Исследование коммутирующих устройств на основе искровых промежутков с предельно высокой частотой коммутации и возможностей их применения тема автореферата и диссертации по физике, 01.04.13 ВАК РФ

/ /■ Л О ' / '1 I п ,J ГУ

и о / / (/ У

Санкт-Петербургский Государственный политехнический

университет

Боль Юрген

Исследование коммутирующих устройств на основе искровых промежутков с предельно высокой частотой коммутации и возможностей их применения

Специальность 01.04.13 - Электрофизика и электрофизические установки Диссертация

на соискание ученой степени кандидата технических наук

Научный руководитель: доктор технических наук, профессор В.В.Титков

УДК 621.374(076.5)

Санкт-Петербург 2003

SAINT-PETERSBURG STATE POLYTECHNIC UNIVERSITY

Electromechanical Faculty Electro-Physics and High-Voltage-Techniques Special Subject 05.14.12, High-Voltage-Techniques Assistance: Prof. Dr. V. Titkov

Investigations of Closing Switches based on Spark

Gaps with Extremely High Repetition Rates and Potential Electromagnetic Compatibility Applications

Juergen Rolf Bohl

List of Contents

1 Introduction 8

1.1 Status of Technology 9

1.2 Phenomenological Observations and Delimitation 12 of the Work

1.3 New Aspects of the Work 13

1.4 Objectives 13

2 Literature 14

2.1 Pulsed Power Technology 14

2.1.1 General Applications 14

2.1.2 Spark Gap Operation 15

2.1.3 General Static Spark Gap Analysis 17

2.2 High Voltage Switches for Pulsed Power Applications 20

2.2.1 Vacuum and Gas Filled Switches 20

2.2.1.1 Spark Gap 20

2.2.1.2 Cold Cathode Switching Tube 22

2.2.1.3 Krytron 23

2.2.1.4 Sprytron 24

2.2.1.5 Thyratron 25

2.2.1.6 Ignitron 26

2.2.1.7 Vacuum Discharge 26

2.2.2 Solid State Devices 27

2.2.2.1 Thyristor 27

2.2.2.2 GTO-Thyristor 28

2.2.2.3 MOSFET 29

2.2.2.4 Insulated Gate Bipolar Transistor (IGBET) 29

2.2.2.5 Photo-Conductive Switches 31

2.2.2.6 Advantages and Disadvantages of 31 Semiconductor Switches

2.3 High Power Generators - Transmission Line Generators 32

2.3.1 Blumiein Generator 32

2.3.2 Self-Matched Transmission Line Generators 33

2.3.3 Pulse Forming Network (PFN) Marx Generators 34

2.4 Spark Gap Applications in Various Technologies 35

2.4.1 Transmitter Technology 35

2.4.2 Radar Technology 37

2.5 Conclusion 40

3 Electrical Model of the Fast Switching Device 41

3.1 Time Dependent Resistance of the Spark Gap 41

3.2 Modelling with Constant Current Source 49

3.3 Modelling with Constant Voltage Source and 54 Charging Resistor

3.4 Cascading of several Fast Switching Devices 57

3.5 Discussion of Numerical Simulation Results 61

4 Experimental Investigations 65

4.1 Test Set for Electrical Investigations 65

4.2 Macroscopic Experimental Investigations 71

4.2.1 Objectives 71

4.2.2 Test Matrix 71

4.2.3 Characteristic Macroscopic Electrical Test Results 73

4.2.4 Test Parameter Dependencies 78

4.2.5 Electron Emission Mechanisms 81

4.2.6 Electro-Technical Model of PRR with two Modes 84

4.2.7 Physical Interpretation of the two PRR Modes 88

4.2.8 Conclusion 90

4.3 Microscopic Experimental Investigations 92

4.3.1 Objectives 92

4.3.2 Test Set for Optical Investigations 92

4.3.3 Test Matrix 95

4.3.4 Characteristic Microscopic Electrical Test Results 96

4.3.5 Characteristic Microscopic Optical Test Results 105

4.3.6 Test Parameter Dependencies 107

4.3.7 Conclusion 108

4.4 Conclusions of Experimental Test Results 109

5 Comparison between Modelling and Experimental 111 Investigation Results

6 Summary 119

7 Outlook 124

7.1 Further Investigations 124

7.2 Theoretical and Practical Applications 125

8 References 128 Annex 133

Annex

A-1:

