16. What the early studies of discharge times measured

Two authors can both speak of ionization time and start the stopwatch at different instants. To reread the work of the 1940s one has to find the tube type, the shape of the command and the events chosen on the oscillograms.

Sources examined on this page

Harrison (1940). Overvoltage and the spread between specimens

Using a cold-cathode oscillograph, Harrison measures 17 specimens of 12 types, among thyratrons and controlled glow-discharge tubes. To compare them he defines overvoltage as the difference between the applied command and the critical potential. The same generator voltage can therefore represent a different excitation on two specimens. He finds empirically an ionization time inversely proportional to the overvoltage and reports times spread over several orders of magnitude.

The tests include argon, neon and mercury-vapour tubes. For the mercury tubes, in the small-overvoltage regime, the abstract reports no appreciable effect of the mercury temperature on the measured time. Sometimes two specimens of the same type differ more than specimens of different types. To reconstruct a time law for the 2D21 one therefore needs both a threshold measured for the tube and an indication of the spread between specimens.

Caltech, abstract of the AIEE publication · Manuscript held at Caltech

Chance (1946). The case of the G.T.1C

The 1946 abstract gives a numerical result for the G.T.1C. Between 5 and 25 mA the ionization time is about 1.4 ± 0.4 µs. In the range explored, the authors of the abstract report little dependence on anode current and voltage and on the rate of change of the grid potential. This last variable is a time slope. The overvoltage studied by Harrison is instead an amplitude referred to the threshold. The two results can therefore coexist.

For deionization Chance considers the negative bias for several anode voltages and currents. At fixed current and bias, the abstract reports that varying the amplitude of the anode–cathode step used to extinguish the discharge does not change the observed time. The definition of the 0.4 µs uncertainty and the limits of the other parameters have yet to be checked in the original.

Journal of Scientific Instruments 23, 50–52 · Contemporary IRE abstract, p. 620

Mullin (1946). A physical feedback at the start of the discharge

The model starts from the electrons that pass when the grid is taken beyond the threshold. Collisions produce positive ions, some of which return towards the cathode and modify the space-charge distribution. It then becomes possible to extract more electron current. The mechanism reinforces itself up to the emission limit of the cathode in the model. The delay observed at the plate includes the time needed for this multiplication to become appreciable.

Mullin reports agreement between the calculated current-versus-time curves and experiment, as well as the possibility of estimating the ionization time. For the historical comparison this documents a time model of the growth, more specific than an instantaneous threshold. The abstract is not enough to reconstruct the initial conditions or to transcribe the equation. The semi-empirical relation published six years later by Woodford and Williams must remain attributed to them.

Physical Review 70, 401–405 · APS, author’s abstract

Romanowitz (1947) and Romanowitz–Dow (1950). What counts as recovery

Romanowitz’s bulletin of December 1947 is cited by Reich as a study of the measurements and of the statistical nature of the deionization time of a mercury-vapour tube. This is the historical point the citation makes it possible to document. The identification of the KU-627, the dimensioned geometry and the detailed sequence of tests come from the later publication signed with Dow. Retrospectively attributing all the 1950 data to the bulletin would require comparison with the library copy.

In 1950, having fixed the initial conditions, the authors change the time between extinction and reapplication of the voltage and observe the outcome of repeated tests. The intermediate frequencies give information on the width of the transition from frequent to rare re-ignition. The limit of fifty consecutive tests without re-ignition is an operating criterion tied to the number of attempts. Even if the true probability were small but positive, that result would be possible. It is this difference that disappears when a catalog or a simulator gives a single time.

AIEE paper of 1950, pp. 368–379 · Reich, bibliographic reference to the 1947 Kentucky bulletin

Glasoe and Lebacqz (1948). The switch in the pulse-forming network

The volume Pulse Generators devotes §§ 8-11, 8-12 and 8-13 to the general characteristics of the hydrogen thyratron, to the anode circuit and to the grid circuit. The previous chapter deals with the discharge of the line and compares calculated and actual pulses. The following one deals with recharging. This sequence makes it possible to reconstruct a complete cycle. The command starts conduction, the network and the load determine the transferred pulse, and the subsequent recharge requires the switch to be able to hold off voltage again.

