This is an article from our archives that is no longer part of our active site.

The Gibson GA-5 was created in 1954 as a low-power practice amp to accompany the Les Paul Junior guitar. It has three vacuum tubes: a 5Y3 full-wave rectifier, a 6V6/6AQ5 beam power tetrode, and a 6SJ7 pentode. The GA-5 is a highly regarded studio amplifier. Moreover, its power amp design provides an excellent example of how output transformer primary impedance affects power and distortion. Let's take a closer look.

Here is the power amp and power supply.

According to an early GA-5 schematic, the plate is at 345 volts relative to ground and the screen is at 290 volts. The cathode voltage is 17 volts relative to ground, so the plate-to-cathode and screen-to-cathode voltages are 328 and 273 volts, respectively. The blue curve shown here on the transfer characteristics is our estimate for a 273 volt screen.

The green line represents a grid voltage of minus 17 volts. From where it intersects the curve we get an idle plate current estimate of 38mA. By examining the transfer characteristics for screen current at the same grid voltage, shown below, we observe that the idle screen current is approximately 3mA.

This makes the total current through the cathode at idle equal to 41mA, so by Ohm's Law the voltage drop across the cathode resistor is (41mA)(470) = 19 volts, only 12 percent higher than what the curves for an average 6V6 tell us. This is reasonable. The schematic shows a 70-volt drop across the 22k power supply resistor, indicating that it carries only 3mA, which includes screen current and preamp supply current. (The original GA-5 preamp has a 6SJ7 operating at a screen voltage of only 20 volts, drawing only meager amounts of current, so there is no surprise here.)

There is a 20 volt drop across the 470-volt power supply resistor, suggesting that it carries about 43mA. Close enough. Finally, an RCA data sheet for the 5Y3 full-wave rectifier tube indicates that for a current load of 41mA the output voltage for a 340-0-340 power transformer is slightly less than 400 volts, which is close to the 380 volts shown in the GA-5 schematic. Plate dissipation at idle is (328)(38mA) = 12 watts, right at the limit of the 6V6.

According to our RC Ripple Filter Calculator there is 11dB of ripple attenuation for 60-Hertz AC across the 470 ohm power supply filter resistor and 44dB attenuation across the 22k resistor.

At the cathode there is 470 ohms in parallel with the reciprocal of 6V6 transconductance, which is about 4mA/V. The effective cathode impedance is therefore 163 ohms. This makes the -3dB cutoff frequency for the 20uF cathode bypass capacitor equal to 49 Hertz. So the cathode is fully bypassed for guitar frequencies. This means that the input signal amplitude needed to drive the power amp to full power is equal to the DC grid bias, so we conclude that the GA-5 power amp has an input sensitivity of about 20 volts.

Here are the 6V6 plate characteristic curves for a grid voltage of zero.

The blue curve depicts an estimate for a 273 volt screen and the DC operating point is represented by the red dot. The load line for an 8k output transformer (the red line) intersects the diode line well below the knee of the curve. When the grid voltage rises from an idle value of minus 17 volts to a maximum of zero volts, the plate voltage drops by 305 volts and the plate current rises by 38 milliamps, so the output power before transformer losses is (38mA)(305)/2 = 5.8 watts. We can expect 4 to 5 watts under realistic conditions. The right side of the graph beyond 500 volts is not shown, but we can imagine that fairly symmetrical conditions prevail on negative grid voltage swings.

For comparison, here is a recommendation found in a GE data sheet, where the screen is at 225 volts.

This design is typical for tubes like the 6V6, because beam power tetrodes, unlike power pentodes, have well defined knees where the steep diode line transitions to a horizontal line. Grid voltage has little effect on plate current below the knee. Plate voltage has little influence to the right of the knee. Since power is equal to the product of voltage swing and current swing, maximum power is achieved when the load line is placed through the knee. Above it the current swing remains nearly constant but voltage swing diminishes. Below it the opposite occurs.

The effect on power amp distortion is quite significant, as demonstrated by the 6V6 plate characteristics shown here.

These curves are for 250-volt screens. At the top there is a wide separation between the zero grid voltage curve and the minus 5 volt curve. (Compare this to the separation of the minus 25 volt and minus 30 volt curves, for example.) A load line placed through the knee generates substantially greater output swing in this region, as shown by the red line. This creates second harmonic distortion - the amplification of input signals swinging positive is greater than for signals swinging negative.

The green line demonstrates that this type of distortion is reduced when the output transformer primary impedance is increased to place the load line below the knee. The zero and minus 5 volt curves are closer together, so there is less difference between the positive and negative extremes. There is, however, less amplification at both the positive and the negative extremes of signal swing compared to near the DC operating point. This produces third harmonic distortion. Result: less second harmonic, but greater third harmonic when the load line intersects below the knee.

When the output transformer primary impedance is selected to place the load line through the knee, maximum output power is achieved along with generous quantities of second harmonic distortion. As the impedance increases beyond this point, output power and second harmonic distortion decrease. The amplitude of the third harmonic, on the other hand, increases and eventually takes over as the largest contributor to distortion.

Copyright © 2018 Amp Books LLC All Rights Reserved