US2835845A - Electro-static methods of storing and recovering information - Google Patents

Description

May 20, T958 6. F. BLAND 2,835,845

ELECTRO-STATIC METHODS OF STORING AND RECOVERING INFORMATION Filed April 9, 1954 2 Sheets-Sheet 1 Puzsz f M Z5 A7211? 2 E 67 w, gig 2% 91475 a AMPL/HEQ' ATTORNEYS May 20, 1958 G. F. BLAND 2,835,845

ELECTROSTATIC METHODS OF STORING AND RECOVERING INFORMATION Filed April 9, 1954 2 Sheets-Sheet 2 1 I l l l l l l I I l I I I l a i .1 4. T INVENTOR 42mm: E 6.4 A ND BY rt? 8: 1

ATTORNEYS ELECTRQ-STATIC METHQDS OF STORING AND RECOVERHNG iNFORMATION George F. Bland, New York, N. Y., assignor to international Business Machines Corporation, a corporation of New York Application April 9, 1954, Serial No. 422,056

2 Claims. ((11. 31512) The present invention pertains to improvements in electro-static methods of storing and recovering information.

An object of the invention is to provide an improved method of producing effective differentiation between digits by controlled variation in the density of the primary electron beam used to effect storage and read-out.

A particular object is to provide a method of the above type utilizing a dual-density electron beam to store differing digits for use in binary number computational systems and the like.

A further object is to provide a method of the above type in which the length of time during which the electron beam must remain conducting is renderedshorter than in prior methods, whereby speed of operation and precision of the system may be increased.

Another object is to provide a dual density storage method of the above nature wherein distinction between differing digits may be achieved without requiring beamshift or double-spot bombardment for certain digits, thereby increasing the digit capacity of a given screen area.

A still further object is to provide a greater amplitude difference between the read-out signals for differing digits.

Other objects and advantages of the invention will become evident during the course of the following description in connection with the accompanying drawings, in which Figure 1 is a diagrammatic illustration of a typical cathode ray tube equipped with a suitable pick-up plate for reading out stored digital impulses;

Figure 2 illustrates the behavior of a bombarded spot as similar to that of a space-charge-limited diode;

Figure 3 is a curve showing the static voltage-current characteristics of the theoretical diode comprised by the bombarded spot and its environs;

Figure 4 is a graphical illustration of signal contrasting characteristics produced with low-density read-out of charges stored at low and at high sharp beam densities.

In the development of electronic digital computing devices, involving the storage and subsequent, read-out of a great variety of values representative of different digital combinations, it has been advantageous to adopt the binary system of numbers. This system is based on the common series,

G b +a b +a b in which b=2 and a=either O or 1, and n is the number of significant figures. Thus for example, thedecimal number 19 in asystem employing 5 significant figuresmay be written It will be seen that the above representation of 19 is derived by choice of a as either 1 or 0 in the various ,1.

positions of-the series. Similarly, all integers Within the five significant number system may be represented Patented: May 26, 1958 by-proper choice of the position and alternative values of a. Since n and b are constants in a device employing such a binary system, it follows that the essential necessity for the storage and read-out means per se consists in the ability to establish a in its proper alternative value (i. e., either 1 or 0') in each position of the series, and to distinguish accurately between the two values as each position is'read out. Thus, continuing the illustrative use of the number 19, the positions and values of a, that is 10011, comprise the operational definition of this number, while any different arrangement similarly defines a different number. It is therefore obvious that the reliability of the storage device is dependent upon the accuracy with which it distinguishes between 1 and 0 entries, and furthermore, the speed with which such entries can properly be made and read out, either with serial or parallel operation, is of prime importance in determining the practical utility of the apparatus.

Memory methods of the prior utilize static storage by means of a stream of electrons in a variety of special adaptations of the cathode ray tube. One of the devices and. a method of using it. has been described in detail in an. article entitled A storage system for use with binarydigital computing machines, by F. C. Williams and T. Kilb'urn, published in the bulletin of the Institution of Electrical EngineerstBritish), March 1949, pages 8140 Since the present invention is directed not to a single combination of electronic devices which. per se are'well known in the art, but rather to a new method of operatingsuch combinations by which an improved result is obtained, the general arrangement of theabove-mentioned system is utilized herein for purposes. of illustration.

