The tuned-plate-tuned-grid (TPTG) oscillator, as its name implies, utilizes tank circuits in both the plate and grid circuits.
Feedback from output to input, a necessary function in any self sustained oscillator, is accomplished by using the interelectrode capacitance between the plate and grid. Figs. 1 and 2 show the circuit and the currents flowing for each half-cycle of operation. The circuit components are as follows:
L1-Grid tank inductor.
C1-Grid tank capacitor.
C2-Grid coupling capacitor.
R1-Grid-leak bias resistor.
V1-Oscillator tube.
L2-Plate tank inductor.
C3-Plate tank capacitor.
C4-0utput coupling capacitor.
C5-Decoupling capacitor.
The output voltage of the oscillator is capacitively coupled, via C4, to the next stage. Capacitor C5 acts as a decoupling filter to keep pulses of plate current from entering the power supply and affecting its output voltage. There are three main groups of electron currents whose movements, if analyzed, will lead to understanding the operation of this oscillator circuit. These groups might be labeled as follows:
1. Alternating radio-frequency currents.
2. Unidirectional radio-frequency currents.
3. Pure direct currents.
Fig. 1. Operation of a tuned-plate-tuned-grid oscillator-first
half-cycle.
Fig. 2. Operation of the tuned-plate-tuned-grid oscillator-second
half cycle.
ALTERNATING RADIO-FREQUENCY CURRENTS
There are two tank currents in the alternating RF category.
One is in the grid circuit and is shown in dotted blue; the other is in the plate circuit and is shown in solid blue. The feedback currents, which are driven by the plate tank current, appear in dotted red; and the output current, which provides the driving current and consequently the driving voltage for the next stage, is in gray.
For successful self-oscillation, the two tank circuits must be tuned to approximately the same frequency. When the tube is first turned on, the initial surge of plate current sets up oscillation in the plate circuit at its natural, or tuned, frequency. Re call that even a single surge of current or a sudden voltage change will cause any tuned circuit to oscillate at its natural frequency. Even though the oscillation is not sustained by further voltage or current changes, it will continue for several cycles before the initial energy is expended. The purpose of oscillator circuitry is to continue the oscillation indefinitely by providing such a voltage or current change, usually once each cycle. This repetitive action is provided here by pulses of plate current.
As the initial surge of tube current reaches the plate circuit, its voltage-current conditions will correspond roughly to those in Fig. 2. The voltages at the plate of the tube, the entrance to coupling capacitor C4, and the top of the tuned tank will all be positive, as indicated by the plus signs, but these voltages will be lower in value than the supply voltage present on the other side of the tank.
Once this uneven distribution of current and consequently voltage exists across the tuned tank circuit, the charge will re distribute itself in an attempt to overcome the unbalance. Since the lower plate of tank capacitor C3 is more positive than the upper plate, current will flow from top to bottom. If there were no inductance or resistance in the current path, this redistribution would occur instantaneously. However, the primary characteristic of any inductance is that it tends to oppose any change in current: If no current is flowing, an inductance tends to op pose any build-up; and once current flows, the inductance tends to prevent it from decaying.
These properties of inductance should enable us to see why an oscillation is set up when electrons are redistributed. Instead of taking place instantaneously, the current requires the equivalent of a quarter-cycle of oscillation to build up from zero to maximum. After maximum current is flowing, the inductive effect will try to keep it from dying out. At this instant the voltages are the same on each side of the tank capacitor, meaning both charges have been equally distributed. The current, however, requires another quarter of a cycle to decay to zero. At the end of this cycle the charge again is unevenly distributed, but in the opposite direction. The flow of electrons from top to bottom during the next quarter cycle--after the voltages on the two capacitor plates have been equalized-has charged the bottom plate to a lower positive voltage than the top plate.
The first half-cycle is shown in Fig. 1. The greater number of plus signs on the upper plate indicates that midway in the first half-cycle the upper plate is more positive than the lower one. This is confirmed by the sine-wave representation of tank voltage in Fig. 3, which shows the voltage at the top of the tank-in this case, also the point where output voltage is taken off.
Fig. 3. Current and voltage waveforms in the tuned plate-tuned-grid
oscillator.
Fig. 2 depicts current-voltage conditions during the second half-cycle of oscillation. The voltage unbalance across the tank capacitor will again attempt to neutralize itself in the following manner: Since there is an excess of positive ions on the upper plate, current will be drawn from the lower plate, through tank inductor L2. Again the inductor will oppose both the build-up and the decay of electron current. As a result, redistribution of the electric charge will again go too far-the voltage unbalance across the tank will be reversed a second time, with the top plate now less positive.
In order to see how this oscillation sustains itself indefinitely, we must now consider what has been happening in the rest of the circuit. During the single cycle just described, certain inevitable losses will have occurred. Hence, at the end of the first complete cycle, the voltage difference between the two plates of tank capacitor C3 will be smaller than it was at the beginning. An other way of visualizing this condition is to consider that fewer electrons will complete the cycle than started it, some dropping out because of internal circuit resistances. These losses, which must be replenished before the next cycle begins, are supplied by "turning on" the plate current at the appropriate moment.
