US2899494A - System for the translation of intelligence - Google Patents

Description

Aug 11, 1959 R. E. STURM ET AL 2,899,494

SYSTEM FOR THE TRANSLATION oF INTELLIGENCE AT Low sIcNAL-To-NoIsE RATIos Filed June 2, 1954 5 sheets-sheet 1 llg- 11, 1959 R. E. sTURM ET Al. 2,899,494

sYsTEM EOE THE TRANSLATION 0E INTELLIGENCE AT LOW SIGNAL-TO-NOISE RATIOS 5 Sheets-Sheet 2 INVENTORS BY MMM/.7W

Filed June 2. 1954 ATTORNEY Augl1, 1959 R. E. sTURM ETAL SYSTEM FOR THE TRANSLATION OF INTELLIGENCE AT LOW SIGNAL-TO-NOISE RATIOS 5 Sheets-Sheet 3 Filed June 2, 1954 I N VENTORS ATTORNEY United States Patent O SYSTEM FOR TI-m TRANSLATION OF INTELLI- GENCE AT LOW SIGNAL-TO-NOISE RATIOS Ralph E. Sturm, Pikesville, and Russell H. Morgan, Baltimore, Md., assiguors to Bendix Aviation Corporation, Baltimore, Md., a corporation of Delaware Application .lune 2, 1954, Serial No. 433,955

'5 Claims. (Cl. 178-6.8)

This application relates to the translation of intelligence having a low signal-to-noise ratio, and more particularly to the augmentation of a low level signal without emphasizing noise. The invention has its primary application in the eld of fluoroscopic screen intensification, but as will appear more fully hereinafter, it is by no means limited to this lield.

The use of X-ray and fluoroscopy to examine patients produces an image on the fluoroscopic screen at such a low light level that the examiner is required to darkv adapt prior to the examination in order to see the image. It is highly desirable to brightensthis image by an order of a thousand to ten thousand times in some manner in order to eliminate the necessity for dark adaptation and thereby bring the light level up to where the acuity of the eye is very much better that it is at the low levels ordinarily found in this practice.

' Shortly after X-rays were discovered, it became known that they produce ionization effects on materials through which they pass, and, in particular, they produce deleterious effects on live tissue if used in suicient quantity. Therefore, limits for X-radiation employed in diagnostic Work have been adopted as standard practice. Since X- rays possess an accumulative elfect on tissue, the level of radiation which is incident on the body must be specified for the different procedures in diagnosis. With a satisfactory intensifier it would be desirable, if possible, to lower the presently established limits of radiation to the lowest value at which an acceptable observation might be made.

In normal iluoroscopic work Without the aid of an intensifying system, standard practice is to use an X-ray tube at a distance of approximately 18 inches from the human body, exciting this tube with an electrical potential consistent with the thickness ofanatomical structure being observed (which potential generally falls into the range of 40 to 100 kilovolts) and using a tube current of approximately 5 milliamperes. These conditions usually produce a safe X-ray dosage during a normal fluoroscopic examination.

In lluoroscopy of the chest of a medium size human adult at a tube current of 5 milliamperes, the light produced by the uoroscopic screen is of the order of -2 milliliamberts. If one wishes to observe the antero-posterior abdominal region (that is, fore and aft) the increased thickness of the structure decreases the light obtainable from the fluorescent screen to approximately one-tenth of that obtained from the chest, or l0"3 millilamberts, and examination of the abdomen in the lateral direction (that is, perpendicular to the fore and aft) decreases the light by another factor of about one-tenth, producing only 10-4 millilamberts. Because of these low light levels iluoroscopy is frequently quite unsatisfactory, and the diagnostic information which may be obtained is severly limited.

Accordingly, attempts have been made in the prior art to utilize screen intensification and thereby eliminate the serious deficiencies of uoroscopy at low light levels.

observations of the abdomen either in the antero-pos terior or lateral position have been unsatisfactory be ice Various intensifier arrangements have been employed.

One of these, a closed link television system including av pick-up tube focused on the iluoroscopic screen, would appear to lend itself particularly well to the concept ofA screen intensification. Tests of such a system have pro` duced good usable observations of the chest. However,

cause of fluctuations in the picture caused by the random noise of the system. It has become increasingly evident that the conventional television intensifier is limited primarily by the noise level of the system. Use of the best components available has failed to produce any signi' cant improvement in the performance of the conventional television intensifier.

It has been generally assumed in the prior art that the; noise observed on the viewing screen is the manifesta tion of random noise existing in nature, such as orthicon beam noise, shot effect, thermal agitation or resistor noise,l

etc., and that since such noise is inherent in the system, its effects cannot be eliminated. Contrary to this generally accepted view, we have found that the noise present on the television screen is far in excess of that which would be predicted from classical noise theory, and moreover, instead of being random, such noise has a definite' spectral characteristic, which produces much greater de.

' terioration of the picture than would be expected from theory. More specically, we have discovered thatthe excessive noise present on the viewing Yscreen is causedby shock excitation of underdamped modes of vibra'- tion of the circuitry by noise occurring in the input, which results in the augmentation in amplitude and compression into 'a narrow frequency band of the random noise exist-` ing in nature. The present specilication teaches a method and system by which these phenomena may be substantially eliminated. In consequence, we are able t0 produce far greater screen intensification than has ever been produced before.

