US2433237A

US2433237A - Electronic computing device

Info

Publication number: US2433237A
Authority: US
Priority date: 1944-03-31
Filing date: 1944-03-31
Publication date: 1947-12-23
Anticipated expiration: 1964-12-23

Description

Dec. 23, 1947.

J. A. RAJCHMAN ET AL ELECTRONIC COMPUTING DEVICE s Sheets-Sheet 1 Filed Marqh 31, 1944 SA Tl/RA TION opznmme POINT saw/0411a o emmve cu /mvr:

I mm. H

n ll W CURRENT PEGULA CURRENT I'mventomt; Jan H chm am By George H.M0r

. Raj

Patented Dec. 23, 1947 UNITED STATES PATENT OFFICE ELECTRONIC COBIPUTING DEVICE Jan A. Rajchman and George A. Morton, Princeton, N. J., assignors to Radio Corporation of America, a corporation of Delaware Application March 31, 1944, Serial No. 528,882

10 Claims. 1

This case is a continuation in part of application Serial Number 408,870, filed August 29, 1941.

This invention relates to electronic computing devices and has for its principal object to provide a novel and reliable method and means for obtaining, instantaneously, an electrical value (voltage or current) which is proportional to the product of two given electrical values.

Prior art computing devices of the general character described usually depend for their operation either (1) upon the use of matched electron tubes, or (2) upon the use of two or more thermocouples.

The utility of prior art devices constructed in accordance with the first-mentioned principle has been limited by the practical difficulties incident to the maintenance of duplicate characteristics in the matched tubes, whereas, in the case of computing devices employing thermocouples, the necessary thermal action introduces errors due to radiation effects and a time lag which, however brief, is in some cases highly undesirable.

Accordingly, a related object of the present invention is to obviate the foregoing and other less apparent objections to electronic computing devices constructed in accordance with the principles of the prior art.

The two given electrical values, of which the product is desired, may be constant or variable. The present disclosure is not concerned with the utilization of the said product (voltage or current) though it may here be mentioned that the invention finds useful application not only in devices which may be considered as electronic equivalents of conventional mechanical computing machines but also in electro-mechanical control systems for directing antiaircraft fire, and the like, and wherein computations based on rapidly varying data must, for optimum results, be completed with extreme rapidity.

Other objects and advantages will be apparent and the invention itself will be best understood by reference to the following specification and to the accompanying drawings, wherein Figure 1 shows a family of curves which will be referred to in explaining how the voltage across a diode may be automatically adjusted to satisfy a certain formula,

Figure 2 is a partly schematic arrangement of an apparatus, within the invention, for obtain ing an electrical value which is proportional to the product of two given electrical values,

Figures 3 and 4 are views similar to Figure 2 and showing other embodiments of the invention, Figur 5 is a circuit diagram of an electronic computer, within the invention and showing a current regulator similar to the one shown in U. S. Reissue 21,749 to Arthur W. Vance, and including details of current regulator.

Figure 6 is a similar circuit diagram showing still another type of current regulator.

It is Well known that the electron current flowing between a straight filamentary cathode and a coaxially mounted cylindrical anode can be cut off by applying a sufficiently strong magnetic field parallel to the common axis of said electrodes. Cut-01f occurs when the magnetic field bends the otherwise straight electron paths and causes the electrons to travelin curvilinear or arcuate paths such that the electrons fail to impinge the anode. It can be shown that the magnetic field intensity H and the voltage V at which cut-off occurs are related by In the ideal case of a diode with perfect cylindrical symmetry and no space charge when D is the diameter of the anode in centimeters, V is in volts, H is in gauss and the diameter of the cathode is considered negligibly small. In the case of a nonideal structure, it can be shown that the relation 1 holds, though C is of a different but easily ascertained value.

