High Efficiency AC to DC Flyback Converter
Introduction to Our High Efficiency Flyback Converter Design
The Ubiquity of Switch-Mode Power Supplies (SMPS)
In the modern world, nearly every electronic device—from smartphone chargers to industrial machinery—requires an efficient method of converting the ubiquitous alternating current (AC) from the wall outlet into a regulated, low-voltage direct current (DC) for its internal components. While linear regulators are simple, they are prohibitively inefficient for this task, dissipating excess power as heat. This is where Switch-Mode Power Supplies (SMPS) dominate, achieving high efficiency by rapidly switching a transistor between its on and off (saturated and cut-off) states, minimizing the time spent in the high-power-loss transition region.
The Flyback Converter: A Workhorse for Isolation
Among the myriad SMPS topologies, the flyback converter stands out as one of the most popular and cost-effective solutions for low-to-medium power applications (typically up to 100W). Its key advantage lies in its ingenious use of a coupled inductor, which functions simultaneously as an inductor for energy storage and a transformer for galvanic isolation. This isolation is a critical safety requirement, physically separating the dangerous high-voltage AC primary side from the user-accessible low-voltage DC secondary side, preventing electric shock.
The Core Principle of Operation
The operation of a flyback converter is an elegant two-step process, orchestrated by a single controlling switch (a MOSFET):
Energy Storage Phase (Switch ON): When the MOSFET is turned on, current ramps up through the primary winding of the transformer, storing energy in its magnetic core. During this time, the polarity of the secondary winding is arranged to reverse-bias the output diode, preventing any power transfer to the load. The output capacitor alone supplies the load.
Energy Delivery Phase (Switch OFF): When the MOSFET turns off, the magnetic field collapses. This collapse reverses the voltage polarity across all windings. The secondary winding now forward-biases the diode, allowing the stored energy to be released and transferred to the output capacitor and the load.
This "store-and-release" mechanism, cycling energy through the transformer's magnetic field, is the fundamental principle that gives the flyback converter its name.
Objective of This Document
This document details the design theory, component selection, and operational principles of an isolated AC-DC flyback converter. It is structured section-wise to align with the key functional blocks of the schematic: EMI Filter Section, Rectifier Section, Control Section, Feedback Section, and Output Section. It covers the power stage, the critical selection of the flyback transformer, control circuitry, and additional considerations for efficiency and protection.
Specifications
Before diving into the details, it's essential to define the target specifications for the converter. These guide all design choices and ensure the system meets real-world requirements.
Input Specifications:
Voltage Range: Universal input, 85-265 VAC (to support global mains voltages).
Frequency: 47-63 Hz.
Maximum Input Current: 1 A (at minimum input voltage and full load).
Output Specifications:
Voltage: 5 VDC (regulated).
Current: 4.5 A (continuous), up to 3 A peak.
Power: 22.5 W nominal.
Ripple: < 100 mV peak-to-peak.
Efficiency: > 91% at full load.
Regulation: ±5% over line, load, and temperature variations.
Environmental and Safety:
Operating Temperature: -20°C to +70°C.
Isolation: 5 kV primary-to-secondary.
Compliance: Meets IEC 62368-1 safety, CISPR 32 EMI Class B.
EMI Filter Section: Design and Component Selection
The design part of this filter section has been already explained and posted to our blog by my partner Elif Dilara Deniz, read here, so now we will only focus on the component selection part. In the EMI filter we have two main components, capacitors and the common mode choke. Careful selection is crucial, as both component values and packaging significantly impact performance.
In high-voltage, high-reliability applications, safety is non-negotiable, which is why safety capacitors are essential. These passive components are engineered to mitigate the risks associated with transient voltages and electrical interference, helping protect both users and equipment from hazards, even in the event of component failure.
Class X Capacitors
Class-X capacitors, also known as “across-the-line capacitors,” are used between the wires carrying the incoming AC current. These capacitors offer line-to-line protection, so if there is a failure, a short may occur but there is no risk of shock. This is because an X-capacitor failure usually results in a short circuit, triggering a fuse or circuit breaker to open. In all EMI filters you will see that all X-capacitors are film box type (rectangular metallic) capacitors. There are three main reasons for this:
Primary Concern: Fail Open Circuit: If a capacitor fails, a short circuit across the AC mains will draw huge currents and can cause a fire. Therefore the key requirement for an X-capacitor is to fail open circuit.
Why Film Metal Capacitors Excel (Specially Metallized Polypropylene): These capacitors have a thin metal film deposited on the plastic layer. If a small defect causes a short, the high energy at the fault point vaporizes the tiny surrounding metal layer, isolating the fault and restoring the capacitor's function. The capacitor loses a tiny, negligible amount of capacitance but continues to work safely. This makes them exceptionally reliable.
