17. The recovery of control and the memory between two pulses

When the anode current is interrupted, the device still carries the traces of the discharge. The experiments on these pages make it possible to follow the residual charge through three observable quantities: the grid current, the voltage the tube can withstand again and its sensitivity to the next command.

Sources examined on this page

Malter and Johnson (1950), part I. Reading the plasma through the grid resistor

After extinction the authors keep the source negatively biased and record the grid voltage. For a certain time the grid can sit almost at cathode potential even when the source is at −40 V. The difference falls across the resistor. Knowing its value, one reconstructs the current the residual plasma can still supply to the electrode. A measurement of the supply voltage alone would have hidden the whole phenomenon.

The comparison between resistances shows that the time needed for the voltage on the pin to return negative also depends on the external circuit. For part of the trace the plasma supplies enough charge to sustain the drop across the resistor. When the ion flux decreases, the operating point moves. This mechanism gives rise to the graphical construction later proposed by Reich.

RCA Review, June 1950, part I · Second copy of the journal, oscillograms

Johnson and Malter (1950), part II. The charge keeps a memory of the grid

The second part reverses the order of the authors and puts the hypothesis of instantaneous equilibrium to the test. Instead of simply waiting for the discharge to decay at constant bias, the authors change the grid bias shortly before checking re-ignition. The history immediately preceding the test also affects the result. A charge distribution has to reorganize itself before it corresponds to the new value of the voltage.

For a circuit model the consequence is concrete. A formula that depends only on the instantaneous grid voltage and on the time elapsed since extinction will assign the same response to tests that end at the same point. The late change of bias makes it possible to check whether this approximation is sufficient. One of the types identified in the work is the 2050.

RCA Review, part II, pp. 178–189

Malter and Boyd (1951). One electrode to command, another to recover

The authors build a prototype in which a small-area grid receives the signal and a second, low-impedance grid favors blocking. Reducing the area of the control electrode limits both the emission from the hot grid and the ion charge collected after extinction. The most instructive test is the one in which the two grids are brought to the same potential: the behavior returns closer to that of the conventional tube. The result therefore changes even without physically modifying the prototype.

The comparison of figures 10–12, together with the capacitance measurements of table I, separates two reasons for the improvement. During the decay, the ion current through the grid resistance is what counts. During a sharp anode edge, the capacitively induced current also counts. The two grids of the prototype are defined by the functions the authors assign to them. Their numbering must be read on the drawing before comparing it with G1 and G2 of the 2D21.

Proceedings of the IRE 39, 636–643 · Second scan of the issue

Woodford and Williams (1952). Measuring an interval between two crossings

On a hydrogen 5C22 the authors observe the anode voltage as conduction begins. They replace the first barely visible change on the trace with two levels defined as fractions of the total swing. The resulting relation separates a contribution that depends on the tube and on the supply voltage from another proportional to the L/R constant of the circuit. The fixed-load comparison in figure 5 and the fixed-voltage one in figure 6 make it possible to test the two terms.

The open mounting reproduced on page 7 shows the reason for the coaxial geometry: lowering the series inductance reduces the external term and makes it easier to distinguish the contribution of the discharge. Changing convention from 10–80% to 10–90% changes both coefficients. Two times published without their respective thresholds can therefore differ even for an identical waveform.

Journal of Applied Physics 23, 722–724

Appel and Fünfer (1956). Delay and build-up in triodes and tetrodes

The authors vary the amplitude of the command and distinguish the time preceding the discharge from the subsequent build-up interval. The IRE notice gives orders of magnitude for the delays of about 10⁻⁷ s in the tetrodes and 10⁻⁸ s in the triodes studied, with weak dependence on the initial bias. For the build-up time the dependence on command amplitude is visible in the triodes and much reduced in the tetrodes within the range tested.

This comparison indicates that geometry can change which phase of the transient limits the response. Transferring the orders of magnitude to a particular tube would require the type designations, the voltage limits and the oscillographic criterion of the individual tests. Here the values come from contemporary abstracts of the German paper.

