2
\$\begingroup\$

How does CCM can improve the peak power of normal DCM flyback power supply?

I noticed that some of flyback converter can operate in both DCM & CCM. During this peak power excursion, the converter will change from DCM to CCM using special PWM controller. Example High Efficiency, <30mW Standby QR Adapter with Peak Power Excursion Capability

Question:

  1. Based on that, does this means that this peak power can be achieve without changing the size of the transformer?
  2. From my understanding, if we go above the primary peak power, the transformer will saturate. How this CCM mode overcome that?

Any calculation example should help

Thanks.

\$\endgroup\$
  • \$\begingroup\$ I came into an article explaining the method they use to achieve this function without oversizing the parts, have a look into it, it might reply to your question : nanopdf.com/download/… \$\endgroup\$ – Delphesk Feb 14 at 11:30
  • \$\begingroup\$ I've upvoted your question because I think it's very relevant to ask. Personally speaking I have no great belief in CCM offering any benefit over DCM in flyback converters at all. \$\endgroup\$ – Andy aka Feb 14 at 11:38
  • \$\begingroup\$ Size could remain the same, but you need to change your number of turns for a given switching frequency (and input voltage) to go from DCM to CCM for a given power level. It's also very common to have DCM at light load and CCM at full load in flyback converters. Simulate it and do some back-of-the-envelope calculations as to size and number of tuns and get a feeling for it. It will help you in the long run. \$\endgroup\$ – winny Feb 14 at 11:46
  • \$\begingroup\$ The thing is that a QR converter is a variable-frequency DCM-operated converter meaning that if you want to pass more power, then you need a transformer requiring a higher peak current and operating at a lower frequency --> core volume is up. If you decide to limit the frequency decrease - let's say 20-30 kHz for noise reasons - then you no longer fully demagnetize the core and you enter CCM before hitting the max peak current limit. Which, in my opinion, defeats the purpose of QR which is to approach ZVS and avoid \$t_{rr}\$-related losses in the rectifying section (passive or active). \$\endgroup\$ – Verbal Kint Feb 14 at 12:30
4
\$\begingroup\$

Quasi-square-wave quasi-resonant converters also called QR converters are popular structures in ac-dc chargers for notebooks and portable electronics. They are popular because:

  1. they operate permanently in discontinuous conduction mode (DCM).
  2. they remain a first-order system easy to stabilize despite a right-half-plane zero located in high frequency (versus a low-frequency one in CCM).
  3. you can approach zero-voltage switching or valley-switching operation if you reflect enough voltage to the primary side at turn-off. Excellent for switching losses and also for the radiated EMI.
  4. you can use "lazy" diodes in the secondary side as the diode spontaneously or naturally turns off when the core is demagnetized. This changes from CCM where the hard-blocking event generates losses in the primary and secondary sides due to the reverse recovery charge of the diode. QR avoids this problem and also avoids shoot-through current when a synchronous rectifier is used. QR was popular for this reason some time ago in cathode-ray-tube (CRT) TVs and their 120-150-V high-voltage rail where you could not use a Schottky diode.
  5. they naturally offer an excellent protection against short circuit as the operating frequency will go to an extremely-low value (long demagnetization time), naturally scaling down all switching losses. Something you don't have at fixed \$F_{sw}\$ where you would enter a heavy CCM in short circuit engendering extra losses with possible runaway of the peak current and clamping network.

Now some drawbacks:

  1. variable frequency: there is no clock in the simplest version and a QR converter is a self-relaxing system. The frequency depends on load and line conditions: low frequency at low line and high power, high frequency at high line and low power. Efficiency suffers in light-load conditions and that is the reason you now find systems that fold the frequency back via a voltage-controlled oscillator (VCO) taking the lead in moderate to light-load operation.
  2. large power capability in high line: this is really the plague of the QR operation and I wrote a series of articles in How2Power.com showing how you need to reduce the maximum peak power capability at high line compared to low line (see the below picture):

enter image description here

The first article of this 3-part series is here.

3.large circulating rms currents compared to CCM. The fact that you transmit power in DCM implies large secondary-side current requiring the selection of good-quality capacitors. Same in the primary side with more losses in the copper, the MOSFET but also in the core considering the large flux swing compared to the smaller minor loop in CCM.

So, as you can see, many arguments to adopt or reject a QR converter. Now, back to your questions regarding possible saturation in CCM. No, you can't saturate in the case you mention because the transformer is designed to accept the maximum peak current the controller limits the primary current to. Whether the converter operates in CCM or DCM does not change, the maximum peak won't be exceeded as it is under control cycle-by-cycle. Of course is volt-second balance is violated for extreme duty ratio conditions, saturation risks exist but that is a different story.

