A low linear adjustment rate reference circuit based on neural network and its generation method

Abstract

The invention discloses a neural network-based low linear adjustment rate reference circuit and its generation method. This application aims to solve the problem of poor linear adjustment rate of the prior art reference circuit. The temperature-independent reference voltage is generated by the voltage difference between the electrode and the emitter and the circuit topology; this application generates the control voltage required by the LNR adjustment circuit through the artificial neural network, and uses the inherent nonlinear mapping ability of the artificial neural network itself to generate a reference voltage for the power supply. The voltage is a control voltage with a nonlinear function. The voltage increases the grid voltage of the current mirror in the circuit by adjusting the PSR, so that the local voltage at low frequencies basically does not change with the power supply voltage, thereby significantly reducing the LNR of the reference voltage and overcoming the inability to achieve in the prior art. The control voltage of the nonlinear function is used to adjust the difficult problem of the reference voltage; the application is applicable to the related fields of the reference circuit.

Description

technical field [0001] The invention relates to the technical field of analog integrated circuits, in particular to a neural network-based low linear adjustment rate reference circuit and a generation method. Background technique [0002] A reference circuit usually refers to a highly stable circuit used as a voltage source in an integrated circuit. With the increasing scale of integrated circuits. Especially with the development of system integration technology (SOC), it has also become an indispensable basic circuit module in large-scale, VLSI and almost all digital and analog systems. Precise and stable voltage references are required in many integrated circuits and circuit units, such as digital-to-analog converters (DACs), analog-to-digital converters (ADCs), linear regulators, and switching regulators. Therefore, the reference voltage source occupies a very important position in the analog integrated circuit, which directly affects the performance and precision of th...

AI Technical Summary

Problems solved by technology

[0005] The object of the present invention is: aiming at the problem of poor linear adjustment rate of the reference circuit in the prior art, the present invention provides a low linear adjustment rate reference circuit based on neural network and its generation method

Images