As I know, the major difference between normal flyback converter & flyback quasi resonant converter is for flyback quasi resonant converter, the switching frequency is variable. This means that the switching frequency is based on the output load. When the load decrease, the switching frequency will increase, and when the load is increase, the switching frequency will decrease.

My question as below: 1. Since the switching frequency is always changing, how do I select the switching frequency as my initial value to start the transformer design? 2. Since the switching frequency is always changing, how do I know that my transformer is capable for that switching frequency range. Let say I choose a transformer A with value of 300uH @ 55kHz, and my switching frequency for quasi resonant converter is range from 35kHz to 130kHz. Will this transformer A is capable for that frequency range? As I know, as the frequency increase, the inductance will be smaller & as the frequency decrease, the inductance will be bigger. Is it means that at lower frequency, this transformer A (300uH,55kHz) will change its inductance value at lower frequency(suddenly will increase inductance) & high frequency(suddenly will decrease inductance)? Or this transformer A will have fix inductance value of 300uH at all frequency?



1 Answer 1


As you correctly pointed out, a quasi-square-wave resonant converter colloquially called a quasi-resonant converter (QR) is a self-relaxing converter: there is no clock in the controller except for clamping purposes. The switching frequency varies with operating conditions. It is low at low line and high power, it is high at high line and low power.

QR converters are popular for various reasons:

  1. they always operate in discontinuous conduction mode (DCM) regardless of the operating conditions.
  2. as such, they remain a 1st-order system and are easy to stabilize in current-mode control.
  3. if you insert a deadtime before turning on the power switch, you can wait that \$v_{DS}(t)\$ goes through a minimum or a valley - hence the term valley-switching operation - and reduce switching losses potentially to zero (ZVS) if the valley touches ground.
  4. because of the DCM operation, driving a synchronous rectifier is safer than in continuous conduction mode (CCM) as shoot-through currents are minimized.
  5. excellent behavior in short circuit and the output diode loves it. It was one reasons QR converters were popular in CRT TVs applications with a lazy high-voltage diode and a high-voltage output rail.

They have some drawbacks however:

  1. frequency varies along the operating range.
  2. over-power is a problem in ac-dc application: the available power at low line is more than doubled at high line and overload management is tricky.
  3. high ripple current forcing the selection of costly output capacitors subject to a high rms current.
  4. can be noisy in short circuit situations and some customers may not like it (I had the case once).

Here is shot of a typical drain-source voltage where the valley is clearly seen:

enter image description here

In peak current mode control, the loop adjusts the setpoint and indirectly fixes the on-time considering the primary inductance and the input voltage. The off-time slope is constant and depends on the reflected output voltage. Then you add the deadtime DT to ensure valley switching. Below is shot of the operating frequency and peak current as described in the book I wrote on power supplies:

enter image description here

Now that we know a little more about this converter, let's see your questions:

  • How do I select the operating frequency? There is no straight answer as you can imagine but in ac-dc application, people want to stay away from the audible range. They select the primary inductance and transformer turns ratio so that at full power and low line, the converter operates beyond the 30-kHz range. That being said, the power delivered by a discontinuous conduction mode (DCM) flyback depends on the frequency and peak current. Therefore, staying above 30 kHz is one thing but you may want to operate closer to 50-60 kHz to reduce the peak current and limit the transformer size. Then, you know that switching frequency is going to increase as the line goes up but also when the load gets lighter. What maximum frequency you want to limit yourself to? Because we know the losses high-frequency operation will bring but also electro-magnetic interferences (EMI). So you have to go through iterations with these goals in mind and see what transformer size each scenario (what min and max frequencies) brings.
  • Regarding the transformer: no, the primary inductance \$L_p\$ does not change along the operation unless you approach saturation of course. We know transformer losses - copper and iron - will increase as \$F_{sw}\$ and efficiency will suffer. It is clear that if your nominal operating point is 400 kHz, you'll have to select wires and core accordingly. If your nominal point is 70 kHz and you go to 150 kHz in light-load condition before the frequency is clamped, then obviously we are talking about different conditions for the transformer. So once you have run a few iterations on the inductance selection, the transformer turns ratio and so on, you will have a better picture of what the nominal operating frequency is and which wire and core material to favor.

As a final note, carefully look at the controller offer as many different strategies exist. I personally like a free-running operation until a maximum frequency is reached (frequency clamp) so that you limit losses and then gradually reduced by a voltage-controlled oscillator (VCO) as the load current is getting lighter. Plenty of chips to choose from!

  • \$\begingroup\$ thanks for your clear explaination. Now I'm understand that managing the quasi resonant converter is quite tricky compared to normal flyback DCM converter. Your comment has answered all my confusion between normal & QR type. I checked some of QR controller (example NCP1380 from ON-Semiconductor), basically they have valley switching, n=1,2,3,4, VCO. Based on the datasheet, it stated that each valley have a maximum switching frequency, and by using some of the equation, we can get the maximum switching frequency for each valley. \$\endgroup\$
    – Lutz Fi
    Commented Feb 8, 2020 at 5:13
  • \$\begingroup\$ The valley will jump from one to another based on the output power. It will stayed inside the valley (lockout) & only jump when the output power is changed. Based on the curve on the datasheet, valley 1 will have the highest switching frequency at high line and the lowest switching frequency at low line. Based on your explaination, supposedly during n=1,2.3.4. VCO, the primary inductance will stay constant and only effeciency will change? \$\endgroup\$
    – Lutz Fi
    Commented Feb 8, 2020 at 5:21
  • \$\begingroup\$ Correct, in valley lockout mode, the frequency changes by discrete steps depending on the valley number. Best efficiency is obtained when operating in valley 1 but since you need to reduce frequency also for efficiency reasons, there is no other choice than jumping down the valleys. Please, if this is answer you were expecting, thank you for acknowledging it. \$\endgroup\$ Commented Feb 8, 2020 at 7:44
  • \$\begingroup\$ yes thank you so much for all your clear explaination. This answer is exactly what I need. \$\endgroup\$
    – Lutz Fi
    Commented Feb 8, 2020 at 8:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.