US 2281238 A
Description (OCR text may contain errors)
April 1942- Q. E. GREENWOOD 2,281,238
FEEDBACK AMPLIFIER Filed May 1, 1940 IN Cu 7' lNl/EN TOR Q 5. GREENWOOD A T TOR/VEV Patented Apr. 28, 1942 2.281.238 rsunmcx Am'rn Quentin E. Greenwood, Flushing, N. 2., usignor to Bell Telephone Laboratories, Incorporated, New York, N. Y a corporation of New York Application may 1, 1949, Sell No. 332,845
(ill. l79--171) This invention relates to feedback amplifiers. An object of the invention is to reduce modulation' produced by an amplifier, especially modulation components in the amplifier output originating in diflerent stages or cascaded plliying 1,
portions of theamplifler.
In accordance with the invention as applied, for instance, to a two-stage amplifier, the modulation is reduced to a low value by means of mul tipl feedback paths, the reduction being accomplished, for example, as described hereinafter, by so adjusting the feedback circuits that the distortion voltages arising in the first and second stages partially cancel each other at the repeater output.
Other objects and aspects of the invention will be apparent from the following description an claim.
The two-stage amplifier is shown schematically in the single figure of the drawing. Each stage may contain one or more tubes, and the tubes may he cascaded in the stage. v backs through the B1 and Br paths are adjusted to give the low modulation condition while the 5: feedback can he used to give the amplifier a frequency characteristic and the proper stability, and to further reduce the modulation.
Referring to the drawing, the expression for the output distortion voltage Do is found to be and second stages, p1, pi, e: are the losses in thefundamental voltage applied at the input grids of the first and second stages, respectively. These fundamental voltages difier by the factor 7 m thus the ratio A D can he expressed 4 "mfi2) man D2 f2( r) f2( 2) where e1 and e: are the amplitudes of the fundamental voltages at the first and second stag inputs. Substituting this into Equation 2, the condition for zero output distortion becomes,
where A1 and A: are circuit constants. Substituting in Equation 2, the condition for minimum second order distortion becomes mfi:] i "lBl -+P2[ m XA2 Similarly, for a circuit in which the third order terms are relatively large, the condition for minimum distortion would be since the third order terms. are approximately proportional to the input level cubed.
Suitable values for mp1 and e: can bechosen to satisfy Equation or 6 when p4, m, and p:
(or 132) are known. Some examples will be given based on values of #1, m, fil and 92 which are constant in the band. It will be assumed that each stage consists of a single tube and that A1=(+)KA2, K being a positive constant. This assumption means that for the two tubes, the distortion products of a given order are of the same sign but not necessarily the same magnitude (a condition ordinarily satisfied by common types of practical tubes). Then, in Equation 5, taking #1=[/.2=-10 and p2fi2=5, we get FlBl which approximates -.6 when Ai' -Aa. Again, in
which approximates -2.6 .when A1='-Aa. Again, in Equation 5, taking ai= z=1 0 and pz=2 which approximates 0.1 when A1=A2. Now, in Equation 5, assuming that 1=-10, but that (for example by using in the second stage a phase reversing transformer or two tubes in tandem) #2 is made equal to +10, and taking pz}82=-7,
Whether high or low values are chosen may depend upon the particular circuit under consideration. Small parasitic feedbacks may make it undesirable to make 131 and re: small.
When Equation 4 is satisfied by making 151 and are: equal to unity, the feedback at the distortion frequencies is effectively very large. For other values of 161 and #252 which satisfy Equations 5 or 6, the distortion voltages from the first and second stages partially cancel at the amplifier output. In either case the modulation is reduced to a low value.
The gain of the amplifier is given by mm "'#i! i) 'wzfififmmfia Since the ll-lllflfia term in the denominator does not appear in Equations 4, 5 and 6, it can have any desired value that is compatible with stability of the amplifier. It can therefore be used to determine the amplifier gain, to further reduce the moduation and to give the proper stability to the amplifier. The amplifier will be stable if the polar plots of the following quantities for frequencies from zero to infinity do notencircle or pass-through the point 1, 0.
Gain= p (1-75 (1w?) The first and second expressions are the total feedback factors for the first and second stages; the third is the feedback factor at points in the circuit not included in the pi and 132 loops.
What is claimed is:
An amplifier having individual feedback paths of propagations pi and B: respectively around first and second cascaded amplifying portions