Does a single input differential amplifier remove noise?

Question

It is known that the purpose of using a differential amplifier is to remove noise from the input signal, apart from giving some small amplification. To my knowledge, there are two inputs and two outputs in a differential amplifier. Usually, two inputs are given to the circuit, one is the original signal, and the other is the "flipped" version of the same, this is the case in the dual input differential amplifier. But we also have single input differential amplifiers, where one input node is grounded, and the other input is the original signal. Does such a configuration remove the noise from the original signal. If yes, I have another confusion. There are two outputs in a diff-amp circuit. If we measure each output individually (with ground as the reference), does it have noise removed or not? Or is it only when I take the difference of both the outputs when I will see a output without noise? I am very confused.

The context is that I have to construct an Audio Amplifier with a mic input. My stages of the circuit are Diff-amp --> CS Amplifier --> Filter --> Power Amplifier --> Speaker. The issue with taking the difference of the outputs of the differential amplifier is that the input to my CS Amplifier is just one Vin and not the difference of two signals, which makes it hard to design the circuit. Am I missing something or how do I go about designing this?

Answer 1

Once you understand the way that the differential input "removes" noise from the input, the answer to this is clear.

If we call the two input nodes (which are together considered to be a single "input" and I will stay with that convention here) A and B, the differential amp (and a mic amp is a very good example) has a high gain to the input difference (A - B). At the same time, it has low gain (maybe even attenuates) the "common mode" input (A + B).

So if the output is C:

C = G(A - B) + R(A + B)

where G is the "differential gain" and R is the "common mode gain" (or rejection)

What matters here is the difference between G and R, the "common mode rejection ratio".

Let's work through take some typical numbers for a mic amp with a "standard" mic connected (if there is such a thing, but that is another topic):

K might well be 40dB (a voltage gain of 100). R might be 0dB (a voltage gain of 1). So the rejection ratio is 40dB.

Let's say that our mic signal is 5mV differential, but 1mV of common mode noise gets in to our 10 metre microphone cable (which will be a balanced twisted pair with an overall screen connected to 0V at the mic amp end). This means that the noise is 20% of the signal at the input to the amp.

At the output of the mic amp we still have 1mV of noise, but we amplified the differential audio signal by 100, so we have 500mV. At the output of the amp the noise is now only 2% of the signal. We have effected a massive improvement in signal to noise because of our rejection ratio.

Now to answer your question : this improvement relies on keeping the induced noise equal on the two legs of the differential input.

If we connect the B input to 0V for example, so that it no longer receives the same common mode noise as the A input, the noise will be amplified by 100 (as it is on the A input but not the B) and we will not achieve the improvement in signal to noise ratio.

So from this, we can deduce the important factors in amplifying a microphone signal with best signal to noise:

1. a mic amp with excellent common mode rejection ratio (high differential gain, low common mode).
2. a mic amp with well matched impedance to ground on A and B inputs (if this is not the case, the induced voltages from the external noise will differ).
3. a microphone signal which is differential (it does not really matter that much that the A and B legs are equal and opposite, but there must be a difference signal between A and B).
4. the induced external noise should be equal on A and B inputs. (That is why we use a twisted paid - to try to ensure that each leg has the same induced voltage. Screening helps reduce the overall magnitude of noise, but in fact using a twisted pair may be just as important. Older analog professional audio systems often used multiple unscreened twisted pairs to carry multiple line level signals between equipment for this reason.)