US2213871A - Thermionic amplifier - Google Patents

Description

p 1940 c. E. STRONG 2,213,871

THERMIONIC AMPLIFIER Filed Jan. 21, 1938 Fig. 3. v

Ml g 1 2a, 000' wow 9 pa n an VOL 7/16! 000 i //V VE N 70R 6. E. STRONG Patented Sept. 3, 1940 1 mer e STATES THERMIONIG AMPLIFIER Charles Eric Strong, London, England, assignor to International Standard Electric Corporation,

New York, N. Y.

Application January 21, 1938, Serial No. 185,062 In Great Britain January 22, 1937 6 Claims.

This invention relates to an improved method of stabilising high frequency amplifiers by the application of controlled feed-back in a particular manner.

In high frequency amplifiers comprising triode valves a tendency to instability exists arising from coupling between the input and output circuits through the interelectrode capacities of the valves, particularly through the plate to grid capacities.

A known method of overcoming this effect consists in neutralising the unwanted coupling by applying additional coupling having an equal and opposite effect. In the case of a pushpull amplifien'for example it is known to add balancing condensers between the plate of each valve and the grid of the other, these condensers forming, together with the valve plate to grid capacities, a balanced bridge about which the .20 input and output circuits are conjugate to each other. In such a circuit the aim is to secure stabilisation by the elimination within practical limits of all feed-back.

An alternative method of obtaining stability consists not in the elimination of all feed-back but in arranging that feed-back should exist and should be in the reverse sense and so applied that the known criterion ior'stability (as enunciated for example by H. Nyquist in the Bell System Technical Journal, January 1932, page 126) should be met. Any coupling inherent in the system tending to cause instability is swamped by the applied reverse feed-back. This invention relates to a novel application of this method. The object of the invention is to avoid certain practical difficulties which are experienced in the adjustment and operation of balanced am plifiers of the general form described above, such difiiculties, for example, as change of balance 40 with frequency due to unbalanced inductance in the bridge circuit, difficulties in the construction and mounting of suitable balancing condensers in the case of a high voltage, high power amplifier, and instability arising from the presence of resonant circuits made up of balancing condensers and unavoidable inductance in the leads thereto.

The method disclosed in this invention consists in arranging that the input or driving circuit is in series with the plate to cathode output is applied to cancel some of the negative feedback if the latter is unnecessarily great.

A better understanding of my invention may be obtained from the particular description made with reference to the accompanying drawing, in which Fig. l is a diagrammatic illustration of a circuit in accordance with the invention,

Figs. 2 and 3 are equivalent diagrams used for explaining the operation of the circuit of my invention, and

Fig. 4 illustrates a modification'of the grid connections in accordance with another feature of my invention.

One form of circuit arrangement utilising this method and having the advantage of requiring no high voltage balancing condensers is illustrated in Fig. l of the accompanying drawing showing a push-pull amplifier comprising two triode valves.

The valves are arranged with their grids connected together and to ground and the driving voltage is applied between the filaments.

In this circuit negative feed-back is applied if the driving circuit has an appreciable impedance to current of the working frequency since the output current at the working frequency from the plate of one valve must pass through the second valve and through the driving circuit to return to the cathode.

With this arrangement there is no direct coupling between the input and output circuits through the plate to grid capacities and hence the use of plate to grid balancing condensers is avoided. There may, however, be other sources of coupling tending to cause instability but these, according to this invention, are overcome by the negative feed-back applied.

For the purpose of obtaining reverse feed-back by reason of the fact that the driving circuit is in series with the plate to filament output circuit the driving circuit has appreciable impedance to currents of the working frequency, the amount of reverse feed-back depending on the value of that impedance.

It is to be observed that this method of connecting and driving valves characterised by connecting the grids to ground and driving the filaments, a method already known, does not necessarily imply the application of reverse feed-back since none would be obtained if the driving circuit consisted of a generator of negligible impedance.

Having obtained negative feed-back in the M manner described, additional positive feed-back may now be applied as shown in Fig. 1 by means,

; for example, of inductive coupling between the output and input circuits and adjustment of this coupling may be used to regulate the resultant negative feed-back finally applied.

. are two three element vacuum valves in push- Considering Fig. 1' in further detail; I and 2 pull, 3 and 4 are tuned input and output circuits inductively coupled by the coil 5 which'is in series with the input circuit. The filaments are heated by direct or alternating current fed through double wound choke coils 6 and 1. These coils havea'high impedance at frequencies at which the amplifier operates.