Accordingly, it is an object of the invention toprovide a system for and method of translating low levelv signals without emphasizing noise.

Another objectof the invention is to provide a meth,- od of and system for amplifying intelligence having a' low signal-to-noise ratio.

Still another object of the invention is to provide an improved method of and system for intensifying an image on a fluoroscopic screen.

A further object of the invention is to provide a novel amplifier system.

An additional object of the invention is to provide a novel scheme for introducing blanking signals into a tele-v` vision system or the like.

A still further object of the invention is to provide a novel method of and system for operating a cathode follower or the like.

These and other objects of the invention will becomeIr ing blanking signals into the intensier;

Figure 4 is a graphic illustration of the operation of a conventional cathode follower under conditions to be described hereinafter; l y

Figure 5 is a circuit diagram of a modified arrangev f l ment for introducing blanking signals; and

Figure 6 is a block diagram showing the arrangement of the amplifier sections of the invention.

Figure l illustrates the general scheme of the invention. In this embodiment the invention has been applied aitX-ray intensification Vsystem which includes an X- rayV control 10,for *controlling and operating an X-ray tnhe 12,1which projects a beam of X-rays onto a. uo- @scent-'screen 16 through a subject 14. A grid 15VV may beV placed before fluorescent screen l16 to reduce scatter.` Theimage produced on the uorescent screen 16 under the; action of the X-rays is focused by an optical system represented by lens 18'onto the light sensitive element of a pick-up tube 20. This tube may bean image orthicon ofthe type conventionally employed in television prac,- tice;l Block 20 may also include the necessary sweep cuits and controls for the` image orthicon tube.

"The electrical signals corresponding to the image produced on the light'se'nsitive element of the pick-up tube are` applied to a critically damped amplifier 22 which will be described in more detail hereinafter. Block 22 may include controls to set the contrast of the picture produced on the fluorescent screen of a kinescope 24, to whichY is applied the amplified signals from the critically damped amplifier 2,2. A pulse former and Shaper 26 supplies the necessary pulses to initiate the operation of the sweep circuits in the image orthicon system andthe kinescope system at precisely the same time so that the picture which is broken up into small increments by the, irnage orthicon will be re-assembled into exactly the same increments at a greater brightness by the kinescope 24. If desired, the screen of the kinescope 24 may be photographed toprovide a permanent record, of the observation'. 'A suitable power supply (not shown) furnishes all` of the power requirements lfor the intensifying unit. As will appear below, the entire system is at least critically damped, Le., thatpart of the system through which. the video signal passes.

In ordinary television practice, Le., aV 525 lineA interlaced scanning system, amplifiers must be capable of.f

maintaining good amplitude and pliaseresponse over a frequency range of the order of 60 cycles per second up, to` approximately 4 mcgacycles per second. vIn the system of Figure lY this range may be extended to. approxi-- mately 8 megacycles in order'to, be sure of, obtaining the.l best resolution possible, Generally, the useful frequency range may be fromv about 5,0 cycles to 15 mcgacycles per second.

It is well known in the art that ordinary amplifier tubes in, resistance-capacitance coupling will not cover the frequency range required in television practice while producing optimum gain withoutthe inclusion of special peaking circuits which compensate for the input 'capacitance of the tubes as well as the capacitancejassociated with the layout, wiring, and the components Both shuntV and series peaking as well as a combination of the two are employed. Many analyses of such circuits have been published. For reference, twopapers maybe mentioned, one by McLachlan, published in the Philosophical Maga.- zine, volume 22, 1936, page 481, entitled The Requirements of an Amplifier in Order to Extend its Range; and another by Bedford and Fredendall, published in the Proceedings of the IRE, volume 27, 1939, page 2,77, and. entitled Transient Response of` Multistage Video Erequency Amplifiers. The latter article attempts tor set forth the requirements of such amplifiers, using Fourier series to predict the response of a multistage `amplifier to a unit voltagesignal (sometimes called the Heaviside signal).

As` indicated above, it` is standard practice toemploy peaking in amplifiersto compensate for the input-capacitance` of the tubes, wiring capacitance, etc. Peaking is usually accomplished` by-adding theI required amount of nductance to correct for amplitude and phase` angle-dis,` tprtion. In general, peaking renders the circuit oscillatory in one or more modes of vibratiomy andoverslioot,

.M "nascar-194.y

and subsequent oscillation occur because of the oscillatory condition. In order to adequately correct for phase and amplitude distortion without employing an excessive number of tubes, which in turn would increase the noise of the system, it is necessary that the conventional circuits be oscillatory. Even though the circuits may have avery low Q, if a step-function with a sufhciently fast rise time is applied thereto, oscillations at the natural fre-- quency of the circuit will occur, and if' before the latter have completely died down, another step-function is applied, the amplitude of oscillations in general will build up. If a series of such step-functions is applied, in quick V enough succession, the oscillations will build up.

In conventional practice, an amplifier of many stages connected in cascade is used; each stage possesses the above oscillatory characteristic, and each stage oscillates at very nearly the same frequency. Consequently, once the applied step-functions produce oscillations in the first stage, this stage acts as a force function driving voltage for the next stage, andthe latter in turn acts as a force function driving voltage for the stage following it. While in practice, each stage does not operate` exactly at the same frequency, because of variations in wiring capaci'- tance, stagger tuning, etc., in general the oscillating frequencies are close enough together that each stage can be considered a force driving function for the following. It can, therefore, be said that the oscillations built up in the first stage of a conventional amplifier will be largely enhanced by the successive stages.

