Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
WO 94/03918 21413 0 9 pCT/CA93/00311
1
POhSE GENERATOR
Field of the Invention
This invention relates to cold cathode vacuum
discharge tubes and in particular to their use in a field
emission, cold cathode vacuum tube circuit, hereinafter
referred to as a pulse generator, having a large cathode
area and large interelectrode gap which, if properly
triggered, will generate pulsed auto-electronic emissions
in the abnormal glow discharge region.
Background of the Invention
As the current passed through a gas discharge tube
is increased beyond the levels at which normal glow
discharge takes place, such normal gas discharge being
characterized by a negative resistance characteristic
leading to decreasing potential between the cathode and
anode electrodes of the tube, a region of abnormal glow
discharge is entered in which the negative resistance
characteristic changes to a positive resistance
characteristic leading to increasing potential between the
electrodes. Typically this increased potential rapidly
leads to breakdown into vacuum arc discharge between the
electrodes, again characterized by a negative resistance
characteristic. Accordingly, gas discharge tubes have been
operated in the normal glow discharge or vacuum arc regimes
in which stable operation can be achieved by appropriate
ballasting of the tube, the former regime being suitable
for low current applications and the latter for high
current. It is possible to utilize a normal glow discharge
tube in a low frequency oscillator circuit by placing
capacitance in parallel with the tube and in series with
the ballast because such a tube is characterized by a
comparatively high striking potential at which discharge is
initiated, and a lower but still high extinction potential
at which discharge ceases. Operation in such a mode with
vacuum arc devices is difficult because, in order to turn
off the device effectively, the arc must be extinguished or
2141309
WO 94/03918 PCT/CA93/003
2
otherwise interrupted or divested for long enough to
disperse'the intense ionization formed in its path. On the
other hand, the current densities of normal gas discharges
are too limited for use in applications requiring
relatively large currents.
Devices operating in the vacuum arc regime have
other problems, particularly in terms of ensuring adequate
electrode life, which have led to gas diodes and triodes
(thyratrons) being superseded by semiconductor devices in
l0 most applications. A further limitation of such devices is
that the great difficulty in turning them off, except by
terminating current flow through the device for a finite
period, limits their usefulness as control devices to
rectification, current turn-on and low frequency
alternating current applications.
The only prior art of which we are aware which
successfully exploits the abnormal glow discharge regime is
the process described in U.S. Patent No. 3,471,316 (Manuel)
issued October 7, 1969, which we understand is commercially
utilized in forming organic coatings on metal cans. It
relies on the application of externally generated current
pulses to force a discharge tube temporarily into the
abnormal glow discharge region, the pulses being
sufficiently short that no vacuum arc is established.
There is no disclosure of any endogenous pulsed abnormal
glow discharge, the apparatus is dependent upon an external
pulse generator to operate, and its utility is completely
different from the present invention because it uses
externally generated pulses rather than generating such
pulses. U.S. Patent No. 3,471,316 uses externally
generated and limited current pulses to project operation
of a discharge tube in a transient manner into the abnormal
glow discharge region, thus achieving a higher average
current density and accelerating the polymerisation process
beyond the rate attainable using a normal glow discharge.
SUMMARY OF THE INVENTION
The problems associated with the operation of vacuum
arc devices are typically associated with the establishment
r
WO 94/03918 21413 0 9 P~T/CA93/00311
3
of a continuous channel of low resistance ionized plasma
between the electrodes of a device operating in this mode,
accompanied by intense heating of the electrodes. Such a
channel is difficult to interrupt in rapid and predictable
manner once established. We have discovered that it is
possible to set up a stable endogenous pulsed abnormal glow
discharge regime which is characterized by no such
continuous channel having been established, and
predominantly cold-cathode auto-electronic field emission
rather than thermionic emission, these characteristics
providing the ability to control and extinguish the
discharge readily.
We have found that, by use of a suitable design of a
low pressure gas discharge tube, we can sufficiently
inhibit transition from the abnormal glow discharge regime
into the vacuum arc discharge regime that we can
successfully exploit characteristics of the abnormal glow
discharge regime to provide a device having valuable and
controllable characteristics as a high power, pulse
generator when fed from a current source. Such a pulse
generator has useful applications in for example motor
control and other applications requiring high current
pulses. It is a valuable characteristic that the pulse
repetition frequency can be varied over a range, the extent
of which itself varies according to the physical
characteristics of the tube and the environment in which it
is operated. According to circumstances, the frequency may
range as low as 10 pulses per second or range as high as 104
pulses, these figures being exemplary only and not
limitative.
Most prior art vacuum arc discharge has been performed
with devices having short interelectrode gap lengths and
small electrode areas. Prior art devices require the
application of large kilovoltages and amperages, vacuum arc
discharges in those devices being initiated by contact and
separation of the electrodes. We have established that
vacuum devices equipped with cathodes having large surface
areas and having large interelectrode distances will
support field-emission discharges (either of the pulsed
214-1300
WO 94/03918 PCT/CA93/0031
4
abnormal glow or the vacuum arc type) at low DC voltages
(ie. low-field strengths) and low applied currents. This
indicates that the cold-cathode emissions observed (pulsed
abnormal glow discharge and the vacuum arc discharge) in
this new class of vacuum pulse generator are a function of
parameters heretofore ignored or undiscovered.
The present invention provides a pulse generator and
a method of pulse generation as set forth in the appended
claims.
Description of the Drawinc,~s
The invention will be further explained by way of
example only and with reference to the following drawings,
wherein:
Figure 1 is a graph illustrating the current to
voltage relationship exhibited by a notional vacuum
discharge tube;
Figure 2 is a graph illustrating the current to
breakdown, extinction (PAGD) and sustaining (VAD) voltages
of a particular vacuum discharge tube;
Figure 3 illustrates a discharge tube for a pulse
generator having a glass housing and tetrode geometry;
Figure 4 illustrates a tube having a polymer housing
and a triode geometry;
Figure 5 illustrates central cross sections of two
glass housings (Figs. 5A and 58) and a polymer housing
(Flg. 5C);
Figure 6 illustrates volt-ampere linear
characteristics of two distinct cold-cathode discharge
regimes, PAGD and VAD, in the same tube providing the
curves of Figure 2;
Figure 7 illustrates a Fowler-Nordheim plot of the
Vx or Vs values for the PAGD and VAD regimes, respectively,
again in the same tube;
Figure 8 illustrates the pulse per minute rate
variation observed as a function of low current,
anode-supplied constant DC voltage for two pulse generators
WO 94/03918 21413 4 9 PCT/CA93/00311
Figure 9 illustrates the continuous variation of the
pulse per minute rate as a function of anode-supplied or
cathode-supplied DC voltage;
Figure 10 illustrates an increase in the pulse
5 frequency per minute as a function of the peak pulse RMS
current;
Figure 11 illustrates a continuous variation of NGD
sustaining/PAGD extinction voltages, from breakdown to glow
extinction, with decreasing pressure, in 4 discharge tubes
l0 having different plate areas;
Figure 12 illustrates a continuous variation of PAGD
frequency with decreasing gas pressure in 3 discharge tubes
having different anode and cathode plate areas;
Figure 13 illustrates a shift of the PAGD regime to
higher pressure regions during pumpdown with a rotary
vacuum pump;
Figure 14 illustrates a shift of the PAGD regime to
lower pressure regions and higher frequencies during
pumpdown;
Figure 15 illustrates the observed reductions in
device pressure (Figure 15A) and in voltage (Figure 15B) as
a function of the increase in plate area factor, for the
three discharge tubes having different plate areas
stimulated with low direct current during argon pumpdown
under the conditions described in Figures 11 and 12;
Figure 16 illustrates observed effects of plate area
upon the PAGD breakdown (Vb) and extinction (Vx) voltages
for 7 separate discharge tubes;
Figure 17 illustrates the effect of plate area upon
input DC and transduced RMS currents in 7 discharge tubes;
Figures 18A, B and C are oscillograms depicting AGD
pulses in different regions of a circuit as shown in Figure
20B;
Figure 19 illustrates an effect of varying the
capacitance of a power supply in parallel with the tube, on
the frequency of PAGD production;
Figures 20A and B show two typical wiring diagrams
of pulse generators in accordance with the invention;
Figure 20A illustrates the circuit used in the tests that
WO 94/03918 PCT/CA93/0031
4
abnormal
214-130
WO 94/03918 PCT'/CA93/0031
6
supplied data for Figures 5 to 15, and 19; Figure 20B
illustrates the circuit used for test results illustrated
in Figures 16 to 18; and
Figures 21A and 21B show alternative configurations
in which the tubes described can be incorporated into a
pulse generator.
Description of the Preferred Embodiments
The context of the invention in terms of vacuum
discharge phenomena will first be discussed with reference
to Figures 1 and 2. Referring to Figure 1, which plots the
potential between the principal electrodes of a vacuum
discharge tube with increasing current, potential being
shown on a linear but arbitrary scale of voltage, and
current on a logarithmic scale in amperes, curve A, below
its intersection with curve B, represents a typical
relationship between current and voltage for cold cathode
discharges, including auto-electronic emissions, whilst
curve B represents a typical relationship for thermionic
glow discharges, including thermionic emissions. The high-
current intersection of the two curves at point E
represents a transition into the vacuum arc discharge (VAD)
region (curve C) with the establishment of a continuous low
resistance plasma channel between the electrodes.
It will be noted that curve A exhibits, with
increasing current from very low levels, an initially
rising voltage or "positive resistance" characteristic,
through the Townsend discharge (TD) region, a flat
characteristic through the constant discharge (CD) region,
a falling voltage or "negative resistance" characteristic
through the transitional region discharge (TRD) and normal
glow discharge (NGD) regions, to a minimum, before once
again rising to a peak of F and then falling to an even
lower minimum, equal to the sustaining voltage for a vacuum
arc discharge, through the abnormal glow discharge (AGD)
region. The rising potential over the first portion of the
AGD region is believed occasioned by saturation of the
electrodes by the glow discharge, which causes the
potential to rise until auto-electronic emission sets in
WO 94/03918 ~ 1413 Q 9 pCT/CA93/00311
7
allowing the potential to fall again as the current rises
further.' In practice, the increasing interelectrode
potential following saturation, and other factors such as
electrode heating, leading to thermionic emission, will
tend in conventional tubes to result in a premature
transition from the AGD into the VAD regime, following a
curve similar to curve D shown in Figure 1.
The present invention relies on the use of gas
discharge tubes designed to avoid premature transition from
the AGD to the VAD regimes, and capable of being operated
in a stable manner in that region of the characteristic
curve of Figure 1 extending between points E and F. Figure
2, which plots test results for just such a tube,
constructed as described below, shows, again on similar
coordinates to Figure 1 (except that the potential units
are defined), the extinction or sustaining potentials of
the tube (the same information as plotted in Figure 1),
together with the breakdown potential (i.e. the potential
required to initiate the autoelectronic discharge). It
will be noted that the breakdown curve shows two
discontinuous portions X and Y, corresponding to the vacuum
arc and abnormal glow discharge regimes respectively. The
intersection of curve X, and curve Z representing the
sustaining or extinction potential is illustrative of the
difficulties inherent in extinguishing a vacuum arc
discharge,' since a decrease in current is accompanied by a
decrease in breakdown voltage until it equals the VAD
sustaining voltage which does not vary greatly in this
region. On the other hand, the combination of a fairly
high and constant breakdown voltage (curve Y) combined with
an extinction potential which rises with decreasing current
in the region E-F (see Figure 1) of the pulsed abnormal
glow discharge regime means that the pulsed abnormal glow
discharge will be extinguished if the current source during
the tube operation ceases to be able to sustain the
increasing current required to maintain the discharge as
the potential between its electrodes drops, at some current
below the intersection of curves X and Z.
2141300
WO 94/03918 PCT/CA93/0031 '
8
If the effective internal resistance of the source
is above"some critical level, then as the current through
the tube rises, the proportion of the source potential
developed across the tube will fallauntil it intersects the
curve Z at a current below the intersection with curve X,
at which point the abnormal glow~~discharge will self
extinguish, and the current flow through the tube will drop
abruptly until the current through the tube combined with
the potential between its electrodes again intersects the
curve A in Figure 1. This permits reestablishment of a
rising current through the tube traversing the abnormal
glow discharge region as the potential across the tube
rises to the peak F and then again falls to a point short
of E. Accordingly, under these circumstances, a pulsed
abnormal glow discharge will be exhibited, accompanied by
high amplitude current pulses through the tube. It should
be understood that the curves of Figure 1 are indicative of
the static behaviour of a nominal discharge tube under
particular current and voltage conditions, and are not
fully indicative of the behaviour of the tube under dynamic
conditions in which tube current and inter-electrode
potential vary with time, nor with changes of the many
other factors which may influence tube behaviour. In
particular, the plasma effects generated in various phases
of tube operation require finite time to form, reform or
dissipate-as the case may be, and in the case presently
under consideration this time factor, combined with time
constants of the external circuit in which the tube is
placed, are determinative of the pulse frequency of the
discharge.
