Note: Descriptions are shown in the official language in which they were submitted.
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
1 ELECTRIC ARC MONITORING SYSTEMS
2 Technical Field
3 The technical field of the invention includes methods and
4 apparatus for monitoring, detecting, indicating, evaluating and
signaling electric arcs or sparks.
6 Background
7 The chaotic electromagnetic emanations manifesting
8 themselves as electric arcs or sparks are closely linked to
9 matter, wherein electromagnetic interactions bind electrons to
nuclei in atoms and molecules and wherein the fundamental unit
1l of electromagnetic r~~diation is the photon.
12 Indeed, spectra of electric arcs and sparks extend
13 practically from DC through the entire radio-frequency spectrum
14 and through microwaves, infrared and light spectra.
Useful exploitations of the electric arc and spark
16 phenomenon include the electric arc lamp, electric weldi~xg, the
17 electric-arc-type of metallurgical furnace, the arc t.ppe of ion
18 generator in satellite thrusters and for propulsion in outer
19 space, the spark-plug-type of ignition in internal combustion
engines, and electric spark ignition in gas appliance;3.
21 Unfortunately, the same quality of the electric arc oc spark
22 that led to electric lighting, electric arc welding and
23 metallurgy, and ignition of internal combustion, has catastrophic
24 effects in electrical faults that cause explosions or devastating
fires through chaotic:: arcing or sparking.
26 By way of example, electric arc monitors would be useful in
27 garages, automobile or motorcar repair facilities, gasoline
28 (British "petrol" ) storage or dispensing facilities an~i in other
29 areas where accidental electric arcing can cause disastrous
explosions.
31 Moreover, fuses and circuit breakers are c~ipable of
32 preventing serious overload conditions, but they are generally
33 ineffective to prevent electrical fires and other d~maQe from
CA 02298789 2003-08-08
2
1 accidental arcs and sparks which typically generate enough heat for a fire
at electric current
2 levels below the level at which the fuse will blow or the circuit breaker
will trip. Reliable arc
3 monitoring would thus be highly desirable in a large number and variety of
electrical circuits.
4 These are, of course, only representative examples of fields where reliable
arc or
spark monitoring could be useful.
6 A major stagnating problem in this respect has been that prior-art
development has
7 run its course in its fear of false alarms. Of course, false alarms are the
bane of alarm
8 systems, as frequent occurrence of false alarms can nullify the utility of
an alarm system.
9 Accordingly, in an effort to reduce the possibility of false alarms arising
from radio
broadcast and radio frequency security system signals, the arc detection
system as disclosed
11 in the International Patent Publication W090/04278, by HAMPSHIRE, Michael
John, rejects
12 frequencies below about 160 kHz and above some 180 kHz of the arc signal
signature,
13 leaving for electrical fault detection only a narrow 20 kHz band at some
170 kHz center
14 frequency. This, however, left a sample for arc detection that was dozens
of times too small
1 S in the 100 kHz range for reliably detecting the occurrence of an arc
signature while at the
16 same time preventing the occurrence of false alarms equally reliably.
17 An arc detection system which avoids that drawback is apparent from
18 PCT/US90/06113, filed 24 October 1990 and published as W092/08143, by
Hendry
19 Mechanical Works, inventors HAM, Jr., Howard M., and KEENAN, James J., and
in its
corresponding US Patents 5,373,241, issued 13 December 1994, and 5,477,150,
issued 19
21 December 1995. Reference should also be had to their corresponding EPO 507
782
22 (90917578.8) and resulting European national patents, and to their
corresponding Australian
23 Patent 656128, Canadian Patent 2,093,420, Chinese Patent 1077031, Japanese
Patent
24 6504113T, Korean Patent 212020, and Mexican Patent 178914 (9201530). That
system
avoids false alarms by converting instantaneous arc signature frequencies into
a combination
26 frequency from which arc-indicative signals are detected in
contradistinction to extraneous
27 narrow-band signals that could cause false alarms.
28 Against this background, a frequency selective arc detection system appears
as a
29 typical representative of the prior -art approach to arc detection. It
accordingly presents a
variety of approaches to arc detection that mainly look at frequencies in the
upper kilohertz
31 range, such as from 100kHz to one megahertz. This, however, covers not only
major
32 portions of the public A.M. radio broadcast band, also known as "longwave"
and "medium-
33 wave" broadcast bands in some countries, but also the
CA 02298789 2003-05-05
1
1 kind of control or security systenos radio frequency barut refewed to in the
above mentioned
2 W090/04278 reference. Depending on location, one thus had to contend with
dozens of
3 extraneous signal interferences.
4 The same in essence applies to another embodiment in that prior-art proposal
that
suggests using a comb filter ar7-an~emont campo;~ed of four handpass tilters
each of which
O has a 50 kHz passband, and thref; csi' which have ~a c.mter l~reep~ncy
oi~225 kHz, 525 kHz, and
7 825 kHz, respectively. In the A.IL9. broadcast arid above mentioned control
and security
8 systems radio frequency hand pardon of that spccarum, 50 kHtz samples can
only represent
9 minor fragments of the chaotic arc; signature, raising the danger of false
alarms from
coincidental extraneous signals. hhis also affecrs the el~fiecrcy of the ~5 kI
Iz bandpass, filter in
1 1 that comb tilter awangement, inasmuch as that prior-art pr-<>posal
continuously rotates its
I 2 detectic>n process among the four filter components of that comb filter
arrangement.
13 A prior effort at arc detectic>n that ventured into low Frequency regions
effected
14 monitoring in various low frequc;ncy bands that were too narrow for
reliable arc detection as
I S apparent from articles by B. D. h:ussell et al., cntitl~°d "An
Arcing FaGrlt Detection ~Tec;hnique
l6 Using I.ow Frequency Current ( on~ponents- Performance lvaluation lJsing
Recorded Field
17 Data " and "Behaviour
Ih
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
4
1 of Low Frequency Spectra During Arcing Fault and Switching
2 Events" (IEEE Transactions on Power Deliverlr, Vol. 3, No. 4,
3 October 1988, pp. 1485 - 1500) indicating lack of success.
4 These developments in retrospect appear largely as a
reaction to the perception of electric arcs as highly random
6 phenomena borne out of the chaotic nature of arc signatures.
7 This prior-art perception, however, ignores the fact that chaotic
8 systems have a deterministic quality, and can be successfully
9 dealt with, if one is able to discover What the underlying
principles are and how they can be put to effective use.
11 Indeed, even chaotic electric lightning displays some self-
12 similarity among its arboresque nocturnal discharges and within
13 the branched configuration of its lightning bolts.
14 In this respect, pioneering work done by Benjamin Franklin
and by Georg Christoph Lichtenberg back in the 18th Century casts
16 a long shadow all the way to the subject invention.
17 Ia particular, Franklin through his famous kite experiment
18 in a thunderstorm proved that lightning is an electrical
19 phenomenon. Lichtenberg thereafter created his famous
~Lichteaberg figures" in 1777 by dusting fine powder, such as
21 sulfur, over insulating surfaces over which electrical discharges
22 had taken place. Many of these Lichtenberg figures of electrical
23 discharge resemble lightning in appearance and otherwise display
24 a striking self-similarity in their patterns of branching lines
and within such branching lines themselves. Manfred Schroeder
26 compared this to diffusion-limited aggregation (DLA) in his book
27 entitled "FRACTALS, CHAOS, POWER LAWS" (W. H. Freeman and Company,
28 1991), pp. 196, 197, 215 and 216. Kenneth Falconer, in his book
29 entitled "FRACTAL GEOMETRY" (John Wiley & Sons, 1990), pp. 270
to 273, also applied the DLA model to electrical discharges in
31 gas.
