Active Loads

With a resistive plate load, the voltage gain formula tells the whole story: Av = µ × R_L / (rp + R_L). Since R_L / (rp + R_L) is always less than 1, the gain Av is always less than µ. For a 12AX7 with µ = 100 and a typical 100kΩ plate resistor, you get Av ≈ 61 — barely 61% of the tube's amplification factor.

The resistor also wastes B+ voltage. At the quiescent point, the resistor drops a significant fraction of the supply. With B+ = 300V and a 100kΩ load, the plate voltage sits around 170V — the output can only swing 170V peak before hitting the supply rail, and even less in the other direction before grid current flows.

Now replace the resistor with a constant current source (CCS). A CCS has infinite impedance at AC — its loadline is perfectly horizontal on the plate curves. Result: R_L → ∞, the gain formula simplifies to Av ≈ µ. The full amplification factor is realized. The voltage swing approaches 2×B+ because the CCS maintains constant current regardless of plate voltage. Power supply rejection improves dramatically because supply ripple creates no current variation through the infinite-impedance load.

The most elegant CCS uses what we already have: another tube. Pentodes make excellent constant current sources because their plate characteristic curves are nearly flat — once past the knee voltage, plate current barely changes with plate voltage. The screen grid voltage programs the operating current.

Suitable pentodes include the EF184 (sharp cutoff, low noise), EL83 (power pentode, handles high current), and the remarkable EL822 (frame-grid, extremely high output impedance). The operating current is set by: Ik ≈ gm × (Vg2 − |Vg1(cutoff)|), with fine adjustment via cathode resistor.

Modern tube designers often use semiconductor CCS devices. The transistor handles the DC task of maintaining constant current while the tube does what it does best: amplify the signal. This hybrid philosophy accepts that the signal path is what matters sonically, and the DC biasing network can be solid-state without compromise.

The simplest semiconductor CCS is the depletion-mode MOSFET (DN2540N5, IXTP01N100D). Connect gate to source through a resistor — that's it. Two components. The DN2540N5 is rated at 400V and can be set from 1mA to 15mA by choosing Rs. I = Idss × (1 − Vgs/Vp)² with Vgs = −I×Rs.

The µ-follower stacks two triodes: the lower tube amplifies while the upper tube acts as a CCS. The upper tube's cathode connects to the lower tube's plate, and its grid is decoupled to ground via a capacitor. The upper tube sees the output signal on its cathode and adjusts its current to maintain constant operation — presenting a very high impedance load to the lower tube.

The result is extraordinary: with a 12AX7 lower and 6J5 upper (µ = 20), the effective load impedance is µ₂ × rp₂ = 20 × 7.7k = 154kΩ. Gain rises from 61 to about 91. PSRR improves by 26dB — from 6dB to 32dB. With a pentode-connected upper tube (µ ≈ 500+), the loadline becomes essentially horizontal and Av approaches the full µ of 100.

Limitations: the µ-follower requires double the heater supply current and has slightly higher output impedance than a simple common cathode stage because the upper tube's rp appears in parallel. It is optimized for maximum gain, not for driving loads.

SRPP looks identical to the µ-follower in schematic — two triodes stacked with a shared plate/cathode node. The crucial difference is intent and loading. The µ-follower is optimized for high gain into a high-impedance input. The SRPP is designed to provide push-pull drive capability into a defined load impedance.

In SRPP, on positive signal swings the lower tube pulls more current (conventional common cathode), while on negative swings the upper tube pushes current into the load through its cathode follower action. This gives genuine push-pull operation from just two triodes. The optimal load impedance for balanced push-pull operation is: R_load = rp / 2.

For a 6922/E88CC with rp ≈ 2.6kΩ, the optimal SRPP load is about 1.3kΩ — which is close to the impedance of a 600Ω balanced line or a reasonable headphone. With a 12AU7 (rp ≈ 7.7kΩ), the optimal load is about 3.8kΩ. If the actual load differs significantly from rp/2, the push-pull symmetry degrades and one triode does most of the work, reducing the benefit over a simple common cathode stage.

A common misconception is using SRPP to drive a high-impedance grid — in that case, the upper tube barely participates and you get no push-pull benefit. Use a µ-follower instead. SRPP shines when you have a defined, moderate-impedance load to drive.

A simple cathode follower has output impedance Zout ≈ 1/gm, typically 200–500Ω. The White cathode follower adds a second tube in a feedback arrangement that multiplies the effective transconductance. The output impedance drops to: Zout ≈ 1 / (gm₁ × gm₂ × Ra).

With two 6SN7 sections (gm ≈ 2.6 mA/V each) and Ra = 47kΩ, the White CF achieves Zout ≈ 3Ω. This is low enough to drive headphones, long cables, or serve as the output stage of an OTL (output-transformerless) amplifier. The White CF exists in two forms: self-contained (both tubes in one envelope) and with external gain stage feedback.

The cascode amplifier stacks two triodes in series: a common cathode lower tube drives a common grid upper tube. The upper tube shields the lower tube from Miller capacitance and effectively multiplies the plate resistance. When you combine a cascode with a CCS load, you get a horizontal loadline on a tube pair with extremely high equivalent µ.

A practical example: 6SN7 (or 7N7) cascode with DN2540 depletion MOSFET CCS. The cascode runs at 8mA+ for maximum linearity. B+ = 400V. The CCS maintains perfectly constant current while the cascode pair provides an effective µ of µ² (20 × 20 = 400 for 6SN7). Gain exceeds 350, with voltage swing approaching 600Vpp.

Active loads transform every stage of a tube amplifier. Here are the applications where they matter most, with practical component values from proven designs.