US2425405A - Electronic computing circuit - Google Patents

Description

Aug. l2, 1947. A W- VANCE ELECTRONIC COMPUTING CIRCUIT Filed Aug. 29, 1942 lmsww. Mk

. Mmkumm y :h w'entl:l ARTHUR WMNCE (Ittorneg Patented ug. l2, 1947 2,425,405 ELECTRONIC COMPUTING CRCUIT Arthur W. Vance, Moorestown, N. J., assignor to Radio Corporation of America, a corporation of Delaware Application August 29, 1942, Serial No. 456,590

This invention relates generally to electronic v' computers and particularly to a circuit for providing an output voltage proportional in amplitude to the product or quotient of two or more applied voltages, the amplitudes of which are representative of selected values of the variables of the equation to be solved.

A thermionic tube ampliiier may be considered to be a natural multiplying or dividing device, since the output voltage is proportional to the input voltage multiplied .by the gain. The gain may be made to depend on the bias applied to the control electrode circuit of the' ampliiier tube. However, in ordinary amplifiers the relation between gain and control electrode bias is nonlinear due to individual tube characteristics, power supply variations, input voltage amplitude, and frequency. It is therefore necessary to compensate for non-linearity in a variable gain amplifier 4before such a device may be utilized as a high accuracy electronic multiplier.

The instant invention contemplates the use of a variable gain amplifier as a multiplying device wherein the multiplicand is applied to the amplifier input as an A.C. potential of selected amplitude and, the multiplier is applied to the amplifier control electrode as a D.C. bias voltage of selected amplitude. The product is derived from the amplifier output circuit as an A.C. potential having an amplitude proportional to the product of the two applied voltages. Compensation for non-linearity of the amplifier gain is accomplished by applying to the amplifier input circuit an additional A.C. potential of a frequency different from that of the first mentioned A.C. potential. The two frequencies are segregated in the amplifier output circuit by means of suitable lter networks, and the second, or control, A.C. potential is rectified and applied to the amplier input through a suitable feedback circuit to provide a compensating bias Voltage. Division of selected values of the variables may be effected by adjusting the potential of the second A.C. potential applied to the variable gain amplifier to a value proportional to a selected value of a variable representing the divisor.

Among the objects of the invention are to provide a novel and improved method of and means for multiplying or dividing electronically two or more variables applied to a thermionic tube circuit in terms of two voltages of amplitude proportional to the selected values of the variables.

Another object is to provide an improved electronic computer for multiplying two quantities of which one quantity is applied to the input of a variable gain amplier as an A.C. potential,

and the other quantity is applied as a D.C. potential to control the gain of the amplifier. Still another object of the invention is to compensate for non-linearity in the gain of a variable gain amplifier by utilizing a control signal, of a frequency substantially different than the A.C. signal which is applied as a multiplicand to the amplifier input, for deriving through a feedback circuit a D.C. control potential which is applied to the amplifier as an automatic gain control voltage. A further object of the invention is to provide an indicator for the computer described heretofore which will directly indicate the product of the two applied voltages which are representative of the individual quantities to be multiplied.

The invention will be described in greater detail by reference to the accompanying drawing of which Figure 1 is a schematic block diagram of the invention, and Figure 2 is a schematic circuit diagram of a preferred embodiment thereof. Similar reference numerals are applied to similar elements throughout the drawing.

Referring to Fig. 1, a source of substantially constant amplitude A.C. potential is derived from a conventional oscillator having a frequency, for example of 25 kilocycles, and applied to the input cir-cuit of a variable gain amplifier 2. The output of a second oscillator circuit 3 having a frequency, for example of l0 kilocycles, is also applied to the input circuit of the variable gain amplifier 2. The voltage amplitude of the 10 kilocycle applied signal may be varied to represent a selected value of one of the quantities to be multiplied. The output of the variable gain amplifier is applied to a constant gain A.C. amplifier 4 which includes a conventional inverse feedback circuit 5. The output of the constant gain amplifier is applied to the input of two conventional filter circuits 6,v 1, which are designed to pass 10 kilocycle and 25 kilocycle signals, respectively.

The 25 kilocycle signal, transmitted by the 25 kilocycle filter l, is applied to a linear rectifier 8 to derive a D.C. potential proportional in amplitude to the 25 kilocycle control signal derived from the variable gain amplifier. This control D.C. potential is subtracted from a second D.C. potential Vi, derived from a source 9, having an amplitude proportional to the selected value of the second quantity to be multiplied. The difference D.C. potential is amplified by a D.C. ampliiier l0 and applied to the variable gain amplifier 2 to control the gain thereof. The 25 kilocycle signal will therefore provide a D.C. control potential component which will compensate -lampliiien 3 for Ordinary non-linearity of the Variable gain amplifier. The D.C. voltage V1 derived from the source 9, will linearly control the gain of the variable gain amplifier 2in accordance With the selected value of the voltage Vi. The kilocycle signal output of the 10 kilocycle lter 6 Will therefore have a potential substantially directly proportional to the product of the 10 kilocycle input potential V2 and the D.C. gain control potential V1 applied to the variable gain amplifier 2.