32 By way of background, fractals are phenomena in the fractal
33 geometry conceived, named and first explained by Benoit
34 Maadelbrot in 1975. Fractal geometry in eff~:ct is a
manifestation of the fact that the natural world does not coafornn
36 to an Euclidean type of geometry. Euclidean geometry is based
37 on characteristic sizes and scaling. The natural world is not
*rB
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
1 limited to specific size or scaling. Euclidean geometry suits
2 man-made objects, but cannot realistically express natural
3 configurations. Euclidean geometry is described by formulas,
4 whereas the mathematical language of natural phenomena is
5 recursive algorithms.
6 Such recursiveness is an expression of~nature throughout
7 destructive if not chaotic influences, manifesting itself, for
8 instance, in a persistent invariance against changes in size and
9 scaling, called self-similarity or self-affinity. Fractals are
self-similar in that each of various small portions of a fractal
11 represents a miniature replica of the whole. Such small portions
12 are herein called "fractal subsets".
13 Electric arc or spark monitoring generally addresses itself
14 to so-called arc signatures which are part of the electromagnetic
spectrum of arcs or sparks situated in frequency bands way below
16 light, heat radiation and microwave spectra.
17 Problems in this area include false alarms from mutual
18 induction among neighboring monitored circuits. In this respect,
19 reference may be had to a standard equation for mutual induction,
such as between a monitored circuit in which an arc is occurring,
21 and a neighboring monitored circuit in which no arc is occurring
22 at the time:
23
24 In -_ 2~rfMI=./Z~ (1)
26 wherein: I" = arc signature current flowing in the monitored
27 circuit where an electric arc occurs at the
2 8 mome:a t ,
29 In = current induced by the arc signature in a moni-
tored neighboring circuit where no arc has
31 occurred at the moment,
32 M - mutual inductance,
33 Zn = impedance of said neighboring circuit, and
34 f - frequency.
36 As between neighboring circuits, the current I" inct~uced in
37 a neighboring monitored circuit by current I" flowing in the
CA 02298789 2000-O1-28
WO 99109424 PCT/US97/14497
6
1 monitored circuit where an arc is occurring, decreases with
2 decreasing frequency of that primary current I". However,
3 electromagnetic arc signatures are characterized by a special
4 shape approximating an inverse frequency (1/f) progression of
their amplitude. If this is put into the above Equation (i) one
6 gets
7
$ In = (2'~fMI"/f) /Zn (2)
9 in which "f" would cancel out, so that one gets
11 I" = 2rMI"/Z" ( 3
12
13 that is, a mutual inductance and a secondary current, I", that
14 are independent of frequency. Such considerations have led to
the prior-art conclusion that lowering the frequency of arc
16 signature bands in which arcs are monitored would not effectively
17 reduce cross-induction and false arc alarms therefrom.
CA 02298789 2000-O1-28
WO 99/09424 PCT/IJS97/14497
7
1 Summary of Invention
2 It is a general object of the invention to provide improved
3 electric arc monitoring systems that employ novel circuitry
4 and/or take advantage of properties of electric arc signatures
not heretofore utilized.
6 It is a related object of embodiments of the invention to
7 permit reliable arc monitoring at distances from electric arcs
8 longer and with less cross-talk or induction than heretofore.
9 In this respect and in general, the expression "monitoring"
is herein used in a broad sense, including monitoring, detecting,
11 indicating, evaluating and/or signaling electric arcs or sparks,
12 whereas the word "arc" is herein used generically to cover
13 electric arcs and sparks interchangeably as being essentially the
14 same phenomenon.
From one aspect thereof, the subject invention exploits the
16 discovery that electric arcs are fractal phenomena not only in
17 the visible luminous portion of their electromagnetic radiation,
18 as heretofore thought, but in fact are fractal phenomena all the
19 way down to the extremely low frequency band of their
electromagnetic emanation into space or along wires of the
21 circuit where the particular arc occurs. Since all essential
22 information that signifies "arc" is thus contained in each
23 fractal subset, it is sufficient for arc monitoring purposes to
24 monitor a fractai subset of the arc's electromagnetic emanation.
The realization according to the subject invention that the
26 fractai nature of the arc is not limited to its visible region,
27 but in fact extends all the way down to a few cycles per second
28 of its signature, adds to the previously known characteristics
29 of electric arcs at least one fundamental characteristic and at
least one criterion; namely, that:
31 1. All essential information for effective eleci~ric
32 arc monitoring is contained in any fractal subset
33 of the arc signature; whereby
34 2. the selection of the monitoring frequency band for
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
8
1 each purpose is liberated from prior-art con-
2 straints and can truly be the result of an optimum
3 tradeoff in sensitivity, speed of detection,
4 prevention or rejection of false signals, desired
length of travel and mode of transmission of the
6 arc signature from the arc to the monitoring
7 circuit in different environments.
8 Pursuant to these principles, the subject invention resides
9 in a system of monitoring an electric arc having an arc signature
typified by a wideband range of frequencies of a chaotic nature
11 in a monitored circuit, and, more specifically, resides in
12 selecting a fractal subset of the arc signature characterized by
13 relatively long travel along the monitored cir:;uit, and
14 monitoring that fractal subset of the arc signature.
The expression "relatively" in this context refers to the
16 fact that the length of possible travel of the arc signal is
17 inversely proportional to the frequency of the arc fiignature.
18 In this respect, reference may be had to the familiar algebraic
19 equation for electric current:
I = E/ [R2 + (2~rfL - 1/2nfC) z] ~' (4 )
21 wherein: I = electric current,
22 E = voltage or potential,
23 R = resistance,
24 f = frequency,
L = inductance, and
26 C = capacitance of the electric circuit.
27 From this equation a related benefit of an embodiment of the
28 invention can be seen; namely, that a selection of t:he lowest
29 frequency or longest wavelength fractal in effect amounts to a
selection of the longest survivor of the different fractals of
31 the arc signature traveling along the monitored circuit. Up to
32 a point, one can say that the monitored circuit itself thus
33 performs the function of a low-pass filter for the arc detection
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
9
1 monitor. Accordingly, embodiments of the invention permit arc
2 monitoring at considerable distances from the occurrence of arcs
3 in the circuit, which is useful in practice for several reasons,
4 including the capability of surveying large circuits, and the
convenience of providing central arc detection monitoring
6 stations for several different circuits.
7 At any rate, at low arc signature frequencies, the possible
8 travel of the arc signal along the monitored circuit is long,
9 relative to higher arc signature frequencies.
It also turns out that false alarms from mutual induction
11 among neighboring monitored circuits is lowest at low arc
12 signature frequencies, quite contrary to what the prior art would
13 have indicated pursuant to Equations (1) to (3) set forth above,
14 wherein the frequency factor "f" in the denominator would cancel
out the "f" in the numerator in Equation (2).
16 However, the fallacy of that conclusion becomes apparent if
17 certain possible radiation effects are considered. In this
18 respect, it is well known that ~/2 and ~/4 antennas constitute
19 excellent Hertzian and Marconi-type electromagnetic radiators.
The wiring in many~telephone exchange, electric power supply and
21 other installations in effect often forms such antennas at the
22 kind of radio frequencies selected by the prior art for electric
23 arc detection purposes. Even where the length of some wiring in
24 an installation in insufficient to constitute a quarter-wave-
length antenna, certain reactances in the circuit can provide the
26 lumped-impedance kind of tuning or "loading" that renders even
27 relatively short conductors effective radiators.
28 In consequence, picked-up electromagnetic arc signatures are
29 transmitted among neighboring circuits, resulting in false
alarms, unless the segment of the arc signature picked up for
31 monitoring is of a very low frequency (VLF) according to one
32 aspect of the invention.
33 Accordingly, lower frequency fractals pursuant to
34 embodiments of the invention induce less spurious signals through
cross-induction in neighboring circuits than arc signatures
36 having higher frequencies. Low frequency fractals more
37 effectively avoid false alarms from mutual inductance among
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
1 neighboring circuits than arc signatures at higher frequencies.
