AMPLITUDE ANALYZERS OF VOLTAGE PULSES

A. A. Sanin

To solve many problems of modern experimental physics, various circuits of amplitude analyzers of voltage pulses are often used. With their aid one obtains the amplitude spectrum of voltage pulses produced by some detector, for example a scintillation or proportional counter, or an ionization chamber.

The exceptionally important significance of such instruments has prompted many research laboratories to take up this problem. In the last few years alone, descriptions of several dozen different designs of amplitude analyzers have appeared in the periodical literature.

A multichannel amplitude analyzer of pulses consists of \(K\) simple differential discriminators. Since, when analyzers are used, it is not necessary to repeat the experiment \(K\) times and to switch the selected levels (thresholds), the measurement time is reduced and the accuracy of the data obtained is increased. The latter is connected with the reduction of level (threshold) drift due to the shortening of the duration of the experiment. However, despite the obvious merits of amplitude analyzers, because of their complexity and bulkiness they have become much less widespread than simple discriminators.

Depending on the content of the physical experiment, different requirements are imposed on the design of an amplitude analyzer. Thus, for example, in cases where the analyzer is used to obtain the energy spectrum of a short-lived isotope, especially stringent requirements are imposed on the resolving time of the instrument.

The principal characteristics of any analyzer are: the number of channels, resolving time, versatility, stability of the discrimination thresholds, linearity, simplicity and reliability of construction. Most of the characteristics are interrelated.

Thus, for example, increasing the number of channels and decreasing the resolving time usually leads to a complication of the instrument design. It is often believed that the number of channels completely determines the permissible amplitude resolution of the instrument. However, this does not correspond to reality if one takes into account that in some instrument designs expanders are used and not the entire spectrum is recorded, but only a portion of it. In most cases, parameters more important than the number of channels are the resolving time of the instrument and the stability of the channel widths. The number of channels and their width are determined by the type of detector with which the instrument operates. Thus, for example, it can be shown that when working with a scintillation counter it is most advantageous to choose a variable channel width, and the latter should increase in proportion to the square root of the amplitude level[^1].

The bulkiest part of a multichannel amplitude analyzer is the counting device. Indeed, suppose that the instrument is designed for a resolving time \(\tau = 10\) μsec; then the resolving time of the recalculating circuit must be \(\tau < 10\) μsec. If the instrument were intended for recording periodic pulses, then the scaling factor of the counting system would have to be \(10^3\) (it is assumed that the resolving time of the mechanical recorder is \(10^{-2}\) sec), i.e., a circuit consisting of at least 3 decades would be required. However, the pulses arriving from most detectors are distributed statistically in time; therefore the scaling factor of the counting system may be reduced to \(10^2\). Even with such a scaling factor, a twenty-channel amplitude analyzer becomes so bulky that the practical use of such an instrument is associated with great difficulties. Moreover, an instrument containing many hundreds of tubes becomes unreliable in operation, and its maintenance requires highly qualified personnel. Therefore, in most published analyzer designs the scaling factor of the counting systems is chosen to be small, of the order of 10–16, as a result of which the resolving power of the discriminating device is not fully utilized. Some authors have tried to simplify the design of the counting systems themselves by replacing the electron tubes in them with cold-cathode thyratrons. In this connection it is necessary to note the great importance of L. N. Korablëv’s work on the creation of simple and reliable scaling circuits using cold-cathode thyratrons[^2].

In recent years several papers have appeared in the periodical press[^3],[^4] devoted to the use in counting circuits of special tubes: trochotrons and dekatron tubes. Scaling circuits using trochotrons or dekatrons contain a smaller number of elements and can be successfully used in amplitude analyzers.

In one of the works[^5] published in recent years, a simple scheme of a 30-channel mechanical analyzer was described. This instrument has a very large resolving time (0.15 sec) and a small channel capacity (100 pulses). The analyzer consists of a plane inclined at an angle of 10°, in which 30 parallel grooves have been cut (Fig. 1). At one of the edges of the plane there is a “gun,” which is loaded with steel balls. The balls are pushed by a metal rod, which is moved by the coil of an electromagnet. The input signal is fed into a special device. This device amplifies and lengthens the input

Fig. 1.

pulses up to 100 μsec. The amplified pulse is fed to the coil of the electromagnet. The metal rod imparts to the ball a velocity proportional to the amplitude of the input signal. The ball moves over the plane along a parabolic trajectory and falls into the corresponding groove. Depending on the amplitude of the input signal, the initial velocity of the steel ball changes, and consequently so does the number of the groove into which the ball falls. After the arrival of a pulse, the input of the instrument is blocked for 0.15 sec, so that the next pulse cannot arrive before a new ball has been “loaded.”

The resolution is somewhat less than the channel width, so that for the same amplitude of the input signals 85% of the balls fall into the corresponding channel, while the remaining 15% are divided equally between the two adjacent channels.

The considered principle of discrimination of voltage pulses is not promising, since reducing the resolving time of a mechanical analyzer is associated with a number of fundamental difficulties.

