Learning razavi's books and videos gives people the feeling that every circuit does not come for no reason, but because of demand.

A basic common source circuit is shown in the figure below. A common source circuit is defined with the input at the gate and the output at the drain. If the channel modulation effect is considered and it is assumed that it works in the saturation region, its equivalent small-signal model is shown in the figure below. gmV1+Vout/r0+Vout/RD=0

Vout/Vin=-gm(r0//RD)

That is, the voltage gain in the above graph is:

Av=-gm(r0//RD)

because r0//RD

It can be seen from the above formula that if you want to increase the gain of the voltage, you can start from three aspects:

(1) Increase RD

(2) Increase r0

(3) Increase gm

First, increase RD. It can be seen from the above common source circuit that if RD increases while keeping ID unchanged, then VDS must drop, which may cause the tube to enter the linear region from the saturation region. Therefore, if you want to increase the voltage gain by increasing the RD, the space is limited.

However, is there a way to make RD tend to infinity? Or, in the small-signal equivalent model, it is equivalent to infinity.

Recall that the small-signal equivalent model of the constant current source is found to be an open circuit, which is equivalent to infinity. Therefore, if a constant current source is used to replace the resistor RD, the wish just made should be realized. Because at this time:

(1) Under DC bias, stable bias current can be provided

(2) Under the small signal model, the open circuit is equivalent to that RD tends to infinity.

At this time, Av=-gm*r0.

This voltage gain is called the intrinsic gain of the MOS tube, that is, the inherent gain of the MOS, which is the maximum gain that the tube can achieve.

But how can this current source be implemented?

Recall that when the MOS tube works in the saturation region, there is a controlled current source between its drain and source, and this controlled current source is close to a constant current source.

Therefore, you can try to use this controlled current source to achieve the above requirements. However, the current source implemented by NMOS and PMOS has certain limitations, that is, the current source implemented by NMOS can only flow from any node to the ground, while the current source implemented by PMOS can only flow from the current source to any node. None of them can implement a current source that flows from any node to another arbitrary node.

With the implementation of the current source, there is the following circuit. This circuit has a name called common-source stage with current source load, which translates to a common-source circuit with a current source load. Because ro1 and ro2 are generally relatively large, the gain of the amplifier is visibly increased.

As can be seen from the above figure, its voltage gain is proportional to gm, one of the expressions of gm, as shown in the figure below. Changes in the factors in the formula will cause changes in gm, which in turn will cause changes in gain. For example:

(1) The mobilities of electrons and holes are both temperature dependent

(2) The change of the supply voltage will also be reflected on the ID

(3) Processing errors, such as the mobility and turn-on voltage between different wafers will be different

(4) Signal amplitude, the small signal model is based on the case where the signal is very small. If the signal amplitude is relatively large, the corresponding IDs under different amplitudes will be different, which will lead to different gains.

When designing circuits, these undesirable effects need to be mitigated.

The circuit constructed below can reduce the influence of these factors.

The basis of this construction circuit is the diode-connected MOS. If the above circuit is used to replace the load resistance RD, unexpected results will be achieved. It can be seen that if a diode-connected MOS tube is used to replace the load resistor RD, the final gain is only related to the size of the tube itself, and has nothing to do with the electron mobility and the size of the current ID. Increased robustness.

A basic common source circuit, want to increase its gain, found that if the value of RD is increased, it will cause the working state of the tube to enter the linear region from the saturation region;

Change the idea and replace RD with a constant current source, so as to ensure the coexistence of bias and infinite resistance; then use MOS tube to realize the constant current source;

It is found that the gain is too affected by the environment, so the diode-connected MOS is used to replace the RD, so that the gain is only related to the size of the tube.

Reviewing Editor: Tang Zihong