LLC
resonant converters are widely used in modern switching power supplies
due to their high efficiency, high power density, and soft-switching
characteristics. Among them, the transformer (more accurately referred
to as an 'integrated component of resonant inductor + transformer') is
the core and a key challenge in the design process.
This article systematically analyzes the key technical points in LLC transformer design through 23 Questions and 23 Answers about LLC transformer design technology.
A: The core difference lies in functional integration and operating mode.
- Ordinary PWM transformers (e.g., flyback, forward
transformers): Their main functions are electrical isolation and energy
transfer. Their magnetizing inductance is usually designed to be as
large as possible to reduce magnetizing current loss.
- LLC 'transformers': They integrate three functions:
1) Transformer (turn ratio conversion); 2) Resonant inductor (Lr); 3)
Magnetizing inductor (Lm). Lm is a crucial parameter involved in
resonance and requires precise design and control. LLC transformers
operate in a sinusoidal resonant state rather than a square-wave pulse
state.
A:
- ZVS (Primary-side): During the dead time of the
switches, the resonant current (Ir) charges and discharges the junction
capacitance (Coss) of the switches. The switch voltage is pulled down to
zero before the switch is turned on, thus realizing ZVS. This process
requires sufficient energy, i.e., a sufficiently large amplitude of the
resonant current.
- ZCS (Secondary-side): When the magnetizing inductor
(Lm) participates in resonance, the magnetizing current rises to match
the resonant current (Ir). At this point, the secondary-side current
naturally drops to zero, enabling the rectifier to turn off under ZCS
conditions and eliminating reverse recovery issues.
A:
The voltage gain of an LLC converter is a curve related to the
normalized frequency (Fn = fsw / fr) and inductance ratio (Ln = Lm /
Lr). The curve has a 'peak' characteristic, with a gain of approximately
1 near the resonant point (Fn = 1).
- When the input voltage is high, the converter operates near Fn ≈ 1 (gain ≈ 1), where efficiency is maximized.
- When the input voltage is low, a higher gain is required. The gain
is increased by reducing the switching frequency (Fn < 1) to
stabilize the output voltage. This frequency-modulated voltage
regulation method is the basis for the LLC converter’s wide input
voltage operation.
A: The essential initial conditions include:
- Input voltage range (Vin_min, Vin_nom, Vin_max)
- Rated output voltage (Vo) and current (Io)
- Rated output power (Po)
- Resonant frequency (fr) or target switching frequency range (usually
selected based on trade-offs between efficiency and size, e.g.,
100kHz–200kHz)
- Maximum operating frequency (fmax, typically limited by the controller)
A:
The turn ratio n is determined by the minimum input voltage (Vin_min)
and the resonant point (Fn = 1). At the resonant point, the circuit gain
M = 1. Therefore:
n = (Vin_min / 2) / (Vo + Vf)
where
Vf is the forward voltage drop of the output diode. The term 'Vin_min /
2' arises because the LLC converter adopts a half-bridge topology, and
the amplitude of the square wave under maximum duty cycle is Vin / 2.
A: Ln is a critical design parameter, typically ranging from 3 to 7.
- Small Ln (e.g., < 3): A small magnetizing
inductor (Lm) leads to large magnetizing current, resulting in high
circulating energy and increased conduction losses. However, it
facilitates ZVS implementation.
- Large Ln (e.g., > 10): A small magnetizing
current reduces conduction losses, but narrows the gain range. A lower
frequency is required to achieve the same gain, which may increase core
losses. Additionally, the energy required for ZVS may be insufficient.
- Empirical value: A value of 5 or 6 is usually chosen as a good starting point, balancing loss reduction and gain range.
A:
Once Ln and the resonant frequency (fr) are determined, the
characteristic impedance (Z0) of the resonant tank and its parameters
can be calculated.
