In squeezing-enhanced system, the stability and squeezing level of the squeezed states directly affect the improvement of quantum enhancement sensitivity and signal-to-noise ratio (SNR). Squeezed states can be generated by an optical parametric oscillator (OPO) based on second-order nonlinearity. At present, Pound-Drever-Hall (PDH) is the most commonly employed method for locking the OPO cavity, and the photodetector plays a key role in extracting extremely weak signals. For PDH locking systems, the useful signals coupled to photodetectors are narrowband signals at the modulated frequency, while the traditional wideband photodetectors amplify signals and noise in the whole frequency band, which is not conducive to improving the SNR. It is worth noting that the seed light is employed in the active stable parametric cavity in the preparation of the bright squeezed state, and the increased seed power will lead to the coupling of the pump noise into the bright squeezed state, thereby resulting in the reduced squeezing level. However, the increased optical power of extracting signal can improve the error signal of the locked parameter cavity. Thus, it is important to design photodetectors with high gain and SNR. Photodiodes have certain junction capacitance, and combined with variable inductance, inductance and capacitance (LC) resonance circuit can be formed to enhance the resonance of specific frequency signals. The detector is named resonant photodetector (RPD). The LC resonance circuit can be equivalent to the parallel resonance circuit and is regarded as a bandpass filter, which only amplifies the required frequency band and suppresses the noise of unnecessary frequency bands. However, the quality factor Q directly characterizes the suppression effect on the external noise signal of the resonant frequency, and the SNR of the error signal directly affects the minimum jitter of the cavity length and phase after locking.

To evaluate the newly designed RPD, this paper builds a test platform to evaluate transfer functions and error signals, as shown in Fig. 4. The laser source is a single-frequency laser of 1550 nm. The half-wave plate HWP₁ is employed to adjust the laser power reaching the modulator, and HWP₂ is to adjust the polarization direction of the laser, perpendicular or horizontal to the modulator. This means that the direction is 45°from the main axis of the electro-optical crystal. The network analyzer sets the start and end frequencies of the test (starting from 1-100 MHz in the experiment, and then being refined according to the resonance frequency). The output signal is divided into two parts, one of which is loaded on the modulator for modulating the laser beam, and the other is returned as a reference signal. The second part is the measurement of the error signal, which adopts the electro-optical modulator (EOM). MC is closely related to the preparation of high level squeezed state. According to PDH technology, the anti-interference ability of locking is proportional to the peak-to-peak value of the error signal, and the larger peak-to-peak value will lead to stronger anti-interference ability. In addition to the incident laser power, the amplitude and SNR of the error signal also depend on the signal extraction capability of the photodetector, so the SNR of the error signal extracted by the photodetector determines the stability of the whole feedback loop. Therefore, the performance of the developed resonant detector is evaluated by the SNR and the stability of the error signal in MC cavity locking.

This paper measures the transfer functions of commercial BPD (THORLABS PDA10D2) and RPD under the same conditions (Fig. 5). At the resonant frequency of 20 MHz, the gain of RPD is about 30 dB higher than that of BPD. The high gain helps to obtain a stable phase locking at lower power, thus improving the stability of the system in the squeezed state without reducing the quantum noise. Through external mixing and integrated circuit design, the 3 dB bandwidth of RPD is 0.285 MHz. The quality factor Q of RPD is 70 and can be calculated from the measurement results by Formula (5). The experimental results are shown in Figs. 6 and 7. The SNR improvement of the newly designed RPD is more obvious than that of BPD, and the SNR is defined as the ratio of the peak-to-peak value to the noise of the error signal. The error signal is a DC signal, which cannot be measured by a spectrum analyzer and can only be recorded by an oscilloscope. At the resonance frequency of 20 MHz, the peak-to-peak value of the RPD error signal is 560 mV, the peak-to-peak value of noise is 42 mV, and the SNR is about 22.5 dB. The peak-to-peak value of the BPD error signal is 35 mV, and the peak-to-peak value of noise is 20.8 mV. The SNR of the newly designed RPD is about 18 dB higher than that of BPD.

Based on the theoretical analysis of the resonance circuit and cross-resistance amplifier circuit, the selection of low noise devices, and the optimized circuit layout, this paper develops a resonant detector with Q factor of 70 and SNR of 22.5 dB. Compared with the traditional broadband photodetector (BPD), the gain of RPD at 20 MHz is about 30 dB higher than that of BPD. By measuring the peak-to-peak value and SNR of error signals, the peak-to-peak value of RPD locking cavity error signals is 16 times that of BPD, and the SNR of RPD error signals is about 18 dB higher than that of BPD under the same condition. This RPD can provide a key device for photoelectric feedback control and the preparation of continuous variable nonclassical light fields.