Diffuse correlation spectroscopy (DCS) has recently been employed for noninvasive acquisition of blood flow information in deep tissues. In this study, we build a new DCS device based on software correlator. Through the intralipid phantom experiment, we obtained the quantitative relationship between the noise of detected optical intensity and the measurement error of Brownian diffusion coefficient. First, the quantitative influences of background noise and correction factor of detectors on actual beta are developed. Second, the intralipid phantom experiments show that the bigger deviation of actual beta to ideal value will result in the more measurement error of Brownian diffusion coefficient. The background noise and correction factor will lead to the deviation of actual beta to ideal value. The best situation to measure Brownian diffusion coefficient is that actual beta should be larger than 0.4 which requests the detected optical intensity by detector larger than 10 times of background noise and smaller than about 3000kcps. Moreover, our result show the collecting ability of optical intensity of probe is related to type of detecting fiber itself other than the lens coupling on it. In conclusion, we report a new DCS device based on the software correlator and show how the noise of detected optical intensity affects the DCS measurement.