Synchronous Rectifier Design for Low-Voltage Power Supplies

Overview of Technical Issues:

The control circuit provides insufficient turn-off timing precision for the synchronous rectification switching element, causing either harmful reverse current conduction when timing is late or harmful body diode conduction losses when timing is early, directly reducing conversion efficiency below target levels and negating the efficiency advantages over traditional diode rectification; the goal is to optimize turn-off timing control to maximize efficiency in low-voltage synchronous rectifier designs across LLC, flyback, and USB-PD topologies, particularly when implementing GaN FET technology.

Solution directions generated for this problem

Problem Direction 1 :

ImproveTurn-off timing precision
VS
ConstraintControl circuit complexity

Inspiration 1 : Cross-domain reference

Application Principle: #26 Copying
Cross-domain applicability Assess applicability
Integrated consumable data management system and platform
Innovative Solution Refine solution

Pre-characterized timing lookup table for synchronous rectifier control

Replace real-time adaptive circuits with pre-characterized timing data
How to solve :
  • Factory characterize optimal turn-off timing across load (10-100%), input voltage (90-265VAC), and temperature (-40°C to 125°C) in 8×6×5 lookup table (240 entries), store in low-cost EEPROM (256 bytes)
  • Use simple 3-input address decoder with on-chip temperature sensor, voltage divider (two resistors), and current sense resistor to index table—replaces high-speed comparators, adaptive delay compensation, and calibration logic, reducing component count by 35 components
  • Implement linear interpolation between table entries using 8-bit ALU (four gates) to achieve ±1.5ns precision across full operating range
Expected Effect : Component count -38%, timing precision ±1.5ns, efficiency >95.2%, BOM cost -$0.45
Risk Control :
  • characterization coverage insufficient across operating space
  • EEPROM data corruption over temperature cycling
  • interpolation error accumulation at boundary conditions

Problem Direction 2 :

ImproveVoltage transition detection speed
VS
ConstraintControl circuit power consumption

Inspiration 1 : Cross-domain reference

Application Principle: #19 Periodic action
Cross-domain applicability Assess applicability
Imaging device
Innovative Solution Refine solution

Predictive window-gated fast comparator for synchronous rectification

Gate fast comparator ON only during predicted zero-crossing window
How to solve :
  • Implement dual-comparator architecture: low-power monitor (0.5mA, 50ns response) continuously tracks drain-source voltage and predicts zero-crossing timing
  • activates high-speed comparator (8mA, 3ns response) only during 30ns window centered on predicted event, keeping fast circuit active <3% of switching cycle
  • Use adaptive window positioning based on previous 8-cycle timing history stored in 4-bit shift register
  • window center = average of last 8 zero-crossings ± 15ns margin, auto-adjusts for load/temperature drift without continuous calibration
  • Fast comparator uses current-starved differential pair with rapid bias injection: standby current 0.2mA, ramps to 8mA within 2ns when gated ON by prediction circuit
  • implemented with GaN-compatible 0.18μm CMOS process, bias control via fast MOSFET switch (Ron <5Ω)
Expected Effect : Detection speed <5ns during active window; average power 12mW vs 100mW continuous operation; efficiency penalty reduced from 1.2% to 0.15% in 20W USB-PD; timing accuracy ±2ns maintained across -40°C to 125°C
Risk Control :
  • Prediction algorithm fails under sudden load transients >50% step
  • bias injection overshoot causes comparator offset drift
  • window timing jitter accumulates over temperature cycles

Problem Direction 3 :

ImproveTurn-off control reliability
VS
ConstraintControl circuit complexity

Inspiration 1 : Cross-domain reference

Application Principle: #35 Parameter changes
Cross-domain applicability Assess applicability
Ultrasonic sealing algorithm with temperature control
Innovative Solution Refine solution

Temperature-compensated passive component timing control for synchronous rectification

