Design of Programmable Voltage Source in Micro-Current Detection System Based on Artificial Neural Network

The cubic images can be achieved by measuring the tunneling current between the scanning tip and the object surface. This current is generally in the scale of nano ampere or even smaller, therefore it is submerged easily in the noise signals. A programming voltage source applying the back-propagation neural network is designed in this paper to generate a precision voltage for the potentiostatic instrument. A lumped function including the voltage error, a constraint condition and an integrated variable is developed as the objective function of network, and the tuning law is derived. Experiments are conducted to validate the effectiveness of the proposed method. Results show that a satisfied output voltage is achieved, and the current errors expressed as the voltage are dramatically reduced compared with the traditional digital-to-analog converter.