DE929743C - Electrical transmission network for generating a desired amplitude and phase curve over a wide frequency range - Google Patents

Description

(WiGBl. P. 175)

ISSUED SEPTEMBER 22, 1955

E 2343 VIII c / 2ig

Electric

The invention relates to methods and means for electrical wave filter assemblies, whereby any desired amplitude and phase curve depending on the frequency can be achieved. she is for example applicable to a filter network according to a predetermined frequency characteristic in train near the cutoff frequency.

When transmitting wide frequency bands, as they are, for. B. in television broadcasts, it is difficult to manufacture the amplifiers so that they can also handle the very low frequency components involved

let through. It is possible to allow a certain amount of damping for these low components and only to raise them again at a later stage, but here the disadvantage arises that strong Disturbances in the low frequency part of the spectrum can be generated. It is therefore also with the Transmission by cable is advantageous, the entire picture spectrum of a television broadcast in one frequency range to transform which is free from disturbance, e.g. B. by modulation with a suitable Carrier frequency. In doing so, however, the

Part of the fact that twice the frequency bandwidth is required, which naturally increases the difficulty in correcting the amplitude and phase to a constant value over the entire frequency range. With such a solution to the above problem, it is therefore desirable to have means by which the transmission of only a single sideband or at least approximately only one sideband is made possible Frequency α is sufficient, modulated onto a carrier frequency c, so that a frequency band from c - a to c + α results. The carrier frequency c is chosen so that the lowest frequency of the resulting band c - α comes to lie somewhat above the highest frequency of the original band α , so that the two bands can be separated by filtering. Another filter will have to be provided which cuts off the band sharply in the region of frequency c , so that only the sideband below c is allowed to pass. In order to avoid the high attenuation that occurs when transmitting very high frequencies, a second modulation is carried out according to the same principles in order to shift the band back to lower frequency ranges, which, however, are not as low as the range of the original band so that even the lowest frequencies can be transmitted without interference. After transmission over the cable and after passing through suitable networks and amplifiers to compensate for the distortions caused by the cable, the tape is shifted back to its old position, in exactly the opposite way by modulating it twice as it was shifted at the entrance of the cable.

In fact, there are significant difficulties in getting an accurate reproduction of the original Band at the end of a transmission channel in which one or more filters are located, namely mainly when working with suppression of one of the collateral ligaments. To a correct one To get playback of the low frequencies, both in terms of amplitude and phase, you need to both the attenuation curves and the phase curves of the filters used for suppression be asymmetrical with respect to the carrier frequency. When this condition is met, the components are of the reproduced tape is identical to that of the original tape, with the exception of a possible one constant change in amplitude and a constant time shift.

The fulfillment of this condition is not easily possible with the usual filter circuits, because if one can also produce a filter which has the desired damping characteristics, one becomes generally find that such a filter does not exhibit the desired phase characteristic. Through the In accordance with the invention, a filter is described by means of which these difficulties can be overcome. The invention is of course not restricted to use in the transmission systems described above.

According to the invention is an electrical transmission network for generating a desired Amplitude and phase progression over a wide frequency range designed in such a way that delay chains of constant wave speed over the frequency range are provided, which so with the input and output terminals of the transmission network are connected and with tapping points or joints of the wave resistance are provided that the sum of the effective admittance values of the transmission paths formed in this way with different amplitude and phase characteristics results in the desired course.

The invention thus creates a transmission network created, which has a plurality of transmission times of different lengths for a signal, where the output of the network is proportional to the sum of the effective admittance values of the different long ways. The sizes of the individual admittance values and the associated time constants the transmission paths are dimensioned so that the transmission network as a whole has a predetermined Has amplitude and / or phase characteristics. Simple examples of networks according to the invention are given in Figs. 5 and 8. Such a network should z. B. Characteristics according to Fig. 1 own. To do this, one must be able to calculate the frequency curve of such a network. It is known that an EMF, which is represented by Heaviside's unit shock, like him Fig. 2 shows, causes a current, which can be calculated from the Fourier integral:

^τ - f

- / 00

In this expression, A (φ) is the transmission factor of the filter with an applied harmonic

EMF of frequency - ^ - and amplitude 1,

2JS ^r

= jm. A (j>) is the admittance at the output terminals with respect to a generator at the input terminals, or vice versa. Generally good:

Α (φ) = α (ω) + j ω · δ (ω), "5

where α (ω) and b (ω) are both even functions in ω. In Fig. 1, curves 10 and ix indicate the functions α (ω) and δ (ω) for a given filter. In the above approach for A (f>) the current iao has the form:

01% 1O ₄

τ τ Γ d .... ,, ifflt.,. [sinoi 1,

J it) = —j - α, (co) · sin (ω t) · dco + --— ο (ω) · ω · αω ,

^JW 2π J fa ^wv . 2π J άω ^κ ' [ t J

where the integration limits indicate the variation ranges of the real and the imaginary part α (ω) and b (ω) of the filter characteristic.

