US2433379A - Generation of electrical impulses - Google Patents

Description

Dec. 30, 1947. M. M. LEVY ET A1.

GENERATION OF ELECTRICAL IMPULSES 1942 5 Sheets-Sheet 1v.

Filed sept. 11

- TIME 05m Y /vfr wo/ex DfLA Y/Vf WRK Dec. 30, 1947. M. M. LEVY ET AL Y 2433379 GENERATION 0F ELECTRICAL IMPULSES Filed Sept. 11, 1942 3 Sheets-Sheet? Dec. 30, 1947. M. M. LEVY ET AL 2,433,379

` GENERATION ELECTRICAL IMPULSES 'Eled Sept. 11, 1942 3 Sheets-Sheet 3 Patented Dec. 30, 1947 7 mxuirrso STATEsj 2,433,379?! GENERATION OFELECTRLIMPUISS or both,I arejmade steeper orfn-ar-reweij ortboth.t

Inf'connectionfwthtlfe testing of television sysiy In another 'embodimentfof-thefinvention arde@ allied' problems in ,whicnpulsesbiione type are.. 1 TshfQWstH Yatn'of Voltage Wmetmee required Ito'be `changed into pulsesof some other. 0f" anorgmaf`pmse"afiseconidary delayed pulse type# I an=d `thefresultant*combined pulse? Anothenobiect yis to apply :the method to pulse-ff eegf 3 .shevs al'eertieme? .example efjtrfeeeml generating-:systems:whereby thepulses=fproduced=.:l meeelleeee e' v.seelee'efereenel pulses and aeseriese may be rconsiderably shortened-.fr s Oe delayed pulses",

whichitwo delay netwcuflsv are-.asso'ciated-.with en themetworkzand thefmodied pulse 'maybe takennals'ofCfcircuf curve gives the relation between the dilierence of potential of the circuit terminals and the time. The dotted curve b shows likewise the outline of a pulse having the same shape as a but of opposite sign and displaced horizontally by a time t seconds which is small compared with the duration of the pulse. If now pulse b be applied to the circuit t seconds after a the resultant eiect will be the full-line pulse c which is obtained by adding algebraically the ordinates of a and b.

In Fig. 2 is shown another example in which the original impulse has a very steep initial rise and in which the time t separating the pulses a and b is much shorter. This enables a substantially rectangular pulse of duration -very short compared with the original pulse, to be obtained.

Another type of modied pulse can be obtained v'similarly by combining with the original pulse a a delayed pulse b of the same shape as a and having the same instead of a reversed sign.

While these methods may be applied to pulses of any form, particular importance is attached to pulses having a steep initial rise. Thus Fig. l indicates how a relatively narrow pulse c may be obtained from a pulse a having a moderately steep initial rise. Fig. 2 shows how a much narrower pulse c may be obtained from a pulse a having a much steeper initial rise by making the time interval t smaller. Generally, as the resultant pulse c is required to be narrower, so the initial rise of the original pulse a must be steeper.

There are several well-known means of producing pulses having a steep initial rise, so that this requirement does not constitute a practical limitation to the application of the process just described.

According to the principle of the invention, the pulse a is generated by some such convenient means and the pulse b is obtained from it by applying a proportion of the voltage to -a delay network from which the pulse b emerges a short time t, later. Pulse b with the same or reversed sign is then combined with pulse aand applied to the circuit terminals, and as explained above, the resultant pulse c will be obtained. By suitably adjusting the delay in the network, the time t can be varied, and the resulting pulse can accordinglybe made to have any desired width. Details of various methods of utilising the delay network will be explained below.

An example is shown in Fig. 3. The series of saw-tooth pulses a is displaced by a time t and reversed in sign, as shownin b. When these two series of pulses are combined at the input of a circuit, the resultant will be the series of square topped pulses shown in c. In the special case when t is equal to half the interval separating the successive peaks of series a, the symmetrical series of square pulses d will be obtained.

