Self-calibration system and method based on neural network

Abstract

The invention discloses a self-calibration system and method based on a neural network. A self-check unit, an adjustable element array and a calculation unit are arranged, the self-check unit measures a low-frequency measurement value M associated with a to-be-calibrated circuit, and the adjustable element array inputs a control word of the calculation unit into the to-be-calibrated circuit; the calculation unit performs calibration based on the self-check unit and the adjustable element array. The electrical performance parameters of the to-be-calibrated circuit can be predicted by training the neural network, the control words of the corresponding calculation units under the expected electrical performance parameters are calculated, and the adjustable element values accessed to the adjustable element array are controlled through the control words. The to-be-calibrated circuit is calibrated to an expected electrical performance parameter range, so that the test cost and complexity of analog, radio frequency and other circuits are effectively reduced, and the calibration efficiency is improved.

Description

technical field [0001] The invention relates to the field of electronic circuits, in particular to a neural network-based self-calibration system and method. Background technique [0002] With the advancement of process nodes in semiconductor technology, the cut-off frequency of CMOS is getting higher and higher, so that the circuit of CMOS process can work in the frequency band of radio frequency / millimeter wave, and the digital, analog and radio frequency circuits can be integrated on one chip. System on Chip (SOC). Analog and RF circuits are easily affected by process PVT and deviate from the expected working state at the time of design. When coexisting with a large number of rapidly changing digital logic on the same substrate, they will be disturbed by additional noise. Designers must consider the robustness of the circuit when designing the circuit, because as long as there is a performance deviation of a radio frequency module in the SOC, the chip performance may not...

AI Technical Summary

Problems solved by technology

The entire calibration process takes a certain amount of time from the algorithm operation to the change of the hardware tuning knob and the change of the sensor output of the self-test unit. It takes a long time to find the optimal setting value, the test cost is high, and the calibration efficiency is low.

Images