The rectangular box style, typically a wrapped film inside a sturdy plastic case, is often enclosed in epoxy thus making it robust, rugged and almost fire proof.
They maintain their capacitance value well over temperature and time, which is important for consistent filter performance.
Class Y Capacitors
Class-Y capacitors, also known as “line-to-ground capacitors” or “line bypass capacitors,” offer line-to-ground protection, which generally means that if a failure with the ground occurs, there is a risk for shock. Class-Y safety capacitors must meet rigorous specifications, and are designed to fail open, minimizing the chance of electric shock. In all the EMI filters we use ceramic disc capacitors as Y-capacitors here is the reason why.
Primary Concern: Electric Shock Prevention: The paramount requirement for a Y-capacitor is high dielectric strength and extreme reliability to prevent a short-circuit to ground under any condition (including voltage surges like lightning strikes).
Why Ceramic Disc Capacitors Excel (Specifically Class-Y Rated): Class-Y ceramic capacitors use special, high-reliability ceramic formulations (like NPO/COG) that are highly resistant to breakdown under high voltage and high-voltage transients. They are designed and tested to withstand impulse voltages of several kilovolts.
While no component is 100% fail-safe, Class-Y ceramics are manufactured to the highest standards to minimize the risk of a short-circuit failure. Their construction is simple and very robust.
EMI noise is very high frequency. Ceramic capacitors have very low parasitic inductance (ESL), which means they remain effective (low impedance) well into the MHz and GHz range, which is crucial for attenuating radiated EMI.
Common Mode Choke (CMC)
There are certain specifications one must consider before selecting common mode choke (CMC) for EMI filter.
The CMC is not only selected for inductance (L), but for impedance (Z) at the noise frequency you need to suppress. CMCs job is to block common mode noise by presenting high impedance. The formula Z = 2πfL is a simplification that ignores core material effects and parasitics.
The impedance across the entire frequency range must be high where noise is problematic (150kHz to 30MHz) for conducting EMI.
The impedance will peak at a certain frequency (SRF) and then fall off. Ensure the SRF (Self-Resonant Frequency) is above your highest frequency of concern. If noise is at 10MHz but choke SRF is at 2MHz it will be ineffective.
Maximum operating current should be less than 50-80% of the rated heating current so that thermal failures can be avoided.
Take precise measurements for calculating the inductance value needed.
A leakage inductance of <5% is acceptable.
Bleeder Resistor
The main job of the bleeder resistor (R_bleed) is to safely discharge the X-capacitor(s) after the AC power is disconnected. In order to prevent shock due to accumulated charge in the X-capacitors. Following things must be considered while selecting bleeder resistor:
Choose a value that gives an RC time constant of ~1 second or less. R ≈ 1 / C.
Calculate power using: P = V_rms² / R.
Choose a resistor with a rating at least 2x the calculated value.
Ensure the resistors voltage ratings are at least 2x the peak input voltage.
Place the resistor as close to X-capacitor as possible.
Additional components like fuses and varistors (for surge protection) are included to enhance safety and reliability.
Rectifier Section: Design and Component Selection
Right after the filter is the bridge rectifier for rectification (conversion) of AC to DC, this is taken care by the rectifier bridge IC in our case (MDB6S), the main purpose of this IC is to convert AC voltage to DC voltage and provide that as an input to the transformer to pass on the energy onto the output side. On the DC output sides of this IC there are capacitors used for smoothening the DC output and filtering any noise found in it.
Flyback Transformer Design and Selection
The flyback transformer is the heart of the converter, serving as both energy storage inductor and isolation transformer. Proper design ensures efficiency, minimizes losses, and prevents saturation.
Core Principle Recap: The transformer operates in discontinuous conduction mode (DCM) for simplicity, where the primary current resets to zero each cycle. Turns ratio (N_p/N_s) determines voltage step-down, while magnetizing inductance (L_m) sets energy storage.
Power Stage Design (MOSFET and Snubber)
The power stage includes the primary switch (MOSFET), which handles high-voltage switching.
MOSFET Selection:
The MOSFET (e.g., N-channel like IPD65R380C6) handles high-voltage switching.
Key Specs: V_DS > 1.5 * V_in_max * √2 (e.g., >600 V for 265 VAC). R_DS(on) <0.5 Ω for low conduction loss. Q_g <30 nC for fast switching.
Why This Matters: Low R_DS(on) minimizes I^2R losses; high V_DS withstands spikes. Gate drive from controller must match (e.g., 10-15 V).
Heat Dissipation: P_loss = I_rms^2 * R_DS(on) + (1/2) * C_oss * V_in^2 * f_sw. Use heatsink if >2 W.