Let us see how this is accomplished.
The oscillating current (shown in solid blue) in the plate tank circuit feeds three external paths, or loads, These paths are de coupling filter capacitor C5 ( also shown in solid blue) , output coupling capacitor C4 (shown in gray), and the feedback path to the plate and grid, shown in dotted red. (The interelectrode capacitance between these two elements couples the feedback to the grid.) During the first half-cycle in Fig. 1, this feedback current draws electrons away from the plate. In Fig. 2 the polarity of the oscillating voltage is reversed and feedback cur rent is driven back toward the plate, as shown by the arrow on the dotted red feedback line.
There are three alternate paths in the grid circuit, and current flows in each one, in response to the feedback current in the plate circuit. These grid components are also shown in dotted red to help tie them in with the feedback current, and also to differentiate them from the grid-driving current (in dotted green) and the grid-leak current (in solid green) which are also flowing in the grid circuit.
Note how all three components of the feedback current are flowing in unison. During the first half-cycle they are all drawn to the right, and during the second half-cycle they are all flowing to the left-both times being driven by the plate tank voltage.
These current components are said to be in phase with the plate tank voltage.
The components of current to the left of the grid capacitor actually "deliver" the feedback pulse to the grid tank circuit and thereby set up an oscillation of appropriate phase to support the plate-circuit oscillation. The oscillation current in the grid tank has been shown in dotted blue.
The oscillation in the tuned-grid circuit will build up to a maximum strength determined by:
1. The strength of the feedback pulses from the plate circuit.
2. The amount of losses during each cycle.
The grid-circuit oscillation will have internal losses due to electrical resistance, dielectric leakage, etc. Normally they will be very small, only a fraction of one per cent each cycle. Thus, this is a "high-Q" circuit as explained previously.
In addition, the oscillation actually drives the control-grid volt age to its two extremes by sending the grid-driving current (shown in dotted green) up and down through grid resistor R1.
This current flows in phase with the oscillating voltage in the grid tank and thus acts as a load on the oscillation by adding to its total losses during each cycle.
The oscillation will build up until the total losses during each cycle are equal to the energy supplied by the feedback pulse.
When energy lost equals energy supplied, the grid oscillation will become stabilized.
UNIDIRECTIONAL CURRENTS
Currents which flow essentially in only one direction are classified as direct, or unidirectional, currents. In our circuit they are the pulsating DC of the plate current (shown in solid red), and the grid-leakage current (shown in solid green). The plate current replenishes each cycle of oscillation, and the grid leak current provides the grid voltage ( also called operating bias) for the tube.
In any vacuum tube, electrons in the tube stream will tend to strike the control grid whenever it is more positive than the cathode. We have already seen how the grid voltage is made positive, midway in the second half-cycle, by the grid tank oscillation. Electrons (shown in solid green) will now be attracted to the grid wires and leave the tube via the control grid. Thus, three separate electron currents are flowing in the grid circuit.
Shown in solid green, dotted red, and dotted green, they represent the grid-leak, feedback, and grid-drive currents respectively.
Fig. 3 shows the waveform for the feedback current as being in phase with the plate tank voltage. This means that during the first half-cycle (Fig. 1) the plate tank voltage, being at its most positive value, draws the electrons to the right, or toward the high positive voltage. Conversely, during the second half-cycle the plate tank voltage has its lowest positive voltage and repels the electrons of the feedback current, moving them to the left as in Fig. 2.
The grid-leakage electrons accumulate on the right plate of the grid capacitor and build up a permanent negative voltage there, as indicated by the solid green minus signs. These electrons drain continuously downward through grid resistor R1, the amount of current depending on the quantity of electrons in storage, the size of the grid capacitor, and the resistance of grid resistor R1.
This current flow through the resistor is pure DC; consequently, it is represented by a solid green line in both Figs. 1 and 2. Grid resistor R1 and grid capacitor C2 form a conventional long time-constant R-C combination so the grid-leak volt age will remain steady in the face of the pulsating electron cur rent coming to it from the tube. This current enters the grid capacitor during the second half-cycle only, when the control grid has been driven positive. (During the first half-cycle the control grid has been driven negative and no grid-leak electrons can leave the tube.) The amount of grid-leak voltage can be computed from two separate formulas. The first one, known as Coulomb's law, states that: Q=CxE where, Q is the quantity of the charge in coulombs, C is the value of the capacitor in farads, E is the voltage in volts.
The second formula-much more widely used-is Ohm's law, which states that: E=IxR where, E is the voltage across a resistor in volts, I is the current flowing through the resistor, as a result of that voltage, in amperes, R is the resistance of the resistor in ohms.
The presence of this fixed biasing voltage at the grid accounts for the fact that the grid voltage is always lower than the grid tank voltage driving it. The voltage at the top of the grid tank fluctuates around zero as a reference point, whereas the voltage at the grid fluctuates around the negative biasing voltage, represented by the solid green minus signs. The control-grid voltage momentarily becomes positive in the middle of each half-cycle and allows leakage electrons to leave the tube via the control grid.
