US2338410A - Push-pull amplifier - Google Patents

Description

Jan. 4, 1944. v. M. cousms PUSH-PULL AMPLIFIER Filed May 13, 1942 F/G/ H) AME 1.040

[a LOAD F/G. Z

e R- m AMP. LOAD INVENTOR lZMCOUS/NS ATTORNQQ Patented Jan. 4, 1944 UNITED STATES PATENT OFFICE,

PUSH-PULL AMPLIFIER Van M. Cousins, Chatham, N. .L, assignor to Bell Telephone Laboratories,

Incorporated, New

York, N. Y., a corporation of New York Application May 13, 1942, Serial No. 442,781

2 Claims. (Cl. 179-171) ing applied signal voltage, a decreasing propor-- tion of the total signal voltage is efiective at the grid and non-linear distortion results. Further non-linear distortion is introduced in other ways such as by variations in. the mu of the tube and in the impedance of the signal source.

The object of this invention is to reduce this non-linear distortion.

According to the invention rectifier elements are associated with the inputs of the tubes in such a manner that as the signal voltage varies the proportion of the total voltage which is efiective at the grid is varied to compensate for the non-linear distortion due to one or more of these causes.

In the preferred embodiment of the invention as applied to amplifiers of the class B or class C types, the mid-point of the signal source is left floating with respect to ground and the input electrodes of each tube are shunted by a rectifier unit poled to conduct toward the grid. When the signal potential is positive at the grid of one tube, the rectifier unit associated with the other tube is conductive and as the grid current in the active tube increases the resistance of the conductive rectifier decreases to increase the proportion of the generated signalvoltage across the input of the active tube. By a proper choice of the type of rectifier unit used or by suitably modifying its non-linearity by means of associated resistors, the proportion of the generated voltage which is effective at the grid may be kept constant or varied in whatever manner is necessary to compensate also for the non-linear distortion due to factors other than the variations in the grid impedance. 1

It will be noted that since the impedance of the rectifier units in their conducting direction may be much less than the grid impedance of the tubes, a large proportion of the total signal voltage may be applied alternately to the tubes thereby giving, in addition to the'reduction of to use the mid-point connection to the signal source, compensation for non-linear distortion may be obtained by connecting one of the rectifier units in series with each grid connection to the signal source. The rectifier units or the tube input circuits, or both, may be shunted by suitable resistors for adjusting the non-linearity of the network to suit the requirements of the particular case.

In the drawing, Fig. 1 shows a system including an amplifier according to the invention, Fig. 2 shows the .use of resistors with the rectifier elements to compensate the distortion more accurately and Fig. 3 shows the use of distortion correcting elements in a conventionally balanced circuit.

In Fig. 1 signals from a source I are amplified in the amplifier 2 and impressed on the grids 3 and 4 of thetubes 5 and 6 by means of any suitable coupling elements such as a transformer 'I' having a secondary winding the mid-point of which is left floating with respect to ground.

The tubes 5 and 6 are preferably of the type in which the plate current is cut off, or very small, at zero grid bias. The plates and cathodes are connected by means of a source of power 9 and an output transformer III to a load I l. The cathodes are grounded at I! independently of the winding 8 and the tube inputs are shunted by high resistances l3 and I4 and rectifiers l5 and I6 which are oppositely poled to conduct from cathode to grid in each case.

When the signal potential is positive at the grid 3, the rectifier I6 is a relatively low impedance shunt across the input of the tube 6 and the greater part of the full potential of the winding 8 is impressed on the tube 5 to drive the grid 3 positive with respect to the cathode of the tube and produce a large increase in the plate current in accordance with the signal. Similarly, when the signal potential is positive at the grid 4 of tube 6, rectifier I5 is a relatively low impedance.

across the input of the tube 5 and the greater part of the full potential of the winding 8 is impressed on resistor 4 and the plate current of the tube 6 is correspondingly increased.

Since the grid-cathode resistance Rg of a tube operating with a positive grid is a function of the grid voltage and decreases as the grid voltage increases and the impedance of the signal source is ordinarily more nearly constant, the proportion of the generated voltage appearing between the grid and' cathode of the tube will decrease as the signal level increases thus causing the generation of non-linear distortion products.

If in Fig. 1 the rectifler units I! and ii are of the type having appreciable initial resistance which decreases as the current through them increases, the non-linearity of these units will tend to correct for the variable grid impedanc of the tubes. For example, when a positive signal pulse is applied to grid 3 and the unit It becomes conducting, the grid-cathode impedance decreases as the flow of grid current increases so that the grid-cathode voltage is less than required for distortionless amplification. However, as the current through the unit I6 increases, its impedance decreases thereby increasing the proportion of the total signal voltage applied to the tube 5. Similarly when the'grid 4 i driven positive the impedance rectifier unit i5 decreases and tends to hold up the grid-cathode voltage of the tube 6.