Snubber Circuit: Design and Component Selection
When the MOSFET turns on the primary side of the transformer stores energy in form of a magnetic field and when the MOSFET turns off this energy stored transfers to the secondary side. But since no transformer is ideal, some of this energy is left in the primary side and forms the parasitic inductance called leakage inductance.
The energy stored in this leakage inductance (1/2) * L_leak * I_pk² if left unchecked would resonate with parasitic capacitances and create a high frequency voltage spike at the MOSFET drain node exceeding the V_DS rating and destroy it.
The Solution: How the RCD Snubber Works
The snubber's job is to safely absorb and dissipate this energy stored in the leakage inductance. Here is how snubber circuit works:
When MOSFET turns OFF the leakage inductance accumulates at the drain node of the MOSFET thus forward biasing the diode and travelling through it, after this spike passes the diode it gets trapped and stored in the capacitor thus clamping the drain voltage. V_clamp ≈ V_out_reflected + V_cap after the spike is over the energy stored in the capacitor gets dissipated through the resistor, resetting the circuit for the next cycle.
Input bulk capacitors (e.g., electrolytic, 47-100 µF, >400 V) hold up during low-line conditions.
Control Section: Design and Component Selection
The control section uses a PWM controller IC (e.g., UCC28C43, a current-mode controller) to regulate switching.
How It Works:
Oscillator: Sets f_sw via RT/CT pins (e.g., R=10 kΩ, C=1 nF for 65 kHz).
Current Sensing: Via sense resistor (e.g., 0.1 Ω) on MOSFET source; limits peak current for protection.
Feedback Input: FB pin receives optocoupler signal; adjusts duty cycle (0-50%) for regulation.
Startup: Internal bias or auxiliary winding provides V_CC (e.g., 12-18 V).
Advantages: Current-mode provides inherent cycle-by-cycle limiting, improving stability over voltage-mode.
Component Selection:
Ensure IC handles V_in up to 20 V; add UVLO for brownout protection.
Additional components include gate drive resistors for switching speed control and bypass capacitors for stable V_CC.
Feedback Section: Design and Component Selection
Feedback Loop: How It All Is Controlled and Stabilized
What is TL431: it is three terminal shunt regulator with output range of 2.5 to 36V set by a voltage divider, with an output impedance of 0.2ohm, active output circuitry providing a very sharp turn on characteristic making it excellent replacement for zener diodes.
So how this actually works is that it acts like a switch (smart, adjustable, efficient) between the LED and GND. The voltage divider resistors create a scaled down version of the output voltage and feed it to the Ref pin of the comparator.
The comparator compares this voltage to its internal 2.5V reference:
If scaled voltage is lower than 2.5V the output is too low, the comparator draws less current and LED in the optocoupler is less bright (dim) sending a low voltage to the FB pin of the voltage control IC pulling FB pin down slightly, IC interprets this as output voltage too low increasing the duty cycle to push more energy to secondary (output) side.
If the scaled voltage is higher than 2.5V, the output is too high then the comparator draws more current and the LED is brighter, sending high voltage to the FB pin, pulling it down hard, the IC interprets it as output voltage too high, decreasing the duty cycle to push less energy to output.
An optocoupler acts as a liaison between the primary (controller) and secondary (compensator) side.
Compensation Network: How It Works
The job of this network is to stabilize and shape the dynamic response of the converter. It is designed to ensure not only the feedback loop is stable meaning that perturbation signals will not generate a sustained oscillation in the system but also that after a transient event the output will meet the target requirement regarding over and undershoot ringing and setting time imposed by the supplied load.
In most of the converters requiring light output voltage regulation an optocoupler is used to transmit the feedback signal across the isolation barrier from secondary to primary side (compensator to controller). The optocoupler thus becomes a part of the control loop and more exactly the compensator block. Therefore the parameters like its current transfer ratio (CTR) and collector emitter parasitic capacitance appear in compensator transfer function modifying the amplitude and phase of the feedback signal for this reason their variations must also be considered in the design otherwise the stability of power supply could be compromised and target transient response not met.
The compensator circuit is necessary to set the target open loop crossover frequency and phase margin as discussed below:
As already discussed the job of the compensation network is to ensure the entire power supply is stable and has good transient response. The ultimate goal of this network is the bode plot and so to understand that, we need to think in terms of loop gain and phase shift across different frequencies. The goal is to shape the loop gain (Bode Plot) to have two features:
Gain Margin: the loop gain should be less than 1 (0db) before the total phase shift reaches -360.
Phase Margin: when loop gain is 1(0db) the total phase shift should be sufficiently far from -360 (>45-60°). This prevents rings and oscillations.
A power supply plant (transformer, MOSFET) naturally has gain that falls at higher frequencies and introduces phase lag this networks job is to counteract this to achieve the goals above.
The network we use is a type two compensation network; it provides one zero to boost phase and two poles to roll off gain.
