Transitron type timebase.
One of the wartime radar developments ^{ [1] Radar Circuit Technique, Wireless World May 1946. Vol. LII.} ^{ [2] IEE Radiolocation Convention Papers} having immediate application to both television and cathoderay oscilloscopes is a singlevalve linear sawtooth oscillator. It is of particular interest because its performance is outstandingly good and it is at the same time both simple and economical.
In essence it is a modification of the transitron timebase which has its origins in the Numans oscillator ^{ [3] The Numans Oscillator, K C Van Ryn, Experimental Wireless (now Wireless Engineer), December, 1924. Vol. II, p.134.}. After lying dormant for over a decade the principle was revived under the new name of transitron ^{ [4] Negative Resistance and Devices for Obtaining it, E W Herold, Proc. Instn. Rad. Eng., October 1935. Vol. 23, p.1201} ^{ [5] The Transitron Oscillator, C Brunetti, Proc. Instn. Rad. Eng., February, 1939. Vol. 27, p.88}, but its applications were envisaged chiefly as a sinewave oscillator using a tuned LC circuit. Its use as a sawtooth generator ^{ [6] Trigger Circuits, H J Reich, Electronics, August, 1939.} was less widely realized, and it was not until 1940 that its capabilities appear to have been fully examined ^{ [7] Singlevalve Timebase Circuit, B C FlemingWilliams, Wireless Engineer, April 1940. Vol. XVII, p. 161} ^{ [8] Versatile Oscillator,W T Cocking, Wireless World, September, 1940. Vol. XLVI, p. 390}.
The wartime adaptation of it has resulted in an enormous improvement. In its present form it is really a combination of two things  the union of the prewar transitron with the wartime Miller integrator  and although it still functions on transitron principles the details of its operation are very different.
Since it is a combination of two circuits it is convenient to consider them separately as far as possible. The Miller integrator will be dealt with first, for this is the heart of the new circuit in as much as it is responsible for the linearity of, the sawtooth waveform. It is called the Miller integrator because it depends for its action upon the wellknown Miller effect. That is, the effect which makes the input capacitance of a valve (1 + A) times its gridanode capacitance, where A is the voltage amplification between the grid and anode circuits.
Fig. 1. The basis of a timebase. A charged capacitor is allowed to discharge slowly through a resistor.
In the normal timebase a capacitance is charged slowly through a resistance from the HT supply, and then discharged rapidly through a valve. Alternatively, a capacitance is charged rapidly through a valve and then discharged slowly through a resistance. Considering this latter form, if C in Fig. 1 is charged to an initial voltage E and is allowed to discharge through R, the voltage e at any time t after the start of discharge is e = Eℇ^{tT}, where T = CR.
The voltage falls exponentially. The slope de/dt = (E/T)ℇ^{tT}.
In practice e is permitted to change only a small amount and for this small change the discharge is considered to be linear. Thus, if a change of slope during the discharge of 2% is permissible ℇ^{tT} = 0.98 as a minimum and t/T = 0.02 as a maximum. Therefore; the change of voltage across C is restricted to 2% of the initial voltage on C.
For timebase purposes, a change of 40 Volts across C is usually desirable, and this demands an initial voltage, and therefore an HT supply, of 2,000 Volts. It is usually uneconomical to provide this.
Fig. 2. The Miller integrator adapted for a timebase.
The Miller integrator in the form adapted for a timebase is shown in Fig. 2. R_{1} is usually very small compared with R, so that to a close approximation the valve and R_{1} can be considered as a generator in series with R and C.
The action is most easily understood by starting with C charged to the full HT voltage and with the valve drawing some anode current, without bothering for the moment about how the capacitance is charged. The valve current produces a voltage drop across R_{1} so that the anode a is negative with respect to + HT,  the point b. Therefore, the grid g is negative with respect to b by the sum of the drop across R_{1} and the HT voltage. But the cathode c is negative with respect to b by the HT voltage. Consequently, g is negative with respect to c by the voltage drop along R_{1}. Grid current, therefore, does not flow.
