Effect of resistor tolerance on the performance of resistor network—An application of the statistical design of experiment

Abstract

Resistors come with different tolerances, but they are generally not seriously considered in circuit simulation as only the exact resistances are used. In actual implementation of a resistor network, the tolerance of the resistors used can affect the performances of the network, especially when the network consists of large number of resistors. This work demonstrates the use of the statistical design of experiment methods combined with circuit simulator to obtain the optimum value of the tolerances for different resistors in a resistor network for an application used in proton therapy. The discretized positioning circuit consisting of 86 resistors of four different values (R1 = 1KΩ, R2 = 600 Ω, R3 = 300 Ω, and R4 = 100 Ω) is analyzed for three tolerances of 0.05%, 0.1%, and 0.5%, respectively, and maximum variation in the output signal is observed for 0.5% tolerance. Resistors with values of R1 and R4 are dominant in numbers (56 with R1 and 22 with R4 out of 86 resistors) in the network, and hence, their effect is observed to be significant on the output as expected. Significant interactions between resistors are also observed.