In a differential amplifier, there is a common emitter transistor configuration (often referred to a "Constant Current Tail") connected between the common point of the symmetric transistor pair and the power supply ground.

The base of the transistor is connected to the positive line of the power supply and the emitter is connected to ground via a resistor.

This is purportedly to minimize the gain for common mode signals at both the + and - inputs to the diff amp. However, I'm looking for an explanation of how this works.

 

Each of the input transistors has to have their emitter connected to ground in some way, right? If you use a resistor, how much voltage does it have to use up to get the bases to the midpoint of the power supply? About half the supply voltage. If you want the input range to be anything except half the power supply, that resistor will change the amount of current in the pair. That's enough reason right there to use a transistor.

If you're going to use a transistor in the ground path, why let it change the idle current? Why not just let it change its voltage from collector to emitter to allow a wide common mode range?

Common mode rejection happens best with a matched pair for the input. An emitter resistor would have no effect on that because it's common for both transistors. Why would a constant current transistor be any better?

That was your question, and I don't know that it is even a true allegation.

This really has me thinking. Improve common mode rejection compared to what? A capacitor to ground? An inductor to ground? The fact that I'm puzzled usually means I'm on the wrong track.

 

To have a high common mode rejection, the operating current of both differential transistors should be unaffected by the common-mode voltage.
To do that you would want the emitters to be connected to a constant-current source (very high resistance source). That way the current is not changed by a change in the emitter voltage (which is basically the same as the base voltages except for the base-emitter voltage drop).
A simple constant-current source is a transistor with an emitter resistor and the base connected to a constant voltage. Alternately a two transistor current-mirror circuit can be used to generate the constant current.
The collector impedance of such a circuit is high, thus making the current largely independent of the collector voltage (a good current source).
This means the common-mode voltage has little effect on the differential current and thus little effect on the differential output voltage. This translates to a high common-mode rejection.

For even better common-mode rejection, a Wilson current-mirror can be used, which uses 3 transistors with local feedback to give an even higher collector impedance (closer to an ideal current source).

 

OK. So it's a constant current source compared to a resistor. So basic that I flew right past it in the first paragraph.
Thanks, crutschow.

 

A constant current source or sink for a differential pair assures that the transistors' operating points do not vary with the DC portion of the input signal. This stabilizes the gain, bandwidth, distortion, and thermal environment of the input transistor pair, which go a long way toward reducing input offset errors and thermal drift, and improving the overall stability of the amplifier across many applications.

ak