As C discharges the voltage across ag falls and so g becomes less negative with respect to c. This increases the anode current and so makes a more negative with respect to b. As V_{ag} falls V_{ba} increases (V_{ag} is the voltage by which a is more positive than g). If the two were equal the voltage V_{bg} would be constant and the current through R would be constant. This current, however, is the capacitance discharge current and a constant current flowing out of a capacitance means a linear fall of voltage across it.
This cannot be achieved for, if V_{bg} were constant, V_{gc} would also be constant and there would be no change of grid voltage to produce V_{ba}. However, by making the amplification large V_{ba} can be much larger than V_{gc} and the linearity can approach perfection.
It is not difficult to show that if A is the voltage amplification (≈g_{m}R_{1}) the discharge current i = (E/R)ℇ^{tT} where T = CR(1+A). As far as current is concerned the circuit behaves as if the capacitance were (1 + A) times its actual value.
The anode voltage of the valve varies in the form
V_{ab} = E(1  A (1  ℇ^{tT} )).
It varies as if C were (1 + A) times its actual value and E were A times the true voltage. With a screened pentode g_{m} may be 6 mA/V and R_{1} may be 10 kΩ, making A = 60. Now A can easily be 250 Volts, so that the effective voltage is, 15,000 Volts. For 2% linearity, 0.02 of this or 300 Volts, would be available as output but for one limitation. The maximum output must be less than the real HT voltage. A linear output is obtained only if the valve is acting as a linear amplifier. The maximum possible output is about 80% of the real voltage, and to give a factor of safety it should be rather less than this figure.
When acting as a selfoscillator, the output obtainable is further restricted by the transitron action and is about 20% of the HT voltage. For a 250 Volt supply the output is about 50 Volts, but for this 50 Volts the linearity is as if 15,000 Volts were acting in the circuit.
The distortion is therefore only (50 × 100)/15,000 = 0.3%
Using a pentode, only the control grid, cathode and anode are needed for this linearizing action and the screen and suppressor grids are available for other use. If they are resistancecapacitance coupled a transitrontype circuit is formed and a selfoscillating linear timebase is obtained. The circuit is shown in Fig. 3.
Fig. 3. The combination of the Miller integrator with a transitron gives a linear sawtooth oscillator.
The action depends upon the ability of the suppressor G_{3} to control the ratio of the anode and screen currents. If the potential of G_{3} is changed negatively from zero it reduces anode and increases screen current, slowly at first and then quite rapidly. At zero Volts the anode current might be 10 mA and the screen current 2.5 mA. When G_{3} is very negative, however, the anode current may be zero and the screen current 1012 mA.
To a first approximation the cathode current is independent of the potential of G_{3} and this electrode acts to control the division of the current between G_{2} and anode.
Now consider Fig. 3 at an instant when C is fully charged and the valve is conducting. At this time G_{3} is at or near cathode potential and the screen current is at its minimum. The voltage on G_{2} is thus a maximum. The valve then functions more or less normally as an amplifier linearizing, the discharge of C in the manner already described.
If the potentials of G_{2} and G_{3} were constant the action would be identical, but because the G_{1} potential is changing positively, the G_{2} current is increasing. This means the potential of G_{2} is falling. The potential of G_{3} does not follow this change of voltage of G_{2} at all well, however, for the timeconstant C_{1} R_{3} is much less than CR (1 + A).
The main effect, therefore, is rather like that of a pentode amplifier without a screengrid bypass capacitor, and a form of negative feedback occurs reducing the amplification somewhat.
As this process goes on the total cathode current is increasing as well as the individual anode and G_{2} currents, but the ratio of anode to screen currents is substantially constant. Now R_{1} and R_{2} are of the same order of magnitude, therefore, the voltage change at the anode is greater than that at the screen and at length the anode potential drops to such a degree that its field acting through G_{3} is no longer sufficient to collect the normal proportion of electrons passing G_{2}. These consequently fall back to G_{2} and increase its current, thus dropping the G_{2} voltage.
This change of voltage is passed to G_{3} through C_{1} and makes it negative. This further reduces anode and increases G_{2} current, and so makes G_{3} still more negative. The action is cumulative and there is a very rapid transition to the flyback state with anode current cutoff, the G_{2} current a maximum and its voltage a minimum and G_{3} highly negative.