In recent years descriptions of three designs of analyzers with diode discriminators have appeared in the periodical literature\(^{6,7,8}\). The first two designs, despite their complexity, have poorer characteristics than the third. The latter is a six-channel analyzer. However, the number of channels can be increased, since the design permits the joint use of several instruments. The resolving time of the analyzer is 10 µsec. The instrument is designed for discrimination of pulses with a minimum rise front of 0.2 µsec and a duration

Fig. 2.

of no more than 600 µsec. In each channel a 10-fold scaling circuit is provided; therefore the instrument permits a maximum counting rate of the order of 150 pulses/sec.

The input signal with a maximum amplitude of about 14 V is additionally amplified to 200 V; this amplifier is followed by 6 diode discriminators. The amplifier circuit is shown in Fig. 2. This circuit has an amplification factor of the order of five. The amplifier contains a cathode follower and a phase inverter. Linear amplification of pulses with a steep rise front to such large amplitudes is a difficult problem.

The anode circuit of the amplifying stage consists of a cathode follower and an active resistance. The control grid of the cathode follower is connected through a capacitance to the anode of the tube, which accounts for the large value of the resistance of the anode circuit. With the aid of such an amplifier circuit it is possible to obtain an output signal with an amplitude reaching 200 V without noticeable nonlinear distortions.

The cathode follower consists of two tubes. It is not difficult to show that the output resistance of such a follower is very small\(^9\). For the circuit shown in Fig. 2, the output resistance is approximately equal to several ohms. Cathode followers using two tubes transmit both positive and negative pulses equally well, even with a fairly large capacitive load.

From the output of the amplifier the negative signal is fed to the diode discriminators (Fig. 3). The positive pulse from the phase inverter is also applied to the diode discriminators through a small coupling capacitance. The capacitance is chosen so that

Fig. 3.

the pulse due to the diode capacitance is minimal. To partially compensate for changes in the discrimination level caused by fluctuations of the filament voltage or by aging of the tube, an additional diode \(L_2\) is included. Complete compensation can be obtained when the parameters of tubes \(L_1\) and \(L_2\) are identical. The biases to the diode discriminators are supplied from helical potentiometers. The pulse after the diode discriminator is amplified by tubes \(L_4\) and \(L_5\) and is applied to the input of a Schmitt circuit (\(L_6\) and \(L_7\)). The output signal from the anode of tube \(L_7\), through a differentiating cell, reaches the control grid of the coincidence tube. When a signal from the screen resistance of \(L_7\) of the next channel arrives at the antidynatron grid of this tube, tube \(L_8\) remains cut off. However, if channel \(n\) has fired and \(n+1\) has not, then tube \(L_8\) opens, and an output negative pulse is produced across its anode load. The pulse from the Schmitt circuit of the \(n+1\)-st channel does not coincide in time with the signal arriving at the control grid of tube \(L_8\). After the arrival

pulse from the \(n+1\)-st channel, the capacitor \(C\) is charged and maintains a negative potential on the antidynatron grid of \(L_8\) until the discharge signal arrives. The discharge signal returns the entire circuit to its initial state. The discharge-signal generator is not shown in the circuit diagram. The output signal from the anode of tube \(L_8\) is fed to a decade counting circuit.

The principal merit of the circuit considered above is the comparatively high stability of the discrimination levels. In addition, such a circuit can operate from pulses of various

Fig. 4.

durations and shapes without additional adjustment. However, such an analyzer has a large number of tubes per channel (10 in all). It should also be noted that the counting factor in the channels is too small; therefore the resolving time of the discriminating section is not matched to the resolving time of the counting system.

Figure 4 shows the circuit of the discriminating device of another original design of an amplitude analyzer\(^{10}\). Such an analyzer has high stability of the discrimination levels (30 mV) and a comparatively short resolving time (50 μsec). The instrument permits a counting rate in the channels of up to 2000 pulses/min. In each channel a tenfold counting device is provided.

The instrument is based on a discrimination principle using a threshold system. The input signal is applied to the control grid of the tube of a differential amplifier. By varying the bias on the control grid of the differential-amplifier tube, different portions of the amplitude spectrum can be selected. From the output of the amplifier the signal is applied to a cathode follower. Part of the output signal, taken from the cathode load, is fed back to the differential amplifier. Lengthening of the input signal is achieved by charging a capacitor through a diode. The effect of the interelectrode capacitance of the diode is neutralized by applying a pulse of the opposite sign from a phase inverter. The voltage on the “memory” capacitor is maintained equal to the amplitude value of the signal taken from the cathode follower for the time required for discrimination. During this interval the discharge current of the capacitor is reduced to a minimum. This is ensured by taking the voltage from the capacitor through a two-stage cathode follower. In addition, the filament voltage of the tubes connected with this capacitor is somewhat below the nominal value (75%). The filament circuits of these tubes are supplied with direct current. The shaped pulse is applied to the sorting device, where it is further amplified and discriminated. The circuit of the discriminating device is shown in Fig. 4. Tubes $L_1$ and $L_2$ amplify the selected portion of the amplitude spectrum. A negative pulse with a maximum amplitude of 330 V is applied to the control grid of $L_6$ and, through diodes $L_3$, $L_4$, and $L_5$, to the control grids of the remaining tubes of the discriminating device. This device consists of five double triodes, the left-hand triodes ($A$) being connected in series. Before the arrival of the input signal, the control grid of each of these triodes is maintained at cathode potential by the grid current flowing through the grid-leak resistance. Thus the potentials of the control grids of the left-hand triodes decrease successively. The right-hand triodes ($B$) are normally cut off. A negative pulse is applied to the input of the discriminating device. This pulse cuts off tube $L_{6A}$. The anode of tube $L_{6A}$ is galvanically connected to the control grid of $L_{6B}$, owing to which the cutting off of tube $L_{6A}$ causes the opening of $L_{6B}$. The operation of the discriminating cell is analogous to that of a Schmitt circuit. If the amplitude of the negative pulse is sufficiently large, one or several discriminating cells are triggered. Cutting off the left-hand half of any double triode causes the current to cease in all the double triodes connected above that tube. The current flowing upward is switched into one of the six anode circuits, depending on the amplitude of the input signal. The discriminating device has five channels, the width of each of which is 50 volts. After the discriminating device, in each channel-