Fig. 3 shows the current that is generated with a low-pass filter is generated with the characteristic according to Fig. ι.

If one uses the above equations for J ^) after t

Jl

α (ω)

Ul ₁

sin ω t · dco +

For t equal to zero this formula is:

(U ₂

2JC

α {ω)

The principle on which the invention is based is that a filter can be constructed which produces a current / _(i ) or J [ _t ) depending on whether the applied voltage is a unit burst or a short pulse.

Fig. 5 shows an example of how this principle can be applied. Here is a circuit shown which has a delay network 12, at whose input terminals 13 the signal voltage is applied, and which is terminated by a terminating resistor 14. At different Points 15 of this delay network are high resistances 16 and 17 in the manner shown connected. The designated 16 resistors are connected to the line 18, which via a comparatively small resistor 19 is grounded. The resistors marked 17 are with the Line 20 connected, which in turn via a phase rotator 21 with the same resistance 19 and above is connected to the earth.

The signal voltage generated at resistor 19 can be picked up at terminal 22.

It is now assumed that the network should have the characteristics shown in Fig. Ι. Such characteristics on their own cannot be physically implemented, as can already be seen from Figs. 3 and 4 above. However, they can be approximated with any degree of accuracy if a delay network with a sufficiently large delay is used. The current / (j) or / fo for the combination is then in a range for t greater than zero. The curve 23 in Fig. 6 is intended to represent the calculated current of an arrangement which has the characteristics according to Fig. 1 when a step-shaped EMF (unit surge) is supplied. The task then is to choose the various delays at points 15 and the sizes of the various resistors 16 and 17 so that when this surge is applied to terminal 13, the current at terminal 22 is essentially that represented by curve 23 Shape corresponds.

After the EMF has been applied, it will take a short, finite time until the disturbance differentiates on the first, resulting in the derivative J [ _t ), which indicates the transfer rate for a pulse-shaped EMF in contrast to the step-shaped EMF of the unit surge. Such a current in the case of a filter with characteristics according to FIG. 1 is shown in FIG. The equation is for values of t greater than zero

d <a

b (ω)

cos cot -

d (x>

b (co)

sin

Tap point 15 of Fig.5 arrives. Then flows in Current in the corresponding first resistor 16 and also in resistor 19. The current appears the terminal 22 in the form of a strongly attenuated reproduction of the original signal with a certain Time Delay. This current is denoted by 24 in Fig. 6. After a while, which corresponds to the time delay up to the second tap point 15 is in the same way over the second resistor 16 generates a further current surge, which is represented by the curve 25. The other curves 26, 27 show the effects of the next two tap points. You can see that it is is possible through a suitable choice of the tapping points 15 and the resistors 16 and 17, which latter one Generate current surge in the opposite direction to simulate the curve 23 with any accuracy.

Various other conditions must be met for the assembly to work properly target. So the parallel resistors 16 and 17 must be like this be large that they do not significantly affect the effect of the delay network. If it doesn't it is possible to use resistances of this size, corresponding series resistances must be used to compensate be switched into the delay network in order to meet Heaviside's condition. The attenuation occurring along the network must then be used when choosing the Parallel resistances are also taken into account. It should also be noted that the resistors 16, 17 and 19 can be replaced by any other impedances, but preferably with one another should be similar.

The time constant network itself must have a very high cut-off frequency, so that the time delay is constant over the entire frequency range of the filter. It is preferable to use a network of the derivative ra-type, which has a substantially constant phase delay over the whole Working area results.

When these precautions are taken, one finds that a very close approximation to a desired one Pair of frequency characteristics can be achieved. Of course it won't be the original Time ratios are reached, but the whole process is postponed by a constant amount of time.

