Compact low-power calibration mini-DACs for neural arrays with programmable weights

Area of Science:

• Electrical Engineering
• Neuroscience Engineering

Background:

• Compact, low-power mini digital-to-analog converters (mini-DACs) are crucial for large neural arrays requiring programmable weights.
• Weak inversion transistor operation is necessary for low power consumption in large-scale integrated circuits but suffers from poor device matching.

Purpose of the Study:

• To assess the viability of mini-DACs in large neural arrays.
• To investigate methods for overcoming transistor mismatch limitations in weak inversion operation.
• To enhance the precision of mini-DACs for applications in neural processing circuits.

Main Methods:

• Utilizing transistors biased in weak inversion to achieve low currents and power consumption.
• Implementing a calibration technique by pairing mini-DACs to compensate for inter-device mismatch.
• Developing a design methodology for achieving target precision in calibrated mini-DACs.

Main Results:

• Initial fabrication of 50 mini-DACs showed an effective precision of 47% (1.1 bits) due to transistor mismatch.
• The proposed calibration method improved precision by a factor of 10, reaching 4.8% (4.4 bits).
• Simulation results demonstrated the successful calibration of an array of digitally controlled integrate-and-fire neurons.

Conclusions:

• Calibrated mini-DACs are viable for low-resolution applications, particularly in massive neural processing circuits.
• The demonstrated calibration technique effectively compensates for transistor mismatch, significantly enhancing DAC precision.
• A practical design methodology is provided for creating precise, calibrated mini-DACs for neural engineering applications.