2 In consequence, embodiments of the invention not anly permit
3 arc monitoring at considerable distances from the occurrence of
4 arcs in a monitored circuit, but also avoid false alarms in
5 aeighboring monitored circuits.
6 According to a related embodiment of the invention, the
7 electric arc is detected from a fractal subset of the arc
8 signature at frequencies below 30 kHz. According to The New IEEE
9 Staadard Dictionary of Electrical and Electronics Terms, Fifth
10 Edition (The Institute of Electrical and Electronics Engineers,
11 1993), this is the upper limit of the very low frequency (VLF)
12 band.
I3 A presently preferred embodiment of the invention restricts
I4 fractal subsets from which the electric arc is detected to the
ELF (extremely low frequency) band which in that IEEE Standard
16 Dictionary is defined as extending from 3 Hz to 3 kHz.
17 Another embodiment of the invention restricts monitored
18 fractals to arc signature frequencies below the voice frequency
19 band (vf) defined in that IEEE Standard Dictionary as extending
from 200 Hz to 3500 Hz.
21 In that vein, a further embodiment of the invention
22 restricts monitored fractal subsets to arc signature frequencies
23 below a first harmonic of a standard line frequency in
24 alternating-current power supply systems.
An embodiment of the invention even selects the monitored
26 arc signature fractal subset from a frequency band on the order
27 of a standard line frequency in alternating-current power supply
28 systems.
29 According to a related aspect of the invention, an apparatus
for monitoring an electric arc having an arc signaturA typified
31 by a wideband range of frequencies of a chaotic nature in a
32 monitored circuit, comprises, in combination, an electric filter
33 having an input coupled to that arc, having a passband
34 corresponding to a fractal subset of the arc signature
characterized by relatively long travel along the monitored
36 circuit, and having an output for that fractal subset of arc
37 signature. Such ap~raratus includes a chaotic wideband signal
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
11
1 detector having a detector input for that fractal subset of the
2 arc signature coupled to the output of the electric fa.ltEr.
3 From another aspect thereof, the invention resides in a
4 method of monitoring an electric arc having-an arc signature
extending over a wideband range of frequencies of a chaotic
6 nature in a monitored circuit. The invention according to this
7 aspect resides, more specifically, in the improvement comprising,
8 in combiaation, processing portions of the arc signature in two
9 paths out of phase with each other, and monitoring thc~ electric
arc from such out of phase portions of the arc signature.
11 From a related aspect thereof, the invention wesides in
12 apparatus for monitoring an electric arc having an arc signature
13 typified by a wideband range of frequencies of a chaotic nature
14 in a monitored circuit. The invention according to this aspect
resides, more specifically, in the improvement comprising, in
16 combination, an electric filter having an input coupled to the
17 arc. having a passband corresponding to portions of the arc
18 signature, and havi~~g an output for such portions of arc
19 signature, an invertvng amplifier having an input connected to
the output of the electric filter, and having an amplifier
21 output, a non-invert:i.ng amplifier having an input connected to
22 the output of the electric filter, and having an amplifier
23 output, and a chaotic wideband signal detector having :,. detector
24 input coupled to the amplifier outputs of the inverting and non-
inverting amplifiers.
26 From another aspect thereof, the invention res:~des in a
27 method of monitoring an electric arc having an arc signature
28 extending over a wideband range of frequencies of a chaotic
29 nature in a monitored circuit, and, more specifically, :resides
in the improvement comprising, in combination, treati:ag the arc
31 signature as a modulated carrier having a modulation indicative
32 of the electric or" and monitoring the electric =arc by
33 monitoring a modulatioa of the modulated carrier.
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
12
1 From a related aspect thereof, the invention resides in
2 apparatus for monitoring an electric arc haviag an arc signature
3 typified by a wideband range of frequencies of a chaoi:ic nature
4 in a monitored circuit, and, more specifically, resides in the
improvement comprising, in combination, a modulated carrier
6 detector having an arc signature input and a carrier modulation
7 output.
8 From a similar aspect thereof, the invention resides in
9 apparatus for monitoring an electric arc having an arc signature
typified by a wideband range of frequencies of a chaotic nature
11 in a monitored circuit, aad, more specifically, resides in the
12 improvement comprising, in combination, combined modulated
13 carrier detectors having arc signature inputs and a combined
14 carrier modulation output.
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
13
1 Brief Description of the Drawing's
2 The subject invention and its various aspects and objects
3 will become more readily apparent from the following detailed
4 description of preferred embodiments thereof, illustrated by way
of example in the accompanying drawings which also constitute a
6 written description of the invention, wherein like reference
7 numerals designate like or equivalent parts, and in which:
8 Fig. 1 is a polar coordinate representation of na electric
9 arc signature spectrum in terms of Wavelength and illustrates
selection of a fractal subset for arc monitoring pursuant to an
11 embodiment of the ixivention;
12 Fig. 2 is a block diagram of an electric arc monitoring
13 system pursuant to an embodiment of the invention;
14 Fig. 3 shows gain vs. frequency graphs illustrating a
presently preferred embodiment of the invention;
16 Fig. 4 is a schematic of circuitry that may be used in the
17 system of Fig. 2 or rtherwise for monitoring an arc according to
18 an embodiment of the invention;
19 Fig. 5 is a schematic of another circuitry that may be used
in the system of Fig. 2 or otherwise for monitoring an arc
21 according to an embodiment of the invention;
22 Fig. 6 is a schematic of a further circuitry that may be
23 used in the system of Fig. 2 or otherwise for monitor~.ng the arc
24 also according to an embodiment of the invention; anc.
Fig. 7 is a circuit diagram of an optical indicator of
26 possible or actual arcing that can be used at various stages in
27 the systems of Figs. 2, 4, 5 and 6, according to a further
28 embodiment of the invention.
29 The accompanying Fig. 1 is copyright ~ as an original
creation under the Berne Convention and all corresponding
31 national laws, with Hendry Mechanical Works, cf Goleta,
32 California, United States of America, beiag the copyright
33 proprietor which understands that this figure will be published
34 by the World Intellectual Property Organization and thereafter
by patent offices throughout the World.
CA 02298789 2000-O1-28
WO 99/09424 ' PCT/US97/14497
14
1 Modes of Carrying' Out the Irwention
2 The drawings ilyustrate some basic modes and also preferred
3 modes of carrying oui: an aspect of the invention. Since yractal
4 geometry is a visual art as much as a mathematical science, Fig.
1 shows the workings of the invention in terms of a logarithmic
6 spiral. This is a novel aspect, since arc signature spectra
7 traditionally have been plotted in Cartesian coordinates and in
8 terms of frequency. To a large extent, thinking and plotting in
9 terms of frequency was justified, since the frequency of the arc
signature is largely independent of the medium throug:~ which it
11 travels, while the wavelength of the arc signature de~~~ends more
12 directly on the traversed medium.
13 The traditional focus of the prior art on frequency ab
14 initio obstructed visualization of arc .signatures as a
i5 logarithmic phenomenon of fractal nature, whereas thinking and
16 plotting in terms of wavelength according to the currently
17 discussed aspect of the invention, leads to visu~ili:~ation,
18 graphic representation and beneficial exploitation of that
19 phenomenon.
In this vein, the polar coordinate representation of Fig.
21 1 arises from the basic equation
2 2 r = exp ( q~") ( 5 )
23 wherein: r = radius,
24 q = growth factor larger than 1, and
~" = polar angle in terms of wavelength in the particTilar
26 medium, such as monitored circuit wires.
27 The polar coordinate plot of Fig. 1 represents a logarithmic
28 growth spiral occurring in innumerable natural objects, including
29 the spiral ammonite-.: that appeared on the earth during the
Devonian period which also brought forth algae and the first
31 terestial plants soma 380 million years ago. These developed in
32 the subsequent Carbc.~niferous period some 260 million years ago
33 to ferns that have very pronounced fractal structur~a wherein
34 each leaf structure is a miniature replica of the branch
structure, and wherein each branch is a replica of the fern plant
36 or bush. Ammonites became extinct at the end of the ~~retaceous
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
1 period some 65 million years ago, but ferns are very much alive
2 along with millions of natural objects of fractal structure.