Fig. 7 shows another way how the various resistances and associated time constants can be determined. The curve 28 shows the function J [ _t ), which by differentiation from the

Curve 23 emerges. However, it can be calculated directly from the characteristics of Fig. 1 if an impulse of very short duration is applied to these characteristics. If such an impulse at the Terminal 13 of Fig. 5 is supplied, appear on terminal 22 due to the different Resistances and associated time constants several pulses of different sizes one after the other. It is clear that by choosing the right time constants and resistances, the curve 28 can be increased in size Can build rapprochement. Where there is a reversal of the sign, the impulses must also be reversed by passing them through a phase reverser. This is done through connection of the respective resistors to the line 20, which through the phase reversal network 21 with the Terminal 22 is connected.

The various elements 24, 25, 26, 27 used to build curve 23 in Fig. 6 and likewise the elements used to construct curve 28 in Figure 7 are for simplicity half shown so that they start at the zero point of the time axis, where the curve rises. In reality the first taps on the time constant network must begin significantly to the left of the zero point of the time axis, so that the oscillations that lie to the left of this axis are also shown. These elements are not shown, but must also be produced to accurately reproduce the desired curve will.

It is evident that certain cases can arise which cannot be solved by the methods given above. These are all those cases in which the functions J ^) or J [^ are not convergent, ie do not strive towards a finite limit value. If, instead of converging, these functions oscillate with time without a decrease in amplitude, the length of the delay network required becomes impracticable so that the invention cannot be used.

In the embodiment of Fig. 5, a certain degree of damping necessarily occurs, which however, this can be offset by later reinforcement. An arrangement is shown schematically in Fig. 8 shows where this amplification is carried out in the various transmission paths themselves. in the Otherwise this circuit works according to the same principles as the circuit according to Fig. 5.

In Fig. 8, terminals 29 and 30 form the input of a symmetrical time constant network 31, which is terminated by the terminating resistors 32, 33. The cross members consist of Pairs of equal capacities, each connected in series, with all connection points are grounded. Some of these capacities are in turn split up again and act as Potentiometer to apply a signal of suitable strength to the grid of a tube. That way is a number of tubes 34, 35, 36, 37 are present, which from different points of the time constant network 31 are fed. So that the arrangement can also work at very low frequencies, the capacitive potentiometers can be bridged by ohmic potentiometers, which are in the same Ratio are divided and consist of relatively high resistances. The degrees of reinforcement of the individual tubes, which preferably all have the same characteristics, can by means of suitable cathode resistors be stabilized. To get signals of opposite signs, you can use the grids of the tubes with tapping points on different sides of the symmetrical time constant arrangement be connected. The anodes of the tubes are at different points in another time constant network 38, which is terminated at its end with a resistor 40. At the exit end 41 of this network, a termination is not absolutely necessary. The time delays and effective conductance values of the individual transmission paths formed in this way are transferred to one of the above-described Wise men for sure. The time delay is of course made up of two parts according to FIG both routes to be traversed in the individual networks together. The grid and anode capacities form parts of the transverse capacitance of the relevant network with which these electrodes are connected. Remaining grid-anode capacitances can be achieved by means of suitable capacitive connections be neutralized with the opposite side of the symmetrical network.

Various modifications to this arrangement are of course possible. So z. B. the network 31 does not have to be designed symmetrically, and phase reversal stages can be used there where they are necessary. The anodes of the tubes 34, 35, 36, 37 etc. can also be connected a pair of lines are laid, such as B. the lines 18 and 20 in Fig. 5, in one of which a Phase reversing device is present. Another possibility is to use a symmetrical vision Network for the network 38, the anodes of the tubes depending on the sign of the desired Voltage surge are applied either to one or the other side of the network.

In a manner similar to that described for low-pass filters, according to the invention High pass filters are constructed as well as by combining high and low passes with overlapping Bandpass filters also remove passbands. By connecting an up and A low-pass filter that does not have overlapping passbands can also use band-stop filters getting produced.

As already noted, the invention applies not only to networks for single sideband transmission, but also to networks for single sideband transmission can also be used for other purposes where certain characteristics are required. For example such a filter can reduce or eliminate interfering components which are below of the transmission range can be used without phase difficulties occurring. Such arrangements will be explained in more detail with reference to the further FIGS. 9 to 18.

