Connectivity of large-scale wireless networks has received considerable attention in the past several years. Different from traditional wireless networks, in Cognitive Radio Ad-hoc Networks (CRAHNs), primary users have spectrum access priority of the licensed bands over secondary users. Therefore, the connectivity of the secondary network is affected by not only the density and transmission power of secondary users, but also the activities of primary users. In addition, the number of licensed bands also has impact on the connectivity of CRAHNs. To capture the dynamic characteristics of opportunistic spectrum access, we introduce the Cognitive Radio Graph Model (CRGM) which takes into account the impact of the number of channels and the activities of primary users. Furthermore, we combine the CRGM with continuum percolation model to study the connectivity in the secondary network. We prove that secondary users can form the percolated network when the density of primary users is below the critical density. Then, the upper bound of the critical density of the primary users in the percolated CRAHNs is derived. Simulation results show that both the number of channels and the activities of primary users greatly impact the connectivity of CRAHNs.