Synchro auto-test system is used to measure the rotation angle of angular displacement sensor synchro. The angle measurement error is the main performance parameter of the system. Based on neural network, this paper analyses and compensates the angle measurement error of the automatic test system of the synchro. In this paper, the principle of angle measurement of such a testing system is introduced first. Then, the amplitude mismatch, phase unorthogonality, phase shift between the input signal of SDC conversion chip in the system. The input noise of SDC caused by environmental noise are analyzed respectively, and the error expression is obtained. In this paper, the error is reduced and compensated from the perspective of hardware and software respectively. In the hardware part, the filter phase shift circuit is designed to eliminate high frequency noise. In the software part, The error compensation measure based on BP neural network is proposed. By analysis the results of experiment, the angle measurement accuracy of synchro automatic measurement system using 16Bit converter is 0.3′. After error compensation, the error of the automatic testing system of synchro is less than ±6′, which is met high-precision measurement requirements of the synchro dynamic angle measurement system.