In Fig. 9, a transmission network 10 for the time delay of signals is indicated schematically, which on one side of one

Signal source ii is fed and which is terminated at the other end with its characteristic impedance 12. Taps T ₁ and T _{2 are} attached at two points in the network. Delay times t ± and t _{z should} occur at these two tapping points. In addition, the two transmission conductances occur, which are functions of the angular frequency ω of the transmitted wave. These are to be designated by F ₁ (Jω), F ₂ (Jw) for the two tapping points. Then applies

F ₁ (Yes)) ^ Ae-Wi,

By addition
Conductance

one obtains the resulting

= 2 Ae • cos - ω (^ ₁ - ^ ₂ )

This therefore represents a reproduction which has a delay according to the average of ^ ₁ and and an amplitude which corresponds to the cosine of ω. varies. By choosing a different tap point, other reproductions of the same kind, but with different delays and faster or slower changes in amplitude, can of course be achieved. In particular, it can be achieved that the delay times are the same for each pair of tapping points, while the amplitude laws are given by the cosine terms of a Fourier expansion. If you set

and

- - {k-h) = rr.

this is the general name of the r-th term

g — joiT. 2 Α _τ · cos r ω χ.

You can see the _A1. easily choose so that they are different for different couples. By suitable choice of the A _T and a suitable number of pairs of tapping points, one is able to produce a desired constant delay for any particular amplitude dependence in a given frequency range, provided, of course, that the amplitude dependence can be represented as an even function of the frequency.

In exactly the same way a dependency can be represented, which by an odd Function of the frequency can be represented if the transmission conductance of the individual pairs is subtracted will. The subtraction gives the resulting apparent introduction value

^ i () ⁰ A - F ₂ (j <o) - e ~ ^iiat · 2 jB _r · sin r ω τ

for the r-th pair, where B _{T is} a coefficient like A _r . You can see that there is still a phase delay here

from - is superimposed. Has at high frequencies

this phase delay has only a small influence on the time delay, at low frequencies it may be necessary to establish a correction network for the Switch on phase. This need does not arise with even functions.

Fig. 10 shows schematically an arrangement which is constructed according to the above principles. The time constant network 10, which can be constructed as an artificial line or as a cable fed from one side with signals from a source 11, while the other side through an impedance 12 is completed without reflection. With 17 a point of the network 10 is referred to, to which the tapping points 13, 13 ', 14, 14', 15, 15 'and 16, 16' are each arranged symmetrically. Only four such tap pairs are shown, but you can can of course also be used a lot more.

In order to achieve the desired amplitude dependency, it is necessary to combine the outputs of the individual tap pairs according to the principles set out above. So you have to add or subtract the outputs of the same amplitudes and set the relative amplitudes correctly. This can e.g. B. done in a simple manner by a cathode ray tube, which has a number of deflection plate pairs to which the individual taps of the transmission network are connected. However, this will not be discussed in more detail. To eliminate the phase effect that can occur with odd functions with sine terms of the Fourier expansion, a phase network of a known type can be used (Fig. Ha). In this circuit, the signals are fed to the grid 19 of a tube 18, the anode circuit and cathode circuit of which each contain a resistor 22 and 23, respectively, which are identical to one another. The anode 21 is connected to the output terminal 28 via a blocking capacitor 27 and a resistor 26 of size R , while the cathode 20 is connected to the same terminal inductance 24 and a capacitance 25 of size L and C , respectively. The time delay as a function of the frequency f, which is caused by this circle, is shown in Fig. 11b for various values of the parameter

= JLl / shown. To achieve the

The desired effect can be achieved by several such networks in a suitable combination of their characteristics to be used. The network described is a special case of the phase rotation network according to the iao British Patent 473 028.

Fig. 12 a shows an example of the application of an arrangement according to Fig. 10. Here an amplitude characteristic is shown against the variable cot . In this example only the cosine term of the ias Fourier expansion occurs, so that the outputs at

added to the individual tap pairs. The amplitudes taken from the individual tap pairs are chosen according to the size of the Fourier coefficients:

12 / I I

I cos ωχ cos 3 ω χ A cos 5 ω τ -

2 π

The constant term of this series, -, denotes the amplitude at tap 17 in Fig. 10. The co-

2

efficient - denotes the amplitude in relation to

the pair of taps 13,13 '. The coefficient is

3 ^π

is the amplitude correct for the tap pair 14.14 ' etc. The sign of the coefficients determines the sense in which the outputs of the individual tap pairs must be combined with each other.

