Problems encountered in the simulation of ultra-low-power energy harvester

Hi all,

I am a PhD candidate and I've stuck on this problem for a few months already. I've been simulating ultra-low-power energy harvester. Basically, it is a rectifier where the input is a very small AC signal and this very small AC signal is rectified, and the output of the rectifier is a DC signal. Typical architecture of such a low power rectifier is variations of Dickson charge pump, such as the topologies in the following paper,

"A 90-nm CMOS Threshold-Compensated RF Energy Harvester", JSSC, Sept 2011

I have stimulated several of these (I used transient simulation), and found many problems with such circuits. One common problem is that the output of the rectifier typically starts from 0, gradually increases as the ac input is being rectifier, eventually reaches a steady state DC output voltage. When I simulated them(transient simulation), the output droops after it reaches a certain output DC value. This happens to a few of the topologies published in JSSC, it's unlikely for all of them to be wrong, so I suspect that there's probably something wrong with my simulation.

In my transient simulation, in the ADE environment, choosing analysis window, I used "conservative" as accuracy defaults, and there's also an "options" button in the transient simulation window. I clicked on it, a window called "Transient Options" opened up. And, there are a few tabs, "Time Step", "Algorithm", "State File", "Output", "EM/IR Output", "Misc". I left all of them blank. Shall I try to set some of them, will that make a difference??? I thought by setting accuracy defaults to "conservative", all those things are already set by cadence automatically.

I would greatly appreciate if anyone could give hint/explanation as to what is going on.

Thank you very much

Menghan