RIT researchers create adaptive computer chip design that saves power in everyday electronic devices

Researchers at RIT have developed a new computer chip design approach that allows electronic systems to automatically adapt to real-world conditions, improving how devices manage power in everyday use.

The results show improvements in how chips are designed and how power is managed for always-on electronic systems from sophisticated biomedical wearables to smart devices such as locks, thermostats, and appliances.

The team at RIT’s RF Analog Mixed Signal Laboratory (RAMLab) introduced an adaptive analog design approach for semiconductor chips that allows power delivery circuits to dynamically respond to real-world variability, enabling more reliable sensing and energy efficient electronic systems. The work addresses key aspects of power management, including how circuits handle changing loads, maintain stable voltage, and suppress noise—disturbances that distort or degrade system transmissions.

Their work improves ultra-low power low-dropout, or LDO, regulator design—a key function in power management of computer chips. The approach was tested on a custom chip designed in-house demonstrating energy efficiency during both low power “sleep” modes and active operation.

‌

Carlos Ortiz/RIT

Daniel Zeznick was part of a graduate and undergraduate student team that helped develop a novel approach regarding power functions and how they improve today’s electronic devices consume and save power.

“Our work is motivated by the need for circuits that can adapt to these changing conditions in real time,” said Teju Das, associate professor in the Electrical and Microelectronic Engineering Department in the Kate Gleason College of Engineering.

An expert in analog, RF and mixed signal integrated circuits, Das, the director of the RAMLab, described the work done by the team over the past 18 months. Results were published in IEEE Transactions on Circuits and Systems, and it received the Best Paper award at the IEEE International NEWCAS Conference in Paris.

“At a high level, the circuit we developed continuously monitors its operating environment, such as changes in load, noise, and signal conditions, and dynamically adjusts its behavior to maintain stable and efficient operation,” said Daniel Zeznick, one of the paper’s authors and a teaching assistant with the RAMLab. He is also a silicon design engineer with AMD, a semiconductor company based in Rochester.

Das’ research also highlights how students within the microelectronic and electrical engineering programs are acquiring the in-demand skills to leverage the benefits of both analog and digital technologies to produce today’s integrated circuits.

“In our field, those with analog and mixed signal integrated circuit skills are in high demand,” said Das. “Analog design in modern electronics pushes the limits of physics and having this background will be essential to solving performance issues.”

Modern electronic systems, especially in areas like wearable health monitors, biosensing systems, and Internet of Things sensors, operate under highly variable conditions. However, most power circuits today are designed assuming relatively fixed operating conditions or require external control, which can lead to inefficiencies, instability, or degraded performance in real-world use.

“Our work introduces a fully analog, self-adaptive approach in which the circuit continuously adjusts its behavior based on its operating environment, without requiring external control or discrete events. Rather than optimizing for a single operating point, this enables reliable performance across a wide range of conditions,” said Das.

This shift from static or externally controlled operation to continuous, low-overhead adaptation allows for more efficient operation, particularly in energy-constrained systems where even idle power consumption matters. The progression of these concepts highlights the nature of analog and mixed-signal chip research, where developing and validating new ideas in hardware requires sustained effort across multiple stages of design, fabrication, and measurement. This work was supported in part by Cirrus Logic Inc., with additional support for chip fabrication from GlobalFoundries through their University Partnership Program.