a coincidence circuit on resistors is included (not shown in Fig. 4). At the output of the coincidence circuit a signal is produced only when the given channel is triggered and a pulse from the generator is applied to it. In this way false starting of the counting devices of the channels, caused by the leading or trailing edge of the shaped signal, is eliminated. A decimal recalculating circuit is connected to each channel. This circuit consists of two counters: a binary one on electron tubes and a quinary one on neon lamps.

The most interesting element in this instrument is the device for centering the amplitude spectrum. Such a device can also be successfully used in other designs of amplitude analyzers. Unfortunately, the author does not give the circuit solution of the proposed principle. Automatic centering of the amplitude spectrum is a very important problem in view of the fact that the parameters of the pulse sensor, the amplifier, and the discriminator circuit itself may change slowly. The system for automatic centering of the amplitude spectrum consists of two count-rate meters and a differential amplifier. The circuits of the count-rate meters are connected into two adjacent channels. The output voltage of a count-rate meter is proportional to the number of pulses entering the given channel per unit time. The voltage from the count-rate meters is fed to the control grids of the tubes of the differential amplifier. Suppose that the amplitude spectrum has a peak in a certain amplitude range. By changing the potential of the control grid of the input tube of the differential amplifier, the amplitude spectrum is shifted so that the top of the peak lies between the channels to which the count-rate meters are connected. From the differential circuit a constant voltage is fed to the input amplifier of the discriminating device. Any shift of the spectrum causes a change in the output voltages in the count-rate-meter circuits and, consequently, in the output voltage at the differential amplifier. This voltage shifts the amplitude spectrum in the opposite direction, thereby compensating for the instability of the parameters of the pulse sensor, amplifier, and discriminator. The stabilization principle considered above can be applied only in the case when a sufficiently large number of pulses per unit time enters both channels to which the count-rate meters are connected.

In recent years several papers have been published^11,12,13 in which an electronic commutator was used as the discriminating element. In its design the electronic commutator resembles a cathode-ray tube and differs from the latter in that, in place of the screen, a row of flat electrodes (dynodes) is soldered in. Like a cathode-ray tube, the electronic commutator has a cathode, a control grid, focusing and accelerating elec-

electrodes, deflecting plates. Amplitude analyzers with electronic switches consist of a shaping device, a sweep generator, output amplifiers, and counting devices. The shaping device converts the input signal into a rectangular pulse. The amplitude of this pulse is proportional to the amplitude of the input signal. The rectangular pulse is applied to the deflecting plates, and the sweep signal to the control grid of the electronic switch. Depending on the amplitude of the cathode pulse, the switch beam strikes different dynodes. The amplitude of the voltage pulse arising at the dynode is too small to start the counting system; therefore the signal is preamplified. If the beam falls on the edge of a dynode, then, depending on the quality of focusing and on the position of the electron beam, the amplitude of the output pulse changes. Owing to this, the apparent position of the discrimination level of the given channel depends on the gain of the amplifier and on the operating threshold of the registering device. For more effective control of this level, an additional discriminator is connected between the amplifier and the counting system. Usually such an increase in the number of tubes in each channel greatly complicates the design of the instrument. It should also be noted that serious difficulties arise in manufacturing electronic switches, since the exact positioning of the dynodes in the electronic switch is a very difficult problem.

In one of the first designs^[11] of analyzers of this type, an electronic switch with 50 dynodes was used. The electronic switch was under continuous pumping. To reduce secondary emission, the dynodes were coated with a layer of aquadag (colloidal graphite). Figure 5 shows the circuit of the shaping device of this analyzer. The input signal is applied to the control grid of a stage in whose cathode circuit an artificial line is connected, loaded by a high-ohmic resistance (100 kΩ). At the anode of this tube a signal of duration 22 μsec is produced. At the initial moment the resistance of the cathode circuit is small and equal to the characteristic impedance of the artificial line; therefore the gain of the stage is greater than unity. Conversely, after a time interval equal to the delay time of the artificial line, the resistance of the cathode circuit becomes very large and the gain of the stage practically falls to zero. From the output of the shaping stage the signal goes to an amplifier with feedback, \(L_2\) and \(L_3\). The amplified and shaped signal is applied to the cathode follower. The capacitance \(C = 1500\ \mu\mu\mathrm{F}\) is charged by the current of the cathode follower through the crystal detector \(KD_2\). Discharge of this capacitance takes place through the crystal detector \(KD_1\). The output pulse has a flat top and an amplitude proportional to the amplitude of the input signal. The sweep generator is assembled on a thyratron. Through the artificial line