As anode current is cutoff, there is no voltagedrop across R_{1}. The voltage across. C is less than that of the HT supply, consequently G_{1} is positive with respect to cathode and C now charges rapidly from the HT supply through R_{1} and the G_{1}  cathode path of the valve. This positive G_{1} potential tends to increase the G_{2} current and so reduce the G_{2} potential further and make G_{3} still more negative.
The time constant C_{1} R_{3} is finite and G_{3} cannot remain negative indefinitely. It is, however, larger than the discharge time constant, comprising only C, R_{1} and the gridcathode resistance of the valve. When C is nearly fully charged the G_{1} grid potential becomes nearly zero and reduces the G_{2} current. There is a rise of G_{2} potential and also of G_{3}. The anode potential is also substantially at the full HT voltage. The rising G_{3} potential thus permits the anode to draw current again and so reduces the G_{2} current. This makes the G_{2} potential rise further and so G_{3} goes still further in the positive direction. Again there is a cumulative action and a very rapid changeover to the initial condition of G_{3} at about zero volts and G_{2} at its maximum.
Fig 4. Voltage waveforms on the various electrodes of the valve in the circuit of Fig. 3.
The various voltage waveforms are sketched in Fig. 4, and well illustrate the action. Synchronization can obviously be effected by applying a negativegoing pulse to G_{2} or G_{3} at a time prior to that at which the discharge would naturally occur.
The writer has used the circuit of Fig. 3 at 50 Hz with a TSP4 valve. With a 220 Volt HT supply an output at the anode of about 40 Volts is obtainable with C = 0.02μF; C_{1} = 0.01 μF, R = 4MΩ, R_{1} = R_{2} = 10 kΩ, and R_{3} = 0.5 MΩ R should be variable as a frequency control. A variable output can be secured at lowfrequencies by making R_{1} a potentiometer, and taking the output from the slider.
The flyback time is governed mainly by the value of C and can be increased by reducing this capacitance while increasing R to keep the product constant. The above values give a scantoflyback ratio of the order of 10:1. The TSP4 is not essential and the EF50 should be equally suitable.
In many cases the fact that the output is a negativegoing sawtooth is a disadvantage of the circuit. A positivegoing output can be obtained from the cathode, but only about onehalf the output can be secured. In addition, the G_{2} circuit must be decoupled to cathode.
Fig. 5. Circuit giving a positivegoing sawtooth output at the cathode.
The arrangement of Fig. 5 is satisfactory at 10 kHz with C = C_{1} = 100 pF; C = 0.1 μF; R = 4 MΩ, R_{1} = R_{2} = R_{3} = 10 kΩ; R_{3} = 0.5 MΩ; R_{4} = 1 MΩ; R_{6} = 20 kΩ. A TSP4 was again used and an output of about 20 Volts amplitude secured at the cathode.
The cathode output is limited by two factors. The first is that R_{6} is virtually in parallel with R_{5} as far as voltage changes are concerned, but they are in series for direct current. If the voltage on G_{3} is to be high enough for satisfactory transitron action, therefore, the values of both R_{5} and R_{6} are limited. The second factor is R_{1}; at least onehalf of the total output of the valve is developed across this resistance and must be if good linearity and transitron action are to be secured.
References
 Radar Circuit Technique, Wireless World May 1946. Vol. LII.
 IEE Radiolocation Convention Papers
 The Numans Oscillator, K C Van Ryn, Experimental Wireless (now Wireless Engineer), December, 1924. Vol. II, p.134.
 Negative Resistance and Devices for Obtaining it, E W Herold, Proc. Instn. Rad. Eng., October 1935. Vol. 23, p.1201
 The Transitron Oscillator, C Brunetti, Proc. Instn. Rad. Eng., February, 1939. Vol. 27, p.88
 Trigger Circuits, H J Reich, Electronics, August, 1939.
 Singlevalve Timebase Circuit, B C FlemingWilliams, Wireless Engineer, April 1940. Vol. XVII, p. 161
 Versatile Oscillator,W T Cocking, Wireless World, September, 1940. Vol. XLVI, p. 390