Structure and description of the single stage of the FSD as differential equation for Matlab/Simulink derived from Kirchhoff equations with the time dependent resistor R(t) and constant current source

A-2:

Structure and description of the single stage of the FSD as differential equation for Matlab/Simulink derived from Kirchhoff equations with the time dependent resistor R(t) and constant voltage source in combination with the charging resistor R1

B-1 to B-7: 3D Test Result Charts of the Microscopic Electrical Investigations with the Fast Switching Device (FSD)

Abbreviations

A Cross Section

a Acceleration

a First Townsend Coefficient (Ionisation Coefficient)

B Induced Magnetic Field

C Capacitance

CCMP Charge Carrier Multiplication Process

d Spark Gap Distance

DC Direct Current

E Electrical Field

e Electron Charge [1.6021E-19 C]

ECM Electronic Counter Measure

EMC Electro Magnetic Compatibility

r| Efficiency Coefficient;

Second Townsend Coefficient (Emission Coefficient)

FSD Fast Switching Device

cp Magnetic Flux

GW Giga Watt

HPM High Power Microwave

I Current

ign Ignition

K Kelvin

k Boltzmann Konstante [1.38E-23 J/K]

L Inductivity

I Length

m Mass

me Mass of an Electron [9.109E-31 Kg]

mn Mass of an Neutron [1.6748E-27 Kg]

nip Mass of an Proton [1.6725E-27 Kg]

MW Mega Watt

\i Magnetic Permeability [|i0=4TiE-7 VsA"1m"1];

Electron Mobility

N Amount of Charge Carrier

v Velocity of Charge Carriers

p Pressure

P Power

PFN Pulse Forming Network

PMT Photo Multiplier Tube

PRR Pulse Repetition Rate

R Resistor

r Recombination Coefficient; Distance

RFI Radio Frequency Interference

S Switch

s Free Travelling Path of an Electron

sat Saturation

a Conductivity

T Temperature

t, I, tau Time

U Voltage

U_Switch Break Down Voltage

UWB Ultra Wide Band

V Volume

v, v Velocity

X, Y, Z Direction, Way

1 Introduction

Emission and generation of electromagnetic fields are in the focus of interest since Marconi first proved the emission of electromagnetic fields according to the theoretical work of Hertz [1]. From this origin the intention of transmitting and receiving terrestrial information rose dramatically, establishing a huge economical important market of wireless communication. Nowadays, living standard is dominated by cellular phones, commercial radio and TV, wireless observation, telemetry systems, GPS, and satellite connection.

The EMC aspects are in the meantime very well defined. All kind of electronic devices have to be qualified with respect to electromagnetic emissions. The EMC test houses and also the companies which are developing and producing electronic devices are equipped with "standard" EMC test equipment.

Nevertheless, the common electromagnetic smog and the latest developed and fielded High Power Microwave (HPM) narrow band radiating sources on the basis of accelerated electrons (electron beam devices) as well as the upcoming Ultra Wide Band (UWB) devises leads to an extent of further test capabilities which have to cover the additional electromagnetic pulsed frequency spectrum (between 100MHz and 3GHz), power density, pulse shape, and the pulse repetition rate.

The focus of this work is on the investigation of spark gaps with respect of an operation mode not for the classical high current (several tenth of kA) and high voltages (several tenth of kV) where the switching rate is limited but on the possible (limitation) achievable switching rates at low current (mA) and low voltage (<2 kV). If is possible to realise stable and robust switching operations at reasonable very high switching rates (several tenth of kHz), there would be the change to generate at least electromagnetic test systems operating in the ultra wide band range depending on the

physical achievable switching rate and the achievable switching velocity of a spark gap configuration itself. Such spark gap configurations could lead to a low cost ultra wide band test source in comparison to the conventional, more or less off the shelf available ultra wide band test sources which are designed and constructed on the basis of semiconductor switches. Such ultra wide band test sources are necessary to verify the functionality of communication devices in a rough electromagnetic environment. These rough electromagnetic environments will be specified in the near future. Further future applications could also be seen in the data communication scene and others.