The large capacitor (C22) (integrator, series cap, low frequency pole, P1), this works with the equivalent series resistor to create a very low frequency pole (P1). A pole causes the gain to decrease at a rate of -20dB/decade and adds -90 phase shift. Its ability to set dominant low frequency roll-off of the error amplifier. It makes the error amplifier act like an integrator at low frequencies ensuring very high DC gain, this is what gives power supply its excellent DC accuracy by precisely adjusting to correct any steady-state error.
The resistor (R20) (Gain and Zero setter) works with the series capacitor to introduce a zero (Z1) into the loop response. A zero counteracts a pole by providing a +20dB/decade gain boost and +90 degree phase boost. The pole (capacitor) adds a dangerous phase lag, the zero that is strategically placed to cancel the phase lag from the power stages pole, providing phase margin right where the loop gain crosses 0dB, bringing stability.
The mid band gain of the error amplifier is also set by this resistor, G = Rg/Rfb.
The capacitor (C21) (high frequency noise rejection pole P2, parallel to resistor) attenuates any noise, ripples and signals above the switching frequency ensuring they dont disrupt the control loop. It adds essential noise immunity.
Summing the efforts we can say that these components “tune” the personality of the error amplifier.
P1 makes it an integrator for perfect DC accuracy.
Z1 gives it a gain and phase boost at crossover frequency to achieve stability.
P2 tells it to ignore everything above switching frequency.
Regulation Network: How It Works
R17: It limits the current through the LED when the comparator (shunt regulator) conducts hence controlling how strongly the optocoupler transfers error signal to the primary controller. Rled = (Vo - Vfled - Vk_tl431) / Iledmax.
R18 (feedback divider/TL431 biasing): This resistor provides a pullup path and defines the operating point for the feedback loop by working in conjunction with the components inside the optocoupler package on the primary side. It sets a bias current for TL431.
Diode Zener (D6): Protection component to prevent the TL431 from transient overvoltage events.
Capacitor (C20) (noise filter/loop compensation): Noise filtering of high frequency noise on the cathode of TL431 and preventing it from being amplified and disrupting the feedback loop. Introduces a pole into transfer function of TL431’s response to very high frequency changes preventing it from reacting to noise and causing oscillations.
Resistor (R19) (TL431 cathode bias resistor): This resistor provides bias current necessary for TL431 to operate correctly. It ensures the TL431 cathode is supplied with enough current to stay in its regulated region. Regardless of the optocoupler or other adjacent components in its loop, this resistor makes sure that TL431 has bias current available to maintain accurate regulation.
Output Section: theory and working
Purpose of the Output Section
This stage takes the rectified and regulated DC voltage from the primary converter (AC → DC + switching stage) and conditions it to produce a stable, low-ripple DC output suitable for powering loads such as USB devices (VBUS).
2. Synchronous Rectification Controller (U2: UCC24612)
The UCC24612 is a synchronous rectifier (SR) controller.
It senses the drain-source voltage of the MOSFET (Q3) and drives its gate to act as an active rectifier instead of using a diode.
Advantage: It reduces conduction losses compared to a diode rectifier, especially at low voltages and high currents.
Operation:
When the secondary winding voltage (Vout side) is positive, U2 turns ON Q3 (NMOS), allowing current to flow with very low Rds(on).
When current tries to reverse (during dead time or negative cycle), U2 quickly turns OFF Q3, blocking reverse conduction.
This improves efficiency compared to a Schottky diode.
3. MOSFET (Q3: NMOS)
Works as the synchronous rectifier switch.
Controlled by U2, it replaces the traditional diode rectifier.
Provides much lower conduction drop (tens of mV vs 0.3–0.7V in diodes), reducing heat losses.
4. RC and Diode Network (R16, R14, C16, C17, D5)
R14 (2.2k) ensures proper biasing and stability of gate drive.
C16, C17 are decoupling capacitors that stabilize the SR controller’s supply (VDD and VS pins).
R16 + D5 form a snubber/protection path to clamp high voltages and protect the MOSFET from transients coming from the transformer leakage inductance.
5. Output Filter (L3 + C18 + C19)
After rectification, the current is pulsating.
L3 (1.5 µH) smooths the current by storing energy and releasing it, reducing ripple.
C18 and C19 (270 µF each) act as bulk capacitors to filter voltage ripple and provide a stable DC.
Together, they form a π-filter (L-C), which ensures a clean, low-ripple DC output.
6. USB Output Connector (J2: USB_A)
Provides the VBUS output (5V nominal) for USB-powered devices.
Vfb is a feedback sense line that typically goes back to the controller (in the primary or regulation stage) to maintain accurate output voltage.
Pin 4 (GND) and Shield provide proper grounding and EMI shielding.
No comments:
Post a Comment