- Calculate the characteristic impedance:
Z0 = sqrt(Lr / Cr) = (8 × Ln² × R_load) / π²
(where R_load = Vo² / Po) - Calculate the resonant angular frequency:
ωr = 2 × π × fr = 1 / sqrt(Lr × Cr) - Solve the above two equations simultaneously to obtain:
Lr = Z0 / ωr
Cr = 1 / (ωr × Z0) - Finally, calculate the magnetizing inductor:
Lm = Ln × Lr
A: The AP Method (Area Product Method) is used to initially estimate the core size. After determining the core, calculate the number of primary turns (Np) as follows:
- According to Faraday’s Law, the minimum number of turns to prevent core saturation is given by:
Np_min ≥ (Vin_max × 10⁸) / (4 × fmin × B_max × Ae)- Vin_max: Maximum input voltage (use Vin_max / 2 for half-bridge topologies)
- fmin: Minimum operating frequency (frequency at maximum gain)
- B_max: Maximum allowable magnetic flux density (in mT), depending on
the core material (e.g., ~300mT for PC95, with a safety margin)
- Ae: Effective cross-sectional area of the core (in cm²)
- Fine-tune Np based on the calculated Lm value: Lm ∝ Np². If the Lm
value derived from the calculated Np does not match the target, adjust
Np.
- Calculate the number of secondary turns: Ns = Np / n
A: High-frequency, low-ESR, and non-polar capacitors must be used.
- Type: C0G (NP0) ceramic capacitors are preferred
due to their extremely low ESR and excellent temperature stability. Film
capacitors (e.g., PPN) are a secondary option.
- Avoid: Ceramic capacitors such as X7R/Y5V (with
significant capacitance variation under DC bias and temperature) and
electrolytic capacitors (high ESR and polar characteristics).
- Voltage stress: The peak voltage of Cr can be estimated as π × Io × n × Z0 / 2, and a sufficient voltage margin must be reserved.
A: The resonant inductor (Lr) is essentially the leakage inductance of the transformer. To precisely control the Lr value, the sandwich winding method is typically used.
- Standard sandwich: Primary (half) → Secondary → Primary (half). This method ensures good coupling and low leakage inductance.
- Increasing leakage inductance: If the design
requires a larger Lr, separate winding (primary and secondary windings
wound separately) or intentionally adding insulation between the primary
and secondary windings can be used to increase leakage inductance.
A:
- Magnetizing inductance (Lm) measurement:
Short-circuit all secondary windings and measure the inductance at the
primary side. At this point, the leakage inductance (Lr) is
short-circuited, and the measured value is Lm.
- Resonant inductor/leakage inductance (Lr) measurement:
Open-circuit all secondary windings and measure the inductance at the
primary side. The measured value is (Lm + Lr). Since Lm is much larger
than Lr, this value is approximately equal to Lm. For greater accuracy,
first measure Lm, then subtract Lm from the open-circuit measurement
result to obtain Lr (Lr = L(open_sec) - L(short_sec)).
A: The main functions of the air gap are to prevent core saturation and adjust inductance.
- In LLC transformers, Lm is subject to DC bias (the average voltage
of the half-bridge is Vin / 2). An air gap must be added to prevent core
saturation.
- A larger air gap reduces Lm and consequently decreases Ln. Adjusting the air gap is a key method to control Lm and Lr.
A:
- Core loss (Pcore): Related to frequency, flux swing
(ΔB), and temperature. Using low-loss core materials (e.g., PC95,
PC200) and reducing ΔB (by increasing the number of turns) can minimize
core loss.
- Winding loss (Pcu):
- Skin effect: High-frequency current tends to
concentrate on the conductor surface, reducing the effective
current-carrying area. Using multi-strand Litz wire or flat copper wire
is an effective solution.
- Proximity effect: Eddy current loss caused by
magnetic fields from adjacent wires. Methods such as using parallel thin
wires and interleaved winding can significantly reduce this loss. For
example, the sandwich winding method (placing the secondary winding
between two primary winding layers) can greatly optimize the AC
resistance of the primary and secondary windings.
A:
- Reason: Under light load, the amplitude of the
resonant current (Ir) decreases, which may be insufficient to charge and
discharge the junction capacitance of the switches during dead time.
This causes ZVS loss, leading to a sharp increase in switching loss.
- Improvement methods:
- Frequency control: Significantly increase the
switching frequency (entering the inductive region) under light load,
but this method has limited effectiveness.
- Burst Mode: Under extremely light load, the
converter operates for several cycles and then enters a sleep state.
