Magnetocardiography (MCG) source imaging allows high-precision three-dimensional localisation of regions with abnormal cardiac activity, playing a crucial role in disease diagnosis. In practical measurements, however, four key types of error in sensor arrays, i.e., crosstalk, gain error, angular error, and position error, can distort magnetic signals or the lead field matrix, thereby compromising imaging performance. To date, their impact on MCG source imaging has not been systematically assessed. In this study, we establish an evaluation framework and systematically investigate the effects of these errors for the first time. A comprehensive evaluation is conducted across three key aspects: reconstruction accuracy, spatial resolution, and the localisation accuracy of correlated sources. Our results indicate that all four error types degrade imaging performance to varying degrees. The detrimental influence of errors on reconstruction accuracy and spatial resolution becomes more severe as the signal-to-noise ratio increases. For the localisation of correlated sources, increases in the number of sources and the inter-source coupling strength further exacerbate the interference caused by array errors. Based on the configuration in this work, we derive conservative safe operating limits: crosstalk <1%, gain error <3%, angular error <1°, and position error <2 mm. Therefore, precise calibration of sensor parameters before MCG measurements is essential to ensure imaging performance. This study highlights that systematically evaluating array errors is critical for designing calibration strategies and supporting the clinical application of MCG systems.