The 2D21 Thyratron: From Historical Curves to SPICE Models
5. Measuring grid current and the control ratio
Before firing: steady-state current and displacement current
Wittenberg uses the 2050 to show how the grid current changes before the discharge as the anode voltage is varied. With a DC anode voltage, in steady state, the capacitive contribution due to the change in that voltage disappears. In the almost horizontal segments the author points to leakage and grid emission as the likely contributions. Approaching breakdown, ion collection emerges. At low anode voltage the electron contribution also becomes visible. This helps to read the manufacturer’s curve as a measured net current, that is, as a sum of contributions that can have different signs and dependencies. Wittenberg, figures 36–38.
In a pulse circuit a capacitance between anode and G1 introduces a current proportional to the derivative of the voltage between the two electrodes. Using the GE value of 0.026 pF, an illustrative edge of 100 V/µs produces about 2.6 µA of displacement current, before considering the rest of the network. It is an order-of-magnitude calculation, useful because the Sylvania data sheet gives steady-state currents of hundredths of a µA before the discharge. The two quantities belong to different conditions and can be compared only by making explicit the edge rate and the actual capacitances of the circuit. GE, interelectrode capacitances · Sylvania, curve before conduction.
A precedent in industrial metrology
Brooker and Ware devote an article to measuring the current immediately before firing. To make the testing of industrial tubes practicable, they present a circuit that recreates the grid condition while limiting the power required at the test bench. The measurement also serves to assess the quality of the tube and the sensitivity of the threshold to the external network. The work shows why this current must be treated as a measurable quantity and why an ideal threshold alone poorly describes a grid driven from a high impedance. Brooker and Ware, Philips Technical Review, August 1954, pp. 43 ff..
Measuring a critical current from two voltages
Brooker and Ware distinguish leakage between electrodes, ions produced by electrons escaping control, thermionic emission from the grid and a possible photoemissive contribution. The switch method brings the tube to operating temperature, then changes the configuration to read the microammeter. In the time needed for the reading the electrodes cool. The voltages that determine the leakage also change, and part of the ionization current is lost. One can therefore obtain a stable reading in a circuit that has meanwhile ceased to represent the desired condition.
The cut-off method instead compares two resistances in series with the same grid. The source voltages are read when the tube just fails to re-fire. Denoting the two tests by A and B, the relation is \[ I_{g,\mathrm{crit}}=\frac{V_{s,B}-V_{s,A}}{R_B-R_A}. \] Subtracting the two equations eliminates the effective critical voltage on the electrode, assumed to be the same. The subscripts A and B identify two readings on G1; G2 does not appear in the difference. The thermal condition remains essential: two waveforms with the same mean anode current can have different peaks and heating. Brooker and Ware, pp. 43–48, eq. (1).
From the firing curve to two local sensitivities
Writing a firing condition locally as \(F(V_a,V_{g1},V_{g2})=0\), where \(F_{g1}\neq 0\), the sensitivities are \(\partial V_{g1,\mathrm{crit}}/\partial V_a=-F_a/F_{g1}\) and \(\partial V_{g1,\mathrm{crit}}/\partial V_{g2}=-F_{g2}/F_{g1}\). Here the subscripts of \(F\) denote partial derivatives. This is a local mathematical reading of the historical families. The slopes must be determined in the same neighborhood, with the same axis convention and the same drive conditions. The supply voltages of the grid network and those directly at the pins become comparable after modeling or measuring the drop across the resistor.
To choose the segment from which to estimate a sensitivity, the GE plates can be set beside pages 2 and 4 of Sylvania. The first Sylvania plate varies the shield in DC; the second compares two grid resistances with the shield at zero and the anode on AC. To verify a law that treats G2 and the grid resistance as independent variables, one needs a set of measurements that varies both, or a model of the measurement circuit with a quantified error. GE, control plates · Sylvania, variable G2 · Sylvania, two resistances.
An example with the control ratios of the two grids
A 2D21 data sheet held by the University of Wisconsin gives approximate ratios of 250 for G1 and 1000 for G2. Interpreting them in the same linear neighborhood and with the same convention, a change of 1 V on G2 would correspond in magnitude, for its effect on the threshold alone, to about 4 V on G1. The quotient compares two nominal sensitivities. Philips specifies for each test the bias of the other grid. An effective compensation between G1 and G2 would require the two slopes measured at the same operating point. To follow the threshold over a wide range of \(V_a\) and \(V_{g2}\), the experimental family and its measurement conditions are needed. 2D21 data sheet, general data, Wisconsin teaching archive.