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.

UCC28950: Idea for prevent light load oscillation

Guru 19645 points
Other Parts Discussed in Thread: UCC28950, UCC28950EVM-442

About UCC28950 circuit, one customer is occurring oscillation on light load condition (nearly no load). 

My idea for prevent oscillation is increase RTMIN value and adjust compensation network.

If there the other ideas or reason of oscillation, please let me know.

Best regards,

Satoshi

  • Hello Satoshi-san

    Without any more details I can only guess.

    Increasing RTMIN will increase the current at which the system enters burst mode - this may suppress the oscillation.
    Adjusting the compensation network may help -
    Noise may be a problem too - if the system is running in peak current mode control then the leading edge spikes on the current sense signal (CS pin) may be causing the PWM comparator to trip early and the control loop has to compensate by increasing the trigger level (COMP) for the subsequent pulse.
    It is also possible that the system is running out of control authority at light loads -

    Please take a close look at Vout during light load - is the oscillation sinusoidal - if it is then you have a control loop problem.
    If the oscillation is at a single frequency but heavily distorted - triangular or sawtooth for example - then there may be a linearity problem. If the Vout variation is chaotic - without any obvious pattern - then the cause is most likely to be noise.

    If the variation is chaotic, then take a look at the CS signal over 15 to 20 periods - does the amplitude or duty cycle jump rapidly between cycles - if it does then the most likely cause is noise.

    Hope this helps - if you post some waveforms I could probably give you better advice - you can send them to me by email if you wish - I'm at colingillmor@ti.com

    Regards
    Colin
  • Colin-san

    Thank you for many advice,
    I confirming to customer, I will update to you if I got waveform data.

    Sorry for additional question,
    When I found triangular oscillation, which is adjust point on external circuit of UCC28950 typical application?
    (UCC28950EVM-442 is also OK)

    Best regards,
    Satoshi
  • Hello Satoshi-san

    The solution depends on what is causing the triangular waveform.

    If the waveform is at the switching frequency, then the solution is to increase the output capacitance.

    If the waveform is at the burst repetition rate then the solution would be to increase the burst repetition rate - or increase the output capacitance

    If the waveform is present when the system is above the burst mode threshold then you may need to look at the loop compensation components and at the operating points of the feedback network.

    Can you post a screenshot of the oscillation and some details about the load condition.

    Regards

    Colin

  • Colin-san

    I confirmed to customer waveform.
    Sorry, it was not oscillation and correct is current spike on current transformer due to the effect of current through.
    (Waveform can not attach because of customer confidential)

    For example of Typical application(circuit) on the datasheet;
    When the timing of QA gate turn off threshold and QB current turn on by light load condition, current transformer was be spike.
    ※Light load spec: 140Vout / 1A, spike is around 10A.

    Is there recommended process to adjust dead time for reduce current spike?
    And, is addition to RC filter in front of current transformer good idea?

    If there better to make new thread for this question, please reply it.

    Best regards,
    Satoshi
  • Hello Satoshi-san

    Let's take this topic off E2E. Can you send me the customer waveform - I'm on colingillmor@ti.com

    In general the leading edge spike can be caused by loss of ZVS which increases the dv/dt rates and gives larger currents in the stray capacitance as it discharges. It also depends on where in the circuit they have places their current transformer - normally TI places it in the HV input line to the switching devices but it is also possible to put it in series with the power transformer.

    Normally a RC filter with 1k / 330pF is sufficient. Increasing the RC filter will reduce the size of the leading edge spike at the CS pin but makes the CS pin less responsive to over current conditions. Please ask the customer to make sure that their RC filter components are located as close as possible to the CS pin of the controller. This is VERY important.

    I look forward to receiving the waveforms by email.

    Regards
    Colin
  • Colin-san

    Thank you for advice,

    I made new thread the same timing you replied.

    I will try to get waveform to customer.

    When I get customer waveform, I will send to your email.

    Best regards,

    Satoshi

  • Hello Satoshi-san
    I agree, let's continue this conversation via email - I'll close this thread now.

    Regards
    Colin