Abstract: With the gradual opening of low-altitude airspace， activities involving low， slow， and small （LSS） targets， represented by unmanned aerial vehicles （UAVs）， general aviation aircraft， and large bird flocks， are becoming increasingly frequent. Characterized by low flight altitude， slow speed， and small radar cross section （RCS）， LSS targets easily merge with complex ground and meteorological clutter， often making traditional surveillance methods inadequate for efficient detection. To address this problem， this study proposes a hardware design scheme for wideband and narrowband cooperative detection radar based on a digital dechirp architecture. Utilizing board-level hardware， including the radio frequency system-on-chip （RFSoC）， optical fiber transmission cards， and field-programmable gate arrays （FPGAs）， the system establishes a cooperative operating mode that combines wide-area broad search （using broad beams and long-time integration） with high-resolution fine detection （using narrow beams and wideband signals）. Meanwhile， for the dechirp radar architecture， a multi-channel calibration method combining hardware and software is proposed. To address frequency-domain inconsistencies in the wideband signal channel response of the radar analog front-end， the method first performs time-domain envelope alignment for single-channel signals within the dechirp domain. This process maps beat-frequency signals corresponding to different echo ranges to a unified time-domain reference， enabling subsequent amplitude and phase equalization for each channel. Second， to address sub-sample-level time mismatch in the multi-channel receiving link caused by analog front-end inconsistencies， radio frequency data converter （RFDC） sampling characteristics， and system clock distribution， fractional delay filters are introduced at the RFSoC stage to achieve hardware-level calibration of signal frequency consistency after dechirping. Subsequently， residual amplitude and initial phase errors of multi-channel signals are finely compensated on the graphics processing unit （GPU）， achieving sub-sample-level multi-channel consistency calibration that meets the precision requirements of the wideband dechirp architecture. Digital hardware implementation is carried out for digital beamforming of dechirp wideband signals. Finally， a complete system comprising the radar front-end， digital boards， optical fiber transmission cards， server display control， and GPU data processing modules is implemented. Experimental results show that the system provides wide-area staring capability in narrowband mode and achieves a range resolution of approximately 0.5 m in wideband mode， validating the engineering feasibility of the proposed cooperative detection scheme.