Q1. What exactly are these three topologies? What is each one naturally good at?

LLC (full-/half-bridge LLC resonant converter): originally a full or half bridge on the primary side, plus a single-sided resonant tank (Lr-Lm-Cr), plus diode or synchronous-rectifier full-wave rectification on the secondary side. It is a naturally unidirectional, high-efficiency topology.

CLLC (CLL-C / CLLLC): a symmetric, bidirectional evolution of LLC. The primary and secondary sides each have a set of resonant elements, while the transformer magnetizing inductance is shared. It can run in both directions and remains efficient in both directions.

DAB (dual active bridge): a primary full bridge, a high-frequency transformer, and a secondary full bridge. It has no resonant tank; instead, leakage inductance or an external inductor acts as the energy-transfer channel. Power is controlled by phase shift, so it is naturally bidirectional and has the most flexible control.

Q2. What is the most intuitive difference in topology?

Q3. What is the fundamental difference in control method?

This is where the three topologies diverge at the control level.

LLC / CLLC: variable-frequency control (PFM). The switching frequency fsw is adjusted, and the converter moves along the gain curve to regulate the output voltage. The duty ratio is essentially locked around 50%; frequency is the main control handle.

DAB: fixed-frequency phase-shift control. The switching frequency is basically fixed, while the phase difference phi between the square waves of the primary and secondary bridges is adjusted. Power is approximately proportional to sin(phi).

Consequence: DAB's fixed frequency makes EMI filter design and current sharing among paralleled modules easier. LLC/CLLC frequency sweeping means magnetic components must tolerate the lowest frequency, and the EMI spectrum is spread over a wider band.

Q4. What does "full resonance" mean? Why is LLC called a resonant converter?

The LLC name comes from three impedance elements that determine the gain:

· Lr (resonant inductance) + Cr (resonant capacitance) form the series resonant frequency fr.

· Lm (transformer magnetizing inductance) introduces the second frequency and defines the inductive-region boundary below fr.

When the switching frequency is close to fr, the current naturally becomes sinusoidal or quasi-sinusoidal. The primary MOSFET turns on when Vds has fallen close to zero, enabling ZVS. The secondary current naturally falls to zero before commutation, enabling ZCS.

This is the essence of resonance: the LC phase characteristic "softens" switching, instead of forcing the device to interrupt current abruptly.

Q5. Can all three achieve soft switching? Where does soft-switching quality differ?

Q6. Who has the highest efficiency, and where is the high-efficiency region?

Measured and engineering experience, assuming the same power class and a competent design:

· LLC has the highest peak efficiency. Near the rated operating point it can readily exceed 98%, because its waveforms are the cleanest and switching/turn-off losses are the lowest.

· CLLC peak efficiency is close to LLC, typically above 97.5-98%+, but it pays for that with the cost and volume of resonant components on both sides.

· DAB is commonly lower by about 0.5-2%. This is not because the theoretical limit is low, but because square-wave current creates high turn-off current, voltage mismatch creates reactive circulating current/backflow power and conduction loss, and loss of ZVS at light load makes efficiency fall quickly.

Caution: under wide-voltage and wide-load conditions, advanced DAB modulation such as TPS can narrow the gap. Conversely, an LLC forced far out of the resonant region can also perform poorly.

Q7. How does the design philosophy of the high-frequency transformer differ?

This point is especially important. The three topologies ask almost opposite things from the transformer.

LLC transformer

· The magnetizing inductance Lm must be precisely controlled, often with an air gap.

· Leakage inductance is preferably used as Lr, or a separate Lr is integrated with the transformer.

· The current is near sinusoidal, so core loss is lower and relatively high frequencies, roughly 200 kHz to 1 MHz, are feasible.

CLLC transformer

· The primary and secondary sides must be symmetric in turns ratio, leakage distribution, and winding structure.

· Lm remains one of the resonant parameters, while leakage inductance is absorbed into the resonant inductance.

· The design is the hardest: both resonant parameter sets must be consistent, and FEA verification is often required.

DAB transformer

· Leakage inductance is the energy-transfer channel. It is either deliberately made large through split windings, air gaps, or winding arrangement, or implemented as an external Lk.