The solid line in Fig. 12 a shows the dependency, if an infinite number of taps were used. This is of course not feasible because it would require an infinite time delay chain. One comes, however already with a finite number to a sufficiently good approximation, as the dotted line shows, which represents the result for three pairs of tapping points.

Fig. 12b shows the form of a dependency for odd functions, if so for the individual ones A subtraction is made to tap point pairs. In this case a phase correction be necessary for the low frequencies if a variable time delay is undesirable. The solid line again shows the ideal case for an infinite series of tap pairs while the dotted line represents the characteristic which is achieved by three pairs of taps.

In the curve according to Fig. 12 a, the first limit is

frequency at the values for ω t = - reached. These

Cutoff frequency is therefore. This can be used for

In certain cases the location of the tapping points is calculated will. The curve in Fig. 12b is the first

Cutoff frequency at.

It can be seen that the larger the number of tap pairs is selected, the sharper the cut-off frequency can be made. The more tap points, the greater the delay time T. It can be shown that in fact, if Af is the frequency interval in which the curve rises,

2 Af

A significant increase therefore necessarily requires a network with a long delay.

Assume that the attenuation in the time constant network is negligible for all frequencies, then the curves of Fig. 12 a and 12 b give a whole series of equal pass bands, which are interrupted by the same blocking bands. All of these bands have relatively sharp cut-off frequencies and can be screened out of the rest of the quantity by ordinary filters. These latter can produce phase distortions, but these distortions can usually be found in such areas in which they do not interfere, or they can be compensated relatively easily. ·

Fig. 12c shows a reproduction curve accordingly a number of sine terms of the following sizes:

sin cot

0.32 nm

0.17 sin

ο, 10 sin

0.053 sin 9 « t

0.025 sin 11 ω t

0.008 sin

No special description is necessary for the delay means according to Figs. 9 and 10, because any known delay networks or low pass filters consisting of concentrated reactances with sufficient high cut-off frequency can be used or cables tapped at suitable points. It is desirable to keep the time constant of the network up to a frequency which is considerably greater than the upper limit frequency of the band is to be made approximately constant. In the example of the first Pass band in the curve of Fig. 12 a, in which 'the first cut-off frequency by the expression is given, in such cases it is where

4 τ ° °

only this one pass band is to be used, advantageous to make the arrangement so that the

Time delay up to an amount of - the

upper limit frequency is constant, i.e. up to - ^ -.

It should be noted that for η pairs of tapping points, the total time delay is η · T.

In the previous one, the task was to get steep edges of the damping curves without simultaneous change in the delay time, at least without such changes that could not be easily compensated.

Another example is discussed below. For example, one would like a total permeability value of general form

where f (co) is an even function and g (co) is an odd function. This conductance can be established by building up f (<o) according to the methods already described, namely by forming a cosine series, and that / g (ω) is established in the same way by forming a sine series, the delays for both functions are the same.

The admittance F (j ω) can then be obtained by simple addition. Two tapping points are to be considered which are equidistant from a tapping point which gives the time delay T. The outputs of these tapping points can be represented as follows:

and

If a fraction a _{r of} the first expression is added to a fraction b _{r of} the second expression, the resulting output can be written in the form:

Ae-1 °> ^T ((a _r - + - b _r ) cos rm τ + j (a _r - b _T ) sin τω τ).

From this you can see that a single combination process can be used to reduce an admittance from the Can build up form:

I -

\ 2

A ₀ 4- 2r iAr ^cos 7 ω τ + / J3 _r sin r ω τ) 1

The coefficients A ₁ , A ₂ , A ₃ or B ₁ , S ₂ . - ^ s etc. are determined by the sizes of the fractions a _r and b _{T used} :

A _T = A (a _r + b _T )
and

B _r = A (a _r -b _r ).

- A ₀ is the amplitude coefficient for that one

Tap point, which gives the time delay T.

If the shape of F (Jm) is known, the coefficients A _r , B _r can be found from the following equations:

= J ⁷ (jm) cos rmxdm,

π J
ο

τ C

^r ~ Ί ™ J

sin rmrdm.

According to another method, the overall conductance F (jm) can be built up as follows, if Φ _{Τ is} the amplitude of the wave developed at a single tap r , by adding over all values of r between - η and + w:

■ F (jm) = ^ _, _r Φ _Τ β ' ^ωΓΤ .