This is currently the most mainstream high-efficiency solution for
light-load conditions.
- Optimize dead time to vary with load.
A:
- Root cause: The actual circuit gain is lower than the theoretical value.
- Solutions:
- Check parameters: Verify whether the Lm and Lr
values are accurate, especially if Lr is larger than the designed value
(excessive leakage inductance).
- Reduce Ln: Appropriately reduce Lm (by increasing
the air gap) to expand the gain range, making it easier to stabilize the
voltage under low input conditions.
- Reduce fr: If permitted by the controller, reduce the resonant frequency, but parameters must be recalculated.
A: High-frequency, low-loss soft magnetic materials must be selected.
- PC95, PC200: The most commonly used Mn-Zn power ferrites, suitable for frequencies from tens of kHz to hundreds of kHz.
- Ni-Zn ferrites: Suitable for ultra-high frequencies above MHz, but with lower saturation flux density.
- Amorphous and nanocrystalline materials: Offer excellent performance but are costly, and are mostly used in high-end applications.
A: The skin effect and proximity effect at high frequencies are the main causes.
- Skin effect: Use multi-strand Litz wire or copper foil.
- Proximity effect: Adopt interleaved winding (e.g.,
P-S-P-S structure). This structure can greatly offset magnetic fields,
reduce proximity effect loss, and is the most effective process method.
A: Insufficient gain. Possible causes:
- The actual Lr of the transformer is larger than the designed value
(excessive leakage inductance), or Lm is smaller than the designed
value.
- The load is heavier than the designed value.
- The actual capacitance of the resonant capacitor (Cr) decreases due to DC bias or temperature (avoid using X7R capacitors!).
- The actual input voltage is lower than the designed minimum input voltage.
A:
- ZVS failure: Insufficient dead time or insufficient
resonant current energy (excessively large Ln or too light a load).
Check the Vds waveform of the switches to see if the voltage drops to
zero before turn-on.
- Conduction loss: Excessively large loop resistance (including MOSFET Rds(on) and transformer winding resistance).
- Switching loss: Even if ZVS is achieved, turn-off loss still exists. Excessively large turn-off current can also cause heating.
A: This is usually caused by core or winding vibration during Burst Mode
(periodic start-stop operation). Ensure that the frequency of Burst
Mode is outside the human audible range (>20kHz), or optimize the
Burst Mode modulation strategy to smooth the oscillation frequency.
A:
The secondary-side rectifier diodes of LLC converters turn off under
ZCS conditions, so there is theoretically no reverse recovery issue.
However, parasitic inductance in the PCB layout and diode junction
capacitance can form resonance, generating voltage spikes.
- Solution: Connect an RC snubber circuit in parallel
across the rectifier diodes. Carefully adjust the R and C values to
suppress spikes without significantly affecting efficiency.
A:
- Inaccurate models: The simulation model fails to
account for PCB parasitic parameters (especially parasitic inductance in
the resonant loop), MOSFET Coss nonlinearity, transformer nonlinearity,
etc.
- Measurement errors: At high frequencies, the ground
loop of the measurement probe introduces significant interference,
causing waveform distortion. A probe with a short ground loop or a
differential probe must be used for accurate measurement.
- Parameter deviations: The actual parameters (Lm, Lr) of the wound transformer deviate from the designed values.
A: Iterative optimization (Tuning).
LLC design is a highly integrated engineering process that combines
electromagnetic theory, core materials, winding technology, and
practical debugging experience. It is almost impossible to achieve a
perfect design on the first attempt. The following steps must be
performed:
- Fabricate a prototype.
- Accurately measure key waveforms (primary-side current, switch Vds, secondary-side current).
- Compare the measured results with theoretical calculations and simulations.
- Fine-tune parameters (e.g., air gap, number of turns, Cr value) or even reselect the core until performance meets requirements.
LLC
transformer design is a systematic engineering task that integrates
in-depth knowledge of electromagnetic theory, core materials, winding
processes, and practical debugging experience. A successful LLC design
starts with rigorous theoretical calculations, is refined through
detailed model simulations, and ultimately relies on patient debugging
and optimization on the test bench.