The traditional Field-Programmable Gate Array (FPGA) implementation method for pulse compression uses Read-Only Memory (ROM) to store the pulse compression coefficients of all linear frequency modulation signals in the waveform table. During the debugging process, if there are adjustments or additions of linear frequency modulation signals with different time widths and band widths, the traditional method requires updating or adding ROM storage pulse compression coefficients and recompiling the program, which affects the debugging cycle seriously. An improved FPGA implementation method for generation pulse compression coefficients of linear frequency modulation signals is proposed. The proposed method supports pulse compression processing of linear frequency modulation signals with arbitrary time width and band width. The FPGA program does not need to be recompiled, reducing the debugging cycle greatly. Finally, the feasibility of the proposed method is verified though joint simulation using Matlab and FPGA.