Electronic & Radio Engineer, entry 654 · Parallel IRE abstract, March 1957

Reich (1957). From instantaneous characteristics to the trace in time

Reich assumes an exponential decay of the plasma density and constructs a current–voltage characteristic at the moment of extinction and at four other equally spaced instants. In figure 1 the external network imposes a load line. Intersecting the same line with the five curves gives five grid voltages. Plotting them against time gives figure 2. Even with a regularly decaying plasma, the voltage on the pin can thus show an almost flat initial segment.

The physical point is the closeness of the characteristics on the side of the knee crossed by a high-resistance line. The intersections change little in voltage until the density has fallen far enough. The curves are a theoretical construction that Reich compares with the experiments of Malter and Johnson. This is why a simulation can be checked both in the current–voltage plane and on the time trace derived from it.

IRE Transactions on Electron Devices, p. 287, figs. 1–2

Olmstead and Roth (1957). The direct comparison on the 2D21

The first series of tests establishes how long a pulse on G1 must last for the 2D21 to fire even if the last part of the command is removed. The second establishes how negative the grid must be to prevent a new discharge after a known anode current. The two procedures check different properties. Figure 4 also measures the grid current during the positive excitation and makes it possible to convert the voltage on the pin into the voltage the generator must supply.

Page 9 works through the RCA calculation for 46.5 kΩ in series. The generator must overcome the initial bias and the drop across the resistor as well as supplying the effective +17 V required in that test. In figures 7–9 the role of the resistor changes: it is the residual current that modifies the grid potential during the pause. These measurements are particularly suitable for a comparison between models because they fix tube, sequence and observable and make it possible to check current and voltage together.

RCA Review XVIII, pp. 272–284, figures 1–9

Adati and Sawa (1959). Recovery depends on the experimental question

Hikoichi Adati and Tomio Sawa study the 2D21 for an electromechanical application. The pulse from a ferrite core must fire the tube to drive the clutch electromagnet of a high-speed printer. They take up Olmstead and Roth’s representation, then vary the pulse repetition and the bias present before the command. Even with a pause longer than the normal recovery time, the previous discharge can modify the new firing. In short pulses the initial bias matters particularly.

Figure 4 presents the experimental characteristics. Figure 5 interprets the shift of the curves through an effective distance between G1 and cathode, which represents the initial condition of the electrons. The parameter belongs to the qualitative model: the physical electrodes do not move. The authors then move from rectangular pulses to the narrow ones produced by the core. Figure 6 introduces a circuit to lengthen them and figure 7 checks the behavior with a command that decays in time.

Two recovery criteria follow. The first asks when the tube can withstand the anode voltage again. The second asks when its sensitivity to another pulse has become reproducible again. A model that passes the first test may retain an error in the second. This study provides an independent check on the same commercial type and identifies a concrete limit of descriptions with a single dead time.

計測 9, 590–596, DOI and abstract · Full text, figures 4–7

Hill (1978). From plasma decay to operating time

The thesis deals with what remains of the discharge between successive pulses. In a diffusive model a characteristic length and the ambipolar diffusion coefficient determine the time constant of a mode. With several regions or several modes, a sum of exponentials can describe a fast initial segment and a slow tail. The shape of the tail matters above all when the period between two pulses becomes comparable with the slowest decay.

The quantity that decays in the physical model is the residual density. An RC network inside SPICE produces a voltage instead. To interpret the latter one must say which terminal current or which re-ignition probability changes with it. The correspondence between two exponential forms is a first mathematical check. The tests of Reich, Olmstead–Roth and Adati–Sawa make it possible to verify, separately, the link between internal state and observation.

Texas Tech University, Analysis of High Repetition Rate Thyratrons

#2D21 #characteristic curves #electron tube #Flash-Matic #GE datasheet #General Electric #grid current #SPICE #thyratron #vacuum tubes

Other steps of this project