—Η

With a given F (/ ω) the amplitudes Φ _Τ can be calculated according to

2ai

Φ, =

2JTJ

F (jm) e-i "» * d (jm).

Fig. 13 shows another arrangement which differs from that of Figs. 9 and 10, but by which an amplitude reproduction in the form of a cosine or sine series can also be generated. 10 is a delay network whose input terminals 30 and 30 'are in series with impedance 31, which is equal to the wave resistance of the network. In addition, there is a signal source not shown connected to terminals 32 and 32 '. This signal source may have a vanishing internal resistance own. If this is not possible, make sure that all of the clamps 30.30 'occurring impedance is equal to the characteristic impedance of the network. The exit will taken from terminals 33, 33 'and is equal to the sum of the currents coming directly from the source and those that are reflected back from the network. Depending on the phase the desired reflection, the terminals 34, 34 'are either short-circuited or left open.

When the terminals 34, 34 'are short-circuited, the current is proportional

where t is the time it takes for a wave to travel from one side to the other of the network. When terminals 34, 34 'are open, the expression applies

g- / u) (j sin ωί.

FIG. 14 shows an arrangement which is the inverse of that of FIG. The impedance 31 is equal to that Characteristic impedance of the network 10 and Hegt parallel to its input terminals 30, 30 '. The connected The source must have an infinite impedance or at least a very large impedance. the Input terminals also serve here as output terminals.

The principle is the same as in Fig. 13. If terminals 34,34 'are short-circuited, that is Network transmission rate from the form

when these clamps are open, of the shape
_e -ju> t _{cos ω} ι _

Fig. 15 shows a small modification of the arrangement according to Fig. 14, the output now being taken from terminals 35, 35 'which are connected to a suitable point in the network, between terminals 30, 30' and 34, 34 '. If the delay time is between 30 and 34 α t and that between output terminals 35 and terminals 34 bt, the current at the output is proportional

_e -jamt cos δω ί,

when the terminals 34 are open, and

for short-circuited terminals 34. Using a number of pairs of output terminals, each of which has its associated value of b , a cosine or sine series of terms is obtained by addition in the correct proportion. In any case, the same can be achieved as with

an arrangement according to Fig. io, with a delay network, which has only half the total time delay.

Fig. 16 shows an arrangement of a simple form, by means of which the already discussed, generally exponential series can be built up. The arrangement is identical to that of Fig. 14, but the terminals 34, 34 'are terminated by the characteristic impedance 36 of the network, so that no reflections occur at this point. Instead, there are discontinuities in the wave resistance at points along the delay network which are equidistant with respect to the time constant, which are shown by the dotted lines 37. These discontinuities are such that they reflect the waves with one or the other sign back to the input terminals 30 in a suitable amplitude. If these discontinuities are small, the effect of ao multiple reflection can be substantially avoided. The reflected waves arriving at the input terminals form the various levels of the above exponential series, the amplitude and sign of each term being given by the coefficient Φ _τ . A discontinuity in which the impedance becomes smaller creates a wave of one sign, and a discontinuity in which the impedance becomes greater creates a wave of the opposite sign. It should be noted that it is necessary to balance the direct incident wave at the clamps 30 by some means.

Figs. 17a, 17b, 17c and τγά show an example in which the invention is used to generate a low-pass effect with a sharp cut-off frequency. As already described, if the pass band is very large, a large number of tapping points would be necessary. This number can be significantly reduced by the simultaneous use of known filters. If such filters are additionally used, then it is only necessary, by using the methods according to the invention, to generate a band-pass characteristic which has a constant time delay over a narrow frequency band, specifically to just above the desired cut-off frequency. Such a characteristic, which can be achieved with very few tapping points, is shown in Fig. 17a. The particular pass band which is to be used is shown hatched. There are of course other passbands, which are indicated in the drawing, but these passbands can be switched off by using a bandpass filter which does not need to have as steep a slope as is necessary for the solution of the problem. The phase also only needs to be constant in the hatched part for the additional filters. In this way the transmission characteristic of Fig. 17b is obtained. If the signals pass through such a characteristic, this tape is cut off from the entire original tape, and the blocking effect results according to Fig. 17c. It is now only necessary to cut off the upper pass band with a second auxiliary filter in order to obtain a low-pass characteristic with a very sharp cut-off frequency, as shown in Fig. 17d.

