I work with the K-LC6 microwave module. Previously, when the microwave module was assembled on discrete elements, the signal-to-noise ratio was 5. Now, with the transition to specialized microcircuits, this ratio has dropped to 2. What are some ways to extract a signal from noise or suppress noise to improve the signal-to-noise ratio?
The useful signal is in the 1 kHz band and takes ~ 1-10 Hz. Noise is evenly distributed in this band (1 kHz).
I tried two methods to improve the signal-to-noise ratio.
I split the 1 kHz band with bandpass filters into 10 bands at 100 Hz. When processing, I select the band with the maximum signal. Since the noise is evenly distributed, and the signal is in one of 10 bands at 100 Hz, I manage to improve the signal-to-noise ratio. The disadvantages of this method are that it requires computational capabilities, and if the signal is at the intersection of 100 Hz bands at this moment, I lose part of the signal.
Since the noise is quasi-stationary, I pass the signal through the LPF 0.01Hz. Or further I get rid of noise according to this formula:
S = √ (A ^ 2-N ^ 2)
A - signal + noise
N - A passed through the low-pass filter of 0.01 Hz.
Thus, I seem to get rid of quasi-stationary noise, but anyway, some movements with a frequency> 0.01Hz can penetrate the final processing. The disadvantages of this method are that any movement> 0.01Hz can easily penetrate a useful signal.
Are there any other methods to improve the signal-to-noise ratio?