1.1 Status of Technology

In any situation where communication (wireless as well as wired one) is used the systems have to work properly also "undefined, electromagnetic rough" environment where suppression of the own communication capability has to be ensured. Since world war two an intense afford to jam any kind of wireless communication and sensor signals is taken and known as jamming [2]. Starting with CW-jammers more and more complicated and efficient robust modulation schemes were introduced to face this threat [3]. Today a wide range of counter measure and counter-counter-measure technology is available not only in the communication field but also in jamming radar and if-sensors in intelligent systems. With further development of radar system technology and rising power levels electron beam devices like magnetrons, relirons or cyclotrons gained so much power that their radiated pulsed power up to the gigawatt region caused electronic failure to exposed electronics [4]. Therefore, these sources were optimised to kill electronics in a far distance. According to their pulsed mono-frequent waveform and their peak power in the

megawatt region these sources are covered under the name of HPM (high power microwave) [5].

The last ten years very short pulses with a rise time of some hundred picoseconds and peak power above 1 GW were detected as effective against some class of electronic devices [6]. These sources are summarised under the name of UWB (ultra wide band).

The kind of electronic device and field coupling, the source power, pulse repetition rate, pulse width, the applied frequency and the kind of intended effect (electronic burn out, reset, distortion) define the actual working distance [7]. Usually the main difference of the type of coupling is front door, or back door coupling. Interference of a nominal sensor or communication signal is intended in the front door coupling case. Here, the interference frequency is equal to the nominal signal frequency and both couple through the nominal antenna to the system. The interference signal has to be a modulated continuous wave (CW) or with a high repetition rate pulsed signal. The actual field level of this interference signals is quite low due to the fact , that the automated gain control of the input stage will be brought into saturation where no longer a discrimination between working signal and interference signal can be realised [8].

In the back door coupling case the external field couples through holes and parasitic antennas inside the housing of an electronic system and travels along wires to the electronic components. The frequencies are therefore tuned to structure resonances. In the disruption case the high frequency components (typically in the lower gigahertz region) are rectified to DC and pulses or demodulated to their low modulation frequency due to the non-linearity of the semiconductors. This low frequency distortion superimposes

to the nominal signal. When UWB or HPM-pulses are applied, the rectified pulses may cause temporary digital circuit failure, latch-up or burn out in the semiconductor [9].

Certain classes of electronic systems are only susceptible to high repetition frequencies above 1 MHz [10]. In the front door case the fast recovery time of the automatic gain control circuits are responsible. In the back door case the hit probability of the susceptible fast clock transition time slot rises with higher repetition rate.

HPM sources rely on the interaction between an electron-beam and the electromagnetic field where energy is transformed from the beam to the field. Their power supply consists of a switched pulse-forming-line [4]. The switch itself consists of a spark-gap which allows a PRR of usually only some hundred Hertz. The same disadvantage show spark-gap switched UWB sources [11]. Semiconductor switches are not usable for HPM sources because of the huge energy (kilojoules) and high voltages (some hundred kilovolt) to be switched. In the UWB source the energy is much smaller (joules) however the semiconductor switches have to be stacked to arrays, which makes such switches extremely expensive. Also reliability aspects of these switches are not solved [12].

Typical jammer systems are based on semiconductor switches [2]. All the power and energy is transferred in a small bandwidth or carrier frequency. For an effective operation, first the target has to be identified. Then the main beam of the high gain antenna is directed towards the target. The frequency is either adapted to the reconnaissanced target frequency or is swept around an estimated frequency range. This devices are expensive and have to be multiplied if a wide frequency spectrum has to be covered.

In many "test" applications a low cost, easy to handle, high repetition pulse test source is desirable. This work investigates some concepts capable to fulfil such a potential requirement.

Therefore, further investigations with respect to the potential application of spark gaps as "fast" closing switches is appropriate. Especially the robustness and the costs of spark gaps usage compared to semiconductor switches is an important factor.

> Where are the physical limits of the pulse repetition rate?

> What does it mean for potential applications compared to conventional semiconductor switch applications?

1.2 Phenomenological Observations and Delimitation of the Work

Common switches based on spark gap technology show fast switching times in the nanosecond or even in the sub-nanosecond regime. But the recovery time of those switches is located in the millisecond regime. This is the reason why typical repetition rates of only a few hundred Hertz are possible.

Especially for quasi unsynchronised oscillating applications where the electromagnetic compatibility is critical and semiconductor applications are not possible, very fast pulse repetition rates achieved with spark gaps could be useful.