Random noise is transient in that no interval of periodicity exists. We have discovered that such noise acts on oscillatory circuits (as found in conventional amplihers) in a mannersomewhat similar to that of the varying square waves asV indicated above. The amplitude of the: fundamental oscillating frequency of the respective vcircuits builds up andtherefore accentuates the noise in addition toproducing a; periodicity which we have foundl to be especially deleterious to television picture quality.` From the noise equation E2==4KTR(Af), where E=voltage, K=Planks constant, T=absolute temperature, R=

resistance in ohms, Vand Af=band width in cycles per sec-- ond, it is evident that there is as much` noise power available between one megacycle and two mcgacycles as there` is in the, spectrumbetween two and three mcgacycles, etc. In other words the power distribution of noise signals, is equal in any given frequency interval of the spectrunn` It can, therefore, be seen then that if one applies a series.` of .square waves re-occurring in a random nature inf the frequency spectrum from 60 cycles to 8 mcgacycles to` an oscillatoy-.circuib and in. particular, if the circuits: fundamental oscillation occurs at approximately 2 mega-f cycles,A everything from about '2; mcgacycles to 8 megacycles4 will-tend to build. up an. amplitude in a narrow interval abouty mcgacycles The'random pulses occurring between` 2 and 8 mcgacycles will tend to be compressed into a narrow band at 2 mcgacycles,v and. consequently the amplitude at this frequency will rise. Thus, random noise when applied to a` conventional peaked amplifier will bev accentuated inA amplitude and com.- pressed into anarrow: spectral band. While the fore:` going analysisV produces results` which appear. to agree. with experimentall observations, there are many other methods of analysis which could be employed, and it is` not essential that this simplified analysis be used to expressthe conditions which, exist in this. type of circuit,

An amplifier isan active, not a passive network. The: standard. amplier having shunt peaking, series peaking or af combinationrof both` may be simplified into a passive element comprising, the-peaking. circuit together with other. passive.components,. and anA active element com.- prising the. vacuumtubes andassociatedequipment .which act as generators for Vthe passive elements. Itmay be shown. that. a passive filter. with a. flat bandv pass. frequency response and= linear; phase characteristic will shockexcite atfits mid-baud frequency when. applied sighals change abruptly, even when such signalsy are entirely outside the pass band. (See pages 477 through 486, vol. II of Communication Networks, by Guillemin, published by John Wiley, 1935.) Furthermore, the filter is not required to have circuit parameters normally considered oscillatory. To minimize this effect, the capacitances and inductances of the circuit must be dealt with in a specific manner, for example, in the manner of the line amplifier of the invention as set forth below.

We have discovered that the solution to the problems set forth above, that is, the elimination of circuit oscillation in response to random noise excitation and the accompanying amplitude accentuation and frequnecy compression of random noise, lies in the use of circuits all of whose vibration modes within the entire operating range of frequencies are at least critically damped. If the standard peaked amplifier circuit were modified so that the gain of each stage were low enough to prevent oscillation, many additional stages would be required to produce the necessary over-all gain, and ultimately the noise introduced by the input tubes and their parameters would defeat the purpose. The critically damped amplifier which forms a sub-combination of the present invention utilizes the long line or distributed constant principle. This general principle of amplifier design is, of course, not new to the art, since at the higher frequencies where the ordinary shunt or series peaking is not effective in correcting phase and amplitude distortion, amplifiers have been built on the theory that each tube is a part of the distributed capacitance of a long line. Such amplifiers operate satisfactorily up to frequencies of several hundred megacycles when not limited by circuit effects outside the tubes. They have been employed to obtain Wide band-Width and high gain, but it is to be noted that no reference is found in the prior art to the adaptation of a line amplifier to prevent the emphasis of noise in the translation of intelligence having a low signal-tonoise ratio. In fact the large number of tubes required by such an amplifier would lead one to believe that line amplifiers are unsuitable for such use, because of the increased noise which would be expected from the employment of such a large number of tubes.

Figure 2 illustrates one section of the amplifier of the invention, Actually the complete amplifier may comprise several sections similar to that illustrated, each section connected to the previous one in cascade, as shown in Figure 6. Each section comprises a plurality of driving devices exemplified by the tubes 28 to 4f). In this particular embodiment, seven pentodes, such as the 6CB6, may be employed. The control grids of the respective tubes are connected to a grid line 46 comprising -a series of coils 48 to 62 and condensers 107 to 120. Successive coils may be wound in opposition to reduce the mutual coupling between adjacent coils to the lowest level possible, but this is not essential. The grid line is terminated at its respective ends in its surge impedance by resistors 64, 66, respectively, and the series of small padding condensers 108 to 120 are employed to correct for variations in tube capacitances and to bring the surge impedance of each section of the line to the correct value. By proper adjustment the grid line may be made substantially reflectionless.

