I need to figure out if an LLC half-bridge resonant SMPS is designed (more or less) correctly. I am not an SMPS designer, and neither is the manufacturer, so I'm just reading a bunch of app notes and learning what I can. It's producing ±35 V 400 W from a constant 400 VDC PFC output.

As I understand from my reading:

  1. The resonant frequency is a property of the LLC transformer network, increasing with load from fp (unloaded resonant frequency) towards f0 (shorted resonant frequency).
  2. The switching frequency is controlled by the driver IC, and decreases with load to regulate the output by riding up the resonant gain peak, getting closer to resonance and increasing the output as the load increases.
  3. The switching frequency must never drop below the resonant frequency at any given load, or it will enter capacitive ZCS mode, which is Bad.

Is this all correct?

They've set the switching frequency to always be below f0. They claim this is ok, because Fairchild AN-4151 says you can do above or below resonant operation:

"Below resonance operation is preferred for high output voltage applications, such as Plasma Display Panel (PDP) TV ... above resonant operation ... can show better efficiency for low output voltage applications, such as Liquid Crystal Display (LCD) TV or laptop adapter"

Would 70 V out be considered low or high voltage in this case? I'm guessing high voltage means hundreds of volts, so this is low voltage. So does that mean it would ideally be operating always above f0? Maybe as long as point 3 is met (ZVS mode), it's not too important? This article says:

Most of these converters are operated above resonance so that at light-load conditions the switching frequency is higher than the resonant design points (region 1) [fs > f0]. As load increases, the switching frequency droops. At maximum load, the converter operates slightly below the LlkCr resonant point [fs < f0] in region 2 [but should never enter region 3].


The Q and M curves show that the resonant frequency is a property of the transformer, the load and the series capacitor. \$f_o\$ isn't necessarily a 'shorted' resonant frequency but a mostly-load-independent second resonant point predominantly controlled by the leakage inductance and series capacitor.

The switching frequency is indeed controlled by the some form of control IC (usually a PWM). The frequency of operation will vary as a function of input line and load, and should stay above the worst-case (minimum load) resonant peak to ensure ZVS over the whole range. You are correct in that operation below the worst-case peak gain forces the converter into an uncontrolled region and should be avoided.

LCD TV and laptop adapters are usually < 24V. If the designer was able to use a schottky diode for the output rectifier, you don't have to rely on forced commutation from the primary to reduce losses. The best way to tell if you're suffering from excessive reverse recovery loss is to put a current probe in the secondary and measure the reverse recovery current. Operating this 70V converter above resonance means they're not soft-commutating the secondary (which isn't necessarily bad) to reduce conduction losses (less circulating current in the primary).

  • \$\begingroup\$ Yes, I edited to fix that. "since the resonant tank includes three reactive elements ... there are two resonant frequencies associated to this circuit. One (f0 = fR1) is related to the condition of the secondary winding(s) conducting, where the inductance Lp disappears because dynamically shorted out by the low-pass filter and the load ... the other resonant frequency (fp = fR2) is relevant to the condition of the secondary winding(s) open, where the tank circuit turns from LLC to LC because Ls and Lp can be unified in a single inductor" \$\endgroup\$ – endolith Apr 26 '11 at 21:18
  • 1
    \$\begingroup\$ So they're opting for soft commutation at the expense of having the switching frequency moving around somewhat. That's a perfectly valid design decision. \$\endgroup\$ – Adam Lawrence Apr 27 '11 at 14:59
  • 1
    \$\begingroup\$ If you're below the load-dependent resonant frequency, the converter will be unable to control the output any longer, and the primary-side losses will be huge. If you're between the two resonant frequencies, you should see discontinuous secondary current (dead time between peaks). If you're above the load-independent frequency, the secondary current will be continuous (see figure 11 in AN-4151). \$\endgroup\$ – Adam Lawrence Apr 28 '11 at 17:11
  • 1
    \$\begingroup\$ Otherwise, you have to remove the transformer and measure the magnetizing inductance and leakage inductances, and follow the derivations in the datasheets to calculate the resonant frequencies. \$\endgroup\$ – Adam Lawrence Apr 29 '11 at 13:14
  • 1
    \$\begingroup\$ The gain characteristic is reversed below the resonant point, so the control law no longer works - lower frequency further reduces the gain. You also lose soft commutation of the primary FETs (no more ZVS) and get body-diode reverse recovery, so yes, acoustic noise and explosions may happen if the parts cannot handle the stress. \$\endgroup\$ – Adam Lawrence May 3 '11 at 15:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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