The definition of any regime of electrical discharge
in a vacuum is usually presented as dependent upon the
major operational parameter being considered, i.e. upon the
variation of direct current passing between the primary
electrodes. For a given optimal vacuum (which must
necessarily be less than perfect) all gas electrical
discharge regimes can be presented as dependent upon this
parameter. Figure 1 is such a presentation and the peak
that characterizes the abnormal discharge region means that
WO 94/03918 214 ~ 3 0 9 P~/CA93/00311
9
within this region, as the applied current is increased
linearly, the resistance of the vacuous medium in the tube
first increases with increasing current, only to
subsequently decrease, still with increasing applied
current, down to the minimum resistance value corresponding
to the sustaining potential of a "vacuum" arc (which is
somewhat above the ionization potential of the gas, or in
fact of the metal vapour, in the enclosure). As the
transition from a normal glow discharge into a "vacuum" arc
to discharge is made either directly (in thermionic devices)
or indirectly, in cold-cathode conditions, via an abnormal
glow discharge that may be more or less precipitous, it is
only in the ideal diode and the ideal vacuum that both
linear functions (corresponding to the regimes that have a
sustaining potential) and nonlinear functions
(corresponding to the transition regions, such as the TRD
and the AGD) appear to depend exclusively upon the input
current. In fact, many factors affect the AGD, foremost
amongst them, pressure, plate distance and plate area.
Hence the peak in the curve of Figure 1 is an idealized
view of events .
Experimental observations show that auto-electronic
emissions characteristic of the pulsed abnormal gas
discharge (PAGD) regime emerge from the NGD, as the current
is increased beyond the point when the cathode glow has
reached plate saturation (if the current is not too low and
the plate area not too large).
The same effect occurs when the pressure is reduced
and the current is kept constant at a suitable level
(neither too high nor too low, exact figures depending on
other factors such as gap distance and plate area, etc.).
If the current is increased further, in either case,
the PAGD regime fully emerges (in other words, in pumpdown
tests, the applied current also has to be sufficient). In
this regime the plate is not so much saturated with a
negative glow (which remains, but is attenuated), as it
exhibits local concentrations of the plasma that arise in a
given area of the cathode as a function of the auto-
electronic emission mechanism. If the applied current is
2141309
WO 94/03918 PCT/CA93/0031'
increased in steps, a stage is reached at which the
extinction potential of the PAGD falls until it meets the
minimum potential of an arc discharge, as demonstrated in
Figure 2. With reference to Figure 1, this means that the
5 current-dependent variation of the PAGD in these devices
passes from a high to a low extinction potential or from a
high to a low electrical resistivity of the medium, and is
thus localized on the descending slope of the peak in
Figure 1. Expressed in terms of resistance
10 characteristics, the regime of the pulsed abnormal glow
discharge spans, as a function of applied current, a
subregion in which a positive resistance characteristic
changes into a leading negative resistance characteristic.
The pulsed regime of the AGD is only sustainable when the
intensity of the applied current is greater than that
needed to rapidly saturate the plates (but not so great as
to set up a VAD), the result being development of auto-
electronic emission with its associated inverted cone-like
discharge and a residual, faint glow of the entire cathode
(rather than a saturated glow discharge).
Each PAGD cycle begins as a singular emission and
performs a cycle of functions whose electrical
characteristics vary accordingly with time. During a
charging process (which eventually leads to emission), the
plate potential rises to a maximum at F (see Figure 1),
while being limited by the maximum virtual value of the
applied current. Any substantial increase in the applied
current is blocked by the insulating properties of the
intervening medium (as if a very large resistance
characterized the device); in the discharge process,
beginning with the initiation of auto-electronic emission
at F, conditions for conduction across the (operational)
vacuum are established and, as a consequence, the
resistance characteristic of the device becomes
increasingly negative until the extinction potential is
reached, at which point the glow discharge ceases. This
endogenous on/off behaviour is exactly what characterizes
the PAGD cycle.
m ~_rw..~.~ __.~__. ..
T
WO 94/03918 21413 0 9 PCT/CA93/00311
11
Two boundary conditions arise. In the first case,
the available current is not quite enough to sustain the
PAGD. In this instance, full escape from the NGD regime
and the characteristics associated with its sustaining
potential will not occur, while any heating of the cathode
will eventually lead to the establishment of a semi-
thermionic cathode glow. In the second instance, there is
a risk of degeneration into a thermionic NGD or a VAD if
the available current is too high or sustained too long.
This degeneration will set in during the second phase of
the PAGD unit cycle, and may lower the resistance of the
device to the point of constant conduction of current
across the vacuum; the result is that the auto-electronic
emission is not quenched, as spontaneously happens in the
PAGD. Thereafter, extinction of the resulting VAD, which
may be promoted by a variety of factors is an unpredictable
event; if the current is available, the arc will burn for
as long as there is energy supplied and as long as there is
cathode material available to consume. A VAD in no way
resembles a regular, cyclic oscillator, which is the
outstanding aspect of the PAGD. Whilst an arc discharge
is, like the PAGD, an auto-electronic emission phenomenon
characterized by intermittences (the apparent constancy of
an arc is the result of the very high frequency of these
intermittences), such an arc does not exhibit the regular
or quasi-regular cyclical nature of the PAGD, nor its
inherent current limiting characteristics.
In order that a stable pulsed abnormal glow
discharge (PAGD) as discussed above may be obtained, the
discharge to be utilized must be capable of repeated
excursions into the region E to F of Figure 1. This
entails that the tube be constructed so that, as the tube
operates and the current through it rises, the potential
across the tube can reach the peak F in Figure 1 and
beyond, without the pulsed abnormal glow discharge
degenerating into a vacuum arc discharge. This will be
influenced, among other factors, by the extent of
thermionic emission from the cathode which will itself be
influenced by resistive heating of the electrodes and their
214130
WO 94/03918 PCT/CA93/0031
12
work function, as well as by their separation and
configuration, and the nature and pressure of gas within
the tube, as well as the presence of auxiliary electrodes
or probes. The influence of these various factors is
extensively exemplified below. Whilst the present
invention is exemplified with reference to its
implementation using certain exemplary tubes, it should be
understood that the invention may be implemented utilizing
any tube capable of sustaining a stable PAGD discharge
without rapid self destruction whether or not of a
structure specifically disclosed. Thus we have been able
to sustain PAGD utilising tubes of diverse configuration;
for example high voltage thermionic diodes with the anode
connected as cathode, the cathode as anode, and the heater
unused. Even fluorescent lighting tubes can be operated
briefly in the PAGD regime although they are unsuitable for
practical use and fail very rapidly since their electrodes
cannot withstand the current densities involved. Even
tubes having electrode structures that can withstand the
currents invoved will not be suitable if they become heated
to a point at which thermionic emission promotes
dgeneration of a PAGD into a VAD.
Figures 3 through 5 of the drawings illustrate the
construction geometry of presently preferred embodiments of
tubes for use in pulse generators in accordance with the
invention. The discharge tubes are assembled using
accepted techniques which are well known to those skilled
in the art of vacuum tube technology.
Figure 3 shows a discharge tube, generally referred
to by reference 50, having a cylindrical housing 52 which
is preferably a glass material. Depending on the
interelectrode spacing of the discharge tube, which in
accordance with the invention may range from about 2 cm to
about 2o cm or more, the glass housing 52 is preferably
Pyrex (trademark), or #~~40 borosilicate (Corning, NY).
Such cylindrical housings 52 are commonly available in
diameters of about 6 to about 11 cm and a variable
thickness of about 0.2 to about 0.3 cm. Other borosilicate
glass, quartz glass or ceramic housings can be employed as
WO 94/03918 21413 D 9 P~/CA93/00311
13
suitable alternatives to Pyrex glass and in sizes outside
these commonly available ranges.
The discharge tube 50 further includes two parallel,
spaced-apart electrodes comprising a cathode 54 and an
anode 56, hereinafter often collectively referred to as
"plates" for brevity and convenience. As noted above, the
anode and cathode in discharge tubes according to the
invention are spaced 2 to 20 cm or more apart. The cathode
54 and the anode 56 may be either flat or curved and are
preferably made of 0.5 to 2.0 mm thick aluminum, nickel or
nickel alloy, zinc or iron. The thickness of the cathode
54 and the anode 56 is not critical and any thickness
within a reasonable range apparent to those skilled in the
art may be used. The surface areas of the cathode 54 and
the anode 56 are preferably~quite large in comparison to
the surface area of an anode/cathode in prior art vacuum
tube devices. Surface areas which range from 16 to 256 cm2
have been tested, as described in the examples hereinafter.
Although the scope of the invention is not believed to be
limited by this range of surface area of values, it was
generally observed that the larger the surface area of the
anode/cathode tested, the more readily the discharge tube
50 elicited PAGD discharges providing other conditions such
as plate material, vacuum, residual gas fill, voltage and
current remained constant.
A preferred material for the cathode 54 and the anode
56 is aluminum. Two specific types of aluminum are
preferred: namely, H34 rolled aluminum available from the
Alcan Company and Alzak (trademark) aluminum available from
the Alcoa company. Other types of aluminum are assumed to
constitute suitable material for cathode 54 and anode 56.
Aluminum is a preferred material because of its low work
function for field emission as well as for its other
qualities such as relative freedom from sputtering, except
when subjected to vacuum arc discharges, and its electrical
conductivity. In all instances, the aluminum used for
cathode 54 and anode 56 were degreased and rinsed in
accordance with published methods familiar to those skilled
in the art.
214130
WO 94/03918 PCT/CA93/0031
14
Each of the cathode 54 and anode 56 are suspended
within housing 52 by a support member 58 which passes
through hermetic seal 60 on opposite sides of the housing
52. The support members 58 are preferably rigid rods of
substantially pure tungsten in a diameter of 1/l6th to
3/32nds of an inch, or any suitable diameter. The material
of choice is round finished PureTung (trademark) available
from Union Carbide.
The discharge tube 50 also includes at least one
axial probe 62 and the discharge tube 50 shown in Figure 3
has a tetrode geometry with two spaced-apart axial probes
62. Substantially pure tungsten rod is also the preferred
material for constructing the axial probe(s). All tungsten
rods used in assembling discharge tubes in accordance with
the invention were repeatediy cleaned with sodium nitrate
and fused with a beaded sleeve of uranium glass #3320
available from the Corning Company or Nonex (trademark)
glass #7720. These glasses are graded seals designed for
high vacuum tungsten/Pyrex junctions. Before the metal
components of the discharge tube 50 are introduced into the
glass housing 52, the housing is annealed at a temperature
of 565°C. After the discharge tube was assembled, it was
connected by a glass constriction tube to the glass
manifold of a vacuum system (not illustrated).
Figure 4 illustrates a further geometry for a
discharge~tube 50 in accordance with the invention. The
discharge tube includes a parallelepiped shaped housing 64
which is assembled using a suitable plastic polymer sheet.
Polymer housings are preferably made from polycarbonate
resin, preferably ultraviolet resistant. The joints of the
rectangular panels are sealed for example with a low vapor
pressure resin Torr Seal (trademark) available from the
Varian Corporation which is applied along the mating edges
to glue the panels, or another adhesive system suitable for
withstanding the implosive forces of very high vacuum. For
very large housings 64 the walls are also preferably
screwed together at spaced-apart intervals. Non-metallic
internal braces can also be used to reinforce very large
housings 64. The polycarbonate housings are cleaned as per
WO 94/03918 21413 0 9 PCT/CA93/00311
manufacturer's instructions and all metal to polymer
support interfaces, such as the hermetic seals 60 where
electrodes 58 and probes) 62 pass through a side wall of
the parallelepiped shaped housing 64, are preferably epoxy
5 resin joints made with Torr Seal resin. The single axial
probe 62 is made of substantially pure tungsten rod.
Figure 5 shows transverse cross-sections of exemplary
constructions of discharge tubes in accordance with the
invention. Figure 5A illustrates a cylindrical housing 52
10 with a flat plate anode 56 and cathode 54. As shown in
Figure 5B, the anode 56 and the cathode 54 may be
elongated, transversely curved sections which are
substantially semi-cylindrical in shape. This
anode/cathode geometry is actually preferred for
15 cylindrical housings. The curved electrodes may be made
from laser quality reflective aluminum foil about 200
microns in thickness. Such electrodes have a current
tolerance of approximately 100 mA of direct current in the
PAGD regime and are easily destroyed by current induced
disruptive slippage into arc discharge. Curved electrodes
of press-formed aluminum plate are therefore preferred over
curved electrodes made from aluminum foil.
