The Cathode Follower and Feedback

Author: N.H. Crowhurst

The cathode follower may be regarded as a special example of feedback - all of the output voltage ap] tears in the input circuit, hence (3 - 1. If the tube has a working gain of 50 with the resistor values used, an input fluctuation of 51 volts will give an output fluctuation of 50 volts. Following the feedback principle, the effective value in the input circuit of any resistor connected between grid and cathode will be multiplied by 51. This also applies to the reactance of any capacitance between grid and cathode of the tube. The reactance of a capacitance is inversely proportional to its capacitance value, hence the feedback divides the effective capacitance by 51.

The cathode follower circuit is a special application of negative feedback.

This characteristic makes the cathode follower a very convenient tool for reducing the output impedance of the amplifier. The output impedance in the absence of feedback would be the plate resistance in parallel with the plate load resistor, in this case, about 50,000 ohms. However this gets divided by the feedback factor (in i:his case 51) to give an effective resistance of less than 1000 ohms.

Adding extra stages gives more gain, so more feedback could be used, except for one thing: each extra stage also increases the phase shift problem.

As amplifiers use more and more feedback to get the distortion down to even lower proportions, the problem arises of getting enough amplification to "throw away." The more gain required, the more stages that have to be added, and the more possibility of phase shift we encounter.

One way of achieving the extra gain without adding extra stages is to use positive feedback. Care must be taken to see that the beneficial effects of the negative feedback - reduced distortion - are not cancelled in the process. The secret is to use positive feedback to boost the gain in a part of the amplifier that has very little distortion. Then increased negative feedback can be used over the whole amplifier to reduce distortion elsewhere.

Use of positive feedback avoids adding of extra stages

Positive feedback can only be used over one stage in order to avoid oscillation. An easy way to accomplish this positive feedback is to couple cathode bias resistors of two consecutive stages in the earlier part of the amplifier, where the distortion is small.

If the reduction of current through this positive feedback resistor is greater than the increase in the plate current from the first tube, current in this resistor will drop (also the voltage); increasing the voltage fluctuation between grid and cathode and thus causing positive feedback.

A momentary positive fluctuation at the grid of the first stage will produce a momentary negative fluctuation at the plate, which is passed on to the grid of the following stage. This produces another positive fluctuation at the plate of the second tube. At the same time, the negative fluctuation at the plate of the first tube, resulting from increased plate current, will be accompanied by a positive fluctuation at its cathode. Similarly, a negative fluctuation appears at the cathode of the second tube. The fluctuation at the cathode of the second tube is much bigger than that at the first stage. Connecting the resistor between the cathodes will allow some of the fluctuation from the cathode of the second to cancel the fluctuation at the cathode of the first stage.

Using both positive and negative feedback loops

If the voltage fluctuation fed back caused the cathode of the first tube to move negative by as much as the positive initial fluctuation at its grid, oscillation would take place. (A negative fluctuation at the cathode is equivalent to a positive fluctuation at the grid, hence feedback would provide the total input.) If we use two feedback loops, one positive and one negative, things are not so difficult. Without the negative loop, the initial voltage between grid and cathode of the first stage could feed back enough to cause oscillation, but negative feedback supplies a voltage at the grid that opposes the the initial fluctuation, preventing oscillation from taking place.

Assume that we have an initial fluctuation of 1 volt between the grid and cathode of the first stage. The positive feedback at the cathode that could cause oscillation would also be 1 volt. Now suppose that we provide an amount of negative feedback that in the absence of the positive feedback would give a 20-db (10:1) gain reduction. If the positive feedback were not present, we should require a total input of 10 volts instead of 1 volt, the additional 9 volts being required to offset 9 volts negative feedback to the cathode. When both the positive and negative feedback are used, however, the resultant voltage or the feedback effect at the cathode is only 8 volts of negative feedback. This means that the total input need only be 9 volts instead of 10 volts, and that the positive feedback has reduced the amount of gain lost by 10:9 (very nearly 1 db). Alternatively, we could say that when the negative feedback is added, the effect of positive feedback that could increase the gain of the stage to infinity and cause oscillation is reduced to an increase of less than 1 db.

Development of unity coupled output circuits

Until feedback came along, the choice for output tubes was between triodes and pentodes. Pentode operation is much more efficient in terms of audio power output for the power input, but it is far more critical of being operated with exactly the right load resistance value than when the same tubes are triode connected.

This led to two basic variations in output circuits, although many further minor variations have developed. The first, called "unity coupled," can best be thought of as a "half-way" cathode follower. Assume that we use a pentode tube that needs a 12-volt audio input to produce a 150-volt output across the load coupled to the plate. To go wholly cathode follower would require an input voltage of 150 + 12 = 162 volts to get the power represented by 150 volts across the load coupled in the cathode circuit. But by coupling the load so that the plate circuit feeds half the power and the cathode half, an audio voltage of 75 volts will appear at each. Now the input audio voltage needed is only 75 -f 12 = 87 volts.

To work as a pentode, the screen must always be at a constant voltage "above" the cathode. This can be achieved by using a multiple-wound transformer. One push-pull primary connects to the cathodes of the tubes, with its center tap to the ground. The other, of exactly equal turns, connects to the screen, with its center tap to B+. This insures that the audio voltage on the screen is the same as that on the cathode. For the plate to deliver its half of the power, it must produce an equal but opposite voltage, so the plates are "cross-connected."

One practical unity-coupled push-pull circuit

Development of an ultra-linear circuit

The other special output circuit is called ultra-linear. We can visualize the operation as half-way between pentode and triode connection. When a tube works as a pentode, the screen voltage remains steady. The only changing voltages in the tube are the grid and plate potentials. To make the same tube work as a triode, the screen is connected to the plate. This means that when the grid goes positive, the plate current rises, making the plate voltage drop. As the screen is also connected to the plate voltage, this goes negative, tending to offset the plate current rise.

Connecting a pentode to make it work as a triode is like applying negative feedback from the plate to the screen. For ultra-linear operation, the screen is connected to a tapping on the output transformer winding so its audio voltage swings the same way as the plate, but not as much. Thus ultra-linear can be regarded as using less negative feedback from plate to screen than occurs to convert the pentode to triode operation. This means that the advantages, too, are split. Most of the efficiency of pentode working is retained, without its being so critical of having the correct output load resistance.

A practical ultra-linear push-pull circuit