Algorithm of auto-balancing noise-canceling based on noise correlation for high-speed balancing, high-dynamic range, and robustness against DC-offset drift

The influence of the light source noise can be reduced by subtracting the signal of the light source noise (reference signal) from that of the probe light (probe signal). Here, it is essential that the intensities of the signals are equated. To equate the intensities, an auto-balancing method is widely employed, where the gain of the probe signal is feedback-controlled, regarding the DC component in the subtraction as an error signal. However, DC-offset drift causes a deviation from the optimal intensity balance. Additionally, the DC component is often several orders of magnitude larger than the sample signal, which requires a high-dynamic range in the circuitry. Furthermore, if the feedback control is too fast, it cancels out the sample signal. In this study, we formulate a noise correlation auto-balancing method, where the correlation of the reference signal and residual noise in the subtraction is employed as the error signal. With this scheme, all the above problems are avoided. The feasibility of the algorithm was demonstrated by a prototype circuitry and signals emulating the probe and reference signals. It did not suffer from the DC-offset drift, while a 44-dB canceling rate with auto-balancing of a 1.3-MHz cutoff frequency was demonstrated. We foresee, such as in pump/probe measurements, that this scheme improves the robustness, dynamic range, and response time required to follow changes in transmittance and the measurement position of the sample while employing a light source that is advantageous in wavelength selectivity, coherence, and cost but is noisy. Published under an exclusive license by AIP Publishing.

Automated summary

This paper introduces a noise correlation auto-balancing method to reduce the influence of light source noise on signals. The method balances the intensities of signals by adjusting the gain of the probe signal, and uses the correlation between the reference signal and residual noise in the subtraction as the error signal, solving problems such as DC-offset drift and dynamic range.