Processing of the vacua
An oil diffusion/rotary pump combination (E02/E2M2,
Edwards High-vacuum, using Silicone 705 diffusion oil),
equipped with thermocouple and Penning gauges (for rotary
and diffusion vacua, respectively), was used to pumpdown a
large bore glass-metal vacuum system equipped with a baffle
valve, desiccating and cold traps, down to 10-~ Torr (= mm
Hg) pressures. At 10-3to 10~ Torr, the rotary pump was
bypassed, and 500 mm Hg of UHP (ultra high purity,
spectroscopic grade 99.9996% pure) argon was admitted into
the system. The system was then evacuated back to 10~ Torr
and the operation was repeated three more times except the
third time a tension of 25 kV (10 mA) DC was applied to the
plate electrodes when the pressure reached -10 mm Hg. Cold
cathode normal glow discharge (NGD) currents of 10 mA were
used to liberate all adsorbed gases remaining in the
2141309
WO 94/03918 PCT/CA93/0031
16
electrodes and the inner face of the housing (52,64), while
the pressure fell to 10~ Torr. Flame heating of the housing
52 was also performed throughout and most intensely at the
constriction joint. Two external, water-cooled copper RF
coils were then applied at each end of the housing 52 and
operated at 450 KHz, at calibrated temperatures of 400°C to
further facilitate the liberation of occluded gases,
excessive heating being strictly avoided. Alternatively to
the RF induction heater, an electrical tape (eg. Briskheat
l0 (trademark)) controlled at a temperature of 400°C can be
applied to the glass housing. After about 30 minutes, the
RF induction heater was turned off and a 100 Kc 30 kV Tesla
coil was applied unipolarly to the probe(s). Then, once
more, 500 mm Hg of UHP argon were admitted into the system
and the cycle of evacuation, heating and bombardment was
repeated, except this time the diffusion pump was connected
to the system and the electron bombardment was carried out
to pressures of 5 * 10'5 Torr. At this point, and with the
kV DC still on, weak x-ray production occurred at the
20 plate edges and this could be detected with a sensitive,
mica-window Geiger-Muller tube counter set at a 5 cm
distance from the discharge tube housing 52, 64. This
x-ray production can be sustained indefinitely at these
kilovoltages and at a pressure of 10-6 Torr, without
25 degenerating into a glow discharge (ie. without evolution
of gas and a rise in pressure). The tube was then
considered to be practically clean ('hard' or x-ray
vacuum). It was then pumped down to the 10~'Torr range
until all discharge ceased and maintained at that vacuum
for a further 8 hrs. Seal-off at the constriction joint
involved slowly heating the joint such that the pressure
never rose above 10-6 Torr. The end-processed discharge
tubes 50 were all closed at different values of 'hard'
vacua (10-6 Torr or higher vacua). For discharge tubes 50
closed at lower vacua (medium vacuum), ie. at pressures
higher than 10-6 Torr but lower than 10~ Torr, the desired
pressure was achieved by reintroducing controlled amounts
of UHP argon and adjusting with the diffusion pump on,
after thorough processing, as described above. For
WO 94/03918 2141 ~ 0 9 PCT/CA93/00311
17
discharge tubes 50 closed at low vacuum to medium vacuum (5
to 10~ Torr), the diffusion pump did not need to be turned
on and the procedures of heating and electron bombardment
were followed at the maximal rotary pump vacuum of -7.5 to
5 * 10'~Torr. The desired final pressure was achieved by an
identically controlled re-admission of small quantities of
UHP argon.
Discharge tubes 50 built with polymer-type housings 64
cannot withstand the heating step during pumpdown (nor do
they require annealing). Accordingly, only the electron
bombardment procedure was performed while processing the
vacuum for those discharge tubes, and for suitably longer
periods (up to 1 hour each cycle). Pumpdown times were
also extended under those conditions.
During and after vacuum processing, the vacua were
constantly tested at the electrodes and at the probes with
a unipolar 30 kV Tesla coil, when all other electrical
apparatus were off. At pressures near 10~° Torr only a
faint local bluish fluorescence could be detected, and at
pressures greater than 5 * l0-5 Torr no discharge could be
observed (so-called 'black' vacuum).
The following examples of tests conducted with pulse
generators incorporating discharge tubes 50 illustrate the
character and performance of such pulse generators. The
disharge tubes utilised are listed in Table 1.
For test purposes, the tubes were utilised in test
circuits as shown in Figures 20A and 20B. The circuit of
Figure 20A was used for most tests, the additional features
of Figure 20B being used only for the tests described with
reference to Figures 16-18.
In Figure 20A, a low impedance DC power supply PS
had terminals connected to plate electrodes of a discharge
tube 50, in the case of a first terminal through a ballast
resistor R, of a resistance which is selected according to
the test being performed, and an ammeter capable of
measuring DC current or RMS AC current. The other terminal
was grounded. A DC voltmeter was connected across the tube
50, and a probe electrode of the tube 50 was left
unconnected.
214130
WO 94/03918 PCT/CA93/0031
18
TABLE 1
List of all the devices utilized
# Area, cm2 Plate d,cm Vacuum Figs
:
~
material .' in Torr
1 12 8 H34 Al. 5 ~ T 10-6 5,6,7 ,8,9,
18,19
2 12 8 Alzak 5 10-6 8
3 1 6 H34 Al. 5.5 Variable 11,12,14
4 64 H34 Al. S.5 Variable 11,12
12 8 H34 Al. 5 .5 V ariable 11,12,13
6 12 8 H34 Al. 5 .5 V ariable 1 1
7 1 6 H34 Al. 5 . 0 2 * 10- 16 ,17
6
8 16 H34 Al. S.0 2*10-6 16
9 64 H34 Al. 5.0 2*10-6 16,17
6 4 H34 Al. 5 .0 2 * 10-6 1 6
11 128 H34 Al. 5.0 2*10-6 16,17
12 12 8 H34 Al. 5 .0 2 * 10- 1 6
6
13 256 H34 Al. S.0 2*10-6 16,17
1 6 4 Alzak 5 .0 2 * 10-6 1 7
4
7 7 Alzak 5.0 2 * 10-6 1 7
16 128 Alzak 5.0 2*10-6 17
1 6 4 H34 3.6 2* 10-6 (Table
7 5)
WO 94/03918 21413 0 ~ P~/CA93/00311
19
In Figure 20B, the first terminal was grounded so
that a current waveform developed across the ballast
resistor R could be monitored by an oscilloscope OSC1,
voltage variations across the tube being monitored by the
oscilloscope OSC1 through a capacitor C1. The potential at
the probe electrode was monitored through a further
capacitor C2 by a further oscilloscope OSC2.
It should be understood that the above circuits are
designed for the purpose of testing the invention, and in
many practical applications of the pulse generator of the
invention, it will be necessary to couple the plate
electrodes to a further circuit arm driven by the pulse
generator, the coupling typically utilising capacitors
and/or diodes. The real and imaginary components of any
load applied through such a~coupling will influence the
operating conditions of the pulse generator, and it may
also be necessary to include diodes in series with the
connections to the power supply as shown in Figures 21A & B
in order to help isolate the load from the supply, thus
turning the pulse generator into a double ported device.
EXAMPLE 1
Volt-ampere Characteristics of a Pulse Generator
The tests described in this example were conducted
with a discharge tube 50 (device #1) constructed with H34
aluminum flat plates (128 cmzarea) set 5 cm apart, and
equidistantly from a continuous axial probe 62 in a vacuum
which measured 10-6 Torr at time of seal off. Figure 2
shows that under conditions of a positive, constant DC
voltage applied to the anode 56 of this device, the
volt-ampere curve for both breakdown potential (Vb, shown
as open squares) and for the minimum discharge potentials
(Vs, or VAD sustaining potential and Vx, or extinction PAGD
potential, both shown as closed circles) disclose two
regions or regimes in the operation of this device, a
region of pulsed AGD which spanned from about 10 mA to
about 150 mA RMS (with an applied maximum of 15 mA DC
average), and a region of VAD at RMS current values greater
than 250 mA. PAGD current data was derived from peak pulse
WO 94/03~1~ ~~ ~ ~ ~ PCT/CA93/0031'
RMS values and VAD RMS current data was obtained at
steady-state. Within the range of the pulsed AGD, the Vb
values were high and plateaued at about 850 volts; Vb
values for the VAD regime were generally lower than those
5 of the PAGD and could be raised by an increase in available
current.
As shown in Figure 6, a PAGD regime could also be
equally identified when the supplied DC voltage was
negative and applied to the same cathode plate 54 (see Fig.
10 3), for both PAGD and VAD, Vb and Vx values (closed and
open squares, respectively) at comparable transduced pulse
RMS currents. Utilizing a 10-fold higher direct current
power supply, also earth-grounded at the centertap but
having a parallel supply capacitance of 55 mfd and a slow
15 voltage recovery rate (ie. less than 200 V/sec), the same
discharge tube 50 (device #i) yielded lOx higher peak PAGD
RMS currents (2 A vs. 200 mA) than were obtained under the
same conditions and with the same power supply by a
positive applied voltage of equal magnitude. These
20 findings suggest that, at high applied direct currents,
there is a strong asymmetric response of the discharge
tubes 50 (larger PAGD RMS current values with cathodic
tension than with comparable anodic tension) with respect
to the sign of the plate polarization in reference to
earth-ground.
It is also apparent that the field emission
responsible for the PAGD regime does not obey the
Fowler-Nordheim VAD region law (see Fig. 7): whereas the
VAD graph has the expected negative slope, the slope of the
PAGD graph is positive, contrary to predictions by the
Fowler-Nordheim VAD region law. This constitutes strong
evidence for the existence of auto-electronic emission
discharges that occur at much lower input currents than
predicted by the Fowler-Nordheim field-emission theory.
EXAMPLE 2
Pulse Count Rates in the PAGD Region
Two pulse count studies were done: a first at low
applied direct currents (<1.5 mA) and a second at mid to
WO 94/03918 21413 p ~ PCT/CA93/00311
21
high applied direct currents (1.5 mA to 200 mA). Peak
pulse RMS currents during the second study were as high as
2 A. Two tubes were used, each having 128 cm2 rigid, flat
plate anode/cathode set 5 cm apart, a continual axial probe
and a 10'6 Torr vacua at time of manufacture, but having
different plate materials; namely H34 aluminum (device #1)
and Alzak (trademark) aluminum (device #2) respectively;
the applied direct currents were increased with the voltage
from 0.1 to 1.0 mA, and all measurements were taken with a
1 Mohm ballast resistor.
At low currents (see Fig. 8), using the discharge
tube 50 (device #1) assembled with H34 aluminum plates and
ballasted with a 1 Mohm resistor and a lower pulse
amplitude detection cut-off at less than 25 V, the pulse
per minute counts at the axial probe were observed to
increase as the anode-supplied voltage (and the current,
not illustrated), was incremented from 300 V to 500 V. At
higher voltages, the pulse count plateaued at a somewhat
depressed level (Fig. 8, open squares). Conversely,
utilizing a discharge tube (device #2) assembled with Alzak
plates in an identical vacuum at seal off (10-6 Torr), the
pulse counts increased with applied voltage up to a maximum
voltage applied, the maximum pulse count being about 9
times higher than observed with device #1 (Fig. 8, closed
circles). Reducing the ballast resistance increased the
pulse rate of device #1 to a maximum of 1000 pps or 60,000
PPM with a 0.125 ohm resistor, and increased the pulse rate
of device #2 to 4000 pps or 240,000 PPM. Analysis of the
pulse signals with an oscilloscope showed that, in both
instances, the observed CPM (counts per minute) values at
the axial probe 62 effectively corresponded (about 1:1) to
the PPM (pulses per minute) values at the cathode 54, under
these conditions, for both devices #1 and #2.
At currents higher than 1.0 mA, when the PAGD regime
is fully active, the inverse phenomenon was observed: ie.
the pulse rates increased with a decrease in the value of
the extinction voltage (Vx) (see Fig. 9). Figure 9
illustrates the continuous variation of the pulse per
minute rate as a function of anode-supplied (open squares)
214-130
WO 94/03918 PCT/CA93/003'
22
or cathode-supplied (open circles) DC voltage, at applied
currents that varied from 1 to 200 mA (the higher current
sequence for the anode voltage is on the left); the anode
voltage curve shown on the right side of the figure was
obtained using an intermediate current supply; the two
other slopes were obtained using a high current supply;
capacitance values for both high and intermediate current
supplies were respectively, 50 and 1 mfd, both power
supplies being earth-grounded; the discharge tube (device
#1) had 128 cmz H34 aluminum plates set 5 cm apart and
enclosed a vacuum measured at 10~ Torr at time of seal off.
Pulse rates also increased proportionally to the
transduced pulse RMS current (see Fig. 10). Figure 10
illustrates an increase in the pulse frequency per minute
as a function of the peak pulse RMS current, at applied
currents from 1 to 200 mA; corresponding voltages are shown
in Fig. 9 (not for all points) and the higher current,
anode-supplied curve corresponds to the lower voltage curve
in Fig. 9, all conditions being as described for Fig. 9.