Referring to Figure 1, the numeral 19' designates a cathode ray tube comprising an envelope 1i coated on itsinterior lateral portion with a conducting material 12 such as aquadag, and having its glass end screen portion 13 interiorly coated with a suitable, phosphor M. A conducting plate 15 of metal foil or the lilceis attached to the outside of the glass screen 13 so as to be disposed in capacitative relation with the coating 14.

The plate 15, hereinafter referred to as the pick-up or signal plate, is connected to the input conductor 16 of a signal amplifier 17', and also via a resistor 16:: to ground.

The tube l'll contains the usual cathode 18, control grid 19, focussing anode 21', accelerating anode 2%, and hori- Zontal' and vertical deflector plates 22 and 23 respectively.

Thegrid 19 is connected to and controllable by a suitable pulse generator and control network 24. Similarly, the cathode iil, anodes 21 and 22, and deflector plates. 22

and 23 will be understood to have the usual operative cir art and therefore may properly be omitted from the drawing to avoid unnecessary repetitive complication.

The present invention is based on the behavior of a bombarded'spot on the target'of a cathode-ray tube as the-cathode of a space-chargelimited diode. lnthe illustrated case, the anode of the theoretical diode comprises the-collector lining 1.2 of the tube ill, while the cathode is a spot-area-251 on the fluorescent screen 14. under bombardment by an electron beam 26. Theei'fects taking place at and near the spot Zimay be set forth as follows, referring to diagrammatic Fig. 2:

As the primary beam 26 strikes the spot 25 it supplies electrons: thereto, but duev to the phenomenonof secondary emission, it. also dislodges electrons from the target area. Some of these dislodged secondary electrons have sufficientvelocity toreach the grounded collector 12, Fig. 1, it e., anelectroncurrent-Iflows from the spot'to the collector.

mary electron velocity is such that the secondary emission ratio 6 of the target is greater than unity. Therefore, since for 1 primary electrons striking the target per unit of time, 61 secondary electrons will be released, and the primary beam current and collector current being equal as noted, there is an excess (61)I of secondary electrons which move back and forth between the target and some boundary 27 intermediate the target and collector. The transit time of these electrons is of the order of 105 seconds. These excess secondary electrons establish a space charge in the immediate vicinity of the bombarded spot, the boundary 27 of this charge being the point at which the lower velocity secondary electrons are forced to turn back while the higher velocity electrons have sufficient energy to escape to the collector, as illustrated in Fig. 2 by looped and straight arrows respectively. The actual value of potential at the boundary 27 with respect to the bombarded spot is that which corresponds to the initial potential u of the secondary electrons whosevelocity is zero at the boundary, and is independent of the magnitude or density of the primary beam. The number of electrons leaving the boundary 27 of the space charge equals the primary beam current. Therefore, the boundary 27 may be considered as comprising a virtual cathode in the diode, approximately hemispherical in shape, and located at a distance inversely dependent on the amplitude of the collector current.

In utilizing static storage of alternative digital values, such as 0 and 1 as previously mentioned, a general method is to establish difiering values of static conditions at difiering locations on the target screen. To read out the information, the spots are scanned with an electron beam. As the latter encounters the various spots it produces local potential pulses which differ in response to the differences in the previously established spot potentials. These target pulses produce corresponding signals in the capacitatively coupled pick-up plate, in the present case the exterior plate 15, which latter pulses are sampled by the amplifier 17, amplified, and utilized in any desired computational manner. 7

Figure 3 illustrates the manner in which the static potential V of the spot 25 with respect to the collector 12 varies with different values of primary beam current density in the above described theoretical diode. It will be noted that at low values of current as at point A, the potential V is positive. This is due to the fact that the boundary 27 of the virtual cathode is at or beyond the collector so that the virtual cathode space charge fills the entire space between the spot and the collector. As the density of the primary .beam current is increased the potential of the spot 25 with respect to the collector decreases to zero and then becomes increasingly negative with larger values of current, as the limiting space charge within the virtual cathode .27 increases in density and decreases in radius. 7

In examining the relationship between current density and spot potential it is necessary to consider the currentchange-rate characteristics of the diode. The diode will not respond either upward or downward in accordance with its characteristic curve, Fig. 3, if the primary current rate of change exceeds a certain critical value, this value being determined by the rate at which the space charge can form within the diode. The formation of the space charge is dependent on the capacities associated with the diode and the transit time of the secondary electrons. Assuming 4 this transit time of about 10- seconds, as previously noted, the maximum for consistent curve compliance may be taken about as 10 1 amperes per second, where I is the total current charge of the primary electron beam 26.

