Most servo control systems generally adopt incremental optical encoders for speed detection when considering cost and performance requirements. For a fixed sampling period, this kind of encoder along with the generally used so-called M method, may degrade the response or even cause the system to become unstable in a low-speed operating region because of the resulting speed detection delay. In this article, a reference model improves low-speed responses; parameter identification by recursive least square error algorithm overcomes the problem of parameter variations and an adaptive proportional-integral control strategy based on the parameter identification results further justifies the proposed method. A digital signal processor based permanent magnet synchronous motor drive will be used to carry out the experimental results, which show the effectiveness of the proposed method.