Gm-cell Design

[ahmedatef0]

The Gm-cell must have high linearity and at the same time it must have high Rout thus high Voltage gain. How is this possible if the voltage gain is high transistors will go in triode for small differential input voltage thus linearity will be bad. So how can I design a Gm-cell that is linear and have a high Rout at the same time?

 

[AMS012]

The solution is low Gm. You design low Gm so that you do not saturate for small differential operation. This in turn means better linearity as well. So, killing the Gm is the solution. This also gives higher Rout resulting in higher gain. So, a resistor degeneration is the way to go.

 

[ahmedatef0]

The solution is low Gm. You design low Gm so that you do not saturate for small differential operation. This in turn means better linearity as well. So, killing the Gm is the solution. This also gives higher Rout resulting in higher gain. So, a resistor degeneration is the way to go.

I want to build a filter of 0.8GHz using the gm cell.
If I use a small gm I will have to use small caps which will be smaller than the parasitics, so I need large gm and large Rout this large voltage gain at the same time I need the gm cell to be linear how is this possible

 

[frankrose]

Normally at 0.8GHz the capacitor reactance should be much lower than Rout of a transistor, you shouldn't need huge Rout, only for DC feedback in closed loop to set operating point. Don't apply huge voltage gain without feedback. If any of your devices leave saturation region that means you apply bad operating point or too big voltage gain.