Assuming that the maximum beam current is such as to bring the potential of the target spot 25 to B, Fig. 3, it

will be evident that the target potential may then be left at any one of various values between .B and +A, depending on the rate at which the beam is turned off. If the turn-off is slow enough to allow the diode to adjust to each decreasing value of current, i. e., if the turn-off rate is less than the critical the spot 25 will be left with a charge equivalent to +A.

If the turn-off is rapid the'spot will be left at a potential bring the spot potential to point B, as noted. The current is normally turned on slowly enough .for the diode to adjust itself along its characteristic curve to B, regardless of whether the digit to be stored is 0 or 1. If 1 is to be stored, the current is turned ofi rapidly, leaving the spot 25 with a negative charge as described above. If a zero is to be stored the current is turned off slowly, i. e., at a rate well below allowing the diode condition to adjust itself back along the characteristic curve to point +A, thus leaving the spot 25 positively charged. 7

When the beam is subsequently turned on for reading, if the spot 25 is at or about potential +A, it proceeds from A to -13 along the characteristic curve, and since its motion is only in a negative-direction, an initially negative signal is induced in the pick-up plate. In the case of the negative or digit 1 charge on the spot 25, as the reading current is turned on slowly, the spot potential first moves in a positive direction toward point A, adjusts itself to the characteristic curve and then moves along the latter negatively toward point B. An initially positive signal is thus induced in the pick-up plate. By strobos'copically sampling the pick-up plate signals during their initial stages,'it will be evident that negative and positive signals respectively indicative of 0 to 1 storage may be delivered to the computing system via the amplifier 17. i

The above-outlined method, since it is dependent on slow turn-on and on slow turn-off for one of the digital states, has the'characteristic of requiring the primary electron beam to be conducting for relatively long periods of time, typical beam-on times being 4 microseconds to provide for a positive signal and 7 microseconds for the negative signal. This inherent slowness of operation is obviously a severe limiting disadvantage in computional a single spot at sufliciently low current densityto leave it positively charged by excess secondary emission. To store a. 1 digit, the impingi" beam is displaced a short distance, producing a dash or double spot. Bombardment in the second spot substantially removes the positive charge from the first portion of the dash, so that when this first portion is strobosccpically sampled under the reading beam, its output signal is distinctive from that of the single or Zero spot. this method the speed of turn-on and turn-oil has relatively little effect on the ability to store information, but it shares the above noted disadvantage of comparatively slow speed, for while the turned-on time for a dot" may be about 1 microsecond, a dash requires approximately 5 microseconds. A further disadvantage is the fact that the clashes require more room on the target, thus reducing the capacity for storage in a given area.

The method of the present invention, instead of employin a single basic beam current density and relying on carrying the spot potential more or less gradually to difierent locations along the above-described characteristic curve to achieve distinctive digit charges, utilizes two distinct basic current densities applied siarply for short and substantially equal periods of time. A fixed beam density of low value is employed to record a Zero, while a fixed high beam density is used to eilect a one recording. These two densities are achieved by square-wave potential pulses respectively of low and high amplitude applied to the control grid 19, as indicated in Fig. 1. In a typical procedure, using an accelerating potential of 1000 volts, grid bias of -57.5 volts, the zero grid pulse may be of the order of 22.5 volts and the one pulse from 50 to 75 volts, it being understood, of course, that these particular values are given merely as examples illustrative of the method. For each type of digit entry, both turn-on and turn-0E are very rapid, i. e., the rates of change are greater than the critical Referring to Fig. 3, when the low density beam is initially turned on, the potential of the spot 25 quickly assumes a positive position A on the characteristic curve and remains there, as the fixed low beam current does not permit further progress up the curve. When the beam is turned off, the spot charge potential remains at or very close to +A.

