Digital Control Concepts For Power Supply Engineers. Robert V. White Staff Engineer Worldwide Technology Group

Transcription

1 Digital Control Concepts For Power Supply Engineers Robert V. White Staff Engineer Worldwide Technology Group

2 Presentation Overview Control Techniques Traditional Analog Traditional Digital Modern Digital Examples Of Digital Advantages Dead Time Optimization Automatic Compensation Tuning Digital Control Digital Power Management

3 Analog PWM

4 Analog PWM Control Loop Characteristics Voltage Scaling Switch Timing Control Reference Error Determination

5 Classical Digital PWM

6 Classical Digital PWM Typical DSP/Processor => Lots Of MIPs => $$$ Resolution+ Accuracy + Speed = $$$ Counter Approach + High Precision = Very Fast Clock = $$$

7 Re-Thinking Digital Control For Power

8 Step 1: Digitize Only The Error

9 Step 1: Digitize Only The Error How Many Bits Are Needed? 3? 5? 7?

10 Thinking About Error Bins

11 Non-Linear Error Bins +10 DELTA Opportunities For: Lower Cost Digitization? Improved Nonlinear Control? VOUT NOMINAL +3 ERROR BIN +2 ERROR BIN +1 ERROR BIN ZERO ERROR BIN +5 DELTA +2.0 DELTA +0.5 DELTA 1 ERROR BIN 2.0 DELTA

12 Step 2: Simplify Calculation Engine Lookup Table, Fixed Configuration Filters, State Machines $10 DSP

13 Step 3: Low Cost Digital PWM Example: Combination Counter And Delay Line

14 The Point? Re-Thinking Digital Control For Power Conversion Is Lowering Cost And Simplifying Design

15 Avoiding Limit Cycling COARSE RESOLUTION DPWM POSSIBLE OUTPUT VOLTAGES OUTPUT VOLTAGE D/A ERROR BINS +3 ERROR BIN +2 ERROR BIN +1 ERROR BIN ZERO ERROR BIN 1 ERROR BIN 2 ERROR BIN 3 ERROR BIN

16 Avoiding Limit Cycling N+1 DUTY CYCLE N PWM VOUT NOMINAL

17 Avoiding Limit Cycling FINE RESOLUTION DPWM POSSIBLE OUTPUT VOLTAGES OUTPUT VOLTAGE D/A ERROR BINS +3 ERROR BIN +2 ERROR BIN +1 ERROR BIN ZERO ERROR BIN 1 ERROR BIN 2 ERROR BIN 3 ERROR BIN

18 Avoiding Limit Cycling FINE RESOLUTION DPWM POSSIBLE OUTPUT VOLTAGES OUTPUT VOLTAGE D/A ERROR BINS +3 ERROR BIN Rule: Minimum ΔVOUT < Error Bin Size +2 ERROR BIN +1 ERROR BIN For A Buck Regulator: ZERO ERROR BIN Maximum Setpoint Error 1 ERROR = ±0.1% BIN => Duty Cycle Resolution < 0.2% => 2 ERROR BIN 9 Bits Time Resolution Per Cycle 3 ERROR BIN

19 Sampling the Output Once Per Cycle Can Give Good Information On Average Value Computation Time! Oversampling Is Not Necessarily Helpful

20 Everyone Is Terrified Of Z Transforms Vc z = vc n z [ ] [ ] n n= Programming main(){ printf( hello,world\n ); } REALITY! Converter Designers Will Work Through Computer Based GUI Interfaces

21 Winning Application #1: Efficiency Optimization Digital Control Offers Opportunities To Optimize Operation To Minimize Losses Example: Buck Converter Dead Time (CoPEC) Start With Excessive Deadtime Slowly Minimize While Watching Duty Cycle Minimum Duty Cycle => Minimum Losses Example: On Bus Voltage Monitor Load Of POLs Powering A Board Light Load? Lower Bus Voltage Heavy Load? Increase Bus Voltage

22 Winning Application #2: Self Tuning Control Loops A Converter With A Digital Control Loop Can Act As Its Own Network Analyzer Based On Measurement Of The System, A Device Can Optimize Its Loop Response POLs Compensate For Actual Capacitive Loading Adjust As Components Age Front End Power Supplies Compensation For Initial Conditions Compensate For Configuration Changes Over Time

23 Winning Application #3: Digital Power Management Digital Control Digital Power Management But They Marry Very Well! Passing Digital Values From User s GUI Directly To A Digital Control Loop Will Simplify IC Design And Lower Cost Improved Fault Management Digital Control Enables Graceful Transition Between Normal And Abnormal Operating Modes

24 References For More Information Colorado Power Electronics Center (CoPEC) In My Opinion, Leading The Way In Digital Control And Silicon Integration Research Useful To Industry Much Of This Presentation Based On Their Work Seth Sanders/UC Berkeley Limit Cycling Paper Is Classic

25