3 Fortunately for the insight needed in the subject invention,
4 the area where the pioneering mathematician Jacob Bernoulli lived
5 until 1705, had been a marine environment millions of years
6 earlier. This provided that region with an abundance of
7 petrified ammonites, along with a plethora of other
8 petrifications.
9 Bernoulli was so fascinated by the logarithmic spiral that
10 he devoted his famous treaty entitled "Spira Mirabilis"
11 (Wonderful Spiral) to the same. One of such wondrous properties
12 is that the logarithmic spiral is the perfect f racial in that it
13 persists through various changes. To magnification and reduction
14 it responds elegantly by rotational displacement, thereby
15 preserving its shape unaffected. This and other well-known
16 properties of logarithmic spirals reveals them as truly fractal
17 phenomena to which the electric arc signature seems akin, if
18 viewed in polar coordinates in terms of wavelength, such as in
19 Fig. 1.
In this respect, the ammonite shell had a chambered
21 structure wherein internal chambers were partitioned off by
22 septa, which were a series of spaced plates which were spaced
23 most closely at the center of the shell and the spacing of which
24 increased logarithmically along the growth spiral of the shell.
Accordingly, the size of the chambers between adjacent septa
26 increased logarithmically along the growth spiral.
27 In analogy to asnnonite shell chambers, the size of frequency
28 or wavelength intervals 12 between indicated wavelengths or
29 frequencies 13 also increases logarithmically in terms of
wavelength in the electric arc signature or spectrum 10.
31 The core of tt~e electromagnetic arc emanation has been
32 labelled as <~. at the pole of Fig. 1, indicating wavelengths of
33 less than one micron; that is, signifying the familiar visible
34 light emitted by the arc. In the subsequent logarithmic turn,
the symbol >~, has been shown to indicate infrared radiation and
36 microwaves that can be responsible for the utility of electric
37 arcs in electric welding, metallurgical furnaces, and internal
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
16
1 combustion engine ignition, and contrariwise in the sparking of
2 explosions and startup of devastating fires by electric arcs.
3 Thereafter, Fig. 1 indicates specific portions c:f the
4 electromagnetic arc signature in terms of frequency, including
the following frequency bands with increasing progression:
6 GHz = gigahertz,
7 ~z = megahertz range wherein arc signature detection
has
8 been conducted by the prior art and wherein extraneous
9 signals from television broadcasts and rad:~.a signals
abound,
11 100 kHz - the one-hundred kilohertz range wherein ar~~
12 signature detection also has been conducted
13 extensively by the prior art and wherein radio
14 broadcast signals abound,
30 kHz = a lower limit of prior-art arc detection,
16 VLF = "very low frequency" defined as extending frorx
3 kHz
17 to 30 kHz by the above mentioned IEEE Dictionary.
18 ELF = "extremel~~~ low frequency" defined as extending
from
19 3 Hz to ,4 kIiz by that IEEE Dictionary,
of = "voice fr~:quency" within the range of 200 to 3500
Hz
21 according to that IEEE Dictionary, and
22 ef = "line frequency", i.e. 50 Hz in European Systems,
or
23 60 Hz in American systems.
24 Of course, no patent drawing can actually depict tile chaotic
nature of electric arcs. Rather, Fig. 1 as a minimum has to be
26 viewed in terms of an instantaneous moment in the chaotic
27 occurrence of an arc signature. Nevertheless, Fig. 1 shows the
28 statistical self-similarity of logarithmic fractal subsets of the
29 depicted arc signature.
Summarizing the ammonite analogy, the logarithmic nature of
31 the depicted arc spectrum is seen not only in the evc~lutian of
32 the growth spiral 10, but also to the logarithmically progressing
33 length of frequency intervals 12 in terms of ws.velength,
34 individually delimited by what corresponds to the above mentioned
septa of the amanonite shell. In terms of the electric arc
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
17
1 signature, such septa correspond to radial lines 13 denoting
2 certain frequencies so that the intervals 12 between such
3 frequency points are logarithmically distributed along the arc
4 signature in terms of increasing wavelength or decreasing
frequency.
6 Fig. 1 also depicts the inverse frequency or 1/f dependency
7 of electric arc signatures in terms of amplitude. In the case
8 of Fig. l, this shows as an amplitude increasing with wavelength
9 to a value of am.x~ This is another indication of the fractal
nature of electric arcs.
1i On the subject of 1/f-noise. Dres. rer. nat. Heinz-Otto
12 Peitgen and Dietmar Saupe, have pointed out in their work
13 entitled "THE SCIENCE OF FRACTAL IMAGES" (Springer-Verlag, New
14 York, 1988) pp. 39 to 44, that there are no simple mathematical
models that produce such noise, other than the tautological
16 assumption of a specific distribution of time constants.
17 This quite unlike to white naise on the one hand and
18 Brownian motion on the other hand. In this respect, they relate
19 the discovery that almost all musical melodies mimic 1./f-noise.
This to those serious researchers suggests "that music is
21 imitating the characteristic way our world changes in time".
22 Indeed, music with its numerous variations on a theme is replete
23 with fractals.
24 Dres. Peitgen and Saupe also point out that both music and
1/f noise are intermediate between randomness and predictability.
26 Even the smallest phrase reflects the whole. And so it is with
27 electric arc signatures, with such smallest phrase called herein
28 a "fractal subset".
Z9 As the Leitmotif in music, such fractal subset may have the
nature of an attractor as a limit figure of fractal iteration,
31 as more fully described below.
32 In terms of indications in Fig. 1, a preferred embodiment
33 of the invention detects the electric arc from a fraci:al subset
34 below 30 kHz of the wideband range of arc signature frequencies.
This includes the above mentioned VLF (very low frequency) range
36 and frequencies below that range.
37 More specifically, an embodiment of the invention restricts
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
18
1 fractal subsets from which the electric arc is detected or in
2 which the electric arc is monitored to the ELF (extremely low
3 frequency) band which according to The New IEEE Standard
4 Dictionary of Electrical and Electronics Terms, Fifth Edition
(The Institute of Electrical and Electronics Engineers, 1993) is
6 defined as extending from 3 Hz to 3 kHz.
7 Another embodiment of the invention restricts monitored
8 fractals to arc signature frequencies below the voice frequency
9 band (vf) defined in that IEEE Standard Dictionary as extending
from 200 Hz to 3500 Hz. In that vein, a further embodiment of
11 the invention restricts monitored fractal subsets to arc
12 signature frequencies below a first harmonic of a sta~c:dard line
13 frequency in alternating-current power supply systems. An
14 embodiment of the invention even selects the monitored arc
signature fractal subset from a frequency band on the order of
16 a standard line frequency (8f) in alternating-current power
17 supply systems.
18 In this manner, the invention in its embodiments can select
19 the fractal that will give the best overall performance in a
given situation, all the way to the maximum arc signature
21 amplitude of a,~x for optimum signal-to-noise ratio; with the
22 signal in such case being the 1/f-noise of the arc which we have
23 designated above as ~" - noise. The noise in the expression
24 "signal-to-noise ratio", on the other hand, includes white noise
and Brownian noise, such as produced by the electronic: circuits
26 of the arc monitoring apparatus, and extraneous signals that
27 could engender false alarms or readings.
28 Of course, depending on application, there may be goals
29 other than reaching a~,x, but as Fig. 1 depicts, arc signature
amplitudes at various frequency fractals selected pursuant to the
31 subject invention in the larger.portion of the outer turn of the
32 growth spiral 10 still are significantly better than what the
33 prior art had to work with.