When the high density beam is turned on to eflect a digit 1 recording, the spot potential moves quickly to the point B on the characteristic curve, Fig. 3, dis regarding the lower contour of the curve, since the rapid turn-on rate exceeds the critical rate. As the primary beam is turned on there is a sharp negative swing of the entire target area, due to the formation of a primary beam space charge in the tube, and with high current density and very rapid turn-on as in the present case, this negative swing masks out any initially positive swing. The point B having been reached, the beam 26 is suddenly interrupted. T e turn-off rate being very rapid, the secondary space cloud charge within the virtual cathode 27, Fig. 2, has no opportunity to adjust itself along the diodes characteristic curve. instead, as this cloud disintegrates, some of its electrons escape to the distant collector 12 while the others fall into the spot area 25. These latter electrons, which with high currents may be in the majority due to the close proximity of the virtual cathode 27 to the spot, under these conditions augment the negative spot potential already existing, in other words adding thereto a portion of the potential difference it previously existing between the spot and the virtual cathode. The spot therefore remains charged with a negative potential B falling between B and (B+u), the exact location of course being influenced by the proportionate division of cloud electrons between the collector and the target.

In the above description it has been shown how by the present method of short and sharply-cut primary beam impulses, delivered selectively at either a low or a high pro-determined current density, distinctive positive or negative spot charges respectively may be placed on the target screen la. in operation, any desired pattern of spot charges may be deposited, the beam. as being brought to rest on successive target spots throughout the screen under control of the deflector plates 23 and 22 in a well known manner while the distinctive values of the individual spot charges are determined by selective two-level voltage control of the grid 19 via the pulse generator and control network 24.

The foregoing description has made mention of two space clouds formed in the tube 10, namely, the more or less local secondary space cloud within the virtual cathode 27 and the primary space cloud formed largely by the beam 26 itself, This latter primary cloud is formed whenever the beam is turned on, and causes a momentary negative swing of the target area, as previously noted, the amplitude of the negative swing varying with the density of the beam current. The short duration and relatively low amplitude of the primary spacecloud swing with the low current value and very rapid turn-on utilized in storing a Zero by the present method do not significantly affect the rapidity of storing the desired positive charge, but the space-cloud pulse is utilized in reading out stored Zeros, as hereinafter explained. In each case the formation of this space cloud induces a negative pulse in the capacitatively coupled pick-up plate 15, due to displacement current therein, and similarly, disappearance of the space cloud at turn-ofi produces an equal and opposite positive pulse in the pickup plate.

To read out stored information, the polarity of the stored charges is sensed by subjecting them to the same low density of beam 26 as was used in zero storage. Referring to Fig. 4, the charged spots are subjected to beam pulses controlled by the low-level grid voltage V as shown in the lower curve, the resulting efiects on the pick-up or signal plate appearing in the middle curve. The upper curve illustrates the stroboscopic gating of the amplifier to sample the recorded data.

If the potential of a spot has been left at or about +A, Fig. 3 (Zero storage), when the reading current is turned on no significant change can be produced on the potential of the spot itself, since the impinging beam current is the same as that which originally produced the stored charge. Consequently, the only initial signal appearing on the pick-up plate 15 is the negative pulse S due to the previously noted formation of the primary space-charge in the tube. This initial negative signal is stroboscopically sampled by the amplifier 17, the sampling period z being so gated under control of the network 24 as to confine the observation to the initial portion of the sensing period, so that the terminal positive cloud signal S is ignored. Thus the result of sensing a positively charged or zero spot is a characteristic amplifier output signal indicative of a negative input thereto.

If the potential of the spot to be sensed is at B Fig. 3 (digit one storage), impingement of the sensing beam thereon causes the spot potential to move rapidly in the positive direction toward +A, since the low-density current is not sufficient to maintain potential B. An initially positive signal is thereby coupled to the plate 15. The signal produced in the plate 15 itself is the algebraic sum of the corresponding positive signal and the initial negative space charge signal S but as the latter at the low beam current density is small in comparison to the heavy positive swing of the spot potential, the resultant is a strong positive signal S which when stroboscopically sampled by the amplifier 17, produces a sharp one output of maximum distinction from the zero signal.

If only a single reading of the stored information is required, the low grid potential V is employed throughout, so that all reading is carried out at the low beam density.