Now, how can you pass more power in CCM than in DCM? Actually, for a given design at a fixed switching frequency with a given transformer, you will pass more power in DCM than in CCM. The general power formula for a flyback converter is: \$P_{out}=0.5\times L_p\times (I_p^2-I_v^2)\times F_{sw}\$. For a given maximum peak current set by the controller, you see that if the converter enters DCM at a high voltage input, the term \$I_v\$, the valley current, no longer subtracts from the peak value and you transmit more power. This is the over-power phenomenon I described in my series of articles. For the QR controller you mentioned, it is possible to increase the delivered power by turning the variable-frequency QR controller into a frequency-controlled converter and increase the switching frequency without exceeding the maximum peak current set by the controller: when the switching frequency is too low or when the feedback voltage exceeds a threshold, you decide to quit QR operation and no longer wait for the core total demagnetization to impose new cycles of smaller periods. In the above formula, if the frequency increases, then the converter operates in CCM with a low valley current and you can safely increase the delivered power. But you suffer higher switching losses and lose valley-switching operation. If this is temporary as for an in-rush demand, not a big deal and quite convenient as you do not need to thermally-size the entire converter for a 2x power since it does not last.

|improve this answer|||||
\$\endgroup\$
  • \$\begingroup\$ Upvote on good techy details but, can you underline for me that a flyback design in DCM will always outpower one operating in CCM when we are talking about continuous power delivery? I am yet to see a case that concludes CCM is the best in continuous power for a given core size and core saturation i.e. the numbers never appear to stack up in favour of CCM. \$\endgroup\$ – Andy aka Feb 14 at 14:33
  • \$\begingroup\$ @Verbal Kint, during DCM mode with fixed frequency, Ivalley will always equal to zero for ZVS switching, and due to that, DCM should have more transmit power. For above QR, Ivalley will not equal to zero due to CCM mode but Ipeak will still remains the same. At this stage, by increasing the switching frequency the transmit power can be increased with original peak current. Correct? \$\endgroup\$ – Lutz Fi Feb 14 at 14:56
  • \$\begingroup\$ What I said regarding power is the following: assume you have a fixed-frequency design delivering 100 W at low line and you operate close to the max authorized peak current. Now increase the input voltage and you'll see a reduction of the valley current: this is normal, the duty ratio reduces and there is more time to demagnetize. As the valley current reduces (the peak is constant) and the converter approaches DCM, you transmit more power cycle by cycle. With this QR example, you have to increase \$F_{sw}\$ to transmit more power in CCM. \$\endgroup\$ – Verbal Kint Feb 14 at 15:38
  • \$\begingroup\$ @Verbal Kint, thank you so much for your explaination. Now I'm more clear about that! \$\endgroup\$ – Lutz Fi Feb 15 at 9:55
  • \$\begingroup\$ With pleasure! One of my paper covers QR operation and shows how peak current and frequency evolve with operating conditions. Worth reading if you want to design QR converters. Good luck! \$\endgroup\$ – Verbal Kint Feb 15 at 10:03
1
\$\begingroup\$

A typical flyback design

enter image description here

From my understanding, if we go above the primary peak power, the transformer will saturate. How this CCM mode overcome that?

It's better to talk about reaching a peak in primary current rather than a peak in primary power because it's an excess of current that creates core saturation.

So, if you define \$I_{MAX}\$ as the maximum allowable current to avoid excessive core saturation then, the maximum energy you can extract from that primary is when the secondary current has fallen from maximum to zero amps i.e. all the magnetic energy is removed from the core. Below are a couple of example waveforms of primary and secondary current: -

enter image description here

The lighter load (red and green solid lines) is dispensed less energy per switching cycle and naturally, there is a "hold" in the latter part of that switching cycle. The full load scenario is represented by the dotted lines i.e. the "hold" period is zero and the peak current rises to \$I_{MAX}\$.

So, you can’t rise higher than the saturation point (\$I_{MAX}\$) and, you can’t deliver maximum energy in a single switching cycle unless secondary current (and flux) falls to zero. This is the boundary between CCM and DCM.

Based on that, does this means that this peak power can be achieve without changing the size of the transformer?

If the operating frequency and core are fixed then, this is the maximum power that can be achieved. However, if the switching frequency were (say) doubled, the flyback converter could run at higher power using CCM; the peak curent would still be \$I_{MAX}\$ and it would fall to \$I_{MAX}/2\$ before the next switching cycle began.

The energy dispensed each cycle would be only 75% of what it was when operating at the previous switching frequency but, the power dispensed would be 2 x 0.75 of what was previously shown. So, by doubling the switching frequency, the power throughput is 50% greater in CCM than in DCM at half the switching frequency.

But, if you can "run" at twice the switching frequency, then why not design the flyback transformer to run in DCM with lower primary inductance?

So, if you halved the primary turns you get: -

  • Primary inductance falling by 4.
  • Peak current can rise to \$2\cdot I_{MAX}\$ for the same core saturation levels.

The impact of these is that the energy stored in the first part of the switching cycle is now: -

$$W = \dfrac{1}{2}\dfrac{L}{4}\cdot [2\cdot I_{MAX}]^2 = \dfrac{1}{2} L\cdot I_{MAX}^2$$

And, this is exactly the same energy as was stored in the original DCM design when running at the original frequency but now, because the switching frequency has doubled, that energy is converted twice as often and, twice the power is dispensed into the secondary load.

If you refer back to what I wrote earlier for a CCM design operating at twice the switching frequency, the power improvement is only 1.5.

|improve this answer|||||
\$\endgroup\$

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.