Figs. 18 a to 18 d show a similar application of the invention in the event that a bandpass characteristic is to be achieved in which the upper limit frequency z. B. is ten times as large as the lower limit frequency and at which a sharp one Cutoff frequency for the lower limit should be obtained. Such a characteristic is very useful in connection with single sideband transmissions, however, would without the use of known auxiliary filters just like in the previous example, a large number of tapping points may be necessary. By using the auxiliary filter, the hatched part of the characteristic in Fig. 18a can be masked out so that there is a passband as shown in Fig. 18b. This only needs from a suitable one Low-pass characteristics to be deducted, as they are, for. B. Fig. 18c shows the desired Bandpass characteristic 18 d to be obtained.

The method of using auxiliary filters shown in the last two examples is not only limited to these examples, but can also be used advantageously in connection with networks according to FIG Invention can be used wherever you have filter characteristics with very steep slopes Wants to get cutoff frequency. By their appropriate application it is z. B. possible a delay network to use, which is only in the range below the cutoff frequency as low as possible Distraction is calculated. At higher frequencies one will find in such a case that the transmitted Phases do not completely cancel the signal at the output. By using an auxiliary filter, the limit of which is effective in this area where signals occur again, is the application a much simpler delay network is possible without the sharpness the cut-off frequency would thereby be reduced.

In all of the above examples it was assumed that the attenuation in the delay network is neglected can be. However, this is not a necessary condition because this attenuation can be easily achieved be compensated for by a different choice of the coefficients. It is also possible to be frequency dependent Allow attenuations in the time constant network. This can be done through a filter happen, which is connected in cascade. This filter can either be of the known form or be constructed according to the invention.

Fig. 19 shows an electron discharge arrangement in which the various additions and subtractions, mentioned above can be done very easily. In an evacuated Sheath 40 is a cathode 41, which iao is heated by a heater 42. Two electrodes 43 and 44 are used for rough focusing of the electron current emitted by the cathode, the electrode 44 simultaneously acting as an acceleration electrode works. There are also baffles 45, 45 ', 46, 46', 47, 47 ', 48, 48',

which are shielded from each other by the shield electrodes 49, 50, 51. An anode 52 is provided which is formed by a uniform strip of resistance material, one side 54 of which with a positive voltage source is connected and the other side is connected to the output terminal 53 is. The output is proportional to the distance between the end 54 and the point on which the electron beam occurs, i.e. linearly dependent on the beam deflection. To the subtraction, To achieve the above mentioned, the different pairs of taps with opposite one another Plates connected. Amplitude modifications in the tubes can be made according to the different pairs can be achieved by suitable potentiometers or by setting the plate sensitivity. That the latter can be achieved by suitable spacing between the individual electrodes and by choosing a suitable bias voltage done for the shield electrodes 49, 50 and 51.

It should be noted that in all examples of the invention in which no tubes are used are, an interchangeability of input and output is possible without affecting the Characteristics of the networks and the success achieved something is changed.

Claims (5)

PATENT CLAIMS: i. Electrical transmission network for generating a desired amplitude and Phase curve over a wide frequency range, characterized in that delay chains of constant wave speed over the frequency range are provided, which so with the input and output terminals of the transmission network are connected and with tapping points or joints of the wave resistance are provided that the sum of the effective Admittance values of the transmission paths formed in this way of different amplitude and Phase characteristic gives the desired course. 2. Network according to claim 1, characterized in that that the input terminals are connected to a reflection-free, closed runtime chain which are connected to several tapping points via non-reactive coupling means of high resistance, in particular electron tubes, with the output terminals, possibly via different Tapping points of a second delay chain are connected (Fig. 5 and 8). 3. Network according to claim 2, characterized in that the input-side delay chain is provided with tapping points that are symmetrical in pairs to a fixed point (17) of the chain should preferably be equidistant from each other (Fig. 10). 4. Network according to claim 1, characterized in that that a delay chain whose unconnected end is open, short-circuited or is completed with the wave impedance, as a two-pole is switched on in the network. 5. Network according to claim 1 or the following, characterized in that to achieve a desired low-pass, band-pass or high-pass characteristics, additional filters of known design can be applied. For this purpose 2 sheets of drawings & 509545 9.55