Alternative proposals for reducing substrate noise

Abstract

The focus up to now has been on studying guard rings and substrate bias optimization for reducing noise. We will now analyze the viability of three alternative measures for reducing noise, which may not be advisable for general use due to their cost or drawbacks, but which may be useful in particular cases. First, the use of buried layer in providing a noise return path will be examined. The conditions necessary for maximum efficacy, and the increase in process complexity will be evaluated. Next, we will consider the possibility of producing complementary peaks so they cancel each other out. This technique involves duplicating some circuits — adding dummy circuitry- but even so it would be advisable in the event of noise being produced by a few localized elements. Lastly, we study the viability of using active guards. The technique is similar to the canceling method, but in this case noisy elements are not duplicated, adding instead a circuit which senses noise and creates compensating noise, thus annulling the first one, for instance through negative feedback.