that it resonates - so one has to assume current will be bipolar during its on time. Then a lot of dielectrics will not be acceptable here.

# if one uses logical extrema here, I would select an NPO that can handle the current during resonance with an estimation of the voltage across it( from Iclmp*Zr>> sqrt(L/C)?)

If leakage is , say, 3 uH, with  .33 uF, Z is sqrt(3/.3) hardly more than 3V

Why 250V rating?

Is it because secondary side resonance is a different beast due to Nps?

This is crucial for super compact density- knowing what voltage Cclmp will ever be subjected to.

Does it ever have any dc voltage?

I would like to hear your side of the reasons.

thnx

r

  • Robin

    The voltage seen across the clamp capacitor is limited to the reflected output voltage thru the transformer. A margin to account for any voltage spikes and the capacitance derating for voltage applied across the ceramic capacitor is added to give the 250V rating that was used.

    I recommend watching this video on active clamp flyback operation to see the voltage and current stresses that are seen on the clamp capacitor and rest of the power stage.

    Part 1: training.ti.com/active-clamp-flyback-part-1
    Part 2: training.ti.com/active-clamp-flyback-part-2

    If you are curious as to how parameters effect voltage and current stress, I recommend you use our SIMPLIS models to change parameters to see how it impacts operation.
    GaN based SIMPLIS Model: www.ti.com/.../slum626
    Si based SIMPLIS Model: www.ti.com/.../slum643

    Best Regards,
    Eric
  • Eric

    thnx much.

    got it.

    DO I not need  Simplis installed &/or a license purchased to explore these models?

    -robin

  • Robin

    You need SIMPLIS installed on your computer to run our SIMPLIS model.

    Best Regards,
    Eric
  • Eric

    Yeah thought so.

    I doubt if their "eval" version will run this model, do you know?

    Also: thnx for pointing out the videos. I  had, of course, watched them but did not pay enough attention to Cclmap voltage in the first run.

    What is interesting to realize- it came to me a bit late-- that the body diode of the QHi clamps the  QLow MOSFET to a value & charges the clamp C. During this short time, it is traditional role played by the diode.

    After this delay, in ACF case, the switch is turned ON to begin the ZVS process. Whereas the old scheme simply would keep the  QLow voltage hi & ringing until nex gate ON time. 

    However, GaN has nobody diode.

    So UCC28780 must have a QHi turn ON scheme to replace the self-commutating body diode, right?

    Furthermore, for GaN case, does it not implicitly assume a 3rd quadrant V-I  characteristics almost similar to 1st quadrant? ( pretty well-documented case for GS66508). Does TI GaN have similar behavior? I would certainly assume it to be the case.

    -robin

  • Robin

    I am not sure if our model will run in the trial version of SIMPLIS.

    While GaN does not have a body diode, it does mimic the behavior of a body diode by conducting current when a voltage is applied from source to drain. Since the gate and source of the GaN FET are connected thru the gate driver, when you apply a voltage to the source you apply a voltage to the gate. When the gate to drain voltage exceeds the threshold voltage the GaN FET turns on, allowing current to conduct from the source to drain. You need to be careful that current doesn't conduct for significant period of time though, since the equivalent body diode drop can be several volts. This is why it is critical for GaN based SMPS, like UCC28780, to have minimal dead time. TI GaN has this third quadrant conduction behavior.

    Best Regards,
    Eric
  • Eric

    thnx for the clarification.

    Does it mean that the floating LMG3410 source has to be connected to the GND referenced LMG3410 drain- this node being connected to the left side of the resonant capacitor?

    Because that is the only way on start up the source gets a voltage & the device as you describe can conduct as a diode until the gate drive comes along.

    But in such a connection, the GaN always is conducting in 3rd quadrant where Vds is always higher compared to 1st quadrant Vds. Would it be correct thinking?

    In most cases though, if this was a MOSFET, the source will be towards the load- with DC developed thru rectification.

    The MOSFET always conducts in 1st quadrant excepting on start up.

    r

  • Robin

    When LMG3410 is connected the way that the schematic shows in the UCC28780 datasheet for the FET it will conduct current in the third quadrant when the voltage from source to drain is high enough. For LMG3410R070 the electrical characteristics in the datasheet has "VSD Third-quadrant mode source-drain voltage" parameter to define what the voltage drop will be.

    Best Regards,
    Eric