A method of producing the necessary time delay is indicated in Fig- 4. A delay network I.of thel type well known in the art in which a pulse entering at terminals :ci and :c2 emerges from terminals y1 and y2 after a finite interval depending upon the elements of which it is composed, has an impedance zo preferably equal to its image impedance, connected to its input terminals w1 and r2. The pulse a is applied to the input terminals, and the output terminals y1, y2 are short circuited as shown. The pulse travels intothe network and is reflected at the short circuited output terminals, and returns as the b pulse to the input terminals m1, :r2 after a certain delay. The b pulse will arrive at terminals mi, x2 with the sign opposite to that of the origlnal a pulse. If however the output terminals y1, yz are unconnected the a pulse will be reflected as before and the b pulse will arrive with the same delay at the input terminals m1, :vz but with the same sign as that of the original impulse a.

.Thus in either case the original impulse a and the delayed impulse b will add together at the input terminals mi, m2; and the resultant pulse c will be of the type shown in Figs. 1 and 2 if the output terminals are short-circuited, and will be of the alternative type if they are left unconnected: While for most practical purposes the output terminals y1, y2 of the network will be either short circuited yor left unconnected, the delayed pulse b can be varied in amplitude by connecting to y1 and y2 some impedance Z different from the image impedance Zo, and its sign will depend upon whether Z is greater or less than Zo.

It should be added that the impedance Zo is preferably chosen to be equal to the image impedance of the network in order to prevent additional reections at the input terminals. When the input terminals of the network are connected to a circuit the impedance of that circuit will be effectively in parallel with Zo. Accordingly in this case the value of Zo should preferably be so chosen that when it is combined with the circuit impedance the resultant impedance will be equal to the image impedance of the network. In the circuits described below it will generally be assumed that the value of Zo has been so chosen.

Some circuits showing various methods by which the principles explained above may be carried out will now be described.

In Fig. 5 is shown a simple circuit employing one valve- The valve 2 is provided with the usual anode battery 1, cathode circuit 8 and blocking condenser I0, and suitable pulses are applied between the grid electrode and earth. The ancde load consists of a delay network I shunted by a suitable impedance Zo. The input pulses will be amplied by the valve 2 and a pulses of simllar shape will appear at the input terminals m1, :r2 of the delay network I. As already explained delayed b pulses will also be produced at the terminals 3:1, x2 by reflection at the output terminals y1, y2 and c pulses will accordingly appear at the output terminals ai, a2. If the output terminals y1, y2 of the network I be short circuited as shown in Fig. 5, then the c pulses will be of the type shown in Figs. 1 and 2. If y1 and y2 be left unconnected, then the c pulses will be of the alternative type.

One arrangement whereby the delay network I is connected in the cathode circuit of the valve 2 is shown in Fig. 6. The anode is fed from the battery 1. The output terminals 21 and z2 are connected across the terminating impedance Zo, a blocking condenser Ill being introduced in the connection between the terminals Xl and ZI. By this arrangement the valve is operated as a cathode follower. The output terminals of the delay network I are short circuited, in which case a c pulse similar to Figs. 1 or 2 will be obtained, at terminals ai and 22. The resulting c pulse will be slightly different from that produced by the circuit of Fig. 5, because the use of Zo in the cathode circuit introduces negative feedback which changes somewhat the form of the c pulse.

A second arrangement whereby the delay network I is connected in the cathode circuit of valve 2 is shown in Fig. 7. This differs from Fig. 6 in S that the anodel circuit contains a load impedance. L. across which the output is taken at terminals ai and e2, and that the delay network output terminals are left unconnected. In this case, when a positive pulse is applied to the grid, asimilar positive c pulse appears in the cathode circuit. and the anode. current increases. rlhe reflected b pulse. will arrive at the input terminals x1., 2v2 of the network with the same sign as the d pulse. rthis then causes the cathode. voltage suddenly to increase. and since the grid voltage remains 11n-` ehaneed. the cathode voltage. assumes a large positive value, With respect to the grid, so that the plate current is suddenly cut o. The c pulse ob-` tained in the plate circuit will accordingly be of the type shown in Fig. 2.