In this case the signal plate voltage characteristic follows the solid line, Fig. 4. However, for services wherein the recorded data must be preserved, recourse may be had to regenerative control of the grid pulse generator 24 from the amplifier 17. By this means, when a strong positive pick-up signal S is sampled by the amplifier, the grid potential is raised from the low V to the high V immediately at the termination of the sampling period, as illustrated in dotted lines on the grid potential curve. As a result, the target spot is subjected again to sudden bombardment at the high beam density, rewriting the negative charge condition as previously described. The dotted line on the signal plate voltage curve illustrates the strong negative swing due to the formation of the heavy space cloud as the high beam density is applied, and the corresponding positive swing as the cloud disintegrates at the termination of the pulse. As these swings occur outside the sampling period t they produce no effect on the output of the device. In the case of zero recordings, no special regenerative grid control is necessary, since, as previously noted, the reading by the low density beam automatically maintains the positive spot charges.

To summarize briefly, the present method employs a fixed low beam density for positive charging, a fixed high beam density for negative charging, and preferably the same low density for reading. All entered pulses of either type are made with maximum sharpness both of turn-on and turn-off of the beam, and since there is involved no gradual self-adjustment of the local theoretical diode along the latters characteristic curve, the time necessary to deposit the required distinctive charge is minimized for both types of entries. This bombardment time, while naturally variable to some extent among cathode ray tubes of differing characteristics, may be, as 'an example, of the order of l microsecond or less for each type of entry. Thus it will be evident that the new method presents a highly advantageous increase in speed over the previously noted and various other related methods of the prior art. A second advantage is improved contrasting amplitude and sharpness, of signals, due to the widely contrasting beam density levels and the short and sharp periods of beam conductivity. As an illustration, in the dot-and-dash method of recording previously noted, the maximum difference of swing between dashes and dots is that provided by the potential u of the virtual cathode with respect to the bombarded spot, while the present method provides the much greater diflference between B, and +A, Fig. 3. A third advantage specifically over the dot-and-dash method of recording, is an obvious saving in space, with consequent increased recording capacity for a given target area. The present method normally is operated at fixed recording beam focus, with similarly evident advantages over methods dependent on focusing and de-focusing.

The method has been described for purposes of illustration as applied to a type of cathode ray tube Which has the advantage of simplicity over other types requiring various additional internal screens, plates, etc. However, it'will be obvious that if desired it may be adapted 8 with minor operational adjustments to other types, such as mica target tubes and the like, while still retaining its basic procedure and advantages. In other words, while the method comprising the invention has been described in preferred form, it is not limited to the precise relationships illustrated, as various modifications may be made without departing from the scope of the appended claims. In these claims the Words sharp or sharply, when applied to rates of beam turn-on or turn-off, will be understood to define the rates as being as rapid as prac-' tically possible, that is, substantially instantaneous.

What is claimed is:

l. Electro-static differential data storage technique for use with a cathode-ray tube having a dielectric target surface and a signal plate in capacitative relation to said surface, said tube having a characteristic critical time rate of space charge formation dt when said target surface is bombarded with a cathode ray beam of currelnt density value I, which includes the steps of sharply establishing at a time rate greater than said critical rate a cathode ray bombardment with a predetermined fixed beam density of relatively low current value I on an individual spot on said target surface, mtaintaining said bombardment in stationary relation and fixed focus on said spot throughout a predetermined short interval of time, sharply terminating said bombardment at a time rate greater than said critical rate, whereby a characteristic informational static charge of positive potential with respect to a ground potential associated with said plate may be deposited on said spot in capacitative coupling relation with said plate, sharply establishing at a time rate greater than said critical rate a cathode'ray bombardment with a second predetermined fixed beam density of relatively high current value I on a second individual spot on said target surface, maintaining said second bombardment in stationary relation and said fixed focus on said second spot throughout a second predetermined short interval of time substantially equal to said first time interval and sharply terminating said second bombardment at a time rate greaterthan said critical rate, whereby a second characteristic informational static charge of negative potential with respect to said ground potential may be deposited on said second spot in capacitative coupling with said plate.

, 2. Technique as claimed in claim 1 including the steps of successively subjecting said difi'ering deposited charges to an electron beam of the lower of said two predetermined fixed densities, said beam being applied to each of said spots in stationary relation throughout a predetermined short time interval, whereby characteristically differing signals may be produced in said capacitatively coupled signal plate, and detecting said characteristically differing signals.

References Cited in the file of this patent UNITED STATES PATENTS 2,642,550 Williams June 16, 1953 2,671,607 Williams et al. Mar. 9, 1954 2,675,499 Sears Apr. 13, 1954

Images