34 In this respect, selecting from the arc signature a fractal
that yields an amplitude of a,~,X or an amplitude comparable
36 thereto according to an embodiment of the invention has another
37 advantage where cross-induction of arc signatures could be a
*rB
CA 02298789 2000-O1-28
WO 99109424 PCT/US97/14497
19
1 problem. Take for example the case where several electric
2 circuits are monitored for electric arcs by several corresponding
3 arc detectors. and assume that an arc occurs in one of these
4 circuits and that the arc detector pertaining to that circuit is
to respond thereto.
6 Pursuant to what has been said above at and of ter Equations
7 (1) to (4), by selecting a high-amplitude (i.e. long-wavelength
8 or low-frequency) fractal subset for the arc monitoring process,
9 a preferred embodiment of the invention minimizes if not
practically eliminates the prior-art danger of false alarms from
11 cross-induction among independently monitored neighboring
12 circuits.
13 Fig. 2 shows an electric conductor 20 of electric circuitry
14 21 wherein an electric arc 22 occurs.
By way of example, the circuitry 21 may be part of a
16 telephone exchange or may be another one of a large variety of
17 electric circuits or loads, including the following examples:
18 In internal combustion engine research, develo~merrt, and
19 maintenance, it is important to establish and to maintain the
optimum spark in each cylinder. A reliable spark monitoring
21 system is therefore highly desirable, if not putentially
22 indispensably in cutting-edge internal combustion engine
23 technology.
24 In a similar vein, electric welding is becoming increasingly
robotized and reliable monitoring of the welding arc oz spark
26 would greatly benefit research, development and assembly-line
27 quality control and assurance in automated electric arc welding.
28 Moreover, many modern electric spot welding processes rely on
29 immediate application of electric energy to the work pieces to
be joined, Without intervention of an electric arc. In fact, the
31 occurrence of an electric arc, such as by imperfect contact
32 between the work pieces, degrades the resulting weld in such
33 Joule-effect welding processes. Accordingly, the load at 21 may
34 for instance be a robotic or other spot welding apparatus. In
that case, the electric arc monitor could supervise the spot
36 welding process and could signal when substandard welds are being
37 produced by intervening arcing. This, in turn, would signal the
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
1 need for remedial action, such as including better cleaning of
2 work pieces prior to welding or better compression of the work
3 pieces during welding for more intimate contact.
4 Also, modern descendants of the original electric arc lamp,
5 such as mercury or sodium vapor lamps, could beneøit as to
6 research and development and in the maintenance of high-quality
7 performance from reliable arc monitoring systems, as could
8 electric-arc-types of metallurgical and other furnaces.
9 Similarly, arc and spark monitoring systems would be useful
10 to detect and if necessary eliminate faults in electric circuitry
11 and equipment creating radio interference through excessive
12 sparking or arcing.
13 As a further example, some gas heating appliances have
14 gaseous fuel ignitors that work with an electric spark. In such
15 cases, it is often important to know whether the desired spark
16 has occurred for proper ignition, especially if the thermostat
17 is remote from the heating unit. Also, an electric spark monitor
18 would indicate when the igniter is in need for replacement,
19 before breakdown and costly outage occur.
20 Moreover, electric arcs are used in ionizers: such as
21 ammonia arc and other ion generators that are coming intc use in
22 satellite thrusters and in propulsion systems in outer space,
23 such as for restabilizing satellites in geostationary orbits or
24 for propelling satellites and space probes on their journey. In
such cases, an electric arc monitor would be useful in research,
26 development, maintenance and operation of such ion generators.
27 Alternatively, machinery, circuitry or apparatus at 21 that
28 produces normal sparks in its operation could be monitored for
29 detrimental arcing. One of many examples concerns cummutators
of electric motors that are often damaged when th~:ir carbon
31 brushes wear out, as the rotating commutator then rubs against
32 the metallic brush holder springs. Siace such Wear is
33 accompanied by heavy arcing, an early detection of ;such heavy
34 arcing, as distinguished from regular commutator spar~:ing, would
signal the need for preventive action and could save the
36 equipment from breakdown and severe damage. The same applies to
37 relays and contactors that generate sparks and arcs in their
*rB
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
21
1 normal operation, but are subject to excessive arcing in case of
2 malfunction or excessive wear.
3 Similarly, the recurrence of the electric automobile as an
4 environmentally friendlier vehicle than, the gasoline-drives
automobile or petrol-driven motorcar, renders reliable arc
6 monitoring even more important. In particular, such electric
7 vehicles carry large storage batteries that have to be recharged,
8 typically overnight, and that generate combustible gases, such
9 as oxygen and hydrogen, during such recharging. Electric arcing
obviously could be disastrous in such as atmosphere. Accordingly,
11 monitoring that environment for electric arcing sad shutting down
12 the charging process and giving an alarm immediately upon
13 detection of arcing, could prevent disaster.
14 In all these cases, the invention selects a fractal subset
16 of the signature of the electric arc 22 for the purpose of arc
16 detection. In such selection the invention aims for a relatively
17 long travel of arc signature along the monitored circuit 20
18 (distance between arc 22 and pickup 23), and for low cross-
19 induction among neighboring circuits, including the monitored
circuit 20.
21 According to a preferred embodiment of the invention, the
22 fractal subset of arc signature is selected in a frequency band
23 below 30 kHz, where arc signature amplitudes are higher, arc
24 signature travel along wires (20) is longer, and arc signal
cross-induction among separately monitored neighboring circuits
26 (21, 30) is lower, than at higher frequencies.
27 The invention then detects the electric arc 22 from the
28 fractal subset 16 of the arc signature. A preferred embodiment
29 of the invention detects the electric arc 22 from a fractal
subset of the arc signature below 30 kHz, such as is the ELF
31 (extremely low frequency) band. defined above~as exte;ading from
32 3 Hz to 3 kHz, or even below the voice frequency band (vf)
33 defined above as extending from 200 Hz.
34 By Way of example Fig. 3 shows such a fractal subset of the
arc signature at 16 within a band 15 illustrated on a logarithmic
36 scale. In practice, selection of such a low-frequency fractal
37 subset 16 avoids false alarms by cross-induction, such as between
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
22
1 the circuit 20 where an arc 22 does occur, and any separately
2 monitored neighboring circuits 30, etc., wherein no arc occurs
3 at the moment or, conversely, between any neighboring circuit
4 where an electric arc does occur and is monitored by another
monitoring circuit 19, and the circuit 20 when no arc occurs at
6 that point.
7 Selection of such low-frequency fractai subset 16 also
8 pezmits detection of the arc 22 from a remote location along
9 wires 20 over longer distances than would be possible at high
frequencies. Selectian of such a low frequency band also yields
11 a high amplitude input signal for the detection process according
I2 to the above mentioned 1/f or ~" characterist:ic of the arc for
13 highest signal-to-noise ratio with lowest exposure to false
14 alarms.
Sensitivity to switching transients in telephone exchanges,
16 to multiplexed audio and otherwise. and to harmonics of alterna-
17 ting-current supply frequencies may be practically eliminated,
18 and the permissible travel distance of arc signals between the
19 arc 22 and the pickup 23 may be multiplied as compared to high-
frequency detection systems, by selecting the narrower frequency
21 band 15 on the order of a standard line frequency in public
22 alternating-current power supply systems. Preferably, according
23 to that embodiment, the selected narrower frequency band is below
24 the first-order harm~~nic of that standard line frequency.
According to the embodiment illustrated with the aid of Fig.
26 3, the selected arc signature fractal subset may be a filter
27 passband 15, below line frequency, such as below 50 Hz for
28 European-type systems and below 60 Hz for American-type systems.
29 Accordingly, it may be said that arc monitoring according
to a preferred embodiment of the invention concentrates on the
31 low end of the arc signature spectrum.
32 The fractal subset of the arc signature 16 or the passband
33 15 from which detection of an arc 22 takes place, covers at least
34 a quarter of a logarithmic decade of the wideband range of
frequencies of the electric arc 22.
