The results of substrate cross talk noise investigations using s-parameter measurement on device level are presented. Frequency domain analysis shows that the noise in lightly doped p-type substrate is strongly dependent on layout geometry. Experimental approaches using guard ring (GR), isolation bar (IB), diffusion width (W), physical separation ( S ) , and different dc bias condition are used to reduce the substrate: noise. It is shown in this study that double GR structure with IB implementation can be very effective in shielding the substrate noise.

No abstract