In the literature the stabilisation of very high repetition rates of spark gaps could not be found. Due to the switching time in the microsecond regime the electrical characteristics of the designed and parasitic inductances and capacities are of special interest. It is observed that these elements also influence the fast switching off behaviour of the spark gap arrangement. The coupling of electrical components and physical charge and discharge effects will be considered.

1.3 New Aspects of the Work

New aspects of the investigation of spark gap switches on the basis of the observed behaviour at low switching current (mA) and voltage (<2kV) are:

> Recovery time of some microseconds

> Free oscillating (unsynchronised) spark gap configuration with several hundreds of kilohertz switching events

> Compared to "Trichel Applications" the spark gaps are not voltage driven but current driven

> Stabilisation and synchronisation of the repetition rates due to external electrical circuits

1.4 Objectives

The following major objectives are considered within the investigations of the spark gap configuration operating at lower power and energy but with higher switching rates per time:

> Literature research

> Electrical modelling of a simplified spark gap configuration and determination of the physical limitations and parameter dependencies to achieve high switching rates per time

> Experimental investigations to consider the macroscopic and microscopic switching behaviour of the spark gap configuration and describing of the physical characteristics of the switching behaviour

> Comparison of the experimental test results with modelling

Such high repetitive spark gap closing switches based on free oscillating (not-triggered system) can be used for a wide range of applications -especially where semiconductor systems are too expensive, or where the electromagnetic compatibility is critical, or where electromagnetic fields would destroy semiconductor systems.

2 Literature

2.1 Pulsed Power Technology

The objective of pulsed power is to study the physical and technical basics of generation and application of pulses of high electrical power and pulse energy. The basic element in pulsed power is an energy storage system which is charged slowly to a certain voltage and discharged rapidly using a fast high voltage switch. The electrical energy is stored either by a capacitor for electric fields, or by an inductance for magnetic fields. Generators with capacitive energy storage require closing switches in order to discharge the energy from the capacitor and convert it into a powerful pulse. On the other hand inductive storage requires opening switches to connect the energy to the load. Design and development of high power opening switches is relatively complicated and time consuming. Therefore, most of the pulsed power generators are based on capacitive energy storage systems [13].

2.1.1 General Application

Pulsed power is used for many applications. In the following some typical areas of applications are described [14]:

> Surface treatment:

- drying of paint by electron beams

- paint removal from air planes

> Treatment of food and environment:

- food sterilisation

- treatment of sludge

- water purification

- air sterilisation

> Material treatment:

- fabrication of micro powder

- destruction and crushing of concrete and stones

- magnetic confinement

> Medical application:

- removal of kidney stones

- medical lasers

> Electromagnetic acceleration

- electric gun

- active protection

> Generation of electromagnetic waves for

- disruption and destruction of electronic systems/devices

- suppression of wireless communication

Independent on the specific application typically short current pulses are required with high voltage amplitudes. Dependent on the application repetitions rates vary in a large range. The high current / high voltage application limit the maximal achievable switching rate to about 100 Hz to 1000 Hz depending on the specific system configuration.

2.1.2 Spark Gap Operation

The static spark gap is the simplest type of spark gap as shown in Figure 2.1.2-1. Perhaps the name "Overvoltage Spark Gap" better describes how it operates. The term "static" arises to differentiate it from more sophisticated spark gaps which employ some moving parts like the "Rotary Spark Gaps". Other more advanced spark gap systems are triggered ones to improve the timing or "Quenched Spark Gaps" [15]. The electrodes of the static spark gap

are fixed. In its simplest form the static spark gap consists of two smooth electrodes separated by a few millimetres of air. Firstly, we will consider how a simple static spark gap works.

Figure 2.1.2-1: Static spark gap Operation

When the voltage across the spark gap electrodes becomes sufficiently high, the air in the gap ionises, and an Avalanche effect takes place [40]. The air inside the gap is heated to a very high temperature and becomes a good conductor of electricity. Heavy current flow through the ionised air keeps it heated and maintains the conductive channel. Eventually the current flow will fall sufficiently that the air in the spark gap cools and stops conducting. The avalanche process by which the spark gap starts to conduct is known as Firing or Breakdown and the voltage required to initiate this process is called the Breakdown Voltage. (The breakdown voltage is in the normal case proportional to the distance between the electrodes and also other parameter like pressure, gas, surface, and material). During conduction the arc between the electrodes is not a perfect conductor and posses a little resistance to the flow of current through it. Power dissipated due to the arc resistance is known as conduction loss and it is the power lost here as heat that keeps the arc hot and conductive. The current necessary to keep the arc conductive is often called the holding current. When the arc cools and stops conducting it is known as quenching.