· A larger Lm is preferred to reduce magnetizing reactive power.

· Square-wave excitation increases core delta-B and iron loss, but fixed-frequency operation makes optimization easier.

Q8. Why is DAB said to handle a wider voltage range, while LLC is narrower?

DAB: gain is approximately n x V2 / V1 and is controlled linearly or quasi-linearly by phase shift. It is relatively weakly coupled to load, so large input/output voltage swings can still be handled.

LLC / CLLC: gain is a nonlinear function of frequency. The gain curve has a humped resonant shape. When the voltage deviation is large, the control frequency is pushed out of the comfortable region; the farther it moves away from resonance, the larger the circulating current becomes, and soft switching may be lost.

Therefore, for a battery scenario such as 250-450 V on a 400 V platform, LLC/CLLC can still be acceptable. For a 200-800 V wide range on an 800 V platform, the required frequency swing becomes difficult, and DAB's logic advantage becomes obvious.

Q9. How should EMI and filter difficulty be ranked?

From easiest to hardest in terms of EMI friendliness:

CLLC ~= LLC: the resonant tank naturally filters the waveform, current is close to sinusoidal, and dv/dt is gentler.

DAB: square-wave switching produces stronger di/dt and dv/dt, so snubbers, shielding, and common-mode filtering require more serious design effort.

However, the situation also reverses in one respect:

· DAB operates at fixed frequency, so the EMI spectrum is concentrated and the filter can be designed for targeted frequencies.

· LLC/CLLC operate with variable frequency, so the spectrum spreads and the filter must cover the full frequency band.

In short, both approaches have their own pain points.

Q10. Bidirectional capability: who is naturally bidirectional and who is "made" bidirectional?

For V2G, V2L, and bidirectional energy storage, LLC is usually eliminated unless an independent reverse-stage circuit is acceptable. The practical choice between CLLC and DAB depends on whether efficiency or control flexibility is more important.

Q11. How do device count and cost roughly compare?

A rough comparison, excluding drivers and auxiliary circuits:

· LLC, half bridge: 4 switches + 1-2 resonant L/C components + 2 SR devices; usually the minimum.

· LLC, full bridge: 4-8 switches, depending on whether synchronous rectification is used, plus resonant L/C.

· CLLC: active switches on both sides, typically 8 devices, plus primary/secondary resonant capacitors. Large-current film resonant capacitors are expensive, so CLLC generally has the most components and the highest cost.

· DAB: 8 switches plus leakage inductance or an external inductor, with no resonant capacitor. Cost is usually in the middle.

At high power, especially >=10 kW, the cost and volume penalty of CLLC resonant capacitors becomes very obvious.

Q12. For high-power expansion and multi-module paralleling, which topology is friendlier?

DAB is robust: fixed-frequency phase-shift operation means multiple paralleled modules naturally share the same frequency. Current sharing can be achieved by fine phase-shift adjustment. ISOP, or input-series output-parallel, is very convenient.

LLC / CLLC are more difficult: variable-frequency operation means L/C tolerances among modules can cause unequal power sharing at the same frequency. Current sharing often requires hybrid control or complex digital coordination, which becomes an engineering headache.

Therefore, when a project moves from tens of kilowatts to the megawatt class, such as energy-storage PCS or solid-state transformers (SST), DAB is often nearly irreplaceable.

Q13. What are the usual compensation methods for light load, wide voltage, and loss of ZVS?

Q14. What are the typical application scenarios?

Q15. Final selection method: three sentences to remember

· Unidirectional operation + efficiency priority + controllable voltage range -> LLC.

· Bidirectional operation + extreme efficiency / low EMI + medium- or low-power single module -> CLLC.

· Bidirectional operation + wide voltage / high power / multi-module paralleling and expansion -> DAB.

A more direct version:

· Look at power: below 10 kW, CLLC/LLC are often favored; above roughly 10-20 kW, DAB becomes more attractive.

· Look at voltage: a narrow range is tolerable for a resonant topology; a wide range such as 200-800 V is more stable with DAB.

· Look at cost and paralleling: tight budget or unidirectional output points to LLC; cascaded expansion points to DAB; if the project is willing to buy efficiency with cost, choose CLLC.