The anodes of the respective tubes are connected to a plate line 68 comprising coils 70 to 84, which also may be Wound successively in opposition. Here again a plurality of padding condensers 94 to 106 are employed to adjust the respective sections of the line to the correct surge impedance. The plate line may be terminated at one end by a plurality of resistors 86, 88, 90. The other end need not be terminated in its characteristic impedance, and this arrangement substantially doubles the gain, as is known in the art. As will become more evident hereinafter, reections produced by failure to terminate one end of the plate line in its surge impedance will not greatly affect the operation of the circuit Where pentodes are employed, because of the fact that beyond a certain voltage the plate voltage of a pentode does not substantially determine its plate current. The terminating resistors 86, 88, 90 on the plate line may be quite critical, since in this application they are required to have about 13 watts dissipation with negligible inductance. Ordinary non-inductive resistors of the wire Wound type may not be satisfactory but the type R33 non-inductive resistors produced by the Corning Glass Company, or its equivalent, may be employed satisfactorily.

An input driving device illuustrated by pentode 42, which may be a 6AH6 tube, has its anode connected to inductance 48 of the grid line and its control grid connected to input terminal 156 through a coupling network including coupling condenser 158 and grid return resistor 160. A suitable cathode load resistor 138 is provided. It will be noted that resistors 136 and 138 in series form a cathode load for the line tubes 28 to 40. These resistors are connected to the respective cathodes of the line tubes through lead 134 and are by-passed to ground through condensers 146, 142. The flow of plate current of the line tubes through resistors 136, 138 produces a small positive feed back which results in better low frequency response. The feed back is operative only at the extremely low frequencies, since the higher frequency signals are shunted by condensers 140, 142; control over the feed back is accordingly obtained by selection of the values of condensers 146, 142. Cathode load resistor 138 may be shunted by a small condenser (not shown) to provide high frequency peaking and phase shift, if desired.

The passage of the line tube plate currents through resistors 136, 138 is also utilized to provide well regulated voltages for the grid line 46 and to decouple the grid line from the power supply. In operation, each line tube may have approximately 12 milliamperes flowing through it, and 7 line tubes will give a total current of approximately 84 milliamperes. The resistance of resistor 136 may be approximately a thousand ohms, and the resistance of resistor 138 may have a relatively low value. The combined line tube plate currents passing through these resistors in series produces a regulated potential of approximately 84 volts at the cathodes of the line tubes, and this potential is applied to the plate of input tube 42 through terminating resistors 64, 66. Condensers 140, 142 provide a filter for the input tube plate potential.

The screen grid of tube 42 is fed from the B supply at terminal 122 through variable resistor 146 and fixed resistor 148, and is by-passed to ground by condensers 152, 154-. Resistor 146 may be employed to control the plate current of tube 42. Since the D.C. plate current of the input tube flows through resistors 64, 66, which lie in the control grid to cathode path of tubes 28 to 40, resistor 146 may be employed to control the grid bias on tubes 28 to 40.

An output translating device, which has been illustrated as a triode tube 44 connected as a cathode followeris coupled to that end of the plate line which is not terminated in its characteristic impedance, by a phase corrective network 171 which may comprise variable inductance 170, capacitor 174 and resistances 172, 176. A coupling condenser 166, a grid return resistor 168 and a cathode load resistor 164 are provided for the output tube. The output terminal 92 is connected to the cathode of the tube.

Where large condensers, such as electrolytics, are required in the circuit illustrated, they must be shunted by smaller condensers in order to ensure the desired high frequency response. lt is Wel-l known that an electrolytic capacitor, for instance one having a capacitance of a hundred microfarads is not satisfactory for use at high frequencies. To overcome this each of the large condensers is shunted with a smaller condenser, such as a .0l microfarad. Thus in Figure 2 condensers 130, 140 and k154, which may be large electrolytic capacitors, are shunted. by smaller condensers 1,32, 142 and 152 respectively.r

'IheB supply voltage ted to eaehot Hthe amplitiertubes from :terminal 12.2 should rbe very lcarefully regulated- Since .shoeklexeitation' as well .as standing Waves et high frequencies `mavoeelrr ,ou thelead Wires from the .Bzsupulnladeeouplius network Consisting of e resistor -124 and eeudeuserlo is inserted to deeouple the amplifier from the power supply and toprevent these effects. The value of resistor 1.24 Ais mede large enough so that the induetauee 'of the line A.feeding the amplifier together .with the distributed capacitance will Vnot osellate when shook excited.

The -screen grids of tubes 28 to 40 are fed from the B supply through dropping resistor 128 and condensers 130,I 182, whichforma lter network. A plurality of resistors 1.340 tlll'ollgh 146a are inserted in series with each of the ,screen vgrids of the line tubes. These resistors are empioyed to prevent spurious oscillations due to the inductance and capacitance of Vthe lines feeding the screen grids of the particular tubes, that is, they are employed to ensure at least critical damping. In practice, it may be necessary to insert small resistors (such as resistors 161, `1152, 165 associated with tubes 42, 44) in series with the grids and plates of all tubes except the line tubes per se to counteract any tendency toward oscillation of the in.

ductance of the leads taken in Yconjunction with the distributed and tube capacitances. In this connection it should be noted that for optimum results it may, in some instances, be necessary to insert small resistors in the filanent kleads of the tubes, in the lines between the amplifier sections, and also in the circuits represented by Vblocks and 24 vin Figure 1, including the dynode circuits of the iniage Vorthicon. Damping resistors are inserted wherever an oscillator-'y condition of inductance and capacitance would exist in their absence. These resistors may be of sonic convenient value, Vsuch as of the order of 47 ohms. To ensure the lowest possible noise level in the final kinescope indicator, it is essential that all possible modes of vibration of the circuits Yfrom input to output be at least critically damped. At the frequencies for which the amplifier is employed, even very short cathode leads, for' example, may have sufficient inductance to cause oscillations and other spurious effects. This is particularly true of the cathode circuits, because any inductance here will be very greatly enhanced by the amplification of the tube. ,It was found in one embodiment that an undamped lead length of 4more than half an inch on the cathode of the input tube 42 was suflcient to allow spurious oscillations and degrade the picture on the kinescope because of an apparent enhancement of the random noise.