When considering how a static spark gap operates, the important facts are that it is "turned-on" by sufficiently high voltage, remains on due to primary current flow, and "turns-off" when the current falls too low. Unlike a rotary spark gap, the electrodes of a static gap are always aligned. Since the static gap is voltage triggered it will fire at any time when the voltage goes above its breakdown voltage. This property is good for two reasons:

1. It prevents the primary voltage from getting too high. The static spark gap effectively limits the primary voltage by firing whenever its breakdown voltage is exceeded.

2. If the static gap is set correctly, little energy is wasted since the spark gap only fires when the tank capacitor is fully charged. Firing when the capacitor is less than fully charged would be inefficient as energy is proportional to the capacitor voltage squared (E=0.5*CU2).

One consequence of the constant firing voltage is that it gives rise to rather irregular timing of the spark gap firings.

2.1.3 General Static Spark Gap Analysis

In the following the major dependencies of static spark gaps are described [16].

a) Effect of spark gap spacing on firing events and stored charging capacitor energy

A useful feature of the static gap is that the total gap distance (or the number of small gaps in series,) can be adjusted to change the breakdown voltage. Figure 2.1.3-1 shows in the left diagram the effect of electrode spacing on the average firing rate (switchings per second) of the spark gap. As the static spark gap is made wider its breakdown voltage increases and the charging capacitor has to be charged to a higher voltage before the spark gap can fire.

It takes proportionately longer for the capacitor to be charged to the higher breakdown voltages and therefore the firing rate decreases. The firing rate (pulse repetition rate) is approximately inversely proportional to the breakdown voltage of the spark gap. The right diagram shows how the energy in the 64nF charging capacitor increases as the spark gap distance is made wider. The energy stored in the charging capacitor is proportional to the break down voltage squared, and therefore quadruples every time the break down voltage of the spark gap is doubled.

35 ----

E = O.S*C*U„Swltoh 2

E ■= stored energy JJj

U_Switch := break down voltage C charging capacitor [F)

Break down voltage [kV)

Figure 2.1.3-1: Effect of spark gap spacing on firing rate and capacitor energy (PSpice simulation with 10kV/200mA and 64nF)

b) Effect of spark gap spacing on power throughput

Average power throughput is equal to the firing rate multiplied by the stored capacitor energy (P=EC*PRR) which is released at each firing. Since the firing is proportional "1 / U_Switch" and the storage energy at the charging capacitor is proportional "U_Switch2", then it can be shown that the power throughput is roughly proportional to the break down voltage of the spark gap which is shown in Figure 2.1.3-2 (left diagram). This means that both the break down voltage and the power throughput are directly proportional to the spark gap distance. The deviation from a straight line occur due to the chaotic firing nature of the static spark gap. The right diagram of Figure 2.1.3-2 shows the effect of different charging capacitors on the power throughput.

in the case of the 128nF capacitor (dark blue line), the supply is too heavily loaded by the large capacitor and is unable to charge it to a voltage more than 17kV. If the spark gap is set to a break down voltage in excess of 17kV it does not fire. It should be realised that if the spark gap is set to fire at a voltage in excess of the peak capacitor voltage, no power will be processed into sparks, but considerable supply cord current can be drawn in order to charge the capacitor and can highly stress the set-up transformer if it is allowed to persist.

In the case of the 12nF capacitor (light blue line), the charging capacitor voltage does not grow high enough to fire the spark gap when the gap is set to a break down voltage above 24kV.

Note: The peak voltage and the power throughput can be controlled by adjusting the spark gap distance. The firing rate of the system will adjust itself automatically [50] to process the available power, if the capacitor is set not too small and not too big (within 20nF and 100nF).

