Design Of Process-Compensated Low Voltage Low Power CMOS Voltage Reference

Low-power Internet of Things(IoT)systems are in high demand due to the widespread use of applications such as biomedical devices and environmental sensors.As a ubiquitous and essential block in various circuits and systems,the voltage reference should provide a biasing voltage that is almost insensitive to the process,voltage,and temperature(PVT)variations.To be more efficient in low-power IoT systems,the always-on voltage reference should be able to operate at a low supply voltage while consuming minimal power.Conventional bandgap references(BGRs),utilizing the BJT’s base-emitter voltage(VBE),can provide a PVT-insensitive voltage at the expense of high supply voltage and high power consumption.To achieve low voltage and low power,CMOS voltage references that operate the transistors in the subthreshold region are commonly selected.However,due to the process variation of the threshold voltages(Vth),the process stability of these circuits is compromised.Recent advancements have been made to reduce the power consumption of BGRs;though,the minimum supply voltage remains relatively large(typically higher than 1 V)due to the large value of VBE(～0.7 V).This paper proposed a PVT-insensitive low-voltage low-power CMOS voltage reference.By utilizing the width-induced Vth modulation and self-body-biasing techniques,two Vth-based voltages with opposite process and temperature characteristics are generated,achieving both process and temperature compensation.The core circuit of the proposed reference consists of only three subthreshold-biased PMOS transistors,which significantly reduces the minimum supply voltage.The voltage reference was implemented in a standard 180-nm CMOS process.The power consumption at room temperature is 500 pW,and the active area is 0.0029 mm2.As measured across 16 chips,it provides a mean reference voltage of 288 mV while achieving 0.57%within-wafer inaccuracy,0.23%/V line sensitivity,and 90 ppm/℃average temperature coefficient after batch trimming.This work achieves a competitive overall PVT relative inaccuracy of 2.23%while reducing the minimum supply voltage to an impressive 0.5 V,making it well-suited for low-power IoT systems.