This article considers the differentially private consensus problem of discrete-time second-order multiagent systems with partially measurable states and limited communication channel capacity, where only the integer-value information of agents can be transmitted. To reduce the potential risk of state information disclosure in digital communication, a differentially private consensus algorithm via dynamic encoding-decoding is proposed for the second-order multiagent system to make agents achieve mean-square consensus by transmitting quantized integer values with privacy protection. To deal with the uncertainty of the quantizer saturation, the statistical analysis is given for the boundedness of the input of quantizers. It is shown that the expectation of the minimum memory capacity of quantizers is 2 bits. Finally, some simulation results are given to visualize our conclusions.