a shaped signal is applied to the control grid of the thyratron. An artificial line, connected into the cathode circuit of the thyratron, discharges and produces across the cathode resistance a rectangular pulse of duration 5 μsec. This pulse is fed to the control grid of the electronic switch. The shaped input signal is applied to the deflection plates

Fig. 5.

of the electronic switch. A three-stage amplifier is connected to each dynode. The resolving time of the instrument is mainly limited by the resolving time of the mechanical counters and is equal to 120 pulses/sec.

The amplitude-analyzer circuit considered has a number of serious shortcomings. The most important shortcomings of the instrument are: a short resolving time, the need for constant pumping of the electronic switch, and a relatively large number of tubes. In addition, the instrument is intended for operation together with one type of detector.

A more advanced design of the instrument was developed using a standard electronic switch[^12]. The circuit of the instrument is shown in Fig. 6. The input signal, having a rise front of less than one microsecond (it is assumed that the instrument operates together with a photomultiplier), is applied to the control grid of the cathode follower. Through capacitance \(C_1\) and diode \(D_2\), capacitor \(C_2\) is charged to the amplitude value of the input

pulse. Before the arrival of the input signal, the voltage on capacitor \(C_2\) is fixed by diode \(L_2\). The input signal is amplified and fed, through an artificial line with a delay equal to three microseconds, and a shaping stage, to the discharge pentode \(L_6\). On capacitor \(C_2\) a signal is produced with a flat top and an amplitude equal to the amplitude of the input pulse. This signal is fed through an amplifier to the deflection plates of the electronic commutator. A pulse of duration \(1.5\ \mu\mathrm{sec}\) is applied to the control grid of the commutator. This

Fig. 6.

pulse is generated by a one-shot multivibrator (not shown in the diagram). The one-shot multivibrator is triggered by the input signal, amplified and delayed by one microsecond. Each of the 10 output amplifiers has 2 triode stages and a cathode follower. By adjusting the response threshold of the counting devices, one can make the width of each channel the same. The maximum counting rate is limited by the resolving time of the recording device. Analyzers with electronic commutators, despite certain advantages (absence of anticoincidence circuits, fairly high stability of the discrimination levels, high resolving power), nevertheless have not come into wide use.

A comparatively simple and reliable design of an amplitude analyzer was developed using ordinary one-shot multivibrators[^14]. Such an instrument has twenty channels and a small resolving

time (2.4 μsec). The block diagram of the analyzer is shown in Fig. 7. The analyzer comprises the following units: an expander, monitoring and discriminating devices, and counting circuits.

The signal from the amplifier is fed through an artificial line to the expander. At the output of the amplifier a relaxation circuit is connected, generating short negative pulses. These pulses

Fig. 7.

are generated simultaneously with the beginning of the output signal and are sent to the monitoring device. If pulses from the detector follow one another at time intervals less than 2.3 μsec, then the monitoring device does not switch on the expander and also does not produce the reversal pulse going to the discriminating device. Thus the circuit discriminates only those pulses which are bounded on both sides by a free time interval of 2.3 μsec. The expander selects a part of the spectrum (10–20%) and feeds it to the discriminator. The gain factor of the expander is constant and equal to five. The channel width was chosen equal to 3 V; therefore the portion of the spectrum discriminated by the instrument is 12 V.

The expander circuit is shown in Fig. 8. This circuit consists of a cathode follower, an amplifier with feedback, and a lengthener. The lower boundary of the portion of the spectrum that is fed to

discriminating device, is determined by the setting of the slider of potentiometer \(R_1\). In order that the amplifier not be overloaded, a number of nonlinear elements (diode \(L_3\), crystal detector \(KD_1\)) are included in its circuit. The crystal diodes \(KD_4\), \(KD_5\) and \(KD_6\), and the tubes \(L_7\), \(L_8\), and \(L_9\), are elements of the pulse stretcher. In the quiescent state the control circuit of the stretcher (this

Fig. 8.

circuit is shown in Fig. 9) passes \(3.2\) mA through the crystal diodes \(KD_4\), \(KD_5\) and terminal \(M\). When a negative pulse from the amplifier triggers the controlling device, the current is switched from terminal \(M\) to terminal \(N\). Capacitor \(C_{10}\) is charged by the cathode follower \(L_6\) to the amplitude value of the input signal. After the passage of the signal, the cathode follower and the crystal detector \(KD_6\) maintain a constant voltage at point \(B\), so that the entire reverse potential is applied to \(KD_6\). A voltage of less than \(1.5\) V appears across the crystal detector, since this detector has a very high reverse resistance. The discharge current is very small; therefore the output pulse has a flat top. After \(4.5\ \mu\text{s}\) the controlling device returns the stretcher to its initial state.