The plate current will flow for a longer part of each cycle than the grid current. For every value of plate voltage there is a negative grid voltage below which plate current cannot flow and above which it can.
Note that the tube conducts electrons when the plate voltage is at or near its lowest value. In the middle of the second half cycle, for instance, we see the lowest concentration of plus signs representing positive ions-at the plate and at the top of the tuned tank. This condition is brought about by the oscillating electrons in the plate tank circuit of course, and can be confirmed from the sine waves of voltage in Fig. 3. Each pulse of plate current arrives at the top of the tank and adds to the oscillating electrons concentrated there, thus replenishing the oscillation. The amount of this reinforcement must compensate exactly for the internal resistance losses and the output and feedback loads faced each cycle by the oscillation.
POWER-SUPPLY DECOUPLING
In Figs. 1 and 2, capacitor C5 is placed in parallel with the power supply to sidetrack, or decouple, large fluctuations in cur rent before they reach the power supply. Otherwise, in flowing through the power-supply filters, these currents could cause corresponding voltage fluctuations which would be reflected into other vacuum-tube stages. It was shown previously that the oscillating electrons in the plate tank will flow out along any available path, such as the line to the power supply. When a capacitor is placed in parallel with this line (as C5 has been), the oscillating current will choose this alternate path because of its lower impedance. Thus, most of the current fluctuations are diverted harmlessly into C5.
This decoupling current is shown in solid blue so you can see its relationship to the oscillating tank current driving it. The de coupling network constitutes one more load, or loss, for the oscillating voltage, along with the feedback and output currents de scribed previously. A small decoupling resistor is often added in the power-supply line to provide additional filtering. Even without it, the power supply impedance and the filter capacitor constitute an effective filtering combination.
TANK-CIRCUIT TUNING
In this oscillator circuit the plate tank must be tuned "slightly inductive" with respect to the grid tank. In other words, the plate tank should have a somewhat lower resonant frequency.
One way of accomplishing this is to add more inductive reactance to the plate tank by increasing the inductance of L2.
However, it is also possible to lower the resonant frequency of a tuned circuit by adding capacitance and thereby lowering the capacitive reactance. Now the inductive reactance is greater, and the circuit will again be tuned slightly inductive as before.
Conversely, tuning a circuit "slightly capacitive" means to increase its natural frequency. Here the capacitance must be lowered in order to increase the capacitive reactance of the circuit. As before, the same result would be obtained by lowering the inductance to decrease the inductive reactance in the circuit.
Figs. 1 and 2 give no hint of a frequency difference between the two tanks. If the phase relationships were exactly as shown in these diagrams, the circuit would be unable to support its own oscillation for these reasons: The plate current reaches the plate tank at the precise moment it can give the most support to the oscillation in the tank. However, the feedback current from the plate tank will deliver a pulse to the grid circuit at the wrong instant to support the grid-tank oscillation. In fact, the oscillation will be dampened because the current-voltage combination in the grid circuit is always exactly out of phase with the grid tank voltage. The dotted red arrows in Figs. 1 and 2 represent the feedback current in the grid circuit. As you can see, it is flowing in the opposite direction from the external grid-driving cur rent produced by the grid tank oscillation. It is likely, under these phase conditions, that the oscillation in the grid tank would not be allowed to build up at all.
When the oscillation in the plate tank is lower than the resonant frequency of the grid tank circuit, the feedback current will "see" two different impedances in the grid tank-one in the direction of the tank capacitor, and the other in the direction of the tank inductor. At the lower feedback-current frequency, the inductor will have much lower reactance than the capacitor, so most of the feedback current will be shunted into the inductor path.
Because the two tanks have different natural frequencies, neither operates at its own resonant frequency. The pulses of plate current, released once each cycle at the grid-tank current frequency, will arrive slightly early for maximum reinforcement of oscillation in the plate tank. This will shorten each cycle of oscillation by hastening the end of one and the beginning of the next one. Instead of being a true harmonic waveform, each cycle will be somewhat distorted, and the plate oscillation will occur at slightly higher than the resonant frequency of the plate tank.
By similar reasoning, the grid oscillation is slightly lower than the resonant frequency of the grid tank. This oscillation is sustained by the feedback pulses coupled from the plate to the grid via interelectrode capacitance. The feedback current, in dotted red, is driven by the tank voltage in the plate circuit and must stay in phase with it at all times.
It is impossible to show, in Fig. 1 and 2, how the phase relationship of the feedback current is able to support the grid oscillation. This can be demonstrated graphically and mathematically, but it would require extraordinarily complex waveforms and computations.
It is sufficient to say that the feedback current has a slightly lower frequency than the grid oscillation and can thus provide sufficient "kick" to sustain oscillation during each cycle. In the process the feedback current, itself driven by a distorted current waveform, manages to distort the oscillating current waveform in the grid circuit. Also, the feedback current lengthens each cycle so that oscillation will occur slightly below the resonant frequency of the grid tank.