The pulse rate increase was observed for both positive and
negative polarizations (squares and circles, respectively,
in both Figures 9 and 10) of the 'vacuum', with discharge
tube 50 (device #1). Under these conditions and with a 1
Mohm ballast resistor, rates of 113-124 pps were measured,
the limiting factor being the recovery time of the voltage
regulation of the power supply as the current drain
increased. This phenomenon was exaggerated when no ballast
was employed and the largest peak pulse currents were
observed (not shown). With faster recovery power supplies
capable of delivering the same or higher input currents
(and the same large value of capacitance in parallel with
the plates) much higher pulse rates (>1,000 pps) could be
obtained, along with larger peak pulse RMS currents.
EXAMPLE 3
Detection of the PAGD Region in the Pulse Generator as a
Function of Decreasing Pressure
Argon pumpdown tests were conducted to determine
whether and when the PAGD region of the discharge was
WO 94/03918 214:13 0 ~ P~T/CA93/0031 I
23
apparent utilizing comparably low voltages (up to 2.5 kV).
These tests were performed with both the diffusion pump off
and on. Fig. 11 shows a typical curve of the variation of
the sustaining/extinction voltages at the plates with
decreasing pressure at the rotary pump, from breakdown (at
860 VDC) to glow extinction, for all four discharge tubes
50 examined (device #'s 3 to 6), which were assembled with
H34 aluminum plates having different electrode areas:
device #3, 16 cm2 (small closed squares); device #4, 64 cmz
(open circles); device #5, 128 cm2 (open squares); device
#6, 128 cm2 (large closed squares). Each discharge tube 50
had the same gap distance of 5.5 cm and was assembled with
the same volume of glass envelope. Devices #3 to 5 were
evacuated simultaneously and an identical average direct
current of 1 mA was applied~to each separately, using
comparable power supplies ballasted with a 1 Mohm resistor.
Device #6 was evacuated in a separate test, under the same
pumpdown conditions and at the same applied potential of
860 VDC at breakdown, but was subjected to a 100-fold
higher, average direct current of 500 mA. It is readily
apparent that the continuously varying,
sustaining/extinction voltage curves shown in Figure 11 are
analogous to the Paschen gas breakdown voltage curve and
that throughout most of the voltage range all three low
current curves are parallel. Independent determinations of
the low current breakdown voltage curves for all three
discharge tubes 50 (devices #3 to 5) showed the exact same
relation for all three curves as observed for the
sustaining/extinction voltage curves (results not shown).
The differences between the electrical discharge regimes
observed as a function of decreasing pressure are most
apparent in the larger plate area discharge tube 50 (device
#5). The three regions of the discharge, the transitional
glow, the normal glow and the pulsed abnormal glow, are
clearly distinguishable for that device (see Figure 11).
Figure 11 illustrates a continuous variation of NGD
sustaining/PAGD extinction voltages (Vs/Vx), from breakdown
to glow extinction, with decreasing pressure (at a rotary
pump), in 4 discharge tubes having different plate areas
2141309
WO 94/03918 PCT/CA93/0031'
24
but the same electrode material (H34 aluminum), the same
gap distance and the same potential of 860 VDC prior to
breakdown; all curves except that joining the large closed
squares, were measured using the same low applied direct
current of 1 mA; the curve joining the large closed squares
was obtained with an average direct current of 500 mA; the
three quasi-parallel, low current curves for discharge
tubes (devices #3 to #5) having anode/cathode areas of 16
(small closed squares), 64 (open circles) and 128 cm2(open
l0 squares), respectively, were obtained during a simultaneous
test; the high-current curve for a 128 cmz plate discharge
tube (device #6) was measured during a separate test. In
all cases, pumpdown was performed in an argon atmosphere.
The scale markings for different glow discharge regions
shown at the upper part of the diagram refers only to
observations made with a 128 cm2 area discharge tube device
#5 at low applied currents. In the transitional region
discharge (TRD), the cathode glow is of minimal point-like
size and rapid oscillations of the striations of the plasma
positive column originate quasi-sinusoidal, dampened
sinusoids, ramp-like or noise-like waveforms associated
with sporadic, small amplitude (2 to 15 volts), pulsed
auto-electronic emissions. In this region the voltage
tends to fall, while oscillating erratically at first. As
the pressure further decreases, there follows a stable
normal glow discharge (NGD) region, where conduction of
direct current across the vacuum pre-empts the possibility
of auto-electronic emission. The lowest voltages are
observed in this region. After the recession of the
positive column and upon glow saturation of the plate
areas, just as the cathode glow is beginning to recede, the
intense, large amplitude (>100 V), pulsed auto-electronic
emission characteristic of the PAGD regime emerges. In
this region, the voltage tends to climb until extinction
occurs before the maximum voltage of 860 V is again
attained. In the other two devices at the same low applied
direct current, the borders of the discharge regimes are
blurred. In device #3, the low intensity, small amplitude
auto-electronic emissions develop into a few high
WO 94/03918 21413 0 ~ PCT/CA93/00311
intensity, large amplitude emissions, as they decrease in
frequency and with considerable overlap; the PAGD and NGD
regimes are also mostly mixed, until lower pressures of the
order of 0.01 Torr are attained, at which point the PAGD
5 regime functions alone at low frequency. In device #4, the
NGD regime can be better distinguished from the TRD, and
the PAGD from the NGD, but high intensity, large amplitude
auto-electronic emissions occur early on in the NGD region
as the glow saturates the plates faster than for device #5.
10 There is a dual effect on increasing the average applied
direct current 100-fold (device #6, large closed squares,
shown in Figure 11): the entire ascending arm of the
voltage curve is displaced upward in the pressure scale and
the modal distribution of the voltage variation is
15 compressed. The high applied direct current also abrogates
the two discharge regions that preceded the PAGD. From
breakdown to extinction, the regime of the discharge is
solely that of the PAGD, the positive column of the
discharge weakening with the decreasing pressure.
20 Figure 12 illustrates a continuous variation of PAGD
frequency with decreasing gas pressure in 3 discharge tubes
having different anode and cathode plate areas (16, 64, 128
cmz) but the same cathode material (H34 aluminum) and the
same gap distance of 5.5 cm; all 3 discharge tubes (device
25 #'s 3 to 5) were applied the same potential of 860 VDC
prior to breakdown and were stimulated with the same direct
current of lmA; pumpdown was performed with a rotary vacuum
pump in an argon atmosphere; neither the quasi-sinusoidal
nor the noise-like oscillations observed upon breakdown and
during the transitional discharge region, nor the low
intensity auto-electronic pulsed emission (2 to 15 v
maximum amplitude) observed in the same region, are shown.
In all three devices, the PAGD regime first appeared mixed
together with the NGD regime in the form of pulses that
perturbed the steady-state glow, the pulses increasing in
frequency with the decreasing pressure until a maximum
pulse rate was attained.
Figure 13 illustrates a shift of the PAGD regime to
higher pressure regions during pumpdown with a rotary
214130
WO 94/03918 PCT/CA93/0031'
26
vacuum pump in an argon atmosphere, as a function of a
500-fold increase in applied direct current (1 vs. 500 mA),
at the same starting voltage of 860 VDC and utilizing the
same 128 cmz H34 aluminum plate discharge tube (device #5)
in two separate tests. The higher current displaces the
PAGD region upward in the pressure scale, just as was
observed in the ascending arm of the voltage curve (see
Figure 11). The displacement induced by the applied high
current occurs over a pressure range where, at low current
(1 mA) and with the same applied potential at breakdown,
some weak, low-amplitude, pulsed auto-electronic emissions
are observed during the TRD.
Figure 14 illustrates a shift of the PAGD regime to
lower pressure regions and higher frequencies during
pumpdown with a rotary vacuum pump in an argon atmosphere,
as a function of a higher applied potential; starting
voltages were 860 (closed squares) and 1500 VDC (open
squares). The discharge tube (device #3) had plate areas
of 16 cm2 and the results shown are from separate tests with
the same power supply, at low currents (average 1 mA).
Figure 14 shows the effect of increasing the starting DC
voltage at breakdown by 1.75-fold (from 860 to 1507 VDC).
The increased current displaced the PAGD upper pressure
limit downward in the pressure scale, in opposition to the
current effect and it also increased by a factor of about
8.8 the frequency of the intense, large amplitude,
auto-electronic emissions.
Using the same applied low direct current and
potential magnitude at breakdown (860 VDC) described for
the tests represented in Figures 11 and 12, pumpdown of the
three different plate area discharge tubes 50 (each having
interelectrode distances of 5.5 cm) was performed with the
oil diffusion pump on. While the effect of increasing the
plate area under these conditions remained the same, ie.
lowering the pressure for the same sustaining/extinction
potential and displacing the PAGD region to regions of
higher vacuum, there was a noticeable difference compared
with the same test done with the rotary pumpdown: ie. the
extinction pressure was greatly extended downward in the
WO 94/03918 21 ~-13 0 9 ~'CT/CA93/00311
27
pressure scale for all devices, and, consequently, the PAGD
region was greatly expanded into the medium to high vacuum
ranges. A 128 cm2plate area discharge tube 50 with 5.5 cm
gap, (devices #11 and 12) typically reached PAGD extinction
at 5 * 10'5 Torr, though its peak pulse rate remained
basically unchanged. This overall displacement of the PAGD
phenomenon to higher vacuum regions under conditions of oil
diffusion evacuation may well be due to the migration of
very low vapor pressure oil molecules to the tube ends
(despite the baffle and the cooling trap) and their
interaction with residual gas molecules in the electrical
field of the devices. With the diffusion pump on and
voltages progressively increasing up to 2.5 kV with
decreasing pressure, the PAGD regime in these discharge
tubes 50 operated from 10-3 to 10'5 Torr. Typically a 128 cm2
Ii34 aluminum plate discharge tube 50 (5.5 cm gap) will
operate in the PAGD regime at 2 * 10-5 Torr, with an applied
voltage of 2.2 kV and at a pulse rate of 30 pps. With
higher vacua (<10-5 Torr) and voltages, ultimately the PAGD
regime gives way to the production of cathode rays and very
weak x-rays. From several such diffusion pumpdown tests it
was concluded that the PAGD was facilitated by the use of
Alzak electrode material and, as it will be shown in
Example 4, by larger plate areas.
EXAMPLE 4
The Effect of the Plate Area on the PAGD Characteristics
during Pumpdown
The effect of increasing the plate area of the
cathode 54 and anode 56 of a discharge tube 50 was tested
by two methods: 1) using a pumpdown method of varying the
vacuum by equilibrating of the gas flow against a rotary
pump (as explained below) and 2) using sealed housings 52,
64 enclosing a vacuum of 2 * 10-6 Torr obtained with the
diffusion pump (see Example 5).
The results from the first test is shown in Figures
11 and 12, for the discharge tubes 50 stimulated with low
(1 mA) direct currents, at the same starting potential of
860 VDC at breakdown. A comparison indicates that the
WO 94/03911413 0 ~ pCT/CA93/0031'
28
effect of increasing the plate area in discharge tubes 50
having the same gap distance, and thus the same pd value
(pressure, in Torr, multiplied by interelectrode gap
distance, in cm), and the ,same volume, is to depress the
voltage, particularly in the NGD and PAGD regions and to
displace the auto-electronic pulsed emission characteristic
of the PAGD regime to a higher vacuum range. The peak
frequency of PAGD for each given area is also attained, in
each case, at a vacuum that increases proportionately to
the order of increasing area (16 -> 64 -> 128 cm2) as does
the magnitude of the peak frequency of PAGD for a given gap
distance. The distribution of PAGD frequencies also
narrows its characteristic mode with the larger area
plates, by displacing an upper pressure limit to lower
pressure regions, the most significant shift in this
respect being from the 64 to the 128 cm2 devices (Figure 12,
open circles vs. open squares). This combined compression
of the distribution mode and its shift to the left in the
pressure scale corresponds to a better definition between
the NGD and the PAGD regimes afforded by the discharge tube
50 with the largest plate area employed (128 cm2), as
discussed above in Example 3. Moreover, in accordance with
Paschen's law, the observed area-dependent voltage
reduction effect cannot be explained, inasmuch as the
voltage is predicted to remain the same as long as the
product pd is constant, even if the plate area increases.
Since the interelectrode gap distance was constant for all
devices and as the pumpdown was also performed
simultaneously and the tubes had identical volumes, it is
apparent that there is an electrode plate area effect which
is not accounted for by Paschen's law. The observed plate
area effect appeared to have an effect opposite to current
and in the same direction of increasing potential, as it
displaced the PAGD region downward in the pressure scale to
higher vacuum regionsand increased the PAGD frequency. In
addition, an increase in area also reduces the magnitude of
the potential. From the results shown in Figure 14, and a
comparison with Figure 12, it is apparent that an increase
of 1.75-fold for a given breakdown potential of a 16 cm~
WO 94/03918 21413 0 9 p~'/CA93/00311
29
discharge tube yields the same.pulse rate (about 60 pps) as
does an 8-fold increase in plate area for the same volume
housing (52, 64), but requires a lower pressure.
