The quality of phasor data from micro-Phasor Measurement Units (μPMUs) is critical for smart grid applications. It plays a key role in various aspects of power system management and is essential for the transition to a smarter and more sustainable grid. Recent studies imply that despite having a high level of monitoring features and accurate algorithms, μPMUs are vulnerable to errors in the measurements. Traditional methods for error detection in μPMUs typically rely on direct analysis of voltage signals. While effective to some extent, these methods can struggle with the complex and dynamic nature of power system measurements, especially under varying load conditions and in the presence of noise. To address these challenges, this paper presents a novel approach for error detection in μPMU voltage measurements using a combination of continuous wavelet transform (CWT) and a convolutional neural network (CNN). The proposed detection approach is applied on Assiut university distribution grid sub-feeder. A set of evaluation metrics such as accuracy, recall, precision, and F1 score were used to compare the error detection performance of the proposed CNN model with conventional machine learning (ML) algorithms. The results show that the proposed CNN model outperforms the conventional ML algorithms for detecting errors in μPMU voltage measurements under different load conditions.