The present invention contemplates the use of two magnetrons in the multiplication of two voltages (or currents) at and For an understanding of the principle of the invention, let us suppose that an automatic means serves to adjust the voltage across each magnetron in such a way that the relation 1 is exactly satisfied for any value of magnetic field I-I. Then, by making the magnetic fields HI and H2 in the two magnetrons, respectively,

(where A is merely a constant of proportionality, hereafter omitted, and x and y are the two given voltages), the voltages appearing across the two magnetrons will be V1=c(:I:-|-y) and V2=c(;z:y) (3) Now, if the difference between these two voltages is measured, it will be V1V2=4cxy (4) and will therefore be proportional to the unknown product 3: times 1 To explain how the voltage across a magnetron may be automatically adjusted to satisfy relation 1 for any value of H, reference is made to Figure 1, which shows a family of curves I, as a function of the voltage V, for various values of magnetic field, HI, H2, etc. (The slope of these curves I, between the cut-off and saturation values depends upon the temperature of the filament and upon the accuracy with which the elements of the magnetron are aligned. Their exact slope is of no especial significance insofar as the practice of the present invention is concerned though, as in the usual case, the steeper the curve, the higher the sensitivity.) Now, it will be apparent from an inspection of this family f curves that the relation between the magnetic field and that voltage at which the current through the magnetron is just one-half the saturation value is a relation similar to relation 1, when the constant c in said relation is of an appropriate value. It folloWs that, if a current regulator is provided in series with a magnetron to fix the value of its current at one-half its saturation current, the voltage across the said magnetron will be of the proper value to satisfy the said relation 1, and it will also be apparent that this relation will hold for all values of magnetic field (H) for which the voltage (V) stays within normal operating limits. Therefore, in order to multiply two given voltages (or currents) at and y, it remains merely to provide two magnetic fields H1 and H2 which are proportional, respectively, to ar-l-y and a:y, and to apply said magnetic fields to two diode magnetrons and then read (or otherwise utilize) the difference in voltage which appears across the anodes of the said magnetrons when their cathodes are connected together.

There are, of course, numerous ways to obtain magnetic fields which satisfy the x+y and .r-1/ terms of relation 2. One simple arrangement is shown, by way of example, in Fig. 2 wherein the two voltages :c and y (whose product is desired) are impressed, respectively, upon the grids Gx and Gy of two triodes Ta: and Ty whose cathodes Ca: and Cy are in series with large resistors R11: and Ry and whose anodes A1: and Ay are connected to a network comprising duplicate resistors RI, R2 and R3. With this arrangement, the currents through the triodes Tx and Ty are essentially proportional to the voltages (at and y) applied to the grids Ga: and Gy, due to the degenerative action of the large resistors R2: and Ry in the cathode circuits. Since the currents of both tubes Ta: and Ty flow in the same direction through the resistor RI, the voltage across that resistor is proportional to r+y. On the other hand, the voltage across the resistors R2 and R3 tend to oppose each other; hence, the voltage thereacross is proportional to :z:y.

If desired, the voltage (:c+y) appearing across the resistor RI and the voltage (:n-y) appearing across the resistors R2 and R3 can be applied, respectively, directly to the coils CI and C2 about the magnetrons MI and M2. However, in view of the relatively small amount of energy in the branch circuits B(a:+y) and BUB-ll) which include the said resistors, it is usually desirable to include a suitable amplifier in each of said branch circuits, as shown. In either event, the magnetic fields HI and H2 which are generated by the coils CI and C2 and which are proportional, respectively, to zc+y and x-y, are applied to the magnetrons MI and M2 in a direction parallel to the common axes of their cathodes FI and F2 and anodes AI and A2. The cathodes FI and F2 are connected and may be energized by a common battery B, while a source indicated symbolically at BI furnishes the necessary potential for the anodes AI and A2.

A current regulator (for example, of the type later described in connection with Figs. 5 and 6) CRI, CR2 in the anode supply circuit of each magnetron serves to maintain the current through each tube at one-half its saturation current values, substantially irrespective of the intensity of the magnetic fields generated by the coils CI and C2. Since the magnetic fields supplied by these coils CI and C2 are proportional, respectively, to x+y and x-y, it follows that the voltage between the cathode and anode of the separate tubes is Therefore, if the cathodes of these magnetrons are connected together, as shown in the drawing, the potential difference between their anodes as measured with a high impedance voltmeter Z will be proportional to at times y, in agreement with relation 4. This difference in potential between the two magnetron anodes may, of course, be made available for transmission or utilization in an output circuit U connected between the said electrodes.