It will be apparent that for maximum correction of distortion the rectifier unit must have a relatively high initial resistance, whereas the full signal potential is applied to the positively driven tube only when the impedance of the conducting rectifier unit is relatively very low at all signal voltages. Copper oxide units or their equivalent are therefore more suitable for di tortion correction and electronic tube rectifiers or their equivalent are better adapted for cases where increased gain is th primary consideration.

In Fig. 2 if the total grid impedances R+ and R- are equal at all signal voltages, non-linear distortion due to grid current will not occur provided the impedance R0 of the signal source is negligible. In practice this is rarely the case and hence the impedance of the conducting rectifier unit and associated modifying resistances must decrease faster than the grid impedance of the positively driven tube in order to compensate for the loss of signal potential inRo' and maintain the grid-cathode voltage at its proper value throughout the signal cycle.

The non-linearity of the resistance-current characteristic of copper oxide units follows approximately the inverse square law whereas the non-linearity of the input resistance characteristic of the tubes with grid current is usually of a lower order so that with such units, or their equivalent, it is possible to over-compensate the tube distortion to the extent necessary to correct for the loss of signal potential due to other causes such as the impedance of the source. In cases where the rectifier units alone, or these units with the addition of resistors l3 and I4, do not give the required degree of non-linearity, their effect may be modified, or further modified, as

required by means of shunt resistors l1, I! or series resistors I 8, 18 or both. It will be understood however that these resistors should be kept as small as other considerations permit in order to obtain as much voltage as possibleat the positively driven grid and thereby reduce the power capacity required in the driver stage.

The correct values of the modifying resistors for optimum compensation will depend on the characteristic of the rectifier units used and the various constants and characteristics of the particular circuit in which they are used. In general it will be understood that increasing the value of resistors l 8 (or decreasing the value of resistors l3 and M or H) make the combined resistance or the elements associated with the input circuits less dependent on the signal voltage and increases the voltage required to produceia given degree of corrective non-linearity. The optimum resistance values will therefore be determined empirically in each case. In cases where correction is being applied to compensate only distortion arising from non-linearity not intimately associated with the input circuit, the phase of the distortion may be such as to require that the rectifiers be poled in the direction opposite to that shown in the drawing.

The non-linear distortion correcting feature of the invention may also be used in conventionally balanced input circuits as shown in Fig. 3. In this case the mid-point IQ of thetranstormer 20 is grounded and the rectifier units shunted by suitable modifying resistors 2|, 22 are serially connected in the leads from the signal source to the grids 3 and 4. tubes may be shunted by high resistors 23, 24 and the rectifiers are poled to conduct toward the grids as in the circuit previously described. In the absence of grid current the resistance of the rectifiers is relatively high but when the positive signal potential at either grid becomes great enough to produce grid current the resistance of the rectifier decreases a the grid current increases. By proper choice of the series and shunt modifying resistors, the non-linearity of the network may be adjusted to compensate for both the non-linearity of the tubes and the loss of potential in the driving source impedance so that power is delivered to the load substantially free of non-linear distortion.

While the circuit of Fig. 3 is satisfactory from the standpoint of non-linear distortion, only somewhat less than one half the signal potential ,is effective at each tube and for this reason the circuits of Figs. 1 and 2 will ordinarily be preferred since they require less power capacity in the signal source or driving amplifier.

What is claimed is:

In an amplifying system, two vacuum tubes having inputand output electrodes connected in balanced relation, an ungrounded source 01' signals connected to the input electrodes and a network associated with the input electrodes of each tube including a rectifier element having a nonlinear resistance current characteristic for compensating for non-linear distortion produced by.

the impedance of the source and by the nonlinearity of the impedance of th tubes when current fiows between the input electrodes of the tubes.

' 2. In an amplifier comprising two vacuum tubes having grid and cathode input electrodes and output electrodes connected to form-balanced input and output circuits, said grids being'adapted to be connected to an unbalanced source of signal potentials, two rectifier elements connected in series-opposed relationship between the grids and forming a non-linearly conducting path from the cathode to the grid of each tube for controlling the proportion of the total voltage of the source applied to each tube to reduce the non-linear distortion produced by operating the tubes with grid current.

VAN M. COUSINS.

The input circuits of the

Images