Self-supervised heterogeneous graph neural networks have demonstrated remarkable effectiveness in addressing the challenge of limited labeled data. However, current contrastive learning methods face limitations in exploiting neighborhood information for each node. Some methods only utilize the local information of the target node, ignoring useful signals from deeper neighborhoods. On the other hand, certain methods fail to consider the differences among neighbors at various distances. In this paper, we propose HGNN-DB, a self-supervised heterogeneous graph neural network based on depth and breadth neighborhood encoding. Specifically, HGNN-DB aims to effectively capture features from deep and broad neighborhoods. We introduce a deep neighborhood encoder using a distance-weighted strategy to capture deep features of target nodes. Additionally, we utilize a single-layer graph convolutional network for the broad neighborhood encoder to aggregate broad features of target nodes. We furthermore employ a collaborative contrastive algorithm to learn the complementarity and potential invariance between the two views of neighborhood information. The results demonstrate that HGNN-DB performs better than other state-of-the-art methods on various tasks.