The book explains why, in the modulator literature, some authors measure the signal on the load and others directly on the anode. The overall delay includes the drive and the propagation in the circuit, while the anode–cathode voltage documents the transition of the tube more directly. The comparison with the 2D21 must be made by first choosing the pair of terminals observed and the criterion used to start and stop the measurement.

MIT Radiation Laboratory Series, vol. 5, chs. 7–9

RCA AN-131 (1948). The capacitor decides the time, the tube decides the switching

The note starts from a two-range timer. The capacitor charges slowly through a high resistance and shifts the grid bias. The AC supply offers an opportunity for extinction at every half-cycle. The condition that decides the state of the relay is therefore the one the circuit meets when the anode returns positive. The charging path explains the seconds of delay, while the firing of the tube takes place on a far shorter scale.

In the sequential circuits the firing of the next tube changes the anode voltage of the previous one through a commutating capacitor. Figures 2 and 3 make it possible to follow the order of events moving from two to three stages. The attached errata correct the potentiometer designations. This source documents a concrete use of the 2D21 and 2050 and offers a historical check for anyone wishing to reconstruct the operation from the schematic.

RCA, Electronic Timers Employing Thyratrons 2D21 or 2050, figures 1–3

Wittenberg (1949). A pulse comparison that includes the 2D21

To compare the tubes Wittenberg fixes common conditions. In the test described on p. 119 he chooses a bias of −50 V and a grid resistance of 20 kΩ. He makes the trade-off explicit: reducing the resistance can shorten recovery, but increases the load on the command generator. A bias sufficient to keep a tube off at rest may prove too weak to re-establish control quickly after a pulse.

Table II identifies the 884, the 2D21 and a pair of 2050s separately. This detail also matters in reading the currents: the entry for the two 2050s concerns two specimens. The work relates the pulse response to the dimensions and test conditions of the tubes. For the 2D21 it offers a direct experimental precedent, but the peak values of the bench belong to the pulse service described. The mean current of the solenoid in the Flash-Matic requires other data.

RCA Review, March 1949, table II and figure 2

Knight (1949). Grid construction, arc drop and service life

The contemporary technical review presents Knight’s study as an experimental survey that follows cathode emission, the build-up of the discharge, current capacity and decay after conduction. It notes the comparison between grid structures, in particular the pentode that makes it possible to control high currents from high-impedance sources. The arc drop depends appreciably on the filling pressure and on the current. Ion bombardment can also affect service life before producing an evident change in the electrical characteristic.

These results explain why a typical anode drop, measured for a new tube under a given condition, leaves open both the slope of the characteristic and its evolution. Pages 361–378 belong to the paper and pp. 379–381 to the discussion. The results summarized here are those reported by the abstract service.

Wireless Engineer, entry 504 on the IEE paper and the discussion

Verzuh (1949). Separating the waiting time from the test edge

In the conventional method with a discharge capacitor, changing one parameter can also change the rate at which the voltage returns on the anode. Verzuh uses a second, delayable command and a calibrated time base. He decides when to reapply the voltage, keeps the edge under control and reads on the oscillograph the shortest interval that avoids persistent conduction. In this way two tests can differ by a single chosen parameter.

Among the conditions that the report associates with a shorter recovery are a smaller previous anode current, low impedance of the grid network and a stronger blocking bias. Pressure and plasma geometry also count. Together with Romanowitz and Dow this source explains two complementary aspects of the metrology. The delay set by the bench must be well defined and, once defined, the outcome can still vary from one test to the next.

MIT Research Laboratory of Electronics, Technical Report 101, fig. 1 and conclusions

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

Other steps of this project