Self-Aging-Prognostic GaN-Based Switching Power Converter Using TJ -Independent Online Condition Monitoring and Proactive Temperature Frequency Scaling

Abstract

To mitigate unique aging and failure challenges encountered in GaN technology, this article develops the techniques that reinforce the reliability of GaN power circuits. In particular, an online condition monitoring scheme is developed to prognose the current collapse induced aging and failures in GaN power switches, using the devices' dynamic on-resistance rDS_ON, as the real-time precursor. As rDS_ON is temperature-dependent in general, a gate-leakage based junction temperature TJ sensor is developed to assist in removing temperature effect from rDS_ON reading, achieving TJ-independent condition monitoring. To improve the device longevity, a proactive temperature frequency scaling scheme is implemented to balance the power circuit performance and reliability. To validate the design concepts, a power converter IC prototype was designed and fabricated using an 180 nm high-voltage CMOS process. With all e-mode GaN power switches, it operates at 10 MHz switching frequency with a flexible input supply voltage ranging from 5 to 40 V and delivers a maximum power of 6 W. Experimental results demonstrate that the TJ-independent online condition monitoring reduces the false monitoring rate by 19 times over a temperature range from 0 to 125 °C. Meanwhile, the temperature frequency scaling engine reduces the TJ of the GaN switches by up to 16 °C, while striving for an optimal operation between performance and thermal stress.