Prof. Joseph Shor Develops a Current Sensor Utilizing Differential Current-Controlled Oscillators

Demand for computational performance continues to rise exponentially, while power and thermal budgets impose strict limitations on integrated circuits (ICs). Modern processors employ multiple power domains, each requiring accurate current sensing for efficient power management. Current-sensing techniques such as shunt, Hall-effect, and fluxgate sensors already exist, but none of these is good enough for integration at a scale of tens or hundreds of domains within a single chip. They are either too large, too slow, or too power-hungry. Digital ring-oscillator (RO) sensors have been explored, but they exhibit strong non-linearity, temperature, and supply dependence, which severely limit their accuracy.

Prof. Shor's research, which has earned him the ISF grant, proposes a fundamentally new approach: a differential current-controlled oscillator (CCO) amplifier, which directly converts differential current or voltage into a frequency-domain signal. "Unlike conventional amplifiers, the CCO itself acts as both amplifier and analog-to-frequency converter, providing a highly compact, low-power, and fast solution suitable for power-gate current sensing in advanced processors," he explains.

The proposed sensor targets an accuracy better than 0.2%, conversion time below 2 µs, power consumption under 20 µW, and area smaller than 0.01 mm², with performance significantly surpassing existing state-of-the-art designs. Novel circuit techniques are introduced for non-linearity correction, temperature compensation, and common-mode rejection, including bias trimming, replica oscillators, and chopper stabilization. "Preliminary simulations of a 65nm implementation demonstrate excellent linearity, 60 dB dynamic range, and temperature- and supply-induced variation below 0.02% when using replica compensation," says Prof. Shor.

This project will establish a new class of analog amplifiers based on differential CCOs, enabling fine-grained, energy-efficient current sensing for multi-domain processors, GPUs, and AI accelerators. The outcomes are expected to impact both industrial and academic fields by providing a compact, low-cost, low-power sensing solution for power-management architectures and extending its use to applications such as biomedical monitoring, smart-grid systems, and energy-harvesting circuits.