Magnetic fields, proportional, respectively, to the sum and difference of the voltages (at and y) to be multiplied may be created by means of a special arrangement of coils and without creating the separate sum (:c+y) and difference (x voltages provided by the auxiliary tubes Tat and Ty of the apparatus of Fig. 2. This can be achieved, referring now to the embodiment of the invention shown in Fig. 3, by providing one of the magnetrons with two coils C3, C5 wound in the same direction and the other magnetron with two coils CG, C6 one of which (say C8) is wound in the opposite direction. The two similarly wound coils (C3, C4) about the separate magnetrons are series-connected and have the voltage at impressed thereon, while the other coils C5, C6 are connected together and have the Voltage y impressed thereacross. Thus, the magnetic fields created by the separate windings C3, C5 about the first tube Mi are added together and satisfy the relation H1=r-|-y, whereas, the magnetic fields created about the other tube M2 by its oppositely wound coils C5, C6 are subtracted one from the other, and satisfy the relation H2=a:y.

The operating range of the magnetrons shown in Figs. 2 and 3 is, of course, limited. On the low side it is limited by the lowest value of voltage (between cathode and anode) for which relation 1 exactly holds. For very small values of voltage, space charge effects are very serious. As a practical matter, the lowest permissible value of voltage where a high degree of accuracy is required is of the order of 20 to 30 volts. The upper limit of the operating range is determined essentially by th maximum voltage the power supply source can deliver.

In order that the voltage across each magnetron may remain above the minimum required to accurately maintain relation 1, the magnetic field about each magnetron must have a minimum value determined by H min= (5 where C is th constant of Equation 1, and

V min.- is about 30 volts Since the magnetic field about one of the magnetrons is proportional to :v-y, this magnetic field becomes zero or vanishes when a: and y are equal. Therefore, unless some compensating means are employed, the said magnetron will work inaccurately for those values of a: and y for which :r-y is less than H min. This source of inaccuracy is obviated in accordance with the invention by providing, about each magnetron, a constant auxiliary magnetic field (A) in addition to the variable main fields HI, H2 which, as previously pointed out, are proportional to :c-I-y and :n-y, respectively. Thus,

H1=x+y+A, and H2=a:y+A (6) Since V1=cH 1 and V2=cH2 (in accordance with relation 1) then V1V2=4:cy+4Ay (7) 4my=V1-V2-4Ay (8) It is now apparent that if A is sufliciently larger than the largest value which y (the subtracting term) can assume, H2 will never become less than V min H2 min= =A-y max It appears, however, from relation 7 that the difference in potential between the anodes of the two magnetrons when their cathodes are directly connected (as in Figs. 2 and 3) is no longer proportional to :1: times y in View of the second term (4 Ag) of relation 7. Therefore, means must be provided for subtracting from that difference (i. e., the difference in voltage between the magnetrons) a voltage which is proportional to Ay. Since A is a constant, this can be done merely by subtracting the voltage which is proportional to y.

In accordance with the invention, referring now to Figure 4, the constant auxiliary magnetic field A (referred to above) may be obtained, by way of example, by placing an auxiliary coil about each tube, parallel to said axis. These auxiliary coils C1, C8 are wound in the same direction as the coils C3, C4 and C5 and are connected in a similar manner. One simple way of subtracting a term proportional to Y from VI and V2 (the voltage between the cathode and anode of the respective magnetrons) in accordance with the relation 8 is by inserting a resistor R4 in series both with the windings carrying current y and the cathode FI of the magnetron MI about which the main coils C3, C5 create a magnetic field proportional to w-I-y (i. e., about the magnetron whose main coils are wound in the same direction).

It may be observed that the voltage across resistor R is a function not only of the current y but that it has also an additional relatively small constant-component due to the current flowing through the magnetron. This added constantcomponent may be compensated for by the addition of a battery B2, or other source of constant potential, in series with the utilization circuit which, as in the earlier described embodiments of the invention, is indicated by the meter Z.

