I like the first approach. Honestly I'm very rusty in my control theory, and where I am getting stuck is deriving a transfer function when I don't hold the load resistance constant. I'm not concerned about the power supply itself. At the frequencies of concern, I can assume that it is perfect. I am more concerned about the transient effects of the low pass filter of the line (Rt & C). If replace the load resistance with just a load current, I get (s-1/(RtC))Vc = 1/(RtC) Vs - I, but then I'm stuck on how to reduce this further to get either G(s) or H(s). Note Y = Vc, U = Vs, D = I.