1
\$\begingroup\$

Please consider the differential amplifier (Q1&Q2) with current mirror active load (Q3&Q4) as shown below:

schematic

simulate this circuit – Schematic created using CircuitLab

I know many questions have been asked already regarding this circuit on EE SE, but I didn't find an answer to my question.

My question is: what will the output voltage be when subjected to differential input WITHOUT ANY LOAD AT THE OUTPUT? Framed differently, what is the differential gain with no load?

In my understanding, the currents in the collectors of the 2 right transistors will not agree and thus the output impedances at the collectors will come into play. However, I do not know how to calculate this output impedances and how they will affect the output voltage.

Side note: If you think that asking about the output voltage with no load is meaningless than consider a load whose input impedance is comparable or larger to the output impedance looking into the collectors.

\$\endgroup\$
6
  • \$\begingroup\$ Well, why don't you write down your equations with a generic R as load, and see what happens when said R gets very high? \$\endgroup\$ Nov 24, 2016 at 19:56
  • \$\begingroup\$ My problem is, I don't know how to describe the behavior of the collector under heavy load (comparable to the output impedance at the collector). I saw some places described it as a current source in parallel with some Rout but I do not understand why and how to calculate the said R. \$\endgroup\$ Nov 24, 2016 at 19:59
  • \$\begingroup\$ I think there is a misunderstanding: load your amplifier with a generic resistor R, calculate the gain, take R to infinity and see what happens. \$\endgroup\$ Nov 24, 2016 at 20:00
  • \$\begingroup\$ Oh, sorry I misunderstood you. Anyway I get that when Rc becomes very large the differential gain becomes very large. I'm looking for something more quantitive that will help me understand the limitations of the current mirror active load. \$\endgroup\$ Nov 24, 2016 at 20:03
  • \$\begingroup\$ Actually, it's not clear how to derive it from the Wiki mentioned by Mario. The Gummel Poon model makes it quite clear, as it explains the Early (and Late) Effects from a much more unified and physically sensible approach than the modified Ebers-Moll, which just hacks it in almost arbitrarily. It boils then down to this: \$V_A=\frac{Q_{B0}}{\frac{1}{V_{BC}}\int_0^{V_{BC}} C_{jC}\left(V\right)~\textrm{d} V}\$. \$\endgroup\$
    – jonk
    Nov 24, 2016 at 21:31

1 Answer 1

2
\$\begingroup\$

Assuming that you are talking about the gain at low frequencies ("DC gain") a derivation of the gain will result in an expression like $$ Av = g_m (r_{o2} || r_{o3}) $$ where \$g_m\$ is the transconductance of Q2 or Q3 and the \$r_{o2}\$ and \$r_{o3}\$ is the output resistance of these transistors.

The gm of the transistor is easily found to be $$ g_m = \frac{I_C}{V_T} \qquad V_T = \frac{kT}{q} $$ The output resistance is a property of the transistor and usually described by the Early voltage \$V_A\$. Using the Early voltage the output resistance is $$ r_o = \frac{V_A + V_{CE}}{I_C} \approx \frac{V_A}{I_C} $$ since usually \$V_A \gg V_{CE}\$. Assuming that both transistors have the same ro the gain is $$ Av = g_m (r_{o2} || r_{o3}) = \frac{I_C}{V_T} \frac{V_A}{2 I_C} = \frac{1}{2}\frac{V_A}{V_T} $$ With an Early voltage of 100V and a kT/q of 26mV we get a gain of about 2000.

\$\endgroup\$
3
  • \$\begingroup\$ can you elaborate about the early voltage and the output impedance? or refer me to a source where I can find the relation between the two? \$\endgroup\$ Nov 24, 2016 at 20:56
  • \$\begingroup\$ sorry I meant the derivation of the relation, not the relation itself which is shown here \$\endgroup\$ Nov 24, 2016 at 20:57
  • \$\begingroup\$ You should find everything here: en.wikipedia.org/wiki/Early_effect \$\endgroup\$
    – Mario
    Nov 24, 2016 at 21:01

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.