Traditional envelope demodulation methods (e.g., Fast Kurtogram, Auto-gram) encounter two key challenges in rotating machinery fault diagnosis: (1.) Uncertainty in the mechanical system’s resonance frequency band, limi-tations in the sensor’s high-frequency response, and signal transmission at-tenuation hinder direct use of the resonance band for fault feature extraction; (2.) Under strong attenuation, signal amplitudes below the analog-to-digital converter (ADC) quantization resolution cause information loss, rendering digital-domain methods (including deep learning) ineffective. To address these, this study proposes a fault diagnosis framework integrating analog-domain signal pre-enhancement and digital-domain processing. A tunable resonant circuit amplifies the target frequency band (e.g., bearing fault characteristic frequencies) at the ADC front-end, increasing the signal attenuation cutoff frequency from 10 kHz to  > 50 kHz. This over-comes quantization resolution limitations and restores high-frequency fault signals. Bearing test bench experiments show that enhanced data exhibits vibration waveform pulse intensity and phase characteristics highly consistent with reference data. The Fast Kurtogram envelope spectrum clearly reveals 1 × and high-order fault frequency components, with significantly improved fault feature recognition compared to unenhanced data. This “hardware enhancement + algorithm optimization” cross-domain collaborative approach improves diagnostic reliability for rotating machinery un-der strong attenuation conditions, offering a robust solution for complex scenario condition monitoring.