Considering the operation of the circuit illustrated in Figure 2, a low level signal at input terminal 156 is applied to the control grid of input tube 42 through the coupling network 1 58, 160. For purposes of illustration itis assumed `that the input signal is a square wave with positive polarity. This wave will increase the current in the input tube, which will, by means of resistors ,64, 66, produce a decrease in the voltage at the plate Vof tube 42. For extremely low frequencies, the inductauces 48 through 62 have practically no effect, and resistors 64, 66, are essentially in parallel. However, at high frequencies, these inductances do have substantial effects, and consequently, it can be seen that the input signal produces a negative square wave which proceeds down the grid line 46 toward resistors V64, 66. It is evident from line theory that a line has a finite propagation time depending upon its parameters, that is, it takes a tinite length of time for a voltage wave to move down the line. The propagation time may be controlled by ,the parameters R, C and L including parallel conductances (not shown) in each section of the line. The padding condensers 108 through 120 may be adjusted to .compensate for varying input capacitance of the tubes so that the propagation constant is the same for each sectionlof the line. The propagation constant of the line 3S e Whole Will, therefore, be linear and the negative signal yat the input of ,thelinewill Vmove smoothly and linearly toward resistors 64, 66. A wave which is incident upon either Aof resistors 6.4, 66 .will be completely absorbed, r`since the -line is terminated at each end in its surge impedance.

11i-the plate line;68; the padding condensers v94 through 106may similarly be adjusted to ensure a linear propagation yconstant vfor the plate line Vas a whole. While the termination comprising resistors 86, 88, of the plate line rnay -have a different value from the terminating resistors of the grid line, the propagation constants for the two lines may nevertheless `be made exactly the same. Assuming this lto be the case, when the negative square wave producedat the plate of input tube 42 reaches the control grid `of tube 28, it produces in the plate circuit of this tube a positive pulse which is an amplified inversion of the pulse incident upon the control grid. Current through tube 28, as well as plate current for all of the other line tubes, must flow lthrough the termination 86, 88, 90. Thus, the pulse produced at the plate of tube 28 will start down the plate line toward its ends. Since the propagation constants of the plate and grid lines areidentical, as the negative pulse moves down the grid line, the positive pulse will rnove down the plate line, and each time the negative pulse is incident on the grid of a line tube, the plate of vsuch tube will add a positive pulse to the one already existing from the previous tube. The signal will lbe built up as though all seven tubes had been connected in parallel and all of their trans conductances had been operating on the load resistance, 86, 88, 9,0.

,A grid line signal is `completely absorbed in `the ter minating resistors 64, 66. However, in the plate line, since the far end is not terminated in its characteristic impedance, the signal will be reflected depending on the type of termination. Consequently, a reflected signal will Vstart back down the plate line passing each of the tubes in turn and finally arriving at the terminating resistance 86, 88, 90. Here the reflected signal will be completely absorbed, because the line at this point is terminated in its surge impedance. As indicated previously, the reflected signal will not affect the operation of the circuit, because the plate current of a pentode is substantially independent of its plate potential beyond a certain potential.

The useful signal on the plate line passes through the phase corrective network 171 and is applied to the control grid of cathode follower 44. The signal incident on the grid of the cathode follower produces a signal at the cathode which can be fed to the next amplifier section from a substantially low impedance source. This tube acts las an impedance changer and a decoupling tube, so that whatever is connected to the output of the amplifier section will not have a substantial effect on the characteristics of the plate line for this particular section.

Three sections identical to that illustrated in Figure 2, with the exception of special input and output connections to `facilitate introduction of blanking signals, etc. may be connected in tandem as shown in Fig. 6, producing a maximum gain of from 400,000 to 1,000,000. Such an amplifier system has been tested in a closed link television chain of the type illustrated in Figure l employing a standard image orthicon tube of the 5820 type and it has been found that with critical damping the noise is reduced `by a factor of the order of 20 to 80 times over the system using a standard shunt-series peaked amplifier. Good pictures were obtained on the screen of the kinescopedown to light levels of 10'4 millilamberts, and in particular it was observed at this level that the small amount of noise that remained was entirely random and did not interfere with the resolution of the picture nearly as much as d id the previous periodic noise at the higher light levels.

The fact that the amplifier of the invention, `with its critically damped characteristics, does not accentuate noise also implies, and it is proved in actual practice, that any signal, and in particular of the square wave variety where the rise time is extremely fast, will be reproduced essentially faithfully, without overshoot or ringing, whereas in conventional amplifiers, such as the shunt and series peaked type, this is not the case. This means that much better resolution may be obtained throughout in a picture presented on the kinescope regardless of the source of the signal.

