1 Digitally Controlled Converter with Dynamic Change of Control Law and Power Throughput Carsten Nesgaard Michael A. E. Andersen Nils Nielsen Technical University of Denmark in collaboration with

2Outline Power system specifications The microcontroller Control algorithm and efficiency Analytical redundancy concept Reliability Experimental verification Further work Conclusion

3 Power system specifications Simple buck topology with measurements of input voltage, input current, output voltage and output current Microcontroller for converter control and thermal monitoring

4 The microcontroller 8-bit RISC PIC16F877 microcontroller from Microchip Core features:Uses: 8K 14-bit word flash memory 256 E 2 PROM data memory 10-bit PWM module 8 channel 10-bit A/D converter Single cycle operations 20 MHz clock frequency Algorithm and look-up table Converter control Execution speed

5 Control algorithm and efficiency Simple buck topology with measurements of : Thermal monitoring PWM control law for power throughput above 1.85 W PS control law for power throughput below 1.85 W Look-up table control when operated within specifications Input voltage Input current Output voltage Output current

6 Control algorithm and efficiency Software data flow diagram: Interrupt routine responsible for correct converter control Main loop responsible for temperature measurement, cal- culation of correct control law and type of calculation method (look-up or real-time)

7 Analytical redundancy concept Examples: Converter efficiency is related to system temperature Output voltage is related to the inductor current Result: Continuous converter operation (at a deteriorated level) Analytical redundancy is the concept of deducing a set of variables able to accurately describe the actual system behavior

8 No heatsink The above graph is used to determine converter state h Analytical redundancy concept Minimizing the risk of shutting down a well- functioning converter In the event of a fault in PWM mode:

9 Analytical redundancy concept The system is only partially fault tolerant due to: Resilience towards faults described by the mathematical system Single converter system – one path from input to output Further improves in system reliability require hardware redundancy Example:

10 Analytical redundancy concept Further advantages of analytical redundancy: Fault indicator in hardware redundant systems  Continuously comparing theoretical system constraints with actual system behavior  Enables the system to respond intelligently to unusual system behavior  Increasing the overall system fault resilience

11Reliability Temperature distribution used for reliability assessment: Probability of survival as a function of time: Reliability data found in MIL-217 (assumes a constant failure rate)

12 Failure rates for the two configurations: Analog configuration Digital configuration Failure rate in FIT From a reliability point of view: At temperatures below 120  C an analog controller is preferable At temperatures above 120  C a digital controller is preferableReliability

13 Survivability R(t) for 10,000 hours: Analog configuration Digital configuration The digital configuration is 36 times more likely to fail within 10,000 hours than its analog counterpart.Reliability

14 Converter efficiency: The arrows indicate direction of change in control law The hysteresis loop prevents oscillatory converter behavior when operated close to the optimum point of transition. Experimental verification

15 PWM: PS: Experimental verification Gate-Source voltage Output voltage

16 Inductor current Input voltage PWM: PS: Experimental verification

17 Further work Q000Q L00QI C000V0 40DC Q000Q T 70000P P P0000 Graph theoretical approach is used for thorough system analysis Columns identify the lines interconnecting the individual blocks Line arrows indicate direction of power or data flow Block level buck converter

18Conclusion A buck converter controlled by a low-cost PIC microcontroller has been presented. The system use analytical redundancy, change in control law and thermal monitoring for increased reliability. Also, an introduction to the proposed techniques has been given supported by calculations concerning the pros and cons of the individual techniques. Finally, a set of measurements has verified that the algorithm is indeed capable of performing the required tasks within the timing limitations of the microcontroller.