In this letter, we study the energy saving problem by switching off some macro Base Stations (BSs) under downlink coverage and uplink power constraints. Firstly, we derive the expressions of average coverage probability and users' power consumption with stochastic geometry tools. Then, based on the closed-form results, power adjustment and introducing micro BSs schemes are designed to keep the original coverage performance. In addition, we formulate a BS energy consumption minimization problem considering the users' power constraints, and jointly determine the optimal proportion of sleep macro BSs, transmission power of active macro BSs, and density of introduced micro BSs. Numerical results show that with our schemes, BSs' energy consumption can be significantly reduced while guaranteeing the downlink coverage and user power consumption performance.