The output of the final amplifier section is required to drive a kinescope as indicated in Figure 1. Good design requires that the amplifier be able to drive the kinescope from cut-olf to cut-olf even though this may not actually be done in practice. For the type of kinescope employed in the present system at least a 30 volt signal would be required to accomplish this. In driving the nal cathode follower through a full 30 volts it was noted initially that the low frequencies were handled very well with little or no distortion; however, the higher frequency signals which were impressed on the input were notably distorted. When square waves with extremely short rise time were fed through the circuit there was a noticeable curving off as the square portion of the curve rose, that is, high frequency cut-off or high frequency attenuation was noted. It was found that the capacitance between the cathode and the filament of the cathode follower as well as the capacitance of the associated parts of the circuits delayed the rise time or fall time of the signal applied to the cathode so that it did not follow the grid instantaneously. Under ordinary conditions such a phase lag could be corrected in the circuit if that were the only effect. However, the Ilag of the cathode with respect to the grid causes the grid 'to draw current, upsetting all of the relationships in the circuit and consequently causing a badly deformed Wave in the output. We have discovered that this effect may be overcome by arranging conditions so that the tube has a quiescent bias between cathode and grid which is always equal to or larger than the signal applied to the grid. It had been assumed prior to our discovery that conventional cathode followers would handle signals up to frequencies at which transit time effects become important.

Figure 4 illustrates the phenomenon discussed above. It can be seen that when a square wave is applied to the grid of the cathode follower, the cathode voltage does not rise at the same rate and at time t1, for example, the grid may be positive relative to the cathode by better than 25 volts. The conditions at time t2 indicate that the maximum positive grid-cathode voltage may reach 50 volts in the example given. The result is a badly distorted output signal. The quiescent grid-cathode voltage must, therefore, be chosen so that the grid is at least 50 volts negative with respect to the cathode in order to eliminate the phenomenon discussed. This may be accomplished by choosing the tube, plate voltage and cathode load, so that the quiescent current through the cathode load is sufficient to bias the cathode at least 50 volts positive with respect to the grid, for the example given. The graph makes use of linear curves for simplicity, but in practice these curves are exponential.

As shown in Figure 6, the amplifier comprises three sections, an input section, an intermediate section, and an output section. Each section is direct-coupled, but from section to section resistance-capacitance coupling is employed. This allows the convenient introduction of blanking and shading signals, which are preferably not applied directly to the line tube stages. Figure 3 illustrates a unique way of introducing the blanking signal. This signal is generally employed to cut olf the beam of the kinescope during the return trace of the cathode ray so that the latter does not interfere with the picture, and in the particular system disclosed it is also utilized to set the D.C. black level in association with the circuits that follow so that contrast control is obtained in the nal picture. In the circuit of Figure 3 the video input signal at terminal 202 is applied through a network comprising coupling condenser and grid return Aresistor 182 to the control grid of a triode 178A. The blank signal, which may be fed from a low impedance cathode follower source, is applied from terminal 204 to the control grid of a triode 178B through a coupling network comprising condenser 184 and grid return resistor 186. As indicated in the drawing, tubes 178A and 178B may be constituted by two sections of a dual triode tube. The triodes are provided with a common cathode Iresistor 188 and are connected to a source of B supply v196 through a decoupling network comprising resistor 194 and condensers 198, 200. The latter condenser is of relatively small value and shunts the larger condenser 198 (which may be electrolytic) for the higher frequencies, in the manner set forth previously. Resistor 192 is a plate dropping or load resistor for tube 178B, while resistors 181, 183,190 are small resistors employed to eliminate parasitic oscillations and to ensure critical damping. The video input signal on the control grid of tube 178A is coupled from the cathode of the latter to the cathode of tube 178B. So far as the signal applied to the cathode of tube 178B is concerned, this tube operates as a grounded grid ampli- 'er, which allows operation at a higher frequency, because the input capacitance is quite low. The control grid of tube 178B serves as a mixer grid to which the blank signal is fed, and the output is taken on lead 195 from the plate of this tube. The dual triode 178AB may Vconstitute the output tube corresponding to tube 44 of Fig. 2 for the intermediate amplifier section of Fig. 6. This substitution is indicated in the drawings and may be accomplished by breaking the circuit of Fig. 2 at points X and connecting in place of tube 44, etc., the circuit of Fig. 3 at the points X1. This arrangement operates well up to and including frequencies of 15 megacycles, giving lan appreciable gain, depending on the transconductance of the tubes, Without employing peaking devices. Thus, an amplifying stage is provided which may be used together with the line arnplier to obtain additional gain and to solve the problem of mixing without introducing any deleterious eects on the signal, as would occur with a stage in which peaking were employed in order to proper-ly correct for amplitude and phase distortion. The Ause of a dual triode allows extremely short cathode leads and, therefore, minimizes cathode inductance. If separate tubes were employed, the parameters of the cathode circuits could be adjusted to produce a ltering action, if desired.