To accelerate the discharge of capacitor \(C_{10}\), tube \(L_8\) is included. The output cathode follower \(L_9\) has a capacitive load; therefore the signal appearing at its anode is close in shape to the negative derivative of the pulse. At the moment this pulse ends, a positive pulse is produced at the anode of tube \(L_9\),

which opens the discharge tube \(L_8\). The pulse on the capacitor decays rapidly; however, the output signal has a more gradual trailing edge because of the capacitive load of the output cathode follower.

Figure 9 shows the circuit of the monitoring device. A pulse from the amplifier generator, of duration \(0.4\ \mu\text{sec}\), is applied to the control grid of tube \(L_{10}\). An artificial line is connected in series with the anode load of this tube; therefore the positive signal across resistor \(R_1\) appears after \(2.3\ \mu\text{sec}\).

Fig. 9.

At the anode of tube \(L_{11}\) a negative pulse is produced simultaneously with the input signal. This pulse passes through crystal detector \(KD_1\) to the control grid of the mixer tube \(L_{12}\) and charges the capacitor \(C = 15\ \mu\mu\text{F}\). The capacitance of this capacitor is selected so that tube \(L_{12}\) remains cut off by the control grid for \(2.3\ \mu\text{sec}\). When tube \(L_{12}\) opens at the control grid and a positive signal arrives at its antidynatron grid, a negative pulse is produced at the anode. However, in the case when two signals separated by a time interval of less than \(2.3\ \mu\text{sec}\) arrive at the input of the monitoring device, the first signal does not produce an output pulse. The pulse from the anode of tube \(L_{12}\) triggers the monostable multivibrator \(L_{14}\) and \(L_{13}\). The output signal from this monostable multivibrator has a duration of several hundred microseconds. However, with the aid of tube \(L_8\) the monostable multivibrator is forcibly returned

to the initial state after \(1.8\,\mu\text{sec}\). Indeed, the leading edge of the pulse from the monostable multivibrator is amplified by tubes \(L_9\) and \(L_{20}\) and is delayed by \(1.8\,\mu\text{sec}\) by means of an artificial line connected into the anode circuit of tube \(L_{20}\). This signal is differentiated; the resulting rectangular positive pulse of duration \(0.4\,\mu\text{sec}\) is applied to the control grid of the output stage. Until the arrival of the pulse, crystal detector \(KD_3\) maintains the anode potential of tube \(L_{21}\) equal to \(+271\ \text{V}\). When tube \(L_{21}\) opens, the potential of its anode cannot fall below \(265\ \text{V}\) because of the current of the crystal detectors \(KD_4\). Such limitation of the amplitude of the output signal is necessary for reliable operation of the anticoincidence circuits of the discriminating device. The output signal is fed back through the differentiating circuit \(R\) and \(C\) to the control grid of the trigger tube \(L_9\) of the monostable multivibrator.

Fig. 10.

Figure 10 shows the circuit of the discriminating device. This circuit consists of monostable multivibrators, anticoincidence circuits, and output stages for triggering the counting systems. The use of monostable multivibrators in the discriminating device instead of Schmitt circuits made it possible to greatly simplify the construction of the instrument. The voltage on the control grid of the right-hand triode of each monostable multivibrator is set equal to \(+100\ \text{V}\). The control grids of the left-hand triodes are connected through crystal detectors with high reverse resistance to the taps of a divider with successively decreasing potential. The output signal from the expander is applied to the control grids of the monostable multivibrators. Depending on the amplitude of the input signal, one or several monostable multivibrators are triggered. Crystal detectors \(KD_1, KD_3\)

are included to reduce the dead time of the one-shot multivibrators. The positive signal from the one-shot multivibrators is fed through a resistance of \(10\ \text{k}\Omega\) to the control grid of the output stage. However, this tube opens only if a negative signal from the monitoring device simultaneously arrives at its cathode.

Fig. 11.

If a one-shot multivibrator in the next higher channel is triggered, no signal is produced at the given output stage, since from the anode of the left tube, through the crystal detector \(KD_4\), a quenching negative pulse is applied. The pulses from the output stages are fed to the recalculating devices. In Fig. 11 oscillograms of voltages at various points of the circuit are shown. The analyzer correctly discriminates pulses with a rise time of not less than \(0.2\ \mu\text{sec}\).

In recent years the author has proposed the principle of amplitude-time transformation*). The input signal was first converted into a rectangular pulse in such a way that the duration of the latter was proportional to the amplitude of the input signal. Measurement of the duration of the rectangular pulse was carried out by a mechanical or electronic commutator. It turned out that, using the principle of amplitude-time transformation, one can construct various analyzer systems with a smaller number of elements. In addition, it can be shown that in such systems there is practically no relative drift of channel width.

Fig. 12.