A comparison of breakdown order and pressure, as well
as of peak pps values and peak pps conditions carried out
as a function of plate area for the discharge tubes 50
(devices #'s 3 to 5) represented in Figures 11 and 12, is
shown in Table 2. The discharge tube 50 with the largest
plate area, which was the first to undergo breakdown
(during six separate tests) at the highest pressure of 3
Torr, yields an 8-fold higher PAGD rate than the discharge
tube 50 with the smallest plate area of 16 cmz, at the
lowest pressure (the pressure is 24 times lower than that
of the 16 cmzdevice). This peak pps rate occurs, however,
at a voltage which is about~9.5% greater for the discharge
tube 50 with the largest plate area. These results suggest
that a larger plate area promotes breakdown at higher
pressures (ie. the breakdown pressure decreases inversely
to the order of increasing plate area) and supports lower
sustaining/extinction voltages.
Tables 3 and 4 list sampled data from the tests
shown in Figures 11 and 12. Table 3 shows a comparison, by
fixed voltages, of values on the ascending arm of pressure
dependent voltage curves for three discharge tubes 50
having different plate areas. Table 4 shows the pressure
variation predicted by the Paschen law if the devices had
an interelectrode gap that varied proportionally to a
linear scaling factor k~ (which they do not) and a pressure
that varied inversely to the linear scaling factor kL. In
both Tables 3 and 4 the first six vertical columns describe
(1) the plate area values of the three discharge tubes 50
tested, (2) the linear scaling factors kL for each discharge
tube 50 based exclusively on the linear plate dimensions
(ie. not implying a kLSCaled gap distance), (3) their
respective area scaling factors (kA), (4) the
interelectrode gap distances in cm, (5) the pressure in
Torr and (6) the DC voltage.
Table 3 is horizontally divided into two parts.
Groups A and B represent two sets of theoretical
WO 94/03918 ~ ~ ~ ~ a ~ PCT/CA93/0031 '
predictions derived from the.~haschen law and in conformity
with the Child-Langmui,r~.~~Cheory of the NGD.
Group A represents the model for a linear scaling of
all dimensions, as when the plate factor kLalso applies to
5 the distance between the plates, which thus must increase
by the kLValue. This requires that for the voltage to
remain unchanged, the pressure must decrease by the
reciprocal of the kLValue. Accordingly, a theoretical
pressure reduction factor (prf) is shown in vertical column
10 8 of Table 3, and the predicted pressure values shown in
vertical column 5.
Since in group B, the k,,scaling factor does not
apply to the gap distance, which thus remains constant at
5.5 cm, the theoretical prf is unity and the pressure
15 remains constant if the voltage is to remain the same.
For the horizontal sample groups numbered 1 to 6,
column 5 shows the experimental pressure values obtained
for the same sample voltage and column 8 shows the observed
prf values. Column 7 shows the experimental pps rates. It
20 is readily apparent from the experimental values of Table
3, that identical voltages entail reduced pressures that
diminish as an area dependent effect. Contrary to the
prediction (group B, Table 3), and despite the fact that
the interelectrode gap distance remains the same for all
25 the discharge tubes 50 (device #'s 3 to 5), a constant
voltage is only attained at a lower pressure for the
discharge tubes 50 with larger plate areas. In other
words, the product pd is not constant for a given voltage,
and thus does not conform to the Paschen Law prediction of
30 a prf that equals unity for all discharge tubes 50
regardless of their plate area.
In Table 4 only experimental data is presented. The
predicted voltages shown in vertical column 6 of Table 4,
were obtained using as parameter the experimental values
observed for the discharge tube 50 with the smallest area
(16 cmz) used in these tests, at pressure intervals
determined from arbitrary prf factors (vertical column 9,
Table 4) chosen in accordance with a theoretical model of
the horizontal group A from Table 3, that is, as if the
WO 94/03918
PCT/CA93/00311
31
reciprocal of the kLfactor applied to the pressure of these
discharge tubes, even though the interelectrode gap
remained constant. Using those pressure intervals, and the
actual voltages observed for the other two discharge tubes
(64 cm2 and 128 cm2, vertical column 7, Table 4), the
experimentally observed voltage reductions as % of the 16
cm2 voltage reference (shown in column 6) were determined
and are shown in vertical column 10, Table 4.
The experimental data listed in Tables 3 and 4 was
to then used to calculate relative, average pressure reduction
factors using the fixed voltage series shown in Table 3,
column 8, and percentage voltage reductions using the 1/kL
pressure series shown in Table 4, column 10. Those
calculations interrelated all the discharge tubes used for
the low current pumpdown test with respect to their plate
area factors, or kAvalues: ie. kA= 2, when comparing the
64 cm2 and 128 cm2 discharge tubes; kA= 4, when comparing
the 16 cm2and the 64 cm2discharge tubes; and kA= 8, when
comparing the 16 cmZand 128 cm2 discharge tubes. To
triangulate the data, kA= 8 results for the voltage series
of Table 5 were derived by comparing the pressures obtained
for the 16 cmZand 128 cmz discharge tubes, shown in Table 1.
With respect to the pressure series of Table 5, whereas the
16 cm2device was used as a 100% voltage reference for the kA
- 4 and kA= 8 results, the kA= 2 results were determined by
comparing the percentage voltage reduction for the 64 cm2
and 128 cm2 devices. From the triangulated data,
statistical means and their standard errors were calculated
to determine the regression curves of the area dependent
pressure reduction effect obtained when the voltage is
constant (Figure 15A), and of the area-dependent voltage
reduction as o of the maximum, when the voltage of the 16
cm2discharge tube is taken as a reference voltage and the
pressure is varied arbitrarily in accordance to the
reciprocal of the kLfactor(Figure 15B). Figures 15A and
15B strikingly illustrate the effect of increasing the kA
factor or the plate area in these discharge tubes 50. A
lower pressure is required for the same voltage (ie. a prf
lower than unity), the voltage being depressed when the
214130
WO 94/03918 PCT/CA93/0031'
'~ 3 2
pressure is constant. Within the kArange tested, both
regression curves are linear. Following the regression
curve of Figure 15A, one can predict that a kA = 17 will
reduce the pressure by one order of magnitude. Conversely,
following the regression curve of Figure 15B, one can
predict that for the same pressure and the same
interelectrode gap distance, the voltage will be depressed
by 50% with a kA= -21.5. These predictions, however, will
only hold if the curve remains linear throughout a wider
l0 range of kAvalues.
In conclusion, the effect of increasing the plate
area of discharge tubes stimulated with the same starting
voltage and the same current is to: 1) shift the breakdown
pressure upwards, 2) depress the working voltage, 3)
increase the pulse rate both in the TRD and PAGD regions,
4) shift the PAGD region downwards in the pressure scale
and segregate the discharge regimes more clearly as a
function of decreasing pressure. These observations also
explain why the discharge tubes with smaller plate areas
shift the PAGD up in the pressure scale, as an increase in
current does. Effectively, a smaller plate area not only
concentrates the lines of electrostatic force in a vacuum,
but it also increases the current density per unit area,
with the consequent glow saturation of the plates,
necessary for the abnormal glow discharge region to be
attained, occurring earlier on during pumpdown, than for
discharge tubes with larger plate areas.
-_ 214-13 Q 9
Pcr~ca ~~,'~70311
- / 3 JE/~TEIr~,BEee ~ S 93 ~~.3, o ~'; !~')
-33-
TAB LE 2
Plate Area 16 64 l~
(~z)
Lowest DC 290 226 210
Voltage
(5mA current)
Pressure 0.31 0.68 0.415
(Torr)
Expected Breakdown2.25 2.25 2.25
Pressure
Breakdown 0.75 0.83 1
Pressure
Factor
Breakdown 2.7 t 2 t 0.29 1.3 t
Order 0.37 0.23
MtSEM (n=6)
Peak PPS 7 16 61
Relative PPS 1 2.29 8x
Ratios
1 3.Sx
Peak PPS Pressure0.725 0.220 0.030
Peak PPS Volts307 2?5 332
DC
i . ., ~ ,Y ~ L ~ ~. - ~r tr.9' ~ ~ in to
2141~a9
PCT / CA ~ J / O
l 3 SEP~ Errr,BE~ i f s.3 <i3. c f. s?)
-34-
TABLE 3
AreaPiate Kp d,cm Predicted V PPS TheoreticalGroup
KL, p
(cm2 (Tort) prf
)
16 1 1 5.5 0.125 307 NA NA A
64 2 4 11 0.0625 307 NA 1/KL=1/2 A
128 1.4 2 15.5 0.044 307 NA 11KL=1/1..4A
16 1 1 5.5 0.125 307 NA NA B
64 2 4 5.5 0.125 307 NA 1 B
128 1.41 2 5.5 0.125 307 NA 1 B
AreaPlate Kp d,cm Exptl. V PPS Exptl. Sample
KL p prf
(cml (Tort) #
)
16 1 1 5.5 NA 255 NA NA 1
64 2 4 5.5 0.18 255 14 NA 1
128 1.41 2 5.5 0.092 255 1 1/1.96 1
16 1 1 5.S 0.125 307 4 NA 2
64 2 4 5.5 0.065 307 5 111.91 2
128 1.41 2 5.5 0.035 307 36 l/1.s6 2
16 1 1 5.5 0.0675 350 2.5 NA 3
64 2 4 5.5 0.0475 350 4.5 l/1.a2 3
128 1.41 2 5.5 0.023 350 56 1/2.06 3
16 1 1 S.S O.OS00 407 2 N.A 4
64 2 4 S.5 0.0225 40? 2.5 1/2.22 4
128 1.41 2 5.5 0.0090 407 30 1/2..5 4
16 1 1 5.5 0.0310 450 G.8 NA 5
64 2 4 5:5 0.0091 450 1.S 1/3.4 5
128 1.41 2 5.S 0.0060 450 22 1/1.52 5
16 1 1 5.5 0.0140 500 0.65 NA 6
64 2 4 5.5 0.0060 500 1.2 1/2.33 6
128 1.41 2 5.5 0.0038 500 5 1/1.5s 6
-_ x ...v. . . . ; ~ ,,t., Y ~, .. ._
s! Z .-.'~ ~ a..'.' r L . .~ ~. :.s es d Y 1a r
214.130 PCT/CA ~3~~'~~Z~
~.3 ..f'EPTt~rr,BFR i S93 ~/3, 09. 5?~
'TAB LE 4
Area PlateKA d p V V pp$ ChosenVoltageSample
p
(cm2)KL (cm) (Tort)PredictedObserved FactorReduction#
as %
16 I 1 5.5 0.800 318 318 4 NA NA = 1
I 00%
64 2 4 5.5 0.400 291 253 10 1/2 86.94 1
128 1.41 2 5.5 0.2283Z90 250 0 1/1.4286.2 1
16 1 1 5.5 0.400 291 291 5 NA NA 2
64 2 4 5.5 0.200 295 257 15 112 87.1 2
128 1.41 2 5.5 0.141 309 243 0 1/1.4178.6 2
16 1 1 5.5 0.200 295 295 4 NA NA 3
64 2 4 5.5 0.100 327 260 6.5 1/2 79.5 3
128 1.41 2 5.5 0.070 350 265 9.5 1/1.4175.7 3
16 1 1 5,5 0.0800334 334 3 2'TA NA 4
64 2 4 5.5 0.0400422 365 4.25 I/2 86.5 4
128 1.41 2 5.5 0.0283457 347 58 1/1.4175.9 4
16 1 I 5.5 0.0400422 422 1.5 N.4 NA 5
64 2 4 5.5 0.0200473 415 1.8 1/2 87.7 5
128 1.41 2 5.5 0.0141500 378 33 1/1.4175.6 5
16 1 1 5.5 0.0200473 473 0.8 NA NA 6
64 2 4 5.5 0.0100515 450 1.6 I/2 87.4 6
128 1.41 2 5.5 0.0700555 445 27 1/1.4180.2 6
16 1 1 5.5 0.00800550 550 0.. NA NA 7
64 2 4 5.5 O.OCt400600 535 1.C~ 1/2 89.2 7
128 1.41 2 5.5 0.00283760 8C10 3.C' 1/1.41-5.3 7
;y ... _... ., -
21 x.13 0 ~ PcT ~ ca ~ 3 ~ ~ a ~ n
- 3 6 - ~ 3 ,S C= f TE ~n 6 E,P l 5 5 ~ ~~,~, 0 9, 9.~
TABLE 5
Plate 2 4 g
Kp
Pressure Sample #
reduction
factors
Voltage series
0.51 NA NA 1
0.54 0.52 0.28 2
0.485 0.70 0.34 3
0.4 0.45 0.18 4
0.65 0.29 0.19 5
0.63 0.43 0.27 6
Mean 0.52 0.44 0.24
SENT 0.04 0.06 0.03
Plate 2 4 8
Kp
Voltage Sample #
reduction
as ~'o
of maximum
Pressure Series
99.1 86.9 86.2 1
90.2 87.1 78.6 2
95.2 79.5 75.7 3
87.7 86.5 75.9 4
86.4 87.7 75.6 5
91.7 87.4 80.2 6
Mean 91.7 85.85 78.7
SEM 2.1 1.6 1.8
_,,...