Passive thermal compensation via matched components eliminates active circuits
How to solve :
  • Select gate driver IC and high-speed comparator with matched positive temperature coefficients (+15ppm/°C) that naturally cancel GaN FET threshold drift (-18ppm/°C) across -40°C to 125°C, eliminating temperature sensors and compensation logic
  • Use ceramic capacitor timing network (NP0 dielectric, ±30ppm/°C) with resistor divider (thin-film, ±25ppm/°C) for voltage-dependent delay adjustment that tracks input voltage variation 90-265VAC without active monitoring
  • Factory calibrate turn-off delay at three test points (25°C/120VAC, -40°C/90VAC, 125°C/265VAC) by laser-trimming a single polysilicon resistor (0.1Ω resolution) to ±1ns tolerance, verified via oscilloscope measurement of drain-source voltage zero-crossing alignment
  • quality acceptance requires efficiency ≥94.5% across full operating range with no active compensation
Expected Effect : Circuit complexity -65% (removes 18-25 components including MCU/ADC/temp sensor); timing accuracy ±2.5ns across full range; efficiency 94.8% vs 93.2% baseline
Risk Control :
  • Component temperature coefficient mismatch ±10ppm/°C
  • laser trimming yield <85% at ±1ns spec
  • long-term drift exceeds ±3ns after 5000hrs

Problem Direction 4 :

ImproveTurn-off control reliability
VS
ConstraintControl circuit power consumption

Inspiration 1 : Cross-domain reference

Application Principle: #32 Color changes
Cross-domain applicability Assess applicability
Copper powder and its manufacturing method, as well as the manufacturing method of three-dimensional objects
Innovative Solution Refine solution

Oxide-enhanced magnetic flux sensing for zero-power synchronous rectification timing

Replace electrical voltage sensing with magnetic flux detection using inherently low-power method
How to solve :
  • Deploy magnetoresistive sensor (AMR/GMR) at transformer core air gap to detect flux zero-crossing
  • sensor operates at 5-10mW continuous power, 50× lower than high-speed voltage comparators
  • Apply controlled surface oxidation to sensor magnetic layer (oxygen plasma 30sec, 0.2Torr) forming 2-5nm Fe₃O₄ barrier, increasing sensitivity 40% via exchange bias effect while maintaining <8mW power
  • Sensor output triggers turn-off within 4ns of flux zero-crossing
  • factory calibration maps flux-to-voltage timing offset across -40°C to 125°C (stored in 64-byte lookup table), eliminating real-time temperature compensation circuits
Expected Effect : Control power 8mW vs 100mW baseline; timing accuracy ±3ns across full range; efficiency gain 0.8-1.2%
Risk Control :
  • oxide layer thickness uniformity ±0.5nm required
  • sensor-to-core air gap tolerance ±0.1mm
  • magnetic interference from adjacent components

Problem Direction 5 :

ImproveVoltage transition detection speed
VS
ConstraintMust not deteriorate

Inspiration 1 : Cross-domain reference

Application Principle: #10 Preliminary action
Cross-domain applicability Assess applicability
Ripple monitoring circuit
Innovative Solution Refine solution

Dual-phase detection with predictive blanking window for GaN synchronous rectification

Separate detection into prediction and execution phases across switching cycles
How to solve :
  • Deploy slow filtered detector (50ns bandwidth, 100kHz low-pass) during first 80% of switching cycle to predict zero-crossing timing with ±10ns accuracy, immune to ringing noise
  • Activate fast unfiltered comparator (2ns propagation delay, 300MHz bandwidth) only during final 20ns window centered on predicted crossing, when drain-source voltage oscillations decay below 0.3V threshold
  • Implement adaptive blanking logic using previous-cycle ringing amplitude measurement—if ringing >1.5V, extend blanking to 30ns
  • if <0.5V, reduce to 15ns—updating prediction every 10 cycles to track drift across load/temperature variations
Expected Effect : Detection speed <3ns during clean window; false trigger rate <0.01%; control power 25mW vs 100mW continuous fast detection
Risk Control :
  • prediction accuracy degradation under rapid load transients
  • blanking window miscalibration causing missed zero-crossings
  • comparator mode-switching glitches introducing timing jitter
Patsnap Eureka Solution