On the basis of the proposed principle, the author, together with N. N. Sukhanova, developed the design of a motor analyzer[^15]. Such an analyzer had 30 channels and a resolving time of 0.05 sec. The circuit of the instrument is very simple (Fig. 12). The input signal is applied to the control grid of tube \(L_1\). Through diode \(L_2\), capacitor \(C_1\) is charged to a voltage equal to the amplitude of the input signal. Capacitor \(C_1\) is chosen to have a small capacitance; therefore a pulse having even a short rise-front duration (several microseconds) has time to charge it to a voltage equal to the amplitude of the input signal. To the second cathode follower the pulse is applied through an integrating network, owing to which

*) The principle of amplitude-time transformation was proposed by the author independently of Wilkinson.

the front of the input signal increases in length. By the current of the cathode follower \(L_3\), another capacitor, having a considerably larger capacitance \((0.1\ \mu\mathrm{F})\), is charged through the diode \(L_4\). The cathode of the diode \(L_4\) is connected with the segments of the commutator \(K\) and \(K_0\). The axis of the commutator is rotated by a synchronous motor.

A search contact is mounted on the axis of the commutator. All segments, except \(K\) and \(K_0\), are connected with the control grids of trigger tubes, in whose anode circuits mechanical counters are included. When the search contact of the commutator closes the segments \(K\) and \(K_0\), a positive pulse is produced across the resistor, with an amplitude proportional to the amplitude of the input signal. This pulse is applied to the control grid of \(L_5\). The tube \(L_5\) is connected as a cathode follower. From the output of \(L_5\) the signal is applied to a second cathode follower, which charges the capacitor \(C_2\). The capacitor \(C_2\) is discharged by the pentode of constant current \(L_{10}\). The internal resistance of the tube \(L_{10}\) is sufficiently large; therefore the change of voltage on the capacitor \(C_2\) is close to linear. The duration of the sawtooth pulse is proportional to the amplitude of the input signal and can be regulated by means of the potentiometer \(R_2\). From the anode of the tube \(L_{10}\) the sawtooth pulse is applied to the input of the Schmitt circuit. The rectangular pulse from the anode of the tube \(L_9\) is applied to a differentiating cell. The time constant of the differentiating cell is chosen so that the duration of the output positive pulse is somewhat less than the sliding time of the commutator search contact between segments. From the output of the differential cell the positive pulse is applied to the commutator search contact, while the negative pulse is cut off by the diode \(L_8\). The positive pulse taken from the output of the differentiating cell coincides with the end of the rectangular signal. Depending on the amplitude of the input signal, by the time the positive pulse arrives the commutator search contact will have turned through a different angle relative to the segments \(K\) and \(K_0\). Therefore the positive pulse can enter only that channel on whose segment the commutator search contact was located. Figure 13 shows voltage waveforms at various points of the circuit. To actuate the mechanical counter, in each channel there is connected

Fig. 13.

ion relay of Koroblev. The quenching of unmarked tyratrons is carried out by a common relay.

The considered principle of pulse discrimination is promising, since replacement in the analyzer of the mechanical commutator by an electronic one (for example, of the Zernov type) will make it possible to reduce the resolving time to one hundred microseconds. Replacement of the mechanical commutator by an electronic one entails some complication of the instrument circuit. Thus, for example, an additional tube is used to discharge the capacitor \(C_2\). The moment at which the discharge tube opens is determined by a certain phase of the sinusoidal voltage applied to the circular sweep of the electronic commutator.

Several circuits of amplitude analyzers based on the principle of amplitude-time transformation have been described in foreign periodical literature \(^{16,17,18}\).

Fig. 14.

On the basis of this principle Wilkinson developed a 99-channel amplitude analyzer \(^{16}\) with a resolving time of \(0.1\) sec. In this analyzer the circuit of the time-transformation unit is designed very successfully. The accuracy of operation of the instrument is determined mainly by the quality of the amplitude-time transformation.

A positive input signal is applied to the control grid of the cathode follower (Fig. 14). By the current of the cathode follower through diode \(L_2\), the capacitance \(C\) is charged to the amplitude value of the input pulse. Before the arrival of the signal, the voltage at the anode of tube \(L_3\) is equal to the output voltage of the cathode follower. The control grid of tube \(L_3\) is connected through

the diode \(L_4\) to ground; therefore the potential of this grid during the charging of the capacitor cannot become positive. Owing to the deep feedback, the discharge of capacitor \(C\) follows a linear law. The rate of discharge is determined by the magnitude of the resistance \(R\), the capacitance of capacitor \(C\), and the voltage at point \(M\). For the proper operation of the circuit it is necessary that the input pulse decay faster than capacitor \(C\) is discharged.

