The 2D21 Thyratron: From Historical Curves to SPICE Models
15. From the probe to the grid. Readings from 1926 to 1962
The grid is at once an electrode that modifies the field and a surface that collects charged particles. From the first work on probes to industrial test procedures, the difficulty lies in distinguishing which of the two functions each experiment observes.
Sources examined on this page
- Mott-Smith and Langmuir (1926). Why a grid collects current
- Hull and Langmuir (1929). How the aperture in the plasma closes
- Nottingham (1931). The photocell as a current source
- Seletzky and Shevki (1933). The mean current can hide two opposite flows
- French (1936). The shape of the characteristic before firing
- Puckle (1942). What a time base with an 885 reveals
- Wittenberg (1962). Two experiments for the same grid
- Brooker and Ware (1954). Measuring a current during a transition
Mott-Smith and Langmuir (1926). Why a grid collects current
The authors calculate the current collected by a cylindrical or spherical electrode immersed in an ionized gas. The bias creates a space-charge sheath. At the entrance to the sheath the particles have distributed velocities and directions. Only some trajectories reach the collector. The calculation links the current to geometry, velocity distribution and potential drop, on the assumption that collisions in the sheath are negligible. The authors also treat populations with drift, in addition to the thermal distribution.
Applied to a grid, the model suggests a precise question. If one wants to deduce an electron temperature from the slope of a semi-logarithmic plot, one has to isolate the electron current and know the potential with respect to the plasma. The data-sheet curve instead gives a net current and uses the cathode as reference. At the point where that current crosses zero, the tube can go on collecting both electrons and ions.
Physical Review 28, 727 · Text of the paper
Hull and Langmuir (1929). How the aperture in the plasma closes
Figure 1 shows a grid with a circular hole between the two sides of a mercury arc in a magnetic field. A grid made negative repels electrons and collects positive ions. The region surrounding the edge of the hole is depleted of electrons. When the sheaths fill the whole aperture, the electron beam can be cut off. The drawing on page 219 distinguishes the mechanical hole from the effective aperture left to the charges.
On page 222 the oscillograms also document the limit of control. Increasing the arc current, the authors meet a transition beyond which the beam continues to pass even when the grid is made strongly negative. The table compares the critical grid voltage and the beam drop at the anode for several values of the grid ion current. The work thus gives an experimental origin to the distinction between the ability to prevent the start and the ability to interrupt an already supplied discharge. The bench uses a geometry of its own, earlier than that of the miniature tetrode.
PNAS 15, 218–225 · Full scan, in particular pp. 219 and 222
Nottingham (1931). The photocell as a current source
Nottingham makes clear that the grid circuit may have to supply power even before firing. For a photocell, which delivers very small currents, even this weak draw changes the result. In the phase-shift circuit he adjusts a capacitance with the photocell in the dark. He then measures the mean current in the anode load while the light varies. To calibrate the abscissa he uses a separate measurement of the photoelectric current as a function of the lamp voltage. Figure 11 thus expresses the illumination through a measured current.
On the FG-17, in one of the settings, about 3.5 nA of photoelectric current is enough to produce intermittent firing, observed ten to fifteen times per second. Increasing the capacitance reduces the tendency to intermittent firing and also reduces the sensitivity. The curve describes the interaction between tube, cell and phase network. It is a precise antecedent of the problem met in the Flash-Matic, albeit with a different sensor, tube and topology.
MIT scan, figures 10–11 and text on pp. 287–288 · Publisher’s reference
Seletzky and Shevki (1933). The mean current can hide two opposite flows
The abstract describes AC oscillographic measurements on a mercury-vapour tube. The authors vary grid and plate voltages, grid resistance and relative phase. Within the same period the grid current can show lobes of opposite sign and even of equal area. An ammeter reading zero on average could therefore hide a substantial exchange of charge in both directions. The direction also depends on the anode current and on the amplitude of the grid voltage.
The authors report reverse anode current when the grid current extends into the negative half-cycle of the plate. They attribute the residual component to incomplete deionization. For the 2D21 the result suggests recording the waveforms on the two electrodes simultaneously and stating the phase. The abstract documents the phenomenon on the mercury tube and leaves the scales of the individual oscillograms to be checked in the text.
Journal of the Franklin Institute 215, 299–326 · Abstract available
French (1936). The shape of the characteristic before firing
French compares the FG-17 triode and the FG-95 and FG-98 tetrodes. What matters for reaching firing is how the grid current grows as it approaches, together with the constants of the circuit that feeds it. The abstract reports conditions at reduced temperature and anode voltage in which unusually high currents are needed. In the argon FG-98 the thermal dependence is negligible in the range examined, but an anomaly reappears with a positive shield at low plate voltage.
A particularly instructive result concerns the minimum control current. French associates with the negative-impedance characteristic the need, in the tubes examined, for grid resistances between 100 and 1000 MΩ to reach that minimum. The value describes an experimental objective for those types. To compare it with the Sylvania bands one must distinguish minimum firing current, generator voltage and actual voltage on the electrode. The exact form of the analytical derivation requires the full paper.
Journal of the Franklin Institute 221, 83–102 · Publisher’s abstract
Puckle (1942). What a time base with an 885 reveals
In the sweep generator the observable quantity is the voltage on the capacitor. Charging produces the slow segment. When the operating point reaches the firing characteristic, the gas triode provides the discharge path. Figure 3 gives static characteristics of the 885 and makes it possible to distinguish the point at which the discharge begins from the subsequent conduction branch. Repeating the sweep then requires an extinction condition, obtained by the circuit when the current becomes insufficient or by an external quenching command.
The significance for a model emerges by comparing Puckle with Zabb’s oscillator. Two models can generate the same ramp if they reproduce the threshold and the charging of the capacitor. The fast return and the possibility of starting an immediately following cycle, on the other hand, put the conduction drop and recovery to the test.
Electronic Engineering, March 1942, pp. 664–666
Wittenberg (1962). Two experiments for the same grid
Wittenberg uses a 2050 to discuss the grid before the discharge and a 5563A to illustrate its collection in the plasma. In figures 36–38 for the first regime, contributions with different signs appear. At steady voltages leakage, emission and charge collection can be discussed. If the anode voltage varies, capacitive displacement also comes in. During conduction, on the other hand, a sufficiently negative grid repels electrons and collects mainly ions.
The sheath equation contains the ion mass, the sheath thickness and the drop between plasma and electrode. Changing the discharge current also changes the conditions of the plasma that feeds the sheath. This is why the same local law can go together with several saturation levels in the conduction families. Its most informative application to the GE plates is to recognize the physical region in which to look for a dependence, before estimating its parameters from the digitized points.
RCA, Electron Tube Design, chapter Gas Tube Design
Brooker and Ware (1954). Measuring a current during a transition
What does a microammeter really read if inserting it requires changing the tube connections? The authors show why the answer can change already during the reading. Switching the circuit modifies the distribution of potentials and lets the electrodes cool. The emission one wanted to measure can therefore also change. The industrial circuit studied by the authors seeks to reproduce the conditions immediately before firing without imposing on the bench the full power of normal service.
In the two-resistance method the same limiting condition is found twice and the generator voltages are compared. The voltage difference divided by the resistance difference gives an estimate of the critical current. The verifiable assumption is that the effective threshold at the pin remained the same during the two tests. For a historical review this is a rare case in which the whole chain can be followed, from the error of the simplest method to the quantity eliminated by taking the difference.