Suppose I remove Ibias and just connect to -V, then Avd = (gm1 + gm2)(r01||r04). What would be the common mode gain? It seems like it would be zero since the Q4 current would exactly track with the Q2 current?

What then is the point of Ibias when the load is a current mirror?

 

There's more to consider than just common-mode signal gain.
Look at the transistor bias current with and without Ibias.

 

If the common inputs were a relatively high value to ground, then the power loss would be much more significant. So the Ibias regulates the power consumption, but what else? Would I be correct in assuming the Ibias would actually add a small amount of common mode signal gain?

 

Would I be correct in assuming the Ibias would actually add a small amount of common mode signal gain?

Why would you assume that?

 

Keep in mind how a differential pair works.

The Ibias (ideally) forces the sum of the two branch currents to sum together, so if one goes up the other must go down.

With the active load the current mirror tries to force the currents in the two branches to remain the same.

Something has to give.

The result is that the output voltage has to change significantly in response to a small imbalance in the differential input voltage in order to get enough voltage difference across Q4 for the finite output impedance of it to be enough to change the current flowing in it far enough away from what Q3 is trying to make it output.

It's the contention set up be the two competing currents that results in the very high gain and that contention is a direct result of the bias current.

 

Why would you assume that?

Because the source will have some Rss which is the cause of the common mode amplification. If both or connected to ground, then they will track almost perfectly right?

So Iss gives more gain at the expense at higher common mode gain?

 

So Iss gives more gain at the expense at higher common mode gain?

Yes.
But it also provides the proper bias to the differential stage.
Without Iss, the stage bias would depend upon the common-mode signal and would become very large at high CM voltages.

 

Without Iss, the stage bias would depend upon the common-mode signal and would become very large at high CM voltages.

Riiiiight, I see. Thanks a bunch!

 

You can improve the CM rejection due to the Rss of the Iss source by using a current source with a high impedance, such as a Wilson current mirror.

 

Ibias is there to keep every thing working while allow you to float up VIN1/2 well above the around. Otherwise, VIN1/2 can only be close ground