It will be understood by reference to Fig. 2

that when this amplifier is excited, output current fiows-in a circuit comprising filament to j plate of ,one valve, tuned output circuit, plate to filament of the second valve andfilament of the second valve through the input circuit to the filament of the first valve.

erablyreduced due to negative feed-back, and

if the negative feed-back overcomes the positive feed-back from existing sources to the requisite extent, the amplifier will be stable.

The virtue of the circuit lies in the fact that notwithstanding the presence of two tuned circuits, the phase of the reverse feed-back applied in the manner indicated is not shifted through 180 for any frequency and, therefore, when no added positive feed-back is applied, the circuit does not meet the conditions for instability at anyfrequency even when the gain reduction due to feed-back is high.

The negative feed-back being of such a relatively invariable nature'can effectively overcome and swamp other sources of feed-back which may exist in the circuit and which may change in phase relatively more rapidly with frequency.

For example, in the circuit of Fig. 1 if there should exist stray capacity between the plates and filaments this would give rise to feed-:

back, the phase of which would change rapidly with frequency resulting in a tendency to instability. This feedback could be overcome by negativefeed-back applied in the manner'proposed.

V The fact that the negative feed-back applied in the manner described is relatively stable will be made clear by reference to Fig. 3, representing the case of a single sided amplifier, in which i is a generator representing the source of plate circuit E M. F., 2' is the equivalent internal resistance of the valve, 4 is the tuned output circuit having an impedance at resonance of 8,000 ohms, 3 is the tuned input circuit, 8 is the output circuit load resistance and 9 is the equivalent grid to filament circuit resistance. Typical values are shown corresponding to the case of an amplifier delivering 8 kw. with an efficiency of 50%. age is the voltage due to the current delivered by the generator passing through the driving circuit 3;

The reverse feed-back volt- For the values taken the current delivered by the generator is 1.0 ampere and the reverse feedback voltage is 2,000.

If we suppose the amplification factor of the i exaggerated case in which the output tuned circuit is replaced by an inductance having an impedance of the same absolute value as that of the output circuit at resonance. The generator cur- .'rentnow lags behind the E. M. F. by 34 since the inductance is in series with the valve resistance and driving circuit resistance which togethervare 1.5 timesthe impedance of the inductance. For this case the feed-back voltage is de-phased by 34 only.

Know the driving circuit is also detuned and if we assume the exaggerated case of the circuit having a capacitive reactance equal to its impedance at resonance, the total reactance in the circuit will be 8,0002,000=6,000 ohms inductive, and the current will lag the E. M. F. by 31,

also the voltage across the driving circuit will lag the current by a further 90 giving a total phase shift compared to the E. M. F. of 120 approximately. But instability could only arise from this feed-back if the phase shift was 180.

'In the circuit of Fig. 1 it may arise owing to practical limitations that there'is inductance in the leads connecting the two grids. If that is so and if there is appreciable plate to grid and filament to grid capacity, the inductance forms a coupling element between the input and output circuits. In order to regulate this coupling or if desired eliminate it for any given frequency, a double condenser 0 may be interposed in the grid to grid connection as shown in Fig. 4. By adjustment of this condenser it is possible to cancel out the inductive reactance at a given frequency so that the grids may be effectively connected to ground through negligible impedance.

What is claimed is:

1. A thermionic amplifier which is stable over a large frequency range comprising a valve having grid, cathode and plate; electrodes, means connecting said grid electrode to ground, an input circuit for applying input voltage to said cathode electrode said input circuit having a substantial impedance to waves in, said frequency range, an output circuit coupled to said plate electrode, 7 and completed through said cathode electrode, whereby a feedback in a reverse sense with respect to said input voltage is set up by said output current, and positive feedback means coupling said input and output circuits for regulating the amount of resultant reparticular frequency range, comprising a pair of valves arranged in push-pull, each of said valves having grid, cathode and plate electrodes, means connecting said grid electrodes to ground, an

input circuit for applying input voltages between said cathodes, said input circuit having a high to produce a positive feedback for regulating the resultant amount of reverse feedback.

5. An amplifier according to claim 3 further comprising variable capacity means in said means connecting said grids to ground, for substantially reducing inherent inductive coupling between said input and output circuits due to the inductance in said leads.

6. An amplifier according to claim 3 in which said input and output circuits are tuned approxi- 10 mately to anti-resonance.

CHARLES ERIC STRONG.

Images