I need to find the maximum possible gain for this NMOS cascode circuit.

enter image description here

Here is the s.s. model:

enter image description here


I used nodal-matrix analysis (basis is KCL) to find the gain (\$A_v\$):

$$ \begin{bmatrix} \frac{1}{r_{o1}}+\frac{1}{r_{o2}} & -\frac{1}{r_{o2}} \\ -\frac{1}{r_{o2}} & \frac{1}{r_{o2}}+\frac{1}{(R_{L1}\,||\,50)} \\ \end{bmatrix} \cdot \begin{bmatrix} V_A \\ V_B \\ \end{bmatrix} = \begin{bmatrix} -gm1\cdot V_{in} - gm2\cdot V_A \\ gm2\cdot V_A \\ \end{bmatrix} $$

$$ \Longrightarrow \begin{bmatrix} \frac{1}{r_{o1}}+\frac{1}{r_{o2}}+gm2 & -\frac{1}{r_{o2}} \\ -\frac{1}{r_{o2}}-gm2 & \frac{1}{r_{o2}}+\frac{1}{(R_{L1}\,||\,50)} \\ \end{bmatrix} \cdot \begin{bmatrix} V_A \\ V_B \\ \end{bmatrix} = \begin{bmatrix} -gm1\cdot V_{in} \\ 0 \\ \end{bmatrix} $$

$$ \Longrightarrow V_A = D/C,\quad V_B = -\frac{v_{in}\cdot (r_{o1}\cdot gm1)\cdot (r_{o2}\cdot gm2 + 1)\cdot (R_{L1}\,||\,50)}{r_{o1}\cdot r_{o2}\cdot gm2+r_{o1}+r_{o2}+(R_{L1}\,||\,50)} $$

$$ V_B = V_{out} = -\frac{v_{in}\cdot (r_{o1}\cdot gm1)\cdot (r_{o2}\cdot gm2 + 1)\cdot (R_{L1}\,||\,50)}{r_{o1}\cdot r_{o2}\cdot gm2+r_{o1}+r_{o2}+(R_{L1}\,||\,50)} $$

$$ \Longrightarrow \frac{V_{out}}{V_{in}} = A_v = -\frac{(r_{o1}\cdot gm1)\cdot (r_{o2}\cdot gm2 + 1)\cdot (R_{L1}\,||\,50)}{r_{o1}\cdot r_{o2}\cdot gm2+r_{o1}+r_{o2}+(R_{L1}\,||\,50)} $$

This is the part where I am stuck; I was told that increasing \$gm\$ and \$R_L\$ will reveal the maximum gain but I don't really understand what that entails. Thus far, my interpretation of "increasing \$gm\$ and \$R_L\$" has been:

$$ \lim\limits_{(gm1,gm2,R_{L1}) \to (\infty,\infty, \infty)} A_v $$

However, I've had trouble finding a reliable/straight-forward method for calculating these simultaneous/multi-variable limits, so I assumed that my interpretation was incorrect. For instance, when I apply this interpretation using Wolfram (the only calculator I could find to compute these types of limits), I receive the following output:

(limit does not exist, is path dependent, or cannot be determined)

If you can provide a better interpretation, please enlighten me.


  • I'd like to know the max-gain but I am more interested in being taught a correct/straight-forward method that is applicable to any similar problem.
  • Although this is most likely a question of mathematical understanding, it belongs here b.c. many EE's might have already solved this problem or they might bring enough insight about the circuit (NMOS cascode is recognizable/typical) to derive a solution (or provide the final answer at-least).
  • \$\begingroup\$ Basically it will be zero because Q2's gate is tied to ground. \$\endgroup\$ – Andy aka Oct 1 '18 at 9:10
  • \$\begingroup\$ @Andyaka I don't see how you're correct since, for any reasonable insertion of values, my A_v eq returns a non-zero answer. It seems to tend towards -infinity, as R_L1 and gm's increase. \$\endgroup\$ – Landon Oct 1 '18 at 15:31
  • \$\begingroup\$ @Andyaka And on a fundamental level, a MOSFET's gate being tied to ground doesn't imply that it's output is 0. If the gate and the source were tied to ground (V_GS = V_G - V_S = 0) then the s.s. dep. current source would be 0 & output would be 0. But that isn't the case here. \$\endgroup\$ – Landon Oct 1 '18 at 15:45
  • \$\begingroup\$ I never said your output would be zero; I was answering your title question and saying the gain would be zero. But please sit and wait for that answer. \$\endgroup\$ – Andy aka Oct 1 '18 at 16:08
  • \$\begingroup\$ @Andyaka I doubt it will receive an answer ),; \$\endgroup\$ – Landon Oct 2 '18 at 16:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.