For some purpose extremely narrow pulses are required. already explained. reduction in the width of the c pulse is limited by the steepness of the initial rise of the a pulse. There are sev. eral known methods of producing pulses having a steep initial rise and it` is convenient to cornbine in one circuit the means for generating such pulses with the methods of reducing their width which are the subject of the present invention. One way in which this may be done is shown in Fig. 8. The valve 2 has connected in its anode circuit an anti-resonant circuit consisting of an inductance Il and a capacity I2, which if de-` sired may be represented by the capacity associated with the plate circuit of the valve. Connected in parallel with the inductance is a delay network I having its output terminals y1, y2 short circuited, and having the usual impedance Zo connected to its input terminals x1, m2. "lhe combined impedance of the network I and Zo is of such a value that the antiresonant circuit II, [2, is critically damped. Connected in the cathc-.de circuit of the valve is a second delay network l" having its output terminals y1 y2 left unconnected, and its input terminals mi :rz shunted by the impedance Zo'. The output of the valve circuit is taken from terminals e1, e2 through a blocking condenser I0.

The circuit of Fig. 8 with the omission of def lay networks IA and l" is in essentials the same as one already known for producing pulses having a very steep initial rise. Pulses ai of square topped form are applied to the grid of the Valve 2. After the initial sudden rise of grid potential due to the pulse a1 the anode current increases Very slowly owing to the nature of the anode load and conditions are chosen so that the anode current reaches a maximum before the final fall of the grid potential at the end of pulse a1. When the grid potential falls thev anode current is suddenly cut off and the inductance II discharges very sharply, producing a single peak of the type shown in Fig. 9, the circuit being critically damped.

When the delay networks I and I are added, the operation of the circuit will be as follows:

The pulse a1 applied to the grid of valve 2 will produce across Zu a pulse a of form similar to Fig. 9 as already explained. The network I produces the delayed b pulse which combines with the a pulse to produce the c pulse at the output terminals in the manner already explained with reference to Fig. 5. Since the a pulse has a very steep initial rise, the time delay in the network may be made very small, so that the c pulse can have a width small compared with the width of the d pulse shown in Fig. 9. The delay network I' in the cathode circuit operates in the manner described with reference to Fig. 7, and causes the anode current to be cut off much more sudaA denly on the iinal fall of the pulse ai. As a re-` sult the initial rise of the pulse c (Fie, 9 gis made steeper thanv before. Thus in belief, the network I steepens the initial rise of the pulse a in the anode circuit, and network I produces the narrow pulse c from a by combining with it the delayed and reversed pulse b. It will be clear that the functions of networks I and I are independent and either could beV used alone without the other.

Another application of particular interest is to a. harmonic generator of a known type the circuit of which is shown in Fig. 10. In this circuit a train of pulses is produced from a sinusoidal generator, one pulse being produced for each complete cycle of4 the generator. Referring to Fig. l0, the generator G is connected to a rescriant circuit consisting of an inductance L2 and a variable capacity C2, which is tuned to the frequency of the generator. The potential across the inductance L2 is applied through a high resistance R2 to a condenser CI and a resistance. RI connected in series, and the ouput is taken from terminals e1 and e2 connected across RI. A small inductance LI is connected across R2 to improve the operation of the circuit. The circuit also includes a gas-lled triode V, the plate cathode impedance of which is connected across the series circuit of' RI and CI, and the grid is connected to the generator through a high resistance R3. and a polarising battery B. A delay network l is shown connected with its input terminals 0:1, r2 across. RI with its output terminals y1 and y2 short circuited.

Assuming for themoment that the delay network I is not connected to RI, the circuit produces at the terminals ci and z2 a pulse of the form shown by the full line curve shown in Fig. ll. As the potential produced by the generator rises so also the potential across the series circuit. C-I, RI rises and the circuit elements are so chosen that this rise of potential is very sharp. At the same time the potentials of the grid and the plate of the valve V have risen until a point is reached at which the valve suddenly becomes conducting and the plate-cathode impedance becomes low, thus discharging the condenser CI, and causing the potential across Ri to fall in the manner indicated at Fig. 11.