36 This overcomes a drawback of the prior-art approach
37 manifestation in the above mentioned W090/04278 that missed the
CA 02298789 2003-08-08
23
1 point by limiting the band of detection to within some 20 kHz at
2 some 170 kHz center frequency. this, however, left only a few
3 percent of the arc signature information in the particular
4 logarithmic decade available for detection, considering that a
range of 20 kHz in the 170 kHz area is but a small fragment of
6 the particular logarithmic decade that contains the information
7 signifying "arc" as distinguished from other signals.
8 The situation is not much better in the case of most
9 bandpass filter components of the comb filter disclosed in
W090/0427$. All but the first bandpass filter component have
11 center frequencies belonging in effect to the fifth logarithmic
12 decade covering from 140 kHz to one hertz less than 1 MHz.
13 Since all these components have a 50 kHz bandwidth, they can
14 only pass a small percentage of the arc signature information in
the particular logarithmic decade for detection of an arc as
16 distinguished from other error signals or from picked-up
17 extraneous signals. By rotating detection among the components
18 of its comb filter arrangement, that prior-art proposal even
19 misses an opportunity of making best use of its lowest frequency
component in the 55 kHz area.
21 By way of example, the invention may be practiced with the
22 circuitry shown in Fig. 2. In that circuitry, an electric arc
23 22 having an arc signature typified by a wideband range of
24 frequencies of a chaotic nature, is detected with the aid of an
electric filter 25 having an input 26 coupled to that arc,
26 having a passband corresponding to a fractal subset 16 of the
27 arc signature characterized by relatively long travel along the
28 monitored circuit and low cross-induction among neighboring
29 circuits, and having an output 27 of that fractal subset of arc
signature. Chaotic wideband signal detector circuitry having a
31 detector input for that fractal subset 16 of the arc signature
32 may be coupled to that output of the electric filter, such as
33 disclosed in the further course of Fig. 2.
34 The arc signature pickup 23 may be of a conventional kind,
such as a clamp-on current transformer terminating into an
36 impedance 24 that may be symbolic of a conventional peak-to-peak
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
24
1 limiter for clipping unusually large input transients, such as
2 with the aid of two high-speed diodes connected back to back with
3 one side to ground and a series current limiting resistor.
4 The pickup or transformer 23 may for instance be wound to
be sensitive to frequencies in the 25 Hz to 50 Hz range with a
6 minimum of insertion loss. Such transformer 23 may be clamped
7 around the line 20 to be monitored. Hall effect sensors present
8 another example of arc signal pick-ups that may be employed in
9 the practice of the invention, which extends to the use of other
sensors of a wired or wireless type. Other monitored circuits
11 30, etc., may be provided with like or similar pickups 123 and
12 monitoring circuitry 19.
13 The picked-up arc signature signal or fractal subset 16 is
14 passed through a lowpass filter 25 having an input 26 connected
to the arc signature pickup 13. In a prototype of this
16 embodiment, the configuration of this filter is that of two
17 cascaded 3rd order Butterworth filters; but other configurations
18 and other kinds of filters may be selected within the scope of
19 the invention.
The gain vs. frequency plot of Fig. 3 shows a typical
21 response characteristic 41 of such a filter, having little
22 attenuation in the selected narrower frequency band 15.
23 Within the scope of the invention, the passband 15 could
24 cover an entire logarithmic decade, such as from 10 Hz to 100 Hz .
According to an embodiment of the invention the monitored arc
26 signature fractal 16 or passband 15 covers not more than a
27 logarithmic decade of the wideband range of frequencies of the
28 electric arc 22, inasmuch as the decade of from D.C. to 10 Hz is
29 rather a regular decade than a logarithmic decade.
In this respect, Fig. 3 shows only about half of a
31 logarithmic decade for the upper portion of the filter
32 characteristic 41 at passband 15. This demonstrably has provided
33 reliable arc detection with the illustrated embodiment of the
34 inventiozl. Depending on circumstances, the fractal of arc
signature 16 may be even less than as illustrated in Fig. 3, but
36 should cover at least a quarter of a logarithmic decade of the
37 wideband .range of frequencies of the electric arc, far reliable
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
1 arc detection with simultaneous exclusion ~f false alarms.
2 In principle. aspects of the invention herein disclosed can
3 be applied to frequency bands other than the preferred ELF
4 (extremel~r low frequency) band, and the subject applicants have
5 built models of the circuitry shown in Fig. 2 not only for the
6 ELF band, but also for operation at several kilohertz, as well
7 as in the 10 to 20 kHz region.
8 According to a preferred embodiment, the narrower frequency
9 band 15 0:: fractal subset 16 is selected where there are less
10 extraneous signals than in a remainder of the wideband range of
11 frequencies of the arc signature. In this respect, the
12 embodiment illustrated in Fig. 3 shows the narrow frequency band
13 15 as covering from about 10 Hz to less t~:an 50 Hz where there
14 are no significant switching transients in telephone exchanges,
15 effects of multiplexed audio signals, harmonics of alternating-
16 current supply frequencies, signals from control or security
17 systems and radio broadcast signals. However, not all
18 embodiments of the invention are intended to be limited to
19 operation within and below the very low frequency range.
20 According to Fig. 2, the output 27 of the filter 25 is
21 applied to nonlinear processing in what is herein called a non-
22 linear processor 42. Hg way of example, such non-linear
23 processor may comprise a demodulator also called "modulated
24 carrier diatector" that demodulates signals passed by the filter
25 25, including picked-up electric arc signature segments with
26 chaotically varying amplitudes and frequencies. This and other
27 aspects c~f the invention treat a fractal subset of the arc
28 signature as a modulated carrier having a modulation indicative
29 of an electric arc 22, and monitor the electric arc by monitoring
one or mcxe modulations on such modulated carrier.
31 Apparatus for monitoring an electric arc having an arc
32 signature. typified by a wideband range of frequencies of a
33 chaotic nature in a monitored circuit, include a demodulator or
34 modulate2. carrier detector such as in the non-linear processor
42 having an arc signature input 27 and a carrier modulation
36 output 43.
37 Fey ~~ay of example, the fractal subset of an arc signature
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
26
1 may be treated as an amplitude-modulated carrier, and the
2 electric arc may then be monitored by monitoring a modulation of
3 such amplitude-modulated carrier, such as by recovering the
4 modulatio~a on such amplitude-modulated carrier, aad by then
detecting the amplitude from such recovered modulation.
6 Acco:rdingiy, the non-linear processor 42 may be an Abt
7 detector or demodulator that in response to chaotically varying
8 amplitudes at 27 produces an AC signal at 43 as a function of
9 such chaotically varying amplitudes. More steady signals
erroneously picked up by the circuitry 23 to 27, on the other
1l hand, produce no such AC signal. The non-linear processor 42
12 thus is a .first stage that distinguishes picked-up arc signatures
13 from signals stemming from radio or television broadcasts, radio
14 frequency security systems or other sources except electric arcs.
The non-linear processor 42 thus in effect treats the
16 picked-up arc signature as an amplitude-modulated or AM carrier
17 whose modulation can be monitored by monitoring a modulation or
18 amplitude of such monitored carrier, such as for detection of an
19 electr:i.c arc 22. At least in the case of AM detection, such
'modulation' still includes its 'carrier'. Such carrier may be
21 stripped from its modulation by such means as a bandpass filter
22 44 connected to the output 43 of the non-linear processor or AM
23 detector 42. By way of example, the bandpass filter 44 may
24 comprise a 3rd order Butterworth filter, or a filter with a
similar zesponse. A preferred responsF: of such filter in
26 principle may follow the characteristic 41 shown in Fig. 3,
27 except that the gain may be adjusted as desired or necessary.
28 By way of example, the filter 44 may pass alternating-current
29 signals of frequencies below 20 or 30 Hz and reject spikes and
other fait signals that might be produced or occur within the
31 circuitry up to that point.
