The AC30 push-pull power amp includes four EL84 pentodes, two for each phase.
The plate and screen supply voltages tend to be in the neighborhood of
VPP = 345 volts
VSS = 320 volts
depending on tubes and AC line voltage. The parts values are
CG = 0.15uF
RGS = 1.5k
RG = 220k
RS = 100
RK = 47
CK = 250uF
The stage is self-biased, so negative DC grid bias is created by raising the cathode voltage relative to ground via the cathode resistor RK. The voltage drop across the screen resistors is negligible at idle, so we can assume the screen voltage, measured relative to the cathode, is 320 volts minus the voltage across the cathode resistor. We can get a close estimate of the DC grid bias by using the triode-connected curves.
The power tubes in the AC30 are not triode connected, so the graph is only valid for the unique situation where the plate voltage equals the screen voltage. The advantage of using this plot is that the "plate current" represents plate and screen current combined, which is exactly what is flowing through the cathodes and RK to create grid bias. There are four tubes providing cathode current, so the effective cathode resistor value for one tube is (4)(47) = 188 ohms.
According to Ohm's Law, to create a DC grid bias of -8 volts, the cathode current needs to be
IP = 8 / 188 = 43mA
To create a DC grid bias of -10 volts and -12 volts, the cathode current needs to be
IP = 10 / 188 = 53mA
IP = 12 / 188 = 64mA
This produces the blue grid line shown on the curves. For a grid voltage of minus 10 volts, the screen and plate voltage is 310 volts, as shown by the red line. It intersects the blue line very close to VG = -10, so we conclude that for a screen voltage of 310 volts and a plate voltage of 310 volts, the DC grid bias voltage is minus 10 volts. For a plate supply voltage of 345 volts, the plate voltage is 335 volts, which is close to 310. Moreover, plate and screen currents are much more dependent on screen voltage than plate voltage.
We conclude that the DC operating point is approximately
VGQ = -10
VPQ = 335
VS = 310
IK = 55mA
which represent the quiescent grid voltage, plate voltage, screen voltage, and cathode current.
To determine the proportion of plate and screen current, we can use the transfer characteristic curves.
The solid black curves depict the plate current versus grid voltage for a screen voltage of 250 volts (upper curve) and 210 volts (lower curve). The dashed curves show screen current. We've added a blue line to denote the DC grid bias voltage. The red line is cathode current. Curves for a 310-volt screen aren't shown but we can imagine that they would be in the vicinity of the green lines.
Where the green lines cross the blue line we observe the plate and screen currents
PPQ = 50mA
ISQ = 5mA
which sum to equal the cathode current.
The cathode resistor RK carries a lot of current and at idle it dissipates
P = (4)(10)(55mA) = 2.2 watts
Vox specifies a 10-watt value for this position and ensures that it gets plenty of air.
Plate and screen dissipations are
P = (335)(50mA) = 17 watts
P = (310)(5mA) = 1.6 watts
This compares to the EL84 limits of 12 watts and 2 watts, respectively. The AC30 actually exceeds the plate dissipation limit by about 40 percent.
The transconductance of the EL84 is around 10 milliamps per volt. Two tubes in parallel per side gives us 20 milliamps per volt. The cathode impedance is the reciprocal of this value in parallel with RK, creating an impedance of 24 ohms in parallel with CK. The -3dB bass cutoff frequency is therefore 27 Hertz, so the cathode is fully bypassed, which is typical for Class AB amplifiers.
Here are the plate characteristic curves for a screen voltage of 300 volts, which is close enough to 310 volts for our discussion of output transformer primary impedance.
The DC operating point is represented by the red circle. It is above the 12-watt plate dissipation limit, so unlike a typical 6L6 or EL84 Class AB amp, the tubes run fairly hot at idle. At full power one tube grid reaches zero grid volts as the other reaches -20 volts, putting it into cutoff, so by definition the amp is not pure Class A, although it is certainly closer than a Plexi. Vacuum tubes vary substantially and the curves represent only average values, so it is conceivable that the non-conducting tube could still be conducting a trickle of electrons at minimum grid voltage.
The AC30 output transformer is 4k plate-to-plate, but there are two tubes in parallel per side, creating an effective 8k plate-to-plate per pair of tubes in push-pull. This is in agreement with EL84 data sheet recommendations for a pair of tubes operating in Class AB. The impedance creates 2k for the conducting tube, which is depicted by the red line. If we were to assume pure Class A operation with both tubes conducting at all times, the effective impedance per tube rises to 4k, as depicted by the blue line, which would intersect the zero-grid-bias curve well below the knee. We can conclude that Vox's 8k output transformer impedance was selected to match a Class AB amp.
From the load line we see that a full-power sinewave input signal causes the plate voltage to swing from a value of 335 volts at idle to a minimum of 40 volts, representing a negative swing of 295 volts. The net plate current swings from zero at idle to a maximum of 147 milliamps. The power output is therefore
P = (0.5)(335)(147mA) = 25 watts
measured at the transformer primary.
Here is a screen-shot of the bass response and the bias excursion characteristics based on our Grid Bias Excursion Calculator.
Because there are two tubes per phase, the effective gridstopper resistance is half the value of RGS, or 750 ohms, which is close to the more common part value of 820 ohms. The -3dB bass cutoff frequency is only 4 Hertz because of the unusually large 0.15uF coupling capacitors. This creates a relatively slow bias recovery time of 43 milliseconds.
1Richard Kuehnel, Vacuum-Tube Circuit Design: Guitar Amplifier Power Amps, (Seattle: Pentode Press, 2008).
Copyright © 2005-2015 Amp Books LLC All Rights Reserved