The tube \(L_3\) is cut off at the moment when the input signal begins to decay, and opens after the discharge of capacitor \(C\). Thus, the time during which tube \(L_3\) is cut off depends linearly on the amplitude of the input pulse. The signal taken from the screen resistance of tube \(L_3\) is shaped with the aid of the stages \(L_5\) and \(L_6\). It is then fed to a sinusoidal generator, which produces a sequence of pulses whose number is proportional to the amplitude of the input signal. These pulses are counted by means of two decade ring scaling circuits connected in series. The first circuit determines the units digit, while the second is triggered once for each cycle of the first scaling circuit and determines the tens digit. The circuit includes 20 relays, each of which is actuated by the corresponding tube of the scaling circuits. By means of these relays, voltages are applied to mechanical counters. Suppose that a signal has arrived at the input of the analyzer whose amplitude corresponds to 46 pulses of the sinusoidal generator. In this case one of the terminals of the 46th mechanical counter is connected to ground by the sixth relay, which is supplied by the sixth tube of the units decade. The other terminal of this mechanical counter is connected to the output tube, through which a large pulse current flows after the end of the shaped rectangular signal. This connection is made by means of the fourth relay of the decade scaling circuit that marks the tens. In this instrument the maximum counting rate is limited by the resolving time of the relay and of the mechanical counters. As the author notes, the analyzer has a blocking device which cuts off the input of the instrument for \(0.1\) sec after the arrival of the next signal. The analyzer contains 70 tubes. The considered principle of amplitude discrimination is very promising, since it is possible to reduce the resolving time by replacing the mechanical relays with electronic ones.

In work \(^{17}\) another analyzer circuit is described, likewise based on the principle of amplitude-time transformation. In contrast to the designs of amplitude analyzers considered above, this instrument has a special counting device containing a comparatively small number of tubes. After amplification and lengthening, the input signal is compared with a linearly increasing voltage. This comparison is carried out by means of a dif-

...of the differential amplifier. A constant voltage equal to the amplitude of the input pulse is applied to one input of the amplifier, and a sawtooth signal to the other input. At the moment when the voltages at both inputs of the differential amplifier become equal, an output signal is generated. The time required for the sawtooth signal to rise to the voltage at the other input of the amplifier is proportional to the amplitude of the input pulse. The output signal from the amplifier is fed to the control grid of a Schmitt circuit (trigger). The signal from the trigger is applied to a coincidence circuit. To the other input of the coincidence circuit there is applied a pulse from a generator (Fig. 15).

Fig. 15.

The period of oscillation of this generator is chosen in accordance with the required number of channels. For example, if it is required to have 60 channels, then with a sawtooth pulse duration of 1200 μsec the oscillation period of the generator must be equal to 20 μsec.

Thus, at the output of the coincidence circuit a pulse is created simultaneously with the signal arriving from the generator. This signal is transmitted back to the stretching device. The “memory” capacitor is discharged, and the Schmitt circuit is thrown back into its initial state, owing to which, during the time

during the period of the sawtooth signal, only one pulse can be produced at the output of the coincidence circuit. This pulse is shifted relative to the beginning of the sawtooth signal by \(k \cdot 20\ \mu\text{sec}\), where \(k\) is the channel number. Depending on the amplitude of the input pulse, the number \(k\) may vary from 1 to 60. The data on the amplitude of the input pulse are stored by means of an ultrasonic mercury delay line. This is accomplished in the following way. The signal from the output of the coincidence circuit enters a special device, which passes it onward only in the case when, at that moment, no control pulse is arriving at its other input. If signals are applied simultaneously to both inputs of this device, then at the output a pulse is produced that is shifted in time by one microsecond. Next, the signal from this device starts the modulator of the generator. The oscillation frequency of the generator of the ultrasonic line is 13 MHz. This generator excites a quartz resonator located at one end of the ultrasonic mercury line. At the other end of the line a second quartz resonator is fixed. The propagation time of the signal along the line is slightly more than 1200 \(\mu\text{sec}\). The signal from the second quartz resonator is amplified, detected, and fed to a special device as a control pulse. In the case when a new pulse has not arrived at the input of the instrument which falls into the same channel, the special device passes the pulse from the detector to the modulator. The process becomes periodic.

Thus, in each period of the sawtooth signal a pulse will appear. The ordinal number of the pulse is determined by the magnitude of the amplitude of the signal applied to the input. If a second pulse is applied to the input of the instrument with an amplitude corresponding to the same channel, then the output signal of the special adding device will be shifted relative to the pulse of that channel by one microsecond. Accordingly, the arrival in the same channel of \(2^3\) pulses shifts the output signal of the adding device by 3 \(\mu\text{sec}\). In this way, pulses are counted in all channels. It was noted above that the full period of the sawtooth oscillations is divided into \(m\) equal parts (according to the number of channels). Each time interval corresponding to the width of a channel, in turn, is divided into \(n\) parts, where \(n\) is the coefficient of multiplicity of the binary recalculation. In the example considered, the instrument had 60 channels; therefore, with a sawtooth-signal duration of 1200 \(\mu\text{sec}\), each channel occupies 20 \(\mu\text{sec}\). The input signal opens one of the 60 channel pulses. Suppose that 43 pulses have fallen into the second channel; then in the time interval from 20 \(\mu\text{sec}\) to 40 \(\mu\text{sec}\) there will be a series of signals. The first signal is shifted in time relative to the beginning of the sawtooth

AMPLITUDE ANALYZERS OF VOLTAGE PULSES

pulse at \(20\ \mu\text{s}\), the second at \(21\ \mu\text{s}\), the third at \(23\ \mu\text{s}\), and the fourth at \(25\ \mu\text{s}\), since

\[ 43 = 2^0 + 2^1 + 2^3 + 2^5. \]

Oscillograms of the voltages at various points of the circuit are shown in Fig. 16.

