Coaxial cables are susceptible to faults during long-term operation due to environmental influences and aging effects. Since incipient faults usually cause only slight variations in characteristic impedance and generate weak reflected signals, accurate detection remains challenging. The existing spread spectrum time-domain reflectometry (SSTDR) method provides a wide-ranging capability, but its sensitivity to incipient faults is limited by the chip duration of the modulation sequence. In contrast, stepped-frequency wave reflectometry (SFWR) offers high fault-identification sensitivity but suffers from complex data processing and a limited ranging. To address these problems, this paper proposes a spread spectrum stepped-frequency wave reflectometry (SSFWR)-based cable fault location method combined with time-frequency analysis. First, the variation law of characteristic impedance caused by incipient cable faults is analyzed based on a distributed parameter model and a finite element model of the cable. Subsequently, an SSFWR signal is constructed by integrating spread spectrum modulation with steppedfrequency waveform modulation. Fault reflection information is obtained through cross-correlation processing of the acquired signals, and variational mode decomposition (VMD), short-time Fourier transform (STFT), and reassigned spectrogram (RSP) analysis are further employed to enhance fault features and achieve fault localization. Finally, experimental validations involving open-circuit faults, short-circuit faults, copper shielding layer defects, and conductor core fracture defects are carried out on cross-linked polyethylene coaxial cables. The results show that the proposed method provides a theoretical ranging capability from 3.4 to 3 245.7 m under the designed parameter conditions and can effectively identify both typical hard faults and incipient fault characteristics. Compared with the conventional cross-correlation method, the energy-domain cross-correlation-based time-frequency analysis method can more effectively suppress sidelobe clutter interference, enhance weak fault features, and improve the detection and localization performance of incipient faults, demonstrating the effectiveness of the proposed method for incipient cable fault diagnosis.