Low-Power Digital Temperature Compensation Technique for XO Wake-Up Timers

An accurate wake-up timer is crucial for low-power wireless Internet-of-Things (IoT) devices. Although duty cycling minimizes power consumption, synchronization with a base station or other devices requires periodic activation. Inaccuracies in the wake-up timer extend the active duration around the expected synchronization time, leading to increased energy consumption, which is especially critical in high-duty-cycle operations. This article presents a low-power digital compensation method to enhance the accuracy of a crystal oscillator (XO)-based wake-up timer over varying temperatures. The proposed approach dynamically adjusts the digital counter threshold value at each wake-up to compensate for XO frequency shifts caused by temperature changes. Delta-sigma ( $\Delta \Sigma $ ) modulation further reduces quantization noise for fractional time corrections. This work demonstrates the proposed technique on an existing low-power, miniature IoT system incorporating an exponential temperature sensor and employs two key strategies: 1) approximating the compensation curve as linear segments for simpler calculations and 2) replacing division-heavy operations with a successive approximation (SAR) method. These methods reduce the system’s time-shift error from 64.7 to 3.6 ppm using a single-point room-temperature calibration combined with adaptive temperature compensation, with a power overhead of only 7.92% (8.22 nW compensation +5.58 nW XO). Furthermore, a custom digital circuit for compensation minimizes the system power overhead to 5.07% (3.24 nW compensation +5.58 nW XO). Compared to state-of-the-art temperature-compensated XOs (TCXOs) using analog approaches, the proposed design reduces the power consumption by a factor of $2.37\times $ while meeting the required time-error specification.