.~ w a r . ~.J ...~ ..d i r a.. :~
WO 94/03918 ø 3 ~ PCT/CA93/00311
37
EXAMPLE 5
The Effect of Plate Area on the PAGD Characteristics of
Discharge Tubes Enclosing a High Vacuum
The second method used to test the effect of
increasing the electrode plate area in the design of a
discharge tube 50 made use of glass housings 52 enclosing a
final vacuum of 2 * 10'b Torr obtained with a diffusion pump
on. These tests were performed with high direct currents
(200 mA to 1 A). All discharge tubes tested (device #'s 7
to 13) had an interelectrode gap distance of 5 cm, enclosed
the same volume and the same vacuum, and were assembled
with Ii34 aluminum plates having plate areas which varied
by an area factor of kA= 2, namely: 16, 32 (not
tested),64, 128 and 256 cmz. At a seal off vacuum of 2
10'~Torr, the first two discharge tubes 50 tested in this
series (16 and 64 cmz, device #'s 7 to 10) remained
unresponsive (no signs of discharge). Even when 3.3 kV was
applied, one of the 64 cm2 discharge tubes showed only a
faint glow (also see discussion of results for groups #1
and #4 of Table 6 below). The results for the kA= 2
series are shown in Fig. 16. The results indicate that
when the current, the interelectrode distance and the
pressure are all kept constant, the breakdown potential
(Vb) for the PAGD decreases with an increase in plate area.
For the largest plate area tested (256 cm2), the PAGD
breakdown (287 V) and extinction (Vx = 284 V) voltages
practically coincide, suggesting that larger areas might
depress both Vb and Vx still further. These results were
recorded under identical conditions of applied direct
current (200 mA, closed circles, Figure 17), of peak pulse
RMS current (open circles, Figure 17) and of pulse
frequency (20 pps), using an earth-grounded centertap power
supply with both positive and negative voltages applied
simultaneously to the respective plates. Under the same
conditions of applied total power (same starting voltage,
but higher applied direct current because of their lower
sustaining/extinction voltage), three discharge tubes 50
built with Alzak plates having areas of 64, 78 and 128 cm'-
respectively were tested with the same power supply. As
21413fl9
WO 94/03918 PCT/CA93/0031'
38
shown in Figure 17, these discharge tubes conduct 5-fold
higher DC currents (closed squares, Figure 17), transduce
3-fold higher peak pulse RMS currents (open squares, Figure
17) and yield a 20 to 30-fold increase in pps (from 20 to
600 pps) at similar field strengths, when compared with the
results obtained using hardened aluminum plates.
Table 6 shows the experimental and predicted results
obtained with 4 discharge tubes 50 (device #'s 9, 11, 13
and 17) assembled with hardened aluminum plates, as a
l0 function of scaling the plate area (column D) by a kA= 2
area factor (column E), while varying the interelectrode
distance inversely with respect to the pressure (group #'s
1-3) or, alternatively, keeping these factors constant
(group #'s 4-6), so that in both instances the pd product
is constant. A plate area kAfactor of 2 corresponds to a
plate linear scaling factor kL of 212 = 1.41. Columns A to
C show the scaling of the selected linear dimensions, while
column G shows the vacuum measured when the respective
housings 52 were sealed. The space charge theory of glow
discharge holds that the function V, or the voltage
difference at corresponding points, is the same in kL-scaled
vacuum tubes, the linear dimensions (including the
interelectrode gap distance) of a vacuum tube "b" being k~
times the linear dimensions of a vacuum tube "a". Under
these conditions where the gap distance also increases by
the kL factor, the Poisson term dZ * V (where d2V = r/eo; r =
density of the attracting matter at the point chosen
(charge density) and eo = permittivity of free space) in the
interelectrode space of vacuum tube "a" is k~z times that in
vacuum tube "b", as long as the pressure p changes by 1/kL
so that, to a first approximation, the breakdown voltage
remains the same. As the permittivity of free space is
deemed to be a constant, the charge density r in vacuum
tube "a" is kL2 times that of vacuum tube "b" (the upscaled
device). Consequently, the cathode current density J of
vacuum tube "a" is also expected to be kLZ times that of
vacuum tube "b". We can thus summarize these predictions
as: given a linear factor k~ between "a" and "b", two
vacuum tubes will have the same breakdown voltage if the
T
21413~~
WO 94/03918 PCT/CA93/00311
39
pressure of "b" decreases by 1/kL, with the result that J
should decrease by 1/kL2 and the field strength should also
decrease by 1/kL, while J/pz and E/p (where E = electrical
field strength) both remain constant. Essentially, as the
area factor between the two discharge tubes is kA = kLZ,
both the charge density r and the current density J should
change by 1/kL2 = 1/kA, ie. inversely to the plate area
factor kA. Accordingly, using the results from group #2,
Table 5, as a reference for the fully corrected final value
of the current density J, determined by the autographic
method of emission crater size determination, the expected
values of J were determined for the other groups and these
are shown in column H. Corresponding values of J/p2
predicted are shown in column J. The field strengths
predicted from the Poisson term (E = -dV), are shown in
column L, and their corresponding E/p ratios to be expected
are shown in column N. Pulse rate (column P) was kept low
and constant, for purposes of comparison between the
groups. The experimental values measured at breakdown for
each device are shown in Table 6, columns I (for the
current density J at the emission site), K (for J/pz), M
(for the field strength E) and o (for E/p) . These results
indicate that, for kL(= 1.4)-scaled discharge tubes shown
in group #'s 1 to 3, Table 6, having a kA= 2 and inversely
varying p and d values (the product pd is constant but the
pressure and distance terms obey the kL-scaling inverse
relation) the variation of J and E is nonlinear (in fact,
one would expect group #2 to be just as unreactive as group
#l, Table 6, at these pressures). The Table 6 results for
kL(= 1.4)-scaled discharge tubes (group #'s 4-6,
corresponding to device #'s 9-l0, 11-12 and 13,
respectively) with a kA= 2 but separate constant values for
p and d, also show that J varies discontinuously (it is 0,
even with very high field strengths, in device #9, group
#4, but appears to plateau in device #'s 11 and 13, group
#'s 5 and 6) and not inversely with respect to kL', ie. not
inversely to an increase in area. These results further
indicate that the field strength E necessary for breakdown
at these high applied currents does not remain constant and
214130
WO 94/03918 PCT/CA93/0031
linear, as predicted, but decreases nonlinearly with an
increased plate area, which is the only factor that changed
in the series of group #'s 4-6 (device #'s 9-10, il-12 and
13). It is significant that in this context, the field
5 strength necessary to achieve the same pulse rate fell by
1/3rd (a factor of 2.8x) as the area increased by a factor
of 2, in device #'s 11 and 12 versus 13 (group #'s 5 and
6). This strongly indicates that, in discharge tubes
enclosing a high vacuum obtained under oil diffusion
10 conditions and stimulated with high currents, the plate
area has a synergistic effect on PAGD production. The same
frequency of discrete, intense emission were obtained with
lower field strengths for the plasma discharges triggered
by these auto-electronic emissions. Consequently, in
15 discharge tubes 50, large plate areas promote PAGD behavior
at high vacua and at low field values not predicted by the
space charge theory.
~1413~9
Pcr ~ cA ~ ~ / a G 3 ~ ~
_~l_ l3 sc-~r«nnf,Q ~ s s~ W 3. Q s sa
TABLE 6
A B_ C D_ E_ F G H
No. L Flate Area hp d p J
(cm) (cm) (RL) (c~l~ (cm) (Torr) (A/ttz2)
1 16 4 1 64 1 3.6 2.8 * 3
106 *
105
2 32 4 1.41 128 2 5 2.0 * 1.5
106 *
105
3 32 8 1.41 256 2 7 1.4 * 7.5
106 *
10'~
4 16 4 1 64 1 5 2 * 106 3
*
105
32 4 1.41 128 ~ 2 5 2 * 106 1.5
*
105
6 32 8 1.41 256 2 5 2 * 106 7.5
*
10~
1_ ,I K L_ M N O P_
No.J Jlp2 J/p2 E E E!p E/p PPS
Exptl PredictedExptl PredictedExptl PredictedExptl
1 0 3.75 0 21,500 >97,2227.75 >1.2 *10120
* 1012 * 109
2 1.5 3.75 3.75 * NA 15.'80 NA 7.75 * 20
* 105 * 1012 101'- 109
3 ND 3.75 ND 10.320 ND 7.75 ND ~'D
* 1012 * 109
4 0 7.5 * 0 15,480 >70.0007.75 1.75 * 0
1012 * 109 1012
5 1.5 3.75 3.75 * NA 15,480 NA 7.75 * _'0
* 105 * 1012 1012 109
6 1.2 1.9 * 3 * 1012 15,480 5.600 7.75 2.8 * 20
* 105 1012 * 109 109
NA
=
Not
Applicable
ND
=
Not
Determined
~..d ~:. ~ , d' ~ . : ~.,:- .. .... ...o
2141309
WO 94/03918 PCT/CA93/003
42
A comparison of pulse counts at the axial probe 62
(see Figs. 2 and 4) in discharge tubes 50 and the pulse
counts at the cathode~5~'vshowed:that the axial probe 62
accurately reflects interelectrode events. This
correspondence was confirmed using oscillographic analysis
of the probe waveform, which showed it to be functionally
equivalent to that measured at the cathode 54. Figure 18
shows voltage and current oscillographs over time of AGD
self-generated auto-electronic emission pulses (at 10 pps)
in an H34 128 cmz discharge tube 50 (device #1), registered
as amplitude discontinuities between the anode 56/cathode
54 in Figure 18A, as current pulses at the cathode 54 in
Figure 18B and as dual polarity field direction reversals
of a split axial probe 62 in Figure 18C. Typically, for a
closed high vacuum discharge tube 50 with a plate area of
128 cmz and an interelectrode gap of 5 cm, a breakdown
potential of 668 volts, an average applied current of
500ma, and at 200pps, the pulse amplitude is more than 300
volts. Under rotary pumpdown conditions and for an
identical discharge tube, the pulse amplitude (encompassing
both positive and negative components, the latter being the
prominent value) increases with decreasing pressure, from
60 volts at about 0.5 Torr (with 5 mA DC) to greater than
300 volts at 0.008 Torr. In the closed high vacuum
discharge tube with H34 plates having an area of 128 cm2
(device #1), higher resolution oscillographs taken at the
axial probe 62, show that the negative component precedes
the positive reversal and has a typically higher amplitude
(140 V vs. 80 to 120 V, respectively, for this situation
shown in Figure 18C). Clearly, upon an abnormal glow
discharge pulse, the recovery of the field strength within
these discharge tubes overshoots a 'closed switch state'
(where the current I approaches zero) and results in a net
flow of positive charge past the probe, towards the cathode
(which is the floating ground reference level for these
measurements).
2141304
WO 94/03918 PCT/CA93/00311
43
Example 6
Effect of Capacitance on PAGD Rate
Using the same breakdown voltage of about 668 VDC,
the effect of varying the capacitance of the power supply,
set in parallel with the discharge tube (device #1), on the
frequency of PAGD production was determined while
maintaining all other variables constant (interelectrode
gap, plate area, applied voltage and current levels). The
linear regression in Fig. 19 shows that, under these
conditions, the PAGD frequency is increased by lower
capacitances. The log slope indicates that the pps rate is
doubled as the capacitance decreases by 2/3rds.
Measurements were also taken of the 'non-dynamic'
capacitances of discharge tubes with H34 aluminum plates
having different plate areas. These were insignificant
when compared with the parallel capacitances used in the
power supply in tests illustrated in Fig. 16, and were
observed to vary in accordance with the dielectric law, ie.
doubling the plate doubled the capacitance. Thus for a
plate area of 64 square cm, capacitance was lpF, for 128
sq. cm, 2.05pF, and for 256 sq. cm, 4.lpF.
Optimum Arrangement and Geometry
Prolonged operation of discharge tubes 50 has
provided some geometrical guidelines for promoting PAGD
production:
1) It is advantageous if the discharge does not
wander to the back of the cathode 54 and this is
facilitated by using a semi-cylindrical cathode in
cylindrical housings 52 and a flat cathode (rectangular,
square or circular) in parallelepiped-shaped housings 64
(see Figs. 3 & 5). However, interelectrode gap tests are
best done with flat plates which assure a homogeneous
potential. Moreover, the semicylindrical electrodes are
best made of hardened aluminum, at least 0.5 to 1 mm thick,
and this requires forming them to the right curvature,
given that foil alternatives are not resistant to the
deleterious effect of high-current PAGD transduction at
very high frequencies and do not withstand disruptive VAD
WO 94/03918 21413~~'9~ t PCT/CA93/003?
44
discharges. Nonetheless, a semi-cylindrical electrode
configuration in a housing 52 makes the sheaths (where
ionic recombination occurs during glow discharge) near the
electrodes and the housing wall coincide, and this can be
highly advantageous for sustaining PAGD production. The
same applies to flat plates in flat surface parallelepiped
housings 64.