Pulses of this type periodically repeated at the frequency of the generator G produces a series of harmonics the relative power of which plotted against the frequency is shown in the full line curve of Fig. 12, and it will be seen that the power falls rather rapidly as the frequency of the harmonic increases. The frequency of the generator may be, for example, 4 kc., and the values shown in Figs. ll and 12 are those which can be obtained by a suitable choice of the values of the elements shown in Fig. 10.

When, however, the delay network I is connected to the resistance RI as shown in Fig. 10 the pulse shown in the full line of Fig. 11 Will be reflected at the short-circuited output terminals and will return to the resistance RI with a change of sign, and with a suitable delay. Accordingly, a modified pulse will be obtained at the output terminals 21 and z2 of the type shown in Fig. 2. This pulse has been indicated in Fig. 11 by the dotted line. Such a pulse as this, when periodically repeated at the frequency of the generator, will produce a series of harmonics whose power is shown plotted against the frequency in the dotted line of Fig. 12, and it will be seen that the encareceY 7 power of the harmonics is practically constant up to about 15 megacycles. In other words, assuming the generator produces a frequency of 4 kc., something like 3000 harmonics of equal power will be produced in this way.

It will be understood that no change whatever has to be made to the values of any of the elements of the circuit excepting that the resistance RI should be doubled in value and the delay network should be designed so that its characteristic impedance at the terminals mi and :r2 is also 2Rl. Accordingly, the eective resistance connected across terminals zi and e2 is again RI. Thus it follows that the curves shown in Figs. 11 and 12 represent the improvement obtained with a circuit having elements of predetermined values when the delay network l is added to it.

With the values shown in Figs. 11 and 12 the delay network should have a delay of approximately 0.015 microsecond, thereby producing a time interval between the a and b pulses of about 0.03 microsecond, and, as seen from Fig. 11, dotted line, the width of the c pulse so produced is perhaps 0.07 microsecond. A delay network of an already known type is suitable for a case like this. Such a network consists of a small solenoidal coil with tappings at equal intervals to which are connected a number of equal condensers having capacities of the order of a few auf. The self inductance of the corresponding sections of the coil will be of the order of 10 microhenrys and such a coil might be perhaps 2 or 3 inches long.

Other similar methods of combining delay networks with generators of pulses for use in obstacle detecting and the like, or for producing harmonies will occur to those skilled in the art. For example, a delay network may be connected in the plate circuit of a relaxation oscillator and very narrow pulses may be obtained from the output terminals.

What is claimed is:

A system for producing electrical impulses of short duration, comprising a source of impulses, an electron discharge device including a cathode, control grid and anode pulses from said source into impulses of shorter duration, a circuit connected between the anode for converting the imand cathode of said discharge device including a rst impedance which is also common to the cathode-grid circuit and an anti-resonant circuit comprising a second impedance, a first electrical delay network having its input terminals connected across said first impedance and its output terminals unconnected, a second electrical delay network having input and output terminals and having a delay characteristic such that the total time taken by said converted impulses of shorter duration to travel from said input terminals to said output terminals and after reection at said output terminals to travel back to said input terminals is less than the duration of said converted shorter impulses, means for connecting said input terminals of said second Vdelay network across said second impedance, means for short-circuiting the output terminals of said second delay network, and an output circuit coupled to said anode-cathode circuit.

MAURICE MosE LE'VY. LESLIE ERNEST WEAVER.

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

UNITED STATES PATENTS Number Name Date 2,188,970 Wilson Feb. 6, 1940 2,212,173 Wheeler Aug. 20, 1940 2,212,420 Harriett Aug. 20, 1940 2,212,967 White Aug. 27, 1940 2,221,666 Wilson Nov. l2, 1940 2,226,459 Bingley Dec. 24, 1940 2,236,985 Bartelink Apr. l., 1941 2,265,996 Blumlein Dec. 16, 1941 2,246,534 Peterson June 24, 1941 2,200,099 Nuttall May 7, 1940 2,266,154 Blumlein Dec. 16, 1941 2,310,692 Hansell Feb, 9, 1943 FOREIGN PATENTS Number Country Date 485,989 Great Britain May 23, 1938 479,935 Great Britain Feb. 14, 1938

Images