If the apparatus of the invention is used simply as a multiplying machine, appropriate meters (volt or current) may be employed to measure the values of .r and y. The values of :c and y may then be varied until the meters register the numbers to be multiplied and the result of the 6 multiplication will appear on the meter Z in the common output circuit of the magnetrons. The gradations or indicia on the two input meters can, of course, be made such that the reading on the output or :0 times 1,! meter is actually the product of the readings on the input or x and y meters.

Reference has heretofore been made to the fact that the current regulators (CRI, CR2) employed in carrying the invention into effect may be of any suitable or convenient type and reference has been made to the regulator shown in U. S. Reissue Patent 21,749 to Arthur W. Vance. The current regulators showing in the block diagrams of Figs. 1 to 4 inclusive of the instant case have but two terminals and the Vance regulator employs more than two. Since it may not therefore be entirely obvious how the adaptation is made, it is illustrated in Fig. 5.

In Fig. 5, as in the earlier described embodiments of the invention, the magnetron MI and M2 have their anodes AI and A2 connected in opposition. Thus, the absolute voltage developed across one tube is subtracted from the absolute voltage developed across the other and the resulting voltage is indicated on the voltmeter Z. In this drawing (as in Fig. 3), separate batteries BI and B2 are employed for energizing the filaments FI and F2 of the magnetrons. Similarly, separate

current regulator tubes

29, 29; direct current amplifiers 83, 83', 85; glow tubes I43, I43; and resistors 99, I09, I39, I45, I41; 99, I09, I39, I45, I41, are employed for maintaining the electron current in the separate magnetrons constant substantially irrespective of the values of the magnetic fields applied thereto by the coils CI and C2 respectively. The numerals applied to the several circuit elements of the current regulating system shown in the drawing are the same as those given in Fig. 4 of Vance Re. 21,749 and reference is made to the said patent for a detailed explanation of the operation of the said system. However, it may here be pointed out that the control voltages for the D. C. amplifiers 83, 85 and 33, 85 are provided respectively by means of the resistors I39, I39 in the negative leads of the separate magnetrons MI and M2. Any change in current through the resistor I39 causes a corresponding change in voltage on the input of the tube 83 relative to the fixed voltage standard provided by the gas tube I43. The output of tube 83 is amplified by the D. C. amplifier 85 and in turn impressed upon the grid of the current regulating tube 29. This tube, 29, being in series between the power supply and the load 23, causes an equal and opposite change in current fiow to the said load and compensates for the tendency of the load current to change, thereby maintaining the current through the load substantially constant irrespective of variations in the intensity of the magnetic fields applied to the said load devices MI and M2.

The regulator shown in Fig. 6, like the one shown in Fig. 5 employs separate D. C. amplifier and current regulating tubes for each magnetron (MI and M2). Hence, a description of one unit will serve as a description of both; the corresponding parts of each unit having been given the same reference characters.

As in the earlier described system (of Fig.5), the purpose of the current regulating units shown in Fig. 6 is to keep the electron current flowing between the cathode and anode of each magnetron at a constant and predetermined aeeaesv value (I) independently of the value of themagnetic field created by the current, flowingthrough the; coils X, Ywhich surround the said, tubes. This. is achieved as follows: The power S pply puts. a. voltage (V) between the terminals land 2 which causes a current (I) to flow from the negative terminal (I), between the cathode FI and the anode A! of the magnetron MI through the resistor 3 and between, the cathode 4; and anode. 5 of the regulator tube 6 to the positive terminal 2. The function of the amplifier composed of tubes 1 and 8 and the associatedoircuit istoput a voltage between the cathode 4, and the grid 9 of the regulator tube 4, such that this current (I) be constant and have a predetermined value. This amplifier accomplishes that function as follows:

The voltage between the cathode I and the grid I I of the tube 1 is equal to the difference between the voltage drop (VR) developed in, resistor 3 by the current (I) and the voltage (VS) which is some fixed standard value developed between the terminals I2 and I3 of a battery I4. This. voltage difference is amplified by tube. 1. It is further amplified by tube 8. It. Will be recog, nized that the connections between the tubes! and 8 are of the type of a directly coupled-D. C. amplifier. Therefore, the difference of voltage between the voltage drop VR and the standard voltage VS will be greatly amplified and thisamplified value will appear across a resistor I 5; in the anode circuit of tube 8. This voltage is then applied to the grid 9 of tube 6.