2) The most effective axial probe 62 is either a
single half-length rigid rod or a pair of axial probes 62
separated at the center of the discharge tube 50 by a gap
of more than 1 cm, 4-6 cm being optimum. Whereas an axial
wire will perform satisfactorily as a probe 62, the rigid
rod has the advantage of not yielding to a direct
mechanical transduction of the electrodynamic force
effected upon it by the discharge or to force created by
the acquisition of a constant space charge. A split axial
probe 62 facilitates the exciter function and assures PAGD
operation by preventing a formation of a stable axial
space-charge at high-current operation.
3) A cooling coil (made of rubber, polymer, glass or
copper tubing) surrounding housing 52/64 is useful to
counterbalance the heating of the anode 56/cathode 54 which
promotes the production of semi-thermionic VAD channels and
even thermionic normal glow discharges. A coolant pipe
system that weaves through the plates can also be used for
this purpose, in which case flat plates are preferred.
4) Larger anode 56/cathode 54 surfaces are required
as the interelectrode gap is increased. And inversely,
larger anode 54/cathode 56 surfaces operate best if larger
interelectrode gaps are used; however, the breakdown
voltage also increases with larger interelectrode gaps.
5) One of the limitations of these discharge tubes
stems from their continuous operation at high applied
currents and from eventual slippage into the VAD regime,
both of which promote a deposit of sputtered metal atoms on
the inner walls of the housing 52, 64 thereby making them
conductive. In order to minimize this problem,
electromagnets may be wound longitudinally over the housing
~. .~ .
21413 ~~~~~:
WO 94/03918 ~ . PCT/CA93/00311
52, 64 (one at each end), to limit lateral dispersion of
the discharge vortices.
It is apparent that several factors affect PAGD
production namely: cold cathode work function, voltage,
5 current, parallel capacitance, gas fill, pressure, geometry
plate area and interelectrode gap distance. Except for
capacitance at the high end of the scale, each of these
factors affect the high and low limits of the PAGD, for any
given set of conditions. Heretofore, parameters such as
l0 plate area in vacuum tubes have not been previously
identified as factors which affect the breakdown field
values and the sustaining/extinction potentials of a glow
or an arc discharge. This suggests that the observed
auto-electronic field emission in the PAGD regime is a
15 function of physical factors which to date have been
unrecognized. It further suggests that field emission is
not a property exclusive to the VAD, ie. that it is also a
property of the pulsed operation of an abnormal glow
discharge in low to very high vacua.
20 The present discharge tubes 50 provide a design
capable of transducing high peak pulse currents at very low
field strength, over a wide range of frequencies with
minimal slippage of the PAGD operation into either the NGD
or the VAD regimes.
25 Although the examples described above utilise
discharge~tubes with symmetrical anode and cathode plates
and floating probe electrodes, many other arrangements are
possible. Thus the characteristics of the tubes may be
adjusted by connecting the probe (or probes) through a
30 capacitor to the anode or cathode to form an auxiliary
anode or cathode.
Since only the cathode need be have an extended
surface area, the probe of the tubes described may be
connected as an anode, with the plate electrodes connected
35 as either strapped or independent cathodes. Examples of
such connections are shown in Figures 21A and 21B, which
show how to incorporate a discharge tube operating in the
PAGD regime in an inverter circuit so that the pulse output
may be utilized by a remotely located alternating current
Wo 94/039 81413 0 g p~'/CA93/003'
46
device. The intermittency of the pulses produced by the
arrangements described above are not conducive to efficient
operation of conventional transformers, and a push-pull
circuit arrangement is preferred. While such an
arrangement could utilize two discharge tubes, an
advantageous arrangement utilizes a single tube of the type
described in the parent application, as shown in Figure
21A. In this instance, both plates 8a and 8b of the tube
act as cathodes and are connected to the diode 5, and the
l0 probe or auxiliary electrode, which is typically of
tungsten, acts as a common anode 9 and is connected to the
diode 6. The capacitors l0a and lOb are connected to
opposite ends of a centre-tapped primary winding of a
transformer 26, providing an alternating circuit output
through a secondary winding. The centre-tap of the primary
winding is connected to the electrode 9. The two halves of
the primary winding inductively couple the cathode circuits
in antiphase, thus synchronising the PAGD pulse trains
involving the two cathodes in antiphase.
In a modification of the circuit, shown in Figure
21B, the capacitors l0a and lOb are connected directly to
the electrode 9, and the primary of the transformer 26 is
connected directly between the two cathodes with its centre
tap connected to the diode 5. Whilst this arrangement
bears some superficial resemblance to known inverter
circuits employing VAD devices, it should be noted that the
circuit is completely self-commutating, and does not need
moving external magnetic fields to provide commutation as
in the prior art.
The electrodes themselves may be formed in various
configurations. Provided that both the anode and cathode
electrodes have sufficiently low impedance to sustain the
current densities associated with PAGD without rapid
deterioration or overheating, particularly of the cathode,
and provided that the cathode presents a surface of
extended area to the anode and is sufficiently separated
from it that the cathodic plasma eruptions associated with
AGD do not reach the anode to complete a VAD channel, the
electrode separation and cathode surface area can be varied
WO 94/03918 21413 ~ ~ p~/CA93/00311
47
over a wide range. Large cathode surface areas tend to
reduce the potential required to initiate AGD, and reduced
electrode separations increase the risk of entering the VAD
region. In practice the cathode area should usually be at
least 2 sq. cm and preferably at least 16 sq. cm, and the
electrode spacing should be at least 2cm and preferably at
least 3.5cm.
Particularly when the probe is used as an anode, it
may advantageously be formed as a wire grid or mesh
parallel to one or more plate electrodes acting as a
cathode or cathodes. Cathodes may be arranged on one or
both sides, or surounding a rod anode, or facing a point
anode.
Table 7 shows PAGD frequency results for various
electrode configurations which have been tested.
Configuration sd is a diode with plate electrodes and no
probe, and configuration sd* adds an unconnected axial
probe. In configuration t the probe was connected by a
capacitor to the cathode. Configurations ddl and dd2
use double diode configurations, with plate cathodes and an
intermediate anode, a rod in the first case and a plate in
the second case. Configuration cd used a cylindrical
cathode and an axial rod anode. These tests indicate that
an extended area of both the anode and cathode is
desirable, although the area of the cathode has a greater
influence.
214.1309
Pcr p ca ~ 3 / 0 0 3 ~ ~
_4g_ / 3 Sc~~'TEfrW 3cci? / 5 9,~ ~~,~. ~ s ~ i
TABLE 7
"Broken in" H34 cathodes (128 cmz plates for all configurations except cd; >1
x 106 pulses) at
0.8 Ton pressure, 6 cm distance between plates, 3 cm distance of plate to
axial or intercalated
member, when applicable. V - 540DC; Iav = 0.3A.
Configuration PPS Gap
( @ 1' of running in cm.
where n=30)
sd 10 _+ 2 6 cm.
sd* 6 _+ 3 6 cm.
t 19+4 6cm.
dd ( 65 11 3 cm.
1 )
dd (2) 88 7 3 cm.
cd 121 + 14 3 cm.
sd* = single plate diode with axial member not conected
cd° = cylindrical cathode area of 1,704 cm2
.,. o f ~ . .. d r e~ .:..
WO 94/03918 214 1 ~ ~ ~ PCT/CA93/00311
49
The anode is preferably formed of metal of a
relatively high work function, and the cathode of metal of
a comparatively low work function, although in many cases
the same material may be used for both as exemplified
above. In discharge tubes, where it is desired to be able
to reverse plate polarity,utilisation of the same metal for
both electrodes will be advantageous. For axial probes or
axial anodes, tungsten willgive good results. Hollow or
solid rods of the same metals selected for the cathodes may
also be employed. Axial rods of tungsten or other selected
metals may also be used as cathodes provided that they
present sufficient surface area to th anode.
The best cathode materials identified to date are
aluminum and its alloys, zinc, nickel, soft iron and
silver. Cathodes made of Copper and its alloys, and of
steel, support PAGD but are of much poorer performance.
In Table 8 sampled data from experiments performed
with different cathode metals (elemental or alloys) are
shown. Except for the first two entries in Table 8, which
utilized a perpendicular "surface-to-point " type
configuration (with the "point" being the lateral area of a
tungsten rod utilized as a cathode), all the other entries
were obtained with single diode configurations, utilizing
parallel plates at various gap distances. Most of the
column headings are self-explanatory, but it should be
noted that the voltages shown are breakdown values which,
when indicated, are also minimal breakdown voltages for the
discharge type shown.
As exemplified in entries 1 and 2, for two different
Argon pressures, tungsten cathodes can support the same
cold-cathode type of low-field spontaneous emission
responsible for the PAGD regime which we have identified in
diverse aluminum cathodes. This matches the previous
observation of such discontinuous emissions being aided by
an axial tungsten member when discharge tubes with aluminum
plates are connected in the triode configuration. A higher
potential (750 VDC), however, is needed with this
configuration and tungsten cathodes, than that required by
2141308
WO 94/03918 PCT/CA93/0031
the lowest area aluminum plates tested (cp entries 1 and 2
with 3-7).
In entries 3 to 12, small (4cm2) H34 aluminum plates
are compared for different gap distances (3 vs. 9 cm), and
5 within each group (entries 3 to 7, and entries 8 to 12) for
varying pressure and nature of the residual gas. The net
effect of increasing the gap distance is to increase the
potential needed for electrical breakdown of the vacuum (cp
VDC values for entries 4-7 with 9-12), as expected from
10 Paschen's law. A trend for higher PAGD input currents, as
a function of the higher vacua, is also apparent throughout
this group.
In entries 13 to 28, wider aluminum cathodes (16cm2)
were compared at two different gap distances (2 and 4 cm,
15 respectively, for entries 13 to 20 and 21 to 28), for
different pressures in air or argon atmospheres. At the
same specified battery breakdown voltages utilized, the
shorter gap produces much higher PAGD frequencies (100 PPS
vs 30-52 at 0.8T Argon, respectively, entry 14 and entries
20 25 and 28). The shorter gap also promotes the setting in of
a vacuum arc discharge, when compared to the larger gap at
the same voltage and potential (cp entry 16 with entry 26).
Lastly, it is apparent that the PAGD frequency equally
increases in the presence of argon (entries 24-26) with
25 respect to air (entries 21-23), and with increasing input
current (see entries 17 to 19) within the current range
characteristic of the PAGD at the same pressure (1 Torr).
In entries 29 to 35, and 75 to 76, utilization of
brass cathodes is examined for purposes of PAGD production.
30 Within the PAGD current and voltage ranges determined for
aluminum and other metals, brass cathodes perform poorly,
with low or very low pulsation frequencies (entries 30, 31
and 76) and very erratic bursts of activity (entries 32, 34
and 35). Unlike the typically single aluminum and tungsten
35 PAGD emission foci, brass presents multiple small cathode
spot localizations in the same pulse. The pulsed emission,
however, like those of aluminum, follows the same cyclic
path of abnormal glow saturation, focusing of the discharge
WO 94/03918 ~ ~ ~ ~ PCT/CA93/00311
51
at the emission foci, and subsequent collapse of the
saturated glow.
Results for bronze-aluminum alloy cathodes are shown
in entries 36 to 41, and they indicate that this alloy does
not perform as satisfactorily as aluminum for PAGD
production , but it is certainly utilizable. However, like
brass, in the presence of Argon, erratic bursts of emission
pulses are also observed (cf entry 41), without the
development of a VAD-type regime.
Iron, nickel and zinc cathodes, tested in the
following sequences (respectively entries 42-50, 51-53, and
54-60), proved to be amongst the best cathodes for PAGD
production this purpose. Quasi-regular high PAGD
frequencies are possible utilizing these metals (cp entry
49 & entry 60), which appeared to perform best with argon
rather than air. Iron plates seem to eject the least metal
and can sustain very regular frequencies of pulsed abnormal
glow discharges. Zinc cathodes ejected the most metal and
most easily slipped into a vacuum arc discharge having the
aspect of a meandering flame surrounding the cathodes
spots. With zinc cathodes, the VAD regime would onset in
air at pressure and current values characteristic of the
PAGD regime in Argon (cp entry 55 with entry 60), the
window of the transition between the NGD and the VAD being
rather narrow or absent. Even in Argon, this window
remained relatively narrow in terms of its pressure range.
Tubes for pulse generators were also built with
apposed cylindrical section electrodes made of silver
nitrate directly coated onto the glass inner surface, and
having a cross- section fundamentally identical to that of
Fig. 4. Because of this geometry, the gap distance given
is the average distance between the center and the
extremities of the cylindrical electrodes. Experiments with
such a device showed that silver in greater thickness would
also form a suitable cathode to support PAGD production
(entries 64 and 65). The emission foci in silver form
single cathode spots for each pulse generated, as with
aluminum, tungsten, zinc, nickel and iron cathodes;
however, these intermittent silver cathode emitters travel
2141349
WO 94/03918 PCT/CA93/003'
52
laterally to form quasi-continuous tracks of ejected
cathode material.