The resolving time of the instrument, the capacity of a given channel (the maximum number of pulses that can accumulate in the channel), and the number of channels are uniquely related to one another*).

Increasing the resolving power of the instrument causes a decrease in the permissible number of channels or in their capacity. Thus, for example, with a sawtooth-signal period of \(1200\ \mu\text{s}\) and a number of channels equal to 60, the permissible channel capacity is \(2^{19}\) pulses. Accordingly, with 120 channels the maximum channel capacity is \(2^{10}\) pulses.

Registration of the filling of the analyzer channels is carried out by means of a cathode-ray tube. Sawtooth signals are applied to the sweep and deflection plates of the cathode-ray tube. The duration of the sawtooth pulse applied to the sweep plates is \(1200\ \mu\text{s}\), and to the deflection plates, \(20\ \mu\text{s}\) (for 60 channels). A raster is obtained on the screen of the tube. The number of lines in the raster is set by the required number of channels. Detected signals from the ultrasonic mercury line arrive at the control grid of the tube; therefore a series of luminous points appears on the raster lines. If 43 pulses have entered the second channel, then the first, second, fourth, and sixth points of the second line will glow. The analyzer circuit considered above has a number of substantial advantages. Among these advantages one should first of all mention: simplicity of construction (a comparatively small number of tubes—82 pcs.), reliability in operation, absence of mechanical counters, and a small relative drift of the channels. Along with the advantages noted, the analyzer has many shortcomings.

Fig. 16.

Labels in the figure:

• Starting pulses
• Channel 1; Channel 2; Channel 62; Channel 100
• Trigger output signal
• \(10\ \mu\)
• \(100\ \text{kHz}\) Generator
• \(1\ \text{MHz}\) Generator
• Sawtooth signal
• Input signals
• Stretched input signals
• Coincident marker pulse
• The pulse enters channel 62

* It is assumed that the minimum permissible displacement of the signal in the adding device is some fixed quantity, approximately equal to one \(\mu\text{s}\). This quantity depends on the character of the tube and on a number of other causes.

The main drawback of the instrument is its comparatively low resolving power and its dependence on the number of channels and on their capacitance. Among other shortcomings of the instrument, mention should be made of the bulkiness of the ultrasonic delay line used in it, the difficulty of transporting and manufacturing it, and the difficulty of obtaining small leakages in the circuit of the “storage” capacitor (the insulation resistance must be of the order of \(10^{10}\ \Omega\)).

An analysis of the operation of published circuits of amplitude analyzers of voltage pulses shows that at present it is impossible to point to any instrument design that would combine simplicity, reliability in operation, a comparatively large number of channels, and a short resolving time.

The most promising principle of amplitude discrimination is the principle of amplitude–time transformation. Indeed, in instruments based on this principle, the stability of the discrimination thresholds depends mainly on the comparison element common to all channels. Therefore, in order to increase the accuracy of the instrument’s operation it is sufficient to stabilize a small number of circuit elements.

This review has not considered certain obsolete designs of amplitude analyzers, nor systems whose principles of operation are not new.

REFERENCES

1. G. Kelley, Nucleonics 10, No. 4, 34 (1952).
2. L. N. Korabev, DAN 75, No. 3, 375 (1950).
3. M. I. Elinson and G. A. Baryshevskii, Avtomatika i telemekhanika 13, No. 2, 204 (1952).
4. A. V. Shileiko, Trudy MEI 13, 42 (1952).
5. S. G. Frank, O. R. Frisch, G. G. Scarott, Phil. Mag. 42, 603 (1951).
6. W. F. Freundlich, E. P. Hincks, W. I. Ozeroff, Rev. Sci. Instr. 18, 90 (1947).
7. G. H. Westcott, G. C. Hanna, Rev. Sci. Instr. 20, 181 (1949).
8. N. J. Moody, W. I. Battell, R. H. Taplin, Rev. Sci. Instr. 22, 551 (1951).
9. C. Hammock, Radiation Lab. Rept. 469 (1943).
10. E. H. Cooke-Yarborough, J. Inst. Elec. Engrs. 97, III, 108 (1950).
11. W. E. Glenn, Nucleonics 4, No. 6, 50 (1949).
12. W. E. Glenn, Nucleonics 9, No. 6, 24 (1951).
13. P. A. Watkins, Rev. Sci. Inst. 20, 495 (1949); G. Kelley, Nucleonics 10, No. 4, 34 (1952).
14. A. B. Van Rennes, Nucleonics 10, No. 10, 50 (1952); C. W. Johnstone, Nucleonics 11, No. 1, 36 (1953).
15. A. A. Sanin and N. N. Sukhanova, Vestnik Moskovsk. universit. No. 8, 105 (1953).
16. D. H. Wilkinson, Proc. Cambridge Phil. Mag. 46, 508 (1950); D. H. Wilkinson, J. Sci. Instr. 27, 36 (1950); UFN 43, 148 (1951).
17. G. W. Hutchinson, G. G. Scarrott, Phil. Mag. 42, 792 (1951); G. W. Hutchinson, Nucleonics 11, No. 2, 24 (1953).