In Fig. 2, the 10 kilocycle input signal is derived from a conventional thermionic tube oscillator I3, the output ofvvhich is applied through a variable resistor 23 to the cathode circuit of the variable gaifn amplier l2. The selected value of the first variable X is adjusted by varying the variable resistor 23 to apply the 10 kilo-- cycle signal to the cathode of the variablegain amplifier i2, to provide a voltage which is proportional to the value of X. The 25 kilocycle. sig- 'nal'is also applied to the cathode ofthe variable gain amplifier l2 through a second Variable resistor ZL'Which is connected in the output circuit of the kilocycle therrnionic tube oscillator Il. I

If the`25 kilocycle signal, applied to the cathode of thevariable gain amplifier tube l2,is of just sufficient amplitude to provide thedesired automatic gain control bias voltage to be described hereinafter, the circuit provides multiplication only. If, however, division .is desired, the arnl plitudeof the l0 kilocycle signal may be divided by adjusting the amplitude of the applied 25 kilocycle signal by varying the second variable resistor 2l to provideaV 25 kilocycle potentialori the cathode of the tube l2 which is proportional" in amplitude to the selected divisor.

A constant gain AAC. amplifier l includes the -first and second A. C. amplifier ntube lIll "and l5 and the conventional inverse feedback circuit 5. The loutput ofthe v,variable gain amplifier tube |2vis applied to the control electrode of the first A.C. ampliiier tube i4 through any suitable coupling circuit. The output ofthe second constant gain.A.-.C. amplifier tube .l 5 is applied to the inputs ofthe l0 kilocycle lterf and the 25 kilocycle filter 1.

The D.C. potential S, corresponding to the second variable Y, may be derived, in any .con-

venient manner, from a D.-.C. generator or battery, not shown. rihe desired'value of the llc-C. potential is selected by a third variable resistor i9. The selected D.C. potential corresponding .to the variable Y, and the 25 kilocycleV` signal transmitted by the 25 kilocycle filter 1, are applied to a linear rectiiier; tube 8, Which'inay be a pentode operated on the linearportion of itsjdetection characteristic, The D.C. potential, rep resentative of :the variable Y, and the 25 kilocycle control potentialare effectively subtracted in the linear rectifier 8. The resultant D.C. potential is further amplied in the D.C. amplifier il), which f comprises the first and second fD.C. amplifier tubes 2li and 35. The output of the second D.-C.

' ,amplitude to the non-.linearity of the lvariable gain amplifier and associated circuits. ,.'The 25 kilocycle component will therefore provide effective automatic gain controlof the variable gain irmay be'applied to any suitable indicator l1, to

The: .output of ,the Y1,0 kilocyclegiilter K 4 indicate directly the 10 kilocycle voltage output as a function of the product of the selected values of the variables X and Y.

In order to assure linear detection in the pentode rectifier 8, an additional 25 kilocycle signal is applied to the input of the first constant gain amplifier tube lll in order to obtain la relatively high 25 kilocycle control signal Without applying the full 25 kilocycle signal to the cathode circuit of the variable gain amplifier l2. Compensation for the additional 25 kilocycle signal is effected .by an additional negative bias applied to the rectifier tube 8' from a battery o-r other source I6.

The rectier tube 8 is biased by the battery I6 to operate ysubstantially as a linear rectifier. Naturally, such a rectier is linear only over a relatively narrow range of input voltages. Therefore, the 25 kc. level applied to the rectifier has y positive peaks which have values at about the same level as the bias'battery IB. The actual level Vvariation of the`25 kc. signal is very slight sinceY such variations onlyoccurto provide the necessary A.V.-C. control and the high loop gain of the 'system due to thearnplifiers land I'Urerquires only a small `signal at the grid ofthe rectifier tube 8.' Also due to thevrelatively'highjgain of the D.C. amplifier lil, the levelof the 4DAC. voltage representing the variable vY is at allow value as applied to the grid of therectifier; Any non-linearity in the rectifier and in the amplifier iii is'cornpensated by the A.V.C. control provided by the 25 kc. feedback signal.'l l

A multiplying circuit of the type described can theoretically be made to'provide vany degree of precision over Aany reasonable krang-e, v'if sugiciently high voltages land power are. available. From apractical sta-ndpoint, a `relativelysimple multiplier, of the general vtype described, I nayjibe vbuilt tohave arange of one to oney thousand in the input and outputpotentials, .with anaccuracy yof the order rof one part in tenthousand. 1 The operating timenecessary for therriultiplication mustbe larger'than the period of the lowest'applied frequency.` The `multiplication time'may therefore be considerablyV reduced'b-y increasing the input frequencies.

