4
\$\begingroup\$

As a self imposed exercise, I tried to derive the complete expression of gain in a common emitter amplifier without emitter degeneration. I mean by "complete" that it also takes into account the distortion associated with it. Here is my notations.

schematic

simulate this circuit – Schematic created using CircuitLab

And my attempt.

Derivation attempt

Suppose a tiny \$V_{O}\$ nudge \$v_o\$. We know \$V_O = V_{CC} - R_CI_C\$, therefore, as \$V_{CC}\$ and\$R_C\$ are constant, we get :

$$v_o = -R_Ci_C \Rightarrow i_C = - \frac{v_o}{R_C}$$

This will cause a change in the emitter's intrinsic resistance, \$\Delta r_e\$, defined as : $$\Delta r_e = \frac{V_T}{i_C} = -\frac{V_TR_C}{v_o}$$

By the definition of current gain in a BJT, it will also cause some \$i_B\$ : $$i_B = \frac{1}{\beta}i_C = -\frac{v_o}{\beta R_C}$$ Because \$I_E = I_B + I_C\$, there will be : $$i_E = i_B + i_C = -(\frac{v_o}{\beta R_C} + \frac{v_o}{R_C}) = -\frac{(\beta + 1)v_o}{\beta R_C}$$ By Ohm's law and definition, \$v_i = v_B = i_E\cdot(\Delta r_e + r_e(V_O))\$, so : $$v_i= \frac{(\beta + 1)v_o}{\beta R_C}\cdot(\frac{V_TR_C}{v_o} + \frac{V_TR_C}{\alpha V_O}) = \frac{V_T}{\alpha} + \frac{V_Tv_o}{\alpha^2V_O}$$

From here I am stuck, for I do not know how to handle the "incremental gain" concept. Should I treat it as a gain derivative or as the gain at a particular point ? I do not want this question to exhibit an XY problem, so any pointers toward a solution are appreciated.

I also tried a non-incremental solution, but with it I found \$V_I = V_T/\alpha\$ for all \$V_O\$ which is nonsense.

\$\endgroup\$
  • \$\begingroup\$ \$V_I\$ coming out as a constant makes sense, given that that's how ideal diodes behave. In other words, the derivation should make more sense if you put a base resistor in there. \$\endgroup\$ – Hearth May 13 '18 at 22:13
  • \$\begingroup\$ Since Vbe controls base current , even if β= constant for 5mVp you get a non-linear current. which is why you dont drive Vbe with a voltage source. tinyurl.com/y8684ohw \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 May 13 '18 at 22:19
  • \$\begingroup\$ I haven't seen yet base resistors except with "transistor switches". Are those still used with emitter degeneration ? \$\endgroup\$ – Sachiko.Shinozaki May 13 '18 at 23:09
  • \$\begingroup\$ Tony Stewart - while saying "....you don´t drive Vbe with a voltage source..." do you refer to the DC bias or to the signal input voltage? \$\endgroup\$ – LvW May 14 '18 at 6:42
7
\$\begingroup\$

Assuming active mode operation (\$R_\text{C}\$ doesn't cause saturation), then:

$$\begin{align*}V_\text{OUT}&=V_\text{CC}-R_\text{C}\cdot I_\text{SAT}\left(e^\frac{V_\text{IN}}{V_T}-1\right)\\\\ \text{d}V_\text{OUT}&=-R_\text{C}\cdot I_\text{SAT}\cdot e^\frac{V_\text{IN}}{V_T}\cdot\frac{\text{d} V_\text{IN}}{V_T}\\\\ A_V=\frac{\text{d}V_\text{OUT}}{\text{d}V_\text{IN}}&=-\frac{R_\text{C}\cdot I_\text{SAT}}{V_T}\cdot e^\frac{V_\text{IN}}{V_T} \end{align*}$$

That's it. Note that the gain does in fact depend upon the operating point.


Assume \$I_\text{SAT}=10\:\text{fA}\$, \$R_\text{C}=10\:\text{k}\Omega\$, \$V_T=26\:\text{mV}\$, and the operating point is \$V_\text{IN}=600\:\text{mV}\$. Then the gain would be about \$A_V=-40.5\$ according to that equation.


