Issue with Flyback PWM signals

Discussion in 'General Electronics Chat' started by fchastel, Sep 1, 2015.

  1. fchastel

    Thread Starter New Member

    Aug 25, 2011
    4
    0
    Hi all,

    I've designed a discontinuous current mode Flyback converter. I'm now testing it with fixed loads. At max. load, it works fine: the output voltages are at the expected value and the PWM controller output signal consists of pulses with a constant width. At nominal load, I still get the correct output voltages but the PWM output signal is not as I would expect: it consists of one wide pulse followed by several much narrower pulses. When I look closely at the regulated output voltage, I can see it going up when the wide pulse occurs and then down until the next wide pulse occurs. I've attached a sketch of the signals. I hope it helps.

    Is it a known issue? Where is the issue coming from?

    Thanks in advance for your help! I'm currently completly stuck with this issue.

    Cheers
    Fred
     
  2. pwdixon

    Member

    Oct 11, 2012
    488
    56
    Does the PWM pulse have enough resolution to differentiate between the wide and narrow pulses? ie. is it possible to have an intermediate pulse width?
    Perhaps the error detection window is restricted and it can only generate these two widths?
    Can you post the schematic?
     
  3. crutschow

    Expert

    Mar 14, 2008
    12,973
    3,219
    My crystal ball is a little cloudy but it sounds like you may have a feedback loop instability problem.
    Can't help further without the schematic.
     
  4. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,281
    1,227
    It's hard to tell without a schematic and some parts specs. My guess since it happens at higher current is that the inductor current rating is to low and it is starting to saturate. When that happens to increase in voltage can happen faster than the circuit response.
     
  5. fchastel

    Thread Starter New Member

    Aug 25, 2011
    4
    0
    Hi all,

    thanks for your first feedback. Here's the schematic. Please let me know if you have questions on the schematic.

    Cheers
     
  6. crutschow

    Expert

    Mar 14, 2008
    12,973
    3,219
    You have no diode to generate the -3.3v.
     
  7. fchastel

    Thread Starter New Member

    Aug 25, 2011
    4
    0
    Sorry, I had to quickly draw the schematic and missed the diode... it's also a 1N5822 but in the other direction.
     
  8. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,281
    1,227
    This is only a "SWAG", but try adding a few nf across the 15k going to the op amp to give a little more high frequency gain.
     
  9. umphrey

    Member

    Dec 1, 2012
    39
    1
    Is it a resistive load? If you calculate what the duty cycle should be which does it match (the skinny pulses or the large pulse or somewhere in between)? Can you pin down a range of current or power where it is erratic and the range that it is not?

    I don't know the answer, but DCM power supplies tend to have a lot of ring in the voltages and if the chip reads that it could cause this. I would try to pin down the states of everything when it happens, maybe when you hit 0 inductor current it rings the output voltage and the chip reads that.
     
  10. fchastel

    Thread Starter New Member

    Aug 25, 2011
    4
    0
    At max. load, the theoretical Ton is 2.63us (duty cycle = 26%) and the meas. Ton is 2.68us. It therefore matches quite well.
    At nom. load, the theoretical Ton is 1.03us (duty cycle = 10%) and the meas. Ton is about 2.8us for the large pulse and 0.5us for the small pulses.

    I've attached two snapshots of the drain-source voltage (VDS) of the switching MOSFET, showing the unstability.
     
  11. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,281
    1,227
Loading...