Alternatively, blank signals could be introduced by employing a second pentode in parallel with the input tube 42 in Figure 2, connecting the plates of the tubes together and utilizing separate screen grid, cathode and control grid connections. This arrangement is illustrated in Fig. 5, wherein tube 42 and associated components correspond to those shown in Fig. 2; only that portion of Fig. 2 necessary to the description is repeated. The anode of parallel tube 210 is connected to the anode of tube 42, and the cathode is connected to ground through a bias network including resistor 212 and condensers 214, 216. The screen grid of tube 210 is fed from the B supply through a variable dropping resistor 224 and by-pass condensers 226, 228. Conden'sers 216, 228 may be employed to shunt larger condensers 214, 226, as set forth previously. The blank signal from terminal 222 is coupled to the control grid of tube 210 through resistor- condenser network 218, 220. In operation, the bias of tube 210 is adjusted so as to prevent large shunting of tube 42, thereby preventing substantial loss in gain for tube 42. It has been found that satisfactory operation results if tube 42 carries 80% of the combined plate current, provided the blank signals are sufficiently strong. While this arrangement makes an excellent mixing system, there is some loss of the normal gain of tube 42.

The introduction of a shade signal, indicated in Fig. 6, may be conveniently accomplished by applying the re,-

quired saw .tooth voltage to )the cathode and/.or rthe'oorr trol grid of the ,input tube (corresponding to .tube 4 2 .in Eig- 2) ofetheinterrnediate amplifier section. The nood for such signals is. well knowniin ytelevision practico, .and systems -for .applyingsucb signals .are also Well known- Contrast `control may be Vachieved .by inserting a gain control potentiometer in the input to ,the intermediate amplifier vand a suitable black level setterinthe output of theoutput amplifier section.

From the `equation fornoise referred to previously, ;i t is evidently desirable Ito limit the spectrum of 4the system in vorder to reduce the noise t the lowest IJOSSblt Valli@- This maybe done by limiting both the top and bottgrnof the signal pass Ibandof vthe system, preferably vat itsginput.

As indicated previouslwa system `may be .shock excited by a signal, suchas a spurious oscillation, whichis entirely outside the pass band o f the system. It has been fOllnd that when one controls the phase and amplitude gharacteristics of the limits of the pass band so as to produce the narrowest pass band that is necessary to transmit the intelligence and at `the same time -to prevent oscillations, the lowest possible noise level is obtained.

The-term critically damped as employed in thespeification and claims describes the condition of a circuit-in which the ability of the circuit to oscillate just ,ceases -to exist. vFor example, in .a simple series circuithaving inductance (L), capacitance (C), and resistance (R), critical damping .exists When the solution to thederential equation for the current in the circuit is such ,that the discriminant is equal .to zero, or Where If the left-hand term of this Yequation is greater than the right-hand term, the .circuit is over-damped. In both instances the circuit is non-oscillatory, but if the lefthand term is .smaller than theright-hand term, the circuit is oscillatory. As employed in the specification and claims the term at least critically damped refers .t0 @circuit which is either critically damped or over-damped, i.e., non-oscillatory.

The ideal condition of exact critical damping vis difficult to achieve, land in practice the condition is approached as a limit from the region of over-damping. The amplifier must be at least critically damped through its entire operating range, which includes its pass band .and 'band skirts. Auxiliary circuits of the amplifier through which the signal does not pass, such as power supply leads, leads for inserting blanking signals, etc., should ofcourse be at least critically damped toV prevent noise enhancement, but may be Isubstantially over-damped without detracting from fidelity of reproduction.

It should be evident from the foregoing description of the invention that the use of a system which is at least critically damped for all modes .of vibration within its operating frequency range eliminates the accentuation vof the noise amplitude and the compression .of the noise spectrum into a narrow frequency band, which characterizes the systems of the prior art. We have found that the random noise passed by a non-.oscillatory system is much less deleterious to the final image on a kinescope, for example, than the periodic noise of prior art systems. The objectionable noise phenomenon of the prior art systems occurs even though the damping factor is extremely high. Circuit Qs of even 1, 2 or 3 show the described effects, and it becomes worse as the number of stages is increased, particularly if each stage is tuned or exhibits periodicity effects close to the proceeding stage, since then each stage acts as force-driven generator for those particular frequencies. If the Q of the system is raised, then the noise multiplication is also increased. Therefore, tuned circuits that select frequencies, as. for example, the tuned circuits of radio. receivers or television receivers, are particularly prone to, the .objectionablenoisephenomenon, whatever the source of the noise. 1t isevident that if asystem is employed which is not at least critically damped as a whole, it is very advantageous -to employ in the inputsections of the system amplifiers `which are at least critically damped so that a vrelatively high level low noise signal will be available for appli cation to the remanderof the system. Thus, the use of such an amplifier as the first stage or booster for a conventional television system or other conventional communication system would greatly improve the performance of the latter.

The use of circuits which are at least critically .damped toprevent the accentuation of noise is not limited to the particular system disclosed in Figure l. This method of signal translation `may b e applied to the measurement .of potential existing in vmuscles of the human body. for example, o r to the measurement of potentials occurring in ythe brain as'measured on an electro-encephalograph. lun the past, `very -narrowjfrequencyjands have beenused ip Asuch equipment just to keep the -noise level extremely low. However, these frequency bands could be increased in width, with accompanying increase in resolution, if the principles of the invention set forth above were applied. The principles o f the invention may be applied to telemetering oircuitsrequiring Wide .band Spectrums or highly discriminating tuned circuits, or to systems for analyzing frequency or wave shape, where signal-to-noise ratios are low `and Where highly discriminating periodic circuits are employedto separate frequency components. In the field o f color as -well as monochrome television, practice of the principles of theinvention will allowthe use of considerably lower light levels with accompanying advantages of economy and the reduction of the effects of strong lights on the actors. In general, .the principles ofthe nventionmay be applied to any system for gathering, detecting, transducing, reproducing, or translating, intelligence or signals wherein noise, interference or the like existsat a level approaching the level of -the intelligence or signals, and in particular where large gains or amplification are required. Where associated equipment must be .introduced to compensate `for phase effects, all these circuits must be kept at least critically damped in order to obtain the best reproductions.