P = PRR* E - (n 1 U Switch) * C * U_SwHctl! ■ 0.5*n*C"U_Switch

1 1 / —'

E := stored energy [J[ USwitch creak down voltage C := charging capacitor [F] n ™ arbitrary constant PRR := switching events JHz)

y

0 5 10 15 20 25 30

|~Brcak down voltage [KVI |

2000 1600 1200 800 400

—£=128nF C=90nF C-WflF C-aSnF C=32nF C=12riF

s

I

10

15

20

25

I Break down voltage :-<VJ

Figure 2.1.3-2: Effect of spark gap spacing on power throughput

(PSpice simulation with 10kV/200mA and 64nF and variations)

Remark: These applications consider primarily power transition and not necessarily firing rates (PRR) much greater than 1000Hz.

2.2 High Voltage Switches for Pulsed Power Applications

The most critical element for any application is the high voltage switch. Dependent on various applications, the switch needs to fulfil specific requirements [17]:

high voltage amplitude high currents fast rise time fast recovery time short pulse duration In the following a short overview on various high power switches is given, which are commonly used for pulsed power devices.

2.2.1 Vacuum and Gas Filled Switches 2.2.1.1 Spark Gap

A spark gap usually consists of two electrodes with a hemisphere geometry. One of the electrodes (usually the grounded electrode) may be equipped with a trigger electrode. If a negative high voltage pulse (e.g. -10kV) is applied at the trigger electrode, a spark is ignited towards the cathode. The spark ionises the main gap between the two hemisphere electrodes and ignites the main discharges between cathode and anode (e.g. > 50 kV). With ignition of the main discharge, the capacitive energy storage is discharged over the spark gap into the load. One characteristic of spark gaps are very short formative time lag times on the order of several nanoseconds (e.g. 10ns). A disadvantage of the spark gap is the reasonable electrode erosion with longer duration high current pulses.

Over Voltage Spark Gap

The over voltage spark gap consists of two electrodes with a gap between. When the voltage across the gap exceeds the breakdown voltage of the gas, a spark is ignited and a current is very rapidly established. The voltage at which breakdown occurs is given by the "dynamic breakdown voltage", which is the voltage at which the device will breakdown for a fast rising impulse voltage. This voltage may be as much as 1.5 times greater than the static breakdown voltage ("Paschen breakdown"). This depends almost entirely on the rapid voltage rise. Commutation times for these devices are exceptionally low (sometimes less than one nanosecond). Over voltage gaps are primarily used for protection. But in combination with the other devices mentioned here they are commonly used to sharpen the output pulses (decrease the rise times) of very high current pulses form triggered switching devices e.g. Thyratrons. The size of these devices is almost entirely dependent upon how much current/voltage they are intended to switch.

Triggered Spark Gap

The triggered spark gap is very similar to the over-voltage spark gap. It employs an additional electrode where a high voltage trigger pulse is applied, which initiates the main discharge (spark) between anode and cathode. This trigger pulse may be utilised within the device in a variety of ways to initiate the main discharge, e.g. field distortion trigger, plasma injection trigger, UV-trigger etc.

Most common discharge gases are air, N2, SF6, argon and oxygen. Often a mixture of these materials is employed. Also liquid or solid media (single shot) may be used. Some solid filled devices are designed to switch powers of

10TW such as are encountered in extremely powerful capacitor bank discharges.

Usually Gas filled spark gaps operate in the 20-1 OOkV / 20 to 100kA range. Commonly filled devices have dimensions of a few inch [181]. Though much higher power devices are available (e.g. 3 MA, Maxwell). Electrode pitting being the most common form of damage. Between 1 and 10 thousand shots per device is usually about what is permissible before damage begins to severely degrade performance. The fastest way to switch a triggered spark gap is with an intense pulse of laser light which creates a plasma between the electrodes with extreme rapidity. Triggered spark gaps tend to have long delay times than Thyratrons (their chief competitor, at least at lower energies). However once conduction has started it reaches a peak value exceptionally fast within a couple of nanoseconds.

2.2.1.2 Cold Cathode Switching Tube

Cold cathode trigger tubes are physically small devices designed to switch impulse currents and voltages of relatively small amplitude. Usually they are intended to trigger other larger devices and are designed to switch pulses of a few hundred volts and a few hundred milliamperes. Most trigger tubes have three or four electrodes, anode, cathode (+ve and -ve terminals respectively), a trigger/control electrode and sometimes a priming electrode. A trigger tube performs in a very simple manner as to that of a triggered spark gap, excepting that usually the conduction is not by an arcing but by a glow discharge. Cold cathode trigger tubes rely upon some external or internal source to ionise the gas suitably for conduction to commence ("priming"). Some devices incorporate