Power Analysis for SURVEY data

#1
I am doing secondary data analysis on survey data that has a clustered, complex random design. I am using STATA 9.0. For the Power analysis do I use the design or model based approach? Or, can power analysis even be done with cluster sampling? Epi Info seems to imply power analysis cannot be done.
I have done the analysis on my research questions and have gotten very low powers (1.0).
Kathy