It will lbe vappreciated that the invention is not limited to the use -of line amplifiers or the like, since other amplifiers may be constructed 4having at least critically damped characteristics. Moreover, the principles of the invention maybe applied to transistor circuits as well as vacuum tube circuits. The scope of the invention is not limited to the foregoing embodiments, and such embodiments should, therefore, be taken as exemplary of 'the principles of the invention rather than as restrictive. The'bounds of the invention are set forth in the following claims.

We claim;

-1. In combination, a` source of invisible radiation, means for converting radiation from said source to visible light having a low Vlevel of intensity of the order of l0'-2 millilamberts or less, said apparatus comprising light responsive pick-up means for picking up and converting said low-level light to corresponding electrical signals in the presence of electrical noise having a level of intensity of the same order of magnitude as the intensity of said signals, amplifier means for amplifying said signals, means coupling said amplifier means and said pick-up means, reproducing means for converting said amplified signals to reproduce said visible light, means coupling and reproducing means and said amplifier means, and electrical damping means for at least critically damping all the modes of oscillation of said pick-up means, said amplifier means, said reproducing means, and both said coupling means throughout their range of operating frequencies so that the signal-to-noise amplitude ratio is not .decreased and the randomness of the noise is preserved..

2. Apparatus for intensifying light having a low level of intensity of the order of "2 millilamberts or less, said apparatus comprising light-responsive pick-up means for picking up and converting said low-level light to corresponding electrical signals in the presence of electrical noise having a level of intensity of the same order of magnitude as the intensity of said signals, amplifier means for amplifying said signals, means coupling said pick-up means and said amplifier means, reproducing means for converting said amplified signals to reproduce said visible light, means coupling said reproducing means and said amplifier means, and electrical damping means for at least critically damping all the modes of oscillation of said pick-up means, said amplifier means, said reproducing means, and both said coupling means throughlout their useful frequency range so that the signal-tonoise ratio is not decreased and the randomness of the noise is preserved.

3. A system for producing an intensified lluoroscopic image, comprising a source of X-radiation, means for projecting radiation from said source through a subject and onto a fluorescent screen to produce an image on said screen having a lower level of intensity of the order of 10-Z millilamberts or less, means for picking up said image of low level intensity and for converting said image to corresponding electrical signals in the presence of electrical noise having a level of intensity of the same order of magnitude as the intensity of said signals, amplifier means for amplifying said signals, means coupling said amplifier means and said pick-up means, means for reproducing said image from said amplified signals, means coupling said reproducing means and said amplifier means and electrical damping means for at least critically damping all the modes of oscillation of said pick-up means, said amplifier means, said reproducing means,

and both said coupling means throughout their useful frequency range so that the signal-to-noise ratio is not decreased and the randomness of the noise is preserved.

4. Apparatus for translating low level signals in the presence of electrical noise of the same order of magnitude, comprising pick-up means for picking up said signals, amplifier means for amplifying said signals, coupling means for connecting said pick-up means to said amplifier means, indicating means for indicating the amplified signals, coupling means for connecting said amplifier means to said indicating means, and electrical damping means for at least critically damping all the modes of oscillation of said pick-up means, said amplilier means, said indicating means, and both said coupling means throughout their useful frequency range so that the signal-to-noise ratio is not decreased and the randomness of the noise is preserved.

5. The apparatus of claim 4, wherein said amplifier means comprises a line amplifier.

References Cited in the file of this patent UNITED STATES PATENTS 2,234,806 Ploke Mar. 11, 1941 2,319,712 Williams May 18, 1943 2,422,287 Edwards May 25, 1948 2,555,424 Sheldon June 5, 1951 2,559,515 Pourciau July 3, 1951 2,637,786 Bordewieck May 5, 1953 2,670,408 Kelley Feb. 23, 1954 OTHER REFERENCES Amplifying and Intensifying the Fluoroscopic Image by Means of a Scanning X-Ray Tube, Robert J. Moon, Science, vol. 112, October 6, 1950, pages 389-395.

UNITED STATES PATENT OFFICE CERTIFICATE OF CORRECTION Patent No. 2,899,494 August ll, 1959 Ralph E. Sturm et al.

It is herebjr certified that error appears in the -printed specification of the above numbered patent requiring correction and that the said Letters Patent should read as corrected below.

Column l, line 30, for "better that" Iread better than column 4, line '5, the Word "necessary" should appear italized; column l2, lineA 67, claim l, for "coupling and" read coupling said column 13, line 23, for "lower" read low Signed and sealed this 15th day of March 1960.

(SEAL) Attest:

KARL Ilo AXLINE ROBERT C. WATSON Commissioner of Patents Attesting OHcer

Images