It will be appreciated that this arrangement is of a negative feed-back type and provides a means to regulate the current (I) so that the difference between the voltages VR and VS will be negligibly small, and therefore regulate it at a constant value predetermined by the voltage of the standard battery Id and they value R of the resistor 3, since IR=VR-VS.- To see. that this is actually the case, imagine the current I as increased above the value standard I=VS/ R. That means that the grid I I of tube I will become more negative with respect to its cathode Ifland therefore its anode I6 will become more positive, This will cause the grid II of tube 8 to. become more positive, and therefore, its anode I8 to become more negative. This in turn will cause the grid 9 of tube 6 to become more negative since it is connected to the anode I8 of tube 8 through a protective resistor I9. When grid 9. of the regulator tube 6 becomes negative, it tends to diminish the current I and therefore to counteract the increase of I which was imagined to have occurred. Similarly, if the current I had diminished, the grid 9 of the regulator tube 6. would have become more positive which would tend to increase the current I and counteract the assumed decrease. If the gain of the systemv is very high, that is, if a small change of voltage in the grid of tube I causes, the current I to change a great deal, then it is obvious that the voltages VR and VS will have to remain almost equal.

The voltage V of the main power supply is. alwaysequal to:

since I is constant, VB is constant, so. that by changing the magnetic field surrounding the magnetrons, the voltage VM across the mag netron and VT across the regulator tube 0 change so that their sum remains constant. Anincrease in magnetic field causes VM toincrease andv VT toqdecreaseanda decrease. in that. field has, the converseeffect.

Excellent. results were; achieved when the. circuit. elements of the regulator were as, follows:

Tube 6:, type 811; tubes I and 8, type 1 NSG. The resistors 3, 2D and I5 were each 100,000 ohms. Theprotective grid resistors I9, 2| and 22 were each about. 500,000 ohms. The resistor 23 and capacitor were so chosen as to prevent the tube 8. from. oscillating. In the battery I4, the voltage between the terminal I2 and 2.5 was about, 45,volts; between Band 26- about 9.0 volts, and between 8 and I3 about135.volts.

The heating batteries for the filaments of tubes 1 and 8 have been omitted from the drawing in the interest of simplicity, but it will be understood that each tube is suppliedby a separate battery,

It will now be apparent that the present invention provides a novel and reliable method and means for obtaining, instantaneously, an electrical value which is proportional to the product of two given electrical values, and this too without the use either of thermocouples or matched tubes.

What is claimed is:

1. In an apparatus for obtaining an electrical value which is proportional to the product of two given electrical values, the combination of a pair of electron tubes each comprising a cylindrical anode and a coaxial cathode, means for generating two separate magnetic fields of an intensity proportional, respectively, to the sum and to the difference of said given electrical values and for applying said separate magnetic fields, respectively, to said tubes in a direction parallel to said axes, means for applying voltages to said tubes, means for maintaining a substantially constant flow of electrons between the cathodes and anodes of said tubes substantially irrespective of the intensity of said magnetic fields, and an output circuit effectively connected to said tubes and responsive to the difference of the voltages applied to said tubes.

2. The invention as set forth in claim 1 and wherein said cathodes are connected together and said output circuit is connected between said anodes.

3. The. invention as set forth in claim 1 and wherein means are provided about each of said tubes for applying thereto a constant magnetic field of a value larger than that of either of said two separate magnetic fields.

