I'm using stability analysis and pole-zero analysis in Eldo but I imagine this same question could apply to Spectre;

Stability analysis allows you to plot the loop gain and phase without breaking the loop by adding an inline source and the simulator uses the Middlebrook method. I still seem to need to break the loop though for the pole-zero analysis! The broken-loop bode plots from the pole-zero match the bode plots from the stability analysis as I'm modelling the real circuit. I would not be so confident though when non ideal input/output impedances are used in the real cct. Is there a way of using Eldo pole-zero analysis without breaking the loop?