OK, let's take your -18V bias load line as an example and estimate small signal gain. Let's take a big enough swing to reduce the error reading lines off the chart, say 20V peak grid-to-grid. It is indeed class A operation, but highly asymmetric. For one grid going +10V, the plate swing is about 56V. For the opposite side the grid swings -10V and it's plate swings about 44V. Adding the two plate swings gives about 100V plate signal for 20V grid signal. Pretty close to 14 dB of gain.
Exactly. You've got it. Good explanation.
Hi MeToo2 and thank you for your input. Those links are great, I have started to read about composite load lines for the first time. What is the difference between a long tail pair vs. a short tail pair? You and abbey road d enfer are right about those mu and plate resistance values that I took from the datasheet being useless in this particular case...
Are those to be found in a datasheet? Or do you mean the composite load line I will be drawing myself?
The values on the datasheet assume a particular single ended class A circuit operating at a particular operating point .
You need to use your own load line on the composite characteristics at the operating point in your circuit, or exactly as already MJK stated (as he's already taken into account the circuit topology in his explanation).
The difference between a theoretical long tail pair and short tail is that a long tail pair assumes that the total current through the tail is always pretty much constant irrespective of voltage (perfect constant current sink, so that the common cathode voltage is largely irrelevant.) So if one tube sources more plate current, it automatically robs current from the other tube because the total plate current for both tubes is kept constant by the long tail. A short tailed pair does not behave as a perfect current sink. If one tube sources more current, it also affects Vcathode on both tubes and thus Vg (and thus the total plate current for the 2 tubes also varies)
Small signal models try to predict AC characteristics at one DC operating point and assume that the gain and other transfer characteristics can be characterised by a single linear parameter: e.g. mu = dVo/ dVi.
Large signal models (looking at the actual tube curves and then taking a delta over two or more points) do not make this assumption about linearity. As MJK said, this circuit operates asymmetrically.
A vari-mu tube is a remote cut-off triode, which changes all sorts of parameters across its operating range, not just mu, but transconductance and plate resistance too. The curves are all over the place, which means linear small signal approximations start to break down pretty quickly. Any value of mu or other parameters is only valid at that one DC operating point. Change the DC operating point (by decreasing the value of Vg and/ or Ip or when going into compression) and you will get very different results. That's why it acts a compressor. Because it's non linear.
Also assuming Class A can also be misleading, depending on the load and input signal. A tube can only "turn off," and it cannot source negative current. Which means during class A operation the tubes see a different load impedance than in class B operation (once one tube turns off) That happens at different times for different output voltages and different loads, so the circuit can operate as class A up until a certain amplitude of input and then class B for the peaks. Sine wave in does not mean sine wave out. Bye bye linear model. Hello distortion.