Noise reduction with stereo channel

[PhG]

Hi,
I am designing a mono audio device with one microphone. Since I am using a stereo preamplifier and ADC, I have two separate preamps and ADC with independante noise. My idea is to improve the signal to noise ratio by using both preamps and ADCs to convert the same microphone signal. Since the signal is the same but the noise is independant I can improve the signal to noise ratio by simple building of mean value of both channels.
But this methode does in practise and simulation not improve much. So my question is: Is there a better methode to reduce the noise than simple building of mean value?

best regards
Philippe

 

[barry]

First of all, what makes you think the noise is independent? Both amplifiers get their inputs from the same source. MAYBE the noise generated by each of the amps is different, but how much is that compared to external noise.

Your question is way to broad too answer. Where is the noise coming from? Between the microphone and the amp? Is it a noisy signal into the microphone?

 

[Analog Ground]

If the noise is from the preamps, your idea may work. The noise in the two channels must be uncorrelated. Then, a few things.

1. Attaching the microphone to two preamps may be decreasing the signal. A bad thing which will decrease the signal to noise ratio.
2. The outputs of the preamps should be added. Do not take a mean of the two channels. When added, the signal then increases by a factor of 2 and the noise from the uncorrelated sources increases by a factor of 1.4 (square root of 2). So, the signal-to-noise ratio is improved.
3. The addition should probably be done analog. Not sure about this one. How to do the sampling seems tricky.

 

[zorro]

The idea doesn't work assuming that the preamplifier is designed for maximization of signal to noise ratio (SNR). The total SNR at the output doesn't improve.

Let's assume that the noise of the amplifier does not depend of the source impedance (this is not completely true, but this can be discussed later and does not alter the conclusion). Then, SNR is maximized when the amplifier receives the maximum available power from the source, i.e. when the impedances are matched.
Now, we put two such amplifiers in parallel. The power of the source must be split between them, so the SNR at the output of each amplifier is half of the SNR obtained in the case of a single amplifier.
When the outputs are added, the signal components add together coherently and the noises add together incoherently because the noise contributions are independent, as it was already pointed out. As a result, the SNR after the addition is doubled with respect to the SNR encountered at each aplifier, and we have the same SNR as in the case of the single amplifier.

If the impedance of the amplifier is much higher than the impedance of the source, connecting two amplifiers in parallel makes unnecessary to split the available power of the source in two parts. The result improves in this case because the original amplifier was far to be matched, and connecting two of them in parallel is simply a way to make an amplifier with half the input impedance, i.e., best matched to the source.

I hope it is clear.
Regards

Z