Thus the invention described ."co'rnprisesan electronic computer, for nfnlltiplying.tvvofquarti-V tities, in which the quantities Vare applied respectively Ias AJC. and D.O. potentials to-'agvariable gain amplifier, whereby the D.,C. potential controls the vgain,'and therefore the output,pq--

tential of vthe A.fCQinput, and the output `nuff).

v potential may be applied to the indicatorjtoindicate directly the product of the twoquantities.

A second A.C. input signaLnof ajdifferentffrequency than the 'first Vmentioned vifl.. -Cj. input, s

utilized to Hcorn-pensate` for non-,linearintyl'in "the Y gain. of `thevariable gaingarnplifier by applying va controlloias through av feedback circuitljj n 1 j 'Division .of the quantity represented. v'by fthe applied 10 kilocvcle;potentalmay'beieiegtei'by varying the amplitude ,of .the yapplied 25 jlilocycle potential Yto,the cal'fl'iode of the variablegafllampimer. Thek pctentiaiefthe iiiiocyciesignai thus applied should .beA proportional tothe selected value ofV 'the divisor.V in otlgier vlorldsgthe setting of the control 23 determinesjthe valueY of Vthe dividend, thesetting of "the contro-l 2 l jgleter ininesthe value ofA the divisor, andthe s ,of the D.C'. control lgjdeterm'ines these ,Y

Therefore, the equation ,A Y,

may be solved in one operation, where a, b, and c are separate Variables and n: is the resultant.

I claim as my invention:

1. The combination of a variable gain amplier having a cathode and a control grid, means for applying to said cathode first and second alternating potentials having different frequencies, a constant gain amplifier for amplifying the output of said variable gain amplifier, means connected to the output of said constant gain amplifier for segregating the second of said alternat- -ing potentials, means providing a unidirectional potential, means including a linear rectifier for providing a control potential proportional to the difference between said segregated and unidirect-ional potentials, means for applying said control -potential to said control grid, and means for` deriving from the output of said constant gain amplifier an alternating potential which has the same frequency as that of said first alternating potential and has an amplitude proportional to the product of the amplitudes of said unidirectional and first alternating potentials.

2. The combination of a variable gain amplifier having a cathode and a control grid, means for applying to said cathode rst and second alternating potentials having different frequencies, a constant gain amplifier for amplifying the output of said Variable gain amplifier, means connected to the output of said constant gain amplifier for segregating the second of said alternating potentials, means providing a unidirectional potential, means including a linear rectifier for providing a control potential proportional to the difference between said segregated and unidirectional potentials, means for applying said control potential to said control grid, and means for deriving from the output of said constant gain amplifier an alternating potential which has the same frequency as that of said first alternating potential and has an amplitude proportional to the product of the amplitudes of said unidirectional and first alternating potentials, divided by the amplitude of said second alternating potential.

3. The combination of a variable gain amplifier having a cathode and a control grid, means for applying to said cathode first and second alternating potentials having different frequencies, a constant gain amplifier for amplifying the output of said variable gain amplifier, means connected to the output of said constant gain amplifier for segregating the second of said alternating potentials, means providing a unidirectional potential, means including a linear rectifier for providing a control potential proportional to the difference between said segregated and unidirectional potentials, means for applying said control potential to said control grid, and means for deriving from the output of said constant gain amplifier an alternating potential which has the same frequency as that of said first alternating potential and has an amplitude proportional to the product of the amplitudes of said unidirectional and rst alternating potentials, means for adjusting the amplitude of said first alternating potential to represent different values of a first variable, and means for adjusting said unidirectional potential to represent different values of a second variable.

4. The combination of a variable gain amplifier having a cathode and a control grid, means for applying to said cathode first and second alternating potentials having different frequencies, a constant gain amplifier for amplifying the output of said Variable gain amplifier, means connected to the output of said constant gain amplifier for segregating the second of said alternating potentials, means providing a unidirectional potential, means including a linear rectifier for providing a control potential proportional to the difference between said segregated and unidirectional potentials, means for applying said control potential to said control grid, and means for deriving from the output of said constant gain amplifier an alternating potential which has the same frequency as that of said first alternating potential and has an amplitude proportional to the product of the amplitudes of said unidirectional and first alternating potentials, divided by the amplitude of said second alternating potential, and means for adjusting the amplitudes of said first and second alternating potentials and the value of said unidirectional potential to represent respectively different Values of three different variables.

ARTHUR W. VANCE.

REFERENCES CITED The following references are of record in the file of this patent:

UNITED STATES PATENTS Number Name Date 1,869,209 Mead, Jr i July 26, 1932 2,100,375 Blair Nov. 30, 1937 2,111,607 Black Mar. 22, 1938 2,178,333 Blair Oct. 31, 1939 2,179,915 Blair Nov. 14, 1939

Images