4. In an apparatus for obtaining an electrical value which is proportional to the product of two given electrical values, the combination of a pair of electron tubes each comprising a cylindrical anode and a coaxial cathode, means for generating a voltage which is proportional to the sum of said given values, means for converting said voltage into a magnetic field of an intensity proportional to said voltage and for applying said magnetic field to one of said tubes in a direction parallel to its said axis, means for generating a voltageof a value proportional to the difference of said given values, means for converting said second mentioned voltage into a magnetic field of an intensity proportional to said second voltage and for applying said second magnetic field to the other of said tubes in a direction parallel to its axis, means for applying voltages to said tubes, means for maintaining a substantially constant flow of electrons between the cathode and anode ofeach of said tubessubstantially irrespective of the intensity of the magnetic field applied thereto, and a utilization circuit efiectively connected to said tubes and responsive to the difference of the voltages applied to said tubes,

5. A method for obtaining an electrical value proportional to the product of two given electrical values, comprising producing two potentials and generating two magnetic fields of'intensity proportional respectively to the sum and the difference of said potentials, establishing two voltage developing electron current streams, subjecting one of said streams to the influence of one of said fields and the other of said streams to the influence oi the other of said fields while maintaining said stream currents substantially constant irrespective of the values of said fields, and opposing the voltage developed by one with the voltage developed by the other of said streams.

6. A method of obtaining an electrical value proportional to the product of two given voltages, comprising adding said voltages to produce a first electrical potential equal to the sum of said voltages and subtracting said voltages to produce a second electrical potential equal to the difierence of said voltages, then utilizing said potentials to generate two magnetic fields, establishing two voltage developing electron current streams, subjecting one of said streams to the influence of one of said fields and the other of said streams to the influence of the other of said fields while maintaining said stream currents substantially constant irrespective of the values of said fields, and opposing the voltage developed by one with the voltage developed by the other of said streams.

7. A method in accordance with claim 6 and including the step of indicating the voltage resulting from the opposition of the voltages developed across said electron current streams.

8. A method for obtaining an electrical value proportional to the product of two given electrical values, comprising producing two magnetic fields proportional respectively to two given voltages and combining said fields to produce a resultant field proportional to the sum of said voltages, producing two additional magnetic fields proportional respectively to the said voltages and combining said fields to produce a resultant field proportional to the difference of said voltages, establishing two voltage developing electron current streams, subjecting one of said streams to the influence of one of said resultant fields and the other of said streams to the influence of the other of said resultant fields while maintaining said stream currents substantially constant irrespective of the values of said resultant fields, and opposing the voltage developed by one with the voltage developed by the other of said streams.

9. A method for obtaining an electrical value proportional to the product of two given electrical 10 values, comprising producing two magnetic fields proportional respectively to two given voltages and combining said fields to produce a resultant field proportional to the sum of said voltages, producin two additional magnetic fields proportional respectively to the said voltages and combining said fields to produce a resultant field proportional to the difierence of said voltages, adding magnetic fields to each of said resultant fields of values sufilcient to maintain each of the combined fields above a predetermined minimum value, establishing two voltage developing electron current streams, subjecting one of said streams to the influence of one of said combined fields and the other of said streams to the infiuence of the other of said combined fields while maintaining said stream currents substantially constant irrespective of the value of said combined fields, and opposing the voltage developed by one with the voltage developed by the other of said streams.

10. A method for obtaining an electrical value proportional to the product of two given electrical values, comprising producing two magnetic fields proportional respectively to two given voltages and combining said fields to produce a resultant field proportional to the sum of said voltages, producing two additional magnetic fields proportional respectively to the said voltages and combining said flelds to produce a resultant field proportional to the difference of said voltages, adding magnetic fields to each of said resultant fields of values sufficient to maintain each of the combined fields above a predetermined minimum value, establishing two voltage developing electron current streams, subjecting one of said streams to the influence of one of said combined fields and the other of said streams to the influence of the other of said combined fields while maintaining said stream currents substantially constant irrespective of the values of said combined fields, compensating for the difference in voltages developed by said streams due to said added fields and opposing the compensated voltage developed by one with the compensated voltage developed by the other of said streams.

JAN A. RAJ CHMAN. GEORGE A. MORTON.

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

UNITED STATES PATENTS Number Name Date 1,214,608 Strother Feb. 6, 1917 1,912,856 Pinkel June 6, 1933