32 The recovered modulation signifying "arc" appears at the
33 output 45 of bandpass filter 44.
34 According to an embodiment of this aspect of the invention,
combined modulated carrier detectors having arc signature inputs
36 and a combined carrier modulation output may be used in
37 monitoring electric arcs.
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
27
1 In this respect such combined modulated carrier detectors
2 may be like kind modulated carrier detectors; such as both AM
3 detectors or both FM detectors. An example of this is shown in
4 Fig. 2 ac~:.:ording to which like kind modulated carrier detectors
are series connected.
6 In ~uarticular, the output 45 of bandpass filter 44 is
7 connected to a second demodulator 46 which, for instance, may be
8 an AM demodulator or detector for producing at its output 47 a
9 signal level varying as a function of picked-up arc signature.
With:i.n the scope of the invention, like-kind modulated
11 carrier detectors may be parallel connected, such as within non-
12 linear processor 42.
13 Within the broad aspect of the invention, it should be
14 understood that other types or kinds of demodulation techniques
may be used, such as those developed for frequency modulation,
16 phase modulation or carrier-suppressed or single-sideband
17 modulation, for example.
18 Accordingly, the non-linear processor 42 may, for instance,
19 include an FM demodulator which at output 43 produces a signal
in response to such chaotic variations of phase or frequency as
21 occurring in an electric arc signature. In this respect, the
22 combined modulated carrier detectors may include different kinds
23 of modulated carrier detectors, such as an FM detector at 42 and
24 an AM detector at 46 connected in series.
According to an embodiment of the currently discussed aspect
26 of the invention, an arc signature is treated as a carrier
27 modulated both in a first manner and in a different second
28 manner, and its electric arc is monitored by monitoring first and
29 second modulations of said carrier modulated both in said first
manner. anal in said second manner.
31 By ~~ay of example, different kinds of modulated carrier
32 detectors include an AM detector and an FM detector, and such AM
33 detector and FM detector are connected in parallel.
34 By way of example, the non-linear processor 42 may include
parallel-connected AM and FM demodulators 412 and 413 having 27
36 as their common input, and having individual outputs connected
37 to an ANh-element 414, such as shown in Fig. 4.
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
28
1 AND-element 414 only provides an output signal at 43 if both
2 ~ the AM demodulator 412 responds to chaotic amplitude variations
3 of the picked-up signal at 27 and the FM demodulator 413 responds
4 to chaotic phase or frequency variations of that signal at 27.
This, then, provides a further safeguard against false
6 alarms from such extraneous signals as AM broadcast or control
7 signals and FM broadcast or control signals, and assures that
8 picked-up signals are only signified as stemming from electric
9 arcs if they display not only the chaotic amplitude variation,
but also the chaotic phase or frequency variation, that
11 characterize electric arc signatures.
12 Preferrably, full-wave rectification or detection is used
13 at 42 and 46 instead of a half-wave rectification or detection,
14 in order to improve the speed of detection. Such increased
speed, in turn, permits selection of a lower frequency band, such
16 as 15 shown in Fig. 3, where signal-to-noise ratio is at a
17 maximum with extraneous signals 52 being at a minimum, even in
18 the deliberately exaggerated showing of Fig. 3. In other words,
19 selection of full-wave rectification or detection alleviates the
tradeoff of lower detection speed at lower detection frequencies.
21 The above mentioned higher and less error-affected sensitivity
22 of detection is thus realized without objectionable delays in
23 detection.
24 The arc-indicative signal resulting at an output 47 from the
detection process at 46, is timed and level sensed at 48.
26 Conventional RC-type or other timing circuitry and conventional
27 comparator circuitry may be employed at 47 to prevent the arc
28 detector from responding to switching transients, contact
29 bouncing, ordinary commutator arcing and other transient,
harmless arcs, such as more fully discussed herein with respect
31 to Fig. 6. Alarm circuitry 50 thus responds only to arcs that
32 sustain themselves for a given, dangerous period of time.
33 The circuitry 47 may include a conventional resettable
34 latching arrangement for latching in an alarm condition if an arc
22 occurs and all detection criteria are met as herein described.
36 Various control circuits may be energized at this point, such as
37 for providing an audible alarm, a remote fire alarm, etc., or for
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
29
1 shutting down power is the affected line 20. Since circuits of
2 this kind are known per ~, only a block 50 has been shown to
3 signify the possible presence of such control and alarm circuits .
4 The embodiment of Fig. 2 elegantly avoids false alarms from
extraaeou;~ signals, without narrowing the bandwidths of picked-up
6 arc signals to significantly leas than a fractal subset
7 containing sufficient arc information. Such avoided extraneous
8 signals fir instance include television signals, radio broadcast
9 signals, ~~arious control signals, harmonics and other signals
that ante relatively narrow in bandwidths as compared to the wide
11 band of chaotic arc signals.
12 Some extraneous signals, such as those which are of a
13 chaotic nature themselves, may have to be subjected to some
14 common-mode rejection or other processing in order to avoid
confusion thereof with chaotic arc signatures. Such rejection
16 of extraneous signals elegantly comes about when picked-up
17 signals are beat against themselves, such as in the context of
18 the following embodiments of the invention.
19 In this respect, refinements pursuant to embodiments of the
invention derive from the arc 22 a fractal arc signature subset
21 16 Within a frequency band 15, and convert that fractal arc
22 signature subset 16 to an arc signal 17 in a frequency band 18
23 distinct from that fractal subset or frequency band 15, such as
24 shown in Fig. 3, and detect in that arc signal 17 a chaotic
wideband characteristic typical of an electric arc. A preferred
26 embodiment of the invention subjects the fractal subset 16 to a
27 frequency transformation, such as shown at 17 in Fig. 3, and then
28 detects the electric arc 22 from that fractal subset after
29 frequency transformation. By way of example and not by way of
limitation, the fractal subset 16 may be added to itself and the
31 electric arc 22 may be detected from the fractal subset added to
32 itself, such as in the manner disclosed in Figs. 3 and 5.
33 In particular, a component 142 of the circuitry shown in
34 Fig. 5 prEVents extraneous signals that do occur in the monitored
frequency band from affecting the arc detection process.
36 According to that technique, narrow-band extraneous signals in
37 the monitored fractal subset 16 of the arc signature are
*rB
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
1 diminished in energy relative to a remainder of that fractal
2 before detection of the electric arc from that fractal. Such
3 component may be a frequency converter 142 that has converter
4 inputs 127 and 227 for the arc signature fractal subset 16
5 coupled. to an output 27 of electric filter circuitry 25, and has
6 a converter output 43 for an arc signature segment 17 in a
7 frequency band 18 distinct from the pasabaad 15 of the filter
8 circuitry 25. A chaotic wideband characteristic typical of an
9 electric arc is then detected from such converted arc signal 17,
10 such as w:~th a chaotic wideband signal detector.
11 By way of example and not by way of limitation, the input
12 and output terminals 26 and 45 may be the same in Figs. 2 and 5,
13 whereby the remainder of the circuit may be the same for Fig. 5
14 as in the apparatus of Fig. 2 more fully described above.
15 The ~.llustrated embodiment of the invention even alleviates
16 the inherent tradeoff of slower detection speed at lower
17 frequenci~a. In particular, by converting the selected arc
18 signal fractal subset 16 from the lower frequency band 15 to a
19 higher frequency band 18 such as shown in Fig. 3 or higher, the
20 currently discussed embodiment of the invention realizes the
21 detection Speed corresponding to the higher frequency band 18 for
22 an arc signature fractal subset i6 stemming from the lower
23 frequency band 15 where there are less extraneous signals and
24 where cross-coupling among circuits is lower and actual travel
25 distance of picked-up arc signals is higher than at frequencies
26 above the band 15. At the same time, the illustrated pref erred
27 embodimen~c of the invention realizes for the higher detection
28 speed associated with the higher frequency band 18 the lower
29 cross-coupling among circuits and the longer possible travel
30 distance along the affected line 20 that are associated with the
31 lower frequency band 15. An arc 22 occurring in line 20 thus may
32 be picked up from such line 20 at a considerable distance from
33 that arc, without a release of nay arc alarm condition in
34 neighbori~.~.g lines that are individually equipped with arc pickups
123 and a,.rc monitoring circuits 19 which correspond to the
36 circuitry of Figs. 2, 4, 5 or 6, but in which no arcs are
37 occurring at the time.