Lastly, 64 cm2 wide copper and aluminum plates were
compared, in entries 66-74 and 61-63, respectively. In the
presence of air, under the same pressure, applied voltage
and current conditions, copper cathodes do not support PAGD
production (unless triggered by external proximity of
high-frequency spark gaps, eg a Tesla coil, or a moving
static charge, or a moving magnetic field), whereas
aluminum cathodes do (cp entries 61 and 66). However, in
argon (entries 68, 70-72) or helium (entries 73-74)
atmospheres, copper cathodes readily supported PAGD
production, though they required greater input currents to
attain about half the PAGD frequency observed using
aluminum plates. The PAGD region is also particularly
narrow with the copper cathodes, when the voltage and
current are at threshold levels needed for eliciting the
regime (cf entries 68 and 69). It is easy to see the
slowing down of the PAGD frequency in these cathodes,
apparently due to the rapid heating of their surface (also
observed with brass and bronze alloys): the cycle of onset
of an abnormal cathode glow, followed by a localized
cathode eruption, then by a total or partial collapse of
the abnormal glow onto the electrodes and its subsequent
re- instatement, slows down progressively as the setting in
of the cathode glow becomes more intense in luminosity and
finally ceases giving way to the auto-electronic emission.
A semi-thermionic abnormal glow discharge then sets in.
This transition of the PAGD regime to a semi-thermionic AGD
regime, in tubes using copper cathodes, is all the more
prevalent as higher frequencies and higher currents are
employed to stimulate the pulse generator, and it may
explain the observed erratic behaviour of the
copper-containing alloys, brass and bronze. Finally, the
copper cathode emissions accompanying a single pulse were
not single but multiple, and clustered in a neighbourhood,
as was also observed in brass and bronze-aluminum alloys.
All the above PAGD cathode emitters that had
reasonable characteristics (regular or quasi-regular
2141300
WO 94/03918 PCT/CA93/00311
53
spontaneous pulse discharges) presented a bell-like
distribution for the discharge frequency, with the higher
vacua beyond a given pressure having the effect of
decreasing the frequency of the PAGD emission, while
increasing the input and output peak currents as well as
the cathode voltage drop of each pulse. Lastly, caesium
emitters were also employed at these input currents to
support PAGD production successfully. Other metals
considered promising are bismuth, cadmium and antimony.
2141309
WO 94/03918 PCT/CA93/00'~z
4.'
,. -TAHLE 8
Cathode Area Gas Pressure PPS Discharge gap VDC DCA
Material cm2 in Torr type in
cm
1 Tungsten 0.5 Argoa 0.8 4 PAGD 4 750 0.2
min.
2 Tungsten 0.5 Argon 0.5 18 PAGD 4 750 0.3
3 Aluminum 4 Air 0.9 1 PAGD 3 564 0.2
(H34)
4 Aluminum 4 Argon 0.9 2 PAGD 3 564 0.3
(H34)
5 Aluminum 4 Argon 0.7 1.5 PAGD 3 564 0.4
(H34)
6 Aluminum 4 Argon 0.4 12 PAGD 3 564 0.45
(H34)
7 Aluminum 4 Argon 0.3 0.5 PAGD 3 564 0.5
(H34)
8 Alumiaum 4 Argon 4-0.01 O None# 9 564 O
(H34)
9 Aluminum 4 Argon 0.9 3.5 PAGD 9 850 0.4
(H34) min.
Aluminum 4 Argon 0.~4 20 PAGD 9 900 0.45
(H34)
min.
11 Aluminum 4 Argon 0.2 2 PAGD 9 950 0.5
(H34) min
12 Aluminum 4 Argon 0.1 1 PAGD 9 950 0.5
(H34)
min.
13 Aluminum 16 Argon 2.0 50 PAGD 2 560 0.8
(H34)
14 Aluminum 16 Argon 0.8 100 PAGD 2 560 1.2
(H34)
Aluminum 16 Argon 0.3 10 PAGD 2 560 0.4
(H34)
16 Aluminum 16 Argon 0.3 erratic VAD# 2 560 1.1
(H34)
17 Aluminum 16 Argon 1.0 20 PAGD 2 260 0.05
(H34) min.
18 Aluminum 16 Argon 1.0 100 PAGD 2 260 0.7
(H34)
min.
19 Aluminum 16 Argon 1.0 185 PAGD 2 260 1.6
(H34) min.
Aluminum 16 Argon 0.1 2.5 PAGD 2 350 0.03
(H34)
21 Aluminum 16 Air 0.8 0.4 PAGD 4 560 0.3
(H34)
WO 94/03918 21413 ~ ~ PCT/CA93/00311
22 Aluminum 16 Air 0.5 20 PAGD 4 560 0.5
(H34)
23 Aluminum 16 Air 0.3 1.2 PAGD 4 560 0.5
(H34)
24 Aluminum 16 Argoa 1.0 1.2 PAGD 4 560 0.5
(H34)
25 Aluminum 16 Argon 0.8 30 PAGD 4 560 0.6
(834)
26 Aluminum 16 Argon 0.3 8 PAGD 4 560 0.7
(H34)
27 Aluminum 16 Argon 0.8 10 PAGD 4 550 0.2
(H34)
28 Aluminum 16 Argoa 0.8 52 PAGD 4 550 1.0
(H34)
29 Brass 16 Air 2-0.6 O None * 4 560 O
30 Brass 16 Air O.OI 0.01 PAGD 4 560 ND
31 Brass 16 Air 0.3 0.4 PAGD 4 560 0.3
32 Brass 16 Air 0.2 burstsPAGD # 4 720 ND
33 Brass 16 Argon 2-0.3 O None * 4 560 O
34 Brass 16 Argon 0.9 burstsPAGD # 4 630 0.5
mia.
35 Brass 16 Argon 0.8 -20 PAGD 4 750 0.7
erratic
36 Bronze-
Aluminum 16 Argon 1 O None 4 568 O
37 Bronze-
Aluminum 16 Argon 0.5 0.4 PAGD 4 568 0.3
38 Bronze-
Aluminum 16 Argon 0.3 0.6 PAGD 4 568 0.3
39 Bronze-
Aluminum 16 Argon 0.8 1 PAGD # 4 570 0.35
40 Bronze-
Aluminum 16 Argon 0.8 2.8 PAGD 4 750 0.4
41 Bronze-
Aluminum 16 Air 0.3 5-15 PAGD # 4 750 0.5
42 Iron 16 Air 0.3/0.8 O PAGD 4 560 O
43 Iron 16 Air 0.25/0.18 0.2 PAGD 4 560 0.3
44 Lron 16 Argon 0.8 0.12 PAGD 4 560 0.3
45 Iron 16 Argon 0.6 0.4 PAGD 4 560 ND
46 Iron 16 Argon 0.4 o PAGD 4 560 O
47 Iron 16 Argon 2.0 3 PAGD 4 720 0.4
48 Iron 16 Argon 1.0 1 PAGD 4 750 0.3
2141309
WO 94/03918 PCT/CA93/003°
56
49Iron 16 Argon 1.0 26 PAGD 4 950 1.05
50Iron 16 Argon 0.8 0.1 PAGD 4 720 0.8
51Nickel 16 Argoa 2 1 PAGD 9 1000 0.2
52Nickel 16 Argon 1 2 PAGD 9 950 0.3
53Nickel 16 Argon 0.5 0.5 PAGD 9 1000 0.2
54Ziac 16 Air 2-0.8 O VAD 4 564 1.6-
1.3
55Zinc 16 Air 2-0.8 O VAD 4 564 0.8
56Zinc 16 Air 0.8-0.2 O th. AGD 4 564 0.03
57Ziac 16 Argon 2-0.8 O VAD 4 564 1.8-
1.2
58Zinc 16 Argon 0.9-0.2 O th.AGD 4 560 0.1
/1 000
59Zinc 16 Argon 1.0- 1 PAGD 4 560 0.3
60Zinc 16 Argon 1.0 23 PAGD 4 950 1.0
dlAluminum 64 Air 0.8 5 PAGD 5.5 560 0.4
(H34)
-2Aluminum 64 Argon 0.8 12 PAGD 5.5 560 0.5
(H34)
63Aluminum 64 Argon 0.8 62 PAGD 5.5 540 0.9
(H34)
64Silver 64 Argon 0.8 10 PAGD 5.5 560 0.2
65Silver 64 Argon 0.2 1.2 PAGD 5.5 560 0.7
66Copper 64 Air 4-0.01 O None* 5.5 560 ND
67Copper 64 Argon 4-0.4 O None* 5.5 560 ND
68Copper 64 Argon 0.3 0.44 PAGD# 5.5 560 0.5
69Copper 64 Argon 0.25 O Th. AGD 5.5 560 O
70Copper 64 Argon 0.8 5 PAGD 5.5 580 0.2
min.
71Copper 64 Argon 0.8 33 PAGD 5.5 580 0.9
72Copper 64 Argon 0.1 1 PAGD 5.5 900 0.5
min.
73Copper 64 Helium 0.9 O None* 5.5 560 O
74Copper 64 Helium 0.2 0.5 PAGD 5.5 560 0.7
min.
75Brass 64 Argon 4-0.6 O None* 4 560 O
76Brass 64 Argon 0.5 0.2 PAGD 4 560 0.3
* Single pulsesof dischargecould elicitedby
abnormal be moving
glow static
charges or proximity frequencyalternating
the of currents.
high
# PAGD production erratic.
was
T
214130
WO 94/03918 PCT/CA93/00311
57
Shaping of the cathode is not critical, although it
should be such as to provide a reasonably uniform distance
between different parts of its surface and the anode.
Cylindrical or multiple part cylindrical electrodes may
conveniently be used with rod anodes, or a flat cathode when
the anode is a flat plate or mesh, or the electrode
separation is large. Walls of the discharge tube should be
sufficiently spaced from the electrodes that metal sputtered
from the cathode does not build up a path for arcing or act
1o as an auxiliary electrode modifying the characteristics of
the tube.
The static and dynamic characteristics of the
external circuit which cooperates with the tube must be such
as to prevent the plasma eruptions from the cathode during
1s PAGD from acquiring sufficient energy to reach the anode and
forming a continuous VAD channel. This involves inter alia
the physical dimensions of the tube, its gas content, the
impedance of the supply and any associated ballast resistor,
and the energy storage capacity of any reactive components
2o in the pulse generator circuit, such that the potential
across the tube falls below the AGD extinction potential
before a plasma channel to the anode is established.
Although it may be possible to acieve this with a very small
cathode area or small interelectrode gap, the amounts of
2s energy involved in creating the plasma eruptions and
released by their collapse will likely become too small to
be useful.
In conclusion, we have developed a series of pulse
generators exploiting the abnormal glow discharge regime, as
3o well as a series of low to very high vacuum discharge tubes
which support the production of PAGDs. In testing these
devices we have shown that:
- the low field strengths and typical low emission
current densities observed in the PAGD regime are not
3s predicted by any existing field emission or space-charge
theories;
214130
WO 94/03918 PCT/CA93/003~
58
- the PAGD regime'v.responds asymmetrically to the
polarity of the applied voltage at high applied currents;
- at low applied currents, the PAGD pulse rate
increases with the applied voltage and the current up to an
observed plateau;
-at mid to high applied currents, the PAGD pulse rate
increases with an increase in current and with a lowering of
the extinction potential;
- the PAGD pulse rate also varies with the
1o composition of the cathode material (the pulse rate is
promoted by materials having a low work function) and
increases with a decrease in pressure, during pumpdown, to a
maximal peak rate, thereafter either diminishing to the
point at which the discharge extinguishes or gives way to
1s x-ray production (depending on the magnitude of the applied
potential);
- larger area plates lower the field strength values
needed to elicit comparable PAGD production, displace the
PAGD region downward in the pressure scale and increase the
2o peak PAGD rate;
- higher power supply capacitances slow down the PAGD
rate.
Exploitation of PAGD permits the production of highly
efficient pulse generators for the production of
2s endogenously generated abnormal glow discharge pulses
triggered by intense, cathodic auto-electronic emissions
under conditions of a constantly applied DC potential.
These pulse generators have diverse industrial
applications; directly, they may be used as stroboscopic
so light sources, for vacuum deposition of cathode materials or
cathode coatings (eg. polymer deposition or aluminum
mirroring of target surfaces), detection of ionizing
radiation fields, or electrostatic and electromagnetic
proximity fields, high power noise-signal generation,
35 destructive component testing (transient response) or
destructive testing of-materials in vacua (eg. insulations),
high frequency medium voltage power supply applications
(switching supplies and inverters), as an oscillator or as
T ~ ..
WO 94/03918 ~ ~ ~ ~ ~ ~ ~ .T/CA93/00311
59
part of a pulse forming network. Indirectly, they may be
used for laser pulsing, flash tube pulsing or for research
(eg. chemical reaction triggering) and industrial switching
applications.
