Fast, Self-stabilizing, Hysteretic Boost DC-DC Converter Neeraj Keskar

1. Fast, Self-stabilizing, Hysteretic Boost DC-DC Converter Neeraj Keskar Advisor: Prof. Gabriel A. Rincón-Mora Analog and Power IC Design Lab School of Electrical and Computer Engineering Georgia Institute of Technology September 17, 2004

2. Motivation • Significant dependence of converter frequency response on passive components • Tolerances in capacitor ESR, ESL values • Variations in inductor, capacitor values per design • IC solution for frequency compensation required because • Reduction in design time • Reduction in part count • Reduction in board size, cost • Ease of design • Need to have IC solution that will give frequency compensation independent of external components • Hysteretic control provides a way !

3. Hysteretic Buck Converter • Hysteretic control regulates output voltage ripplevO • vO = vR + vC~ vR=DIL.ESR, if ESR is large • Inductor current ripple regulated resembles current-mode control • Single-pole response, inherently stable operation with LC variations • Other benefits: Fastest transient response, simple design

4. Issues with Hysteretic Control in Boost Converters • Inductor current IL and capacitor voltage VOUT ripples out of phase • VOUT variation during tOFF independent of IL • IL information not completely available from VOUT, no current-mode control • Double-pole response, unstable operation with hysteretic control

5. Hysteretic Control in Boost Converters: Proposed Concept • Regulate IL and VOUT through separate hysteretic control loops • Main switch SM for IL control; auxiliary switch SA for VOUT control • Reference VIREF has to be such that IL>ILMIN =

6. Hysteretic Control in Boost Converters: Obtaining VIREF • If VIREF too large, then power loss , efficiency • Hence, IL kept only 5 % above ILMIN duty-cycle-to-voltage demodulator • I2 = 19.I1, therefore VIREF steady-state when DA = = = 5%

7. Hysteretic Control in Boost Converters: Transient Response • HVQ3 sets the hysteresis window for transient response • If VOUT falls outside HVQ3, switch MPC1 turned on • VIREF raised in single step to VIPK IL to support max designed IOUT • Then, VOUT rises in single cycle of switch SA

8. Experimental Results: Steady-state VOUT VGA Steady-state duty cycle DA ~ 5% VGM IL IL free-wheels during on-time of switch SA VOUT = 5 V, IOUT = 0.3 A, VIN = 3.5 V, L = 6.8 mH, C = 76 mF

9. Experimental Results: Load Transient Response Leading current-mode-control boost Proposed hysteretic boost • DVOUT = 292 mV, Dt = 400 ms • DVOUT = 230 mV, Dt = 50 ms • 0.1 A to 1 A load step • Fast, single cycle response for hysteretic boost converter

10. Experimental Results: LC Variation Limits • CMIN for proposed converter 9 times lower than that for leading conventional boost converter

11. Experimental Results: Power Efficiency • Proposed solution has slightly lower high-load efficiency (by 2 % @ 5 W) compared to leading boost converter • At medium and light loads (less that 2.5 W), proposed solution superior (2 % improvement at 0.5 W)

12. Summary • Need integrated DC-DC converter stable with wide variations in L, C • Hysteretic control in buck converters fastest, simple and w/o compensation • Voltage-mode hysteretic control not been used so far in boost converters • Novel dual-loop technique presented to implement hysteretic control in boost converters • Proposed method can be used with good regulation and fast transient response without using any compensation circuit • Efficiency degraded up to 2 % @ 5 W, but improved by 2 % at 0.5 W • Solution can be used towards an increased degree of integration in DC- DC converters – without an external frequency compensation circuit