You can get \$I_\text{SAT}\$ from a datasheet. Look at the datasheet for the 2N2222A. Figure 4 follows:

enter image description here

Find the point indicated by \$I_C=1\:\text{mA}\$ and \$T=25\:^\circ\text{C}\$. I read off about \$V_\text{BE}\approx 640\:\text{mV}\$. From this I compute:

$$I_\text{SAT}=\frac{1\:\text{mA}}{e^\frac{640\:\text{mV}}{25.8\:\text{mV}}-1}\approx 1.7\times 10^{-14}\:\text{A}$$

Now, technically, \$V_\text{CE}\$ should be the same as \$V_\text{BE}\$ if I were doing a "one-point" approximation. But this chart is fine for as far as it goes.

Some cautions are worth mentioning.

  1. The chart is typical. Individual BJTs will vary by a factor of 2 or 3. I don't consider that to be much of a problem, though, because of the next point.
  2. You can see what the variation of \$I_\text{SAT}\$ over temperature is like by looking at the chart a little further. There are two other curves, in this case. You can work out just how much \$I_\text{SAT}\$ will vary over that temperature range. Here, you may find it varies from \$I_\text{SAT}=2.3\times 10^{-17}\:\text{A}\$ to \$I_\text{SAT}=8.7\times 10^{-10}\:\text{A}\$.

    This is because the saturation current variation over temperature is about a factor of 2.1 to 2.3 for every \$10\:^\circ\text{C}\$ change in temperature.

    The difference between \$150\:^\circ\text{C}\$ and \$25\:^\circ\text{C}\$ will make a difference between \$2.1^{12.5}\approx 11\times 10^3\$ about about \$2.3^{12.5}\approx 33\times 10^3\$. Roughly, that would predict somewhere between \$1.7\times 10^{-14}\:\text{A}\cdot 11\times 10^3\$ or \$\approx 1.9\times 10^{10}\:\text{A}\$ and \$1.7\times 10^{-14}\:\text{A}\cdot 33\times 10^3\$ or \$\approx 5.6\times 10^{10}\:\text{A}\$ for a temperature of \$150\:^\circ\text{C}\$.

    Which isn't far off from the chart!

    The other curve is \$80\:^\circ\text{C}\$ in the other direction, so here a factor of from about 400 to 800 less, rather than more. So it will be the case that the die temperature will be the real issue. Part variation looks really tiny, by comparison.

    To get nearer to part variations the temperature variations need to stay within about \$15\:^\circ\text{C}\$.
\$\endgroup\$
  • \$\begingroup\$ I did not want to use saturation currents as those are rarely , if at all listed on datasheets. But it indeed seems the most straightforward solution. \$\endgroup\$ – Sachiko.Shinozaki May 13 '18 at 23:06
  • \$\begingroup\$ @Sachiko.Shinozaki The gain quite simply depends upon that term and the collector resistor. You can infer it from the datasheet. Find the VBE vs IC chart that includes curves for various temperatures. I'll provide an example. \$\endgroup\$ – jonk May 13 '18 at 23:27
  • \$\begingroup\$ @Sachiko.Shinozaki Take note of the added temperature cautions I added. \$\endgroup\$ – jonk May 14 '18 at 0:18
  • \$\begingroup\$ excuse,where does these formula coming from?@jonk \$\endgroup\$ – lukeluck May 14 '18 at 1:26
3
\$\begingroup\$

Maximum gain for non-emitter-degenerated bipolar transconductor (yes, your standard bipolar transistor is all that), with the output current converted back into a voltage across the Collector resistor, is

Vgainmax = VDD / 0.026.

This assumed the transistor is just barely out of saturation, thus almost all of VDD is dropped across the collector resistor.

For largest voltage swing, bias the bipolar near VDD/2, and the max gain will be VDD / (2 * 0.026).

Thus a 2.6 volt VDD, with collector biased to 1.3 volts, will have gain of 2.6 / 0.052 = 50x, or 34dB.

\$\endgroup\$
  • \$\begingroup\$ Yet the base current being voltage controlled and thus collector current sink is very dependent on Rc for Vcc/2 and rising Vcc allows greater gain with proportional Rc. But of course this may end up being Rc=10M or so +1 \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 May 13 '18 at 23:34

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.