*rB
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
31
1 As a3.ready indicated, the passband of the filter circuitry
2 25 may be located where there are less extraneous signals than
3 is a remainder of said wideband range of frequencies. Such
4 passbaad may be in the ELF (extremely low frequency) band, or
even belong the of (voice frequency) band, or below a first
6 harmon:.c of a standard line frequency in alternating-current
7 power supply systems, or may even be on the order of a standard
8 line frequency in alternating-current power supply systems,
9 and/or ma~~ cover at least a quarter of a logarithmic decade of
the wideb~~nd range of frequencies of the electric arc, so as to
11 provide sufficient information for a reliable detection of the
12 arc sigaa:~ from that logarithmic decade.
13 The frequency converter 142 of Fig. 5 may constitute the
14 son-linear: processor 42 of Fig. 2. By way of example, the
component 142 may include a multiplier having first and second
16 inputs 12~ sad 227 connected to a single line 27 for receiving
17 the filtered picked-up arc signal. This has the net effect of
18 mixing the signal with itself, creating sum and difference
19 products at its output 43. Hy way of example, the component 142
may be a four-quadrant multiplier of the type AD633. However,
21 within the scope of the invention, a diode or non-linear circuit
22 may be us~:d for intermodulation of the selected fractal subset
23 16 with itself.
24 By way of example, Fig. 5 in effect adds the arc signature
fractal subset 16 to itself in mixer 142 so as to double the
26 frequency band 15 of that fractal subset as the distinct
27 frequency band 18 of the arc signal 17 which may be somewhat
28 truncated,. such as by a subsequent filter 144 at the lower end
29 at that higher band 18. That filter may be similar to the above
mentioned filter 44 in the circuit of Fig. 2, but may have a
31 narrower bandpass characteristic, such as shown at 51 in Fig. 3.
32 In apparatus terms, a frequency converter, such as component
33 142, has two converter inputs, such as 127 and 227, for the arc
34 signat~.xre fractal subset 16 coupled to the output 27 of the
electric Filter circuitry 25, and has a converter output 43 of
36 the arc ;.ignature signal in a frequency band 18 double or
37 otherwise higher than the frequency band 15 of the arc signature
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
32
1 fractal subset.
2 The component 142 in effect dilutes re~,gular, man-made, non-
3 chaotic signals that, if picked up and not diluted, might produce
4 false alarms. The multiplier or similar stage 142 does more than
simply double the frequencies of the input. It also beats all
6 the frequencies occurring at one input 127 against all the
7 frequencies occurring at the other input 227. This causes the
8 output to display a summation of all the individual frequencies
9 at one input added and subtracted from all the frequencies at the
other ia~ut. Since both inputs contain a continuum of
11 frequencies, the result is a very rich mix of frequencies at even
12 higher and lower frequencies than those contained at the inputs.
13 Because a man-made signal is usually a discrete single frequency
14 or at wor:~t a narrow spread of frequencies, such signal does not
have the breadth of spectrum to contribute strongly to the output
16 at 43. Such man-made signal constitutes a narrow source beating
17 against the broader naise or arc signature continuum, and the
18 result is a very weak component at the output.
19 In this manner, the relative strength of the continuum
spectrum from the arc is enhanced compared to the strength of any
21 discreet ~r man-made components that are passed by the limiter,
22 such as at; 24, or by the filter circuitry. This is the essence
23 of this embodiment's ability to reject signals that would produce
24 false alarms. However, a fluctuating carrier with no noise at
the input could still be misidentified as an arc if it has
26 sufficient strength, because even though the relative sensitivity
27 to chaotic noise is greater than to modulated carriers, the
28 latter can still reach the output if they are of sufficient
29 strength, and thus can produce a level that may be mistaken as
an arc.
31 Accordingly, the output of the converter or multiplier 142
32 is fed to the input of a bandpass filter 144, that may have a
33 response characteristic 51 as shown in Pig. 3. Response
34 characteristics of the type shown in Fig. 3 at 51 may, for
instance, be realized by operational amplifier type of bandpass
36 filters, quartz filters, LC resonance filters, and other
37 circuitry accomplishing such kind of function.
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
33
1 As can be seen, the characteristic 51 displays minimal
2 response to any fundamental frequency passed by the lowpass
3 filter 25, such as in the band 15, and, in our example, displays
4 response only to frequencies of the picked-up arc signature
fractal subset 16 whose sum or other modulation product falls in
6 the passband range 18 of the frequency-converted arc signature
7 signal 17.
8 Baadpass filter 144 will pass only frequency components that
9 are within a few hertz on either side of the passband, such as
80 Hz, fir instance. This delivers a sample of the higher
11 frequencies in the output of the multiplier 142 to the next stage
12 46. Because the passband is above the 50 8z or other selected
13 cutoff of the first filter 25, this stage 144 is only going to
14 pass signals which have been boosted in the mixing process to
frequencies higher than such cutoff. This i.n effect prevents
16 non-chaotic signals from passing this stage.
17 In principle, the multiplier or stage 142 should not pass
18 any of the original input frequencies (e. g. below 50 8z) if there
19 is no direct-current at either input 127 and 227, and if such
multiplier or stage 142 is perfectly efficient. However, neither
21 assumption is always correct in practice. Accordingly, use of
22 the higher frequency passband filter 144 provides effective
23 rejection. of the unprocessed or unmixed original frequencies.
24 If a higher band of detection frequencies had been selected
within the scope of the invention, such as a band limited between
26 10 and 20 kHz, as an example, an alternate choice for bandpass
27 center frequency would be at some much lower frequency (i.e. 500
28 8z.) detecting a difference component in the output 43 of
29 multiplier 142. This would permit a much wider separation in the
responses of the different filters employed in the circuitry,
31 such as filters 25 and 144.
32 In either case, the frequency conversion or intermodulation
33 at 142 greatly reduces the energy of picked-up extraneous signals
34 52 in the: arc signature segment 16, by intensive common-mode or
similar rejection.
36 The output 43 of the multiplier 142 represents a chaotic
37 frequency sum or difference signal version of the picked-up and
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
34
1 filtered chaotic arc signature fractal subset 16.
2 Within the scope of the invention, the component 142 may
3 also include the above mentioned AM or FM detector or the
4 combined AM and FM~detectors 412 and 413 of Fig. 4, for instance.
In such cases, the detector or detectors detect the amplitude
6 aad/or frequency or other type of modulation in the converted arc
7 sigaature, such as iadicated at 17, for arc monitoring purposes .
8 Fig. 6 shows aaother embodiment of the invention where the
9 energy of extraneous signals is subjected to effective common-
mode or similar rejection by pairing inverting and non-inverting
11 amplifiers 242 and 342 with each other. By way of example, the
12 inverting amplifier 242 has an input 327 connected to the output
13 27 of bandpass filter 25, and the non-inverting amplifier 342 has
14 an input 427 connected to that output 27 of the above mentioned
baadpass filter 25 through which the picked-up arc signal is
16 processed.
17 The inverting amplifier 242 has an output 421 connected to
18 a mixer 442, and the non-inverting amplifier 342 has an output
19 422 connected to that mixer 442.
The electric filter 25 in the embodiment of Fig. 6 may be
21 the same txs the electric filter 25 in the embodiments of Figs.
22 2 to 5. Such a filter may be split in two, providing between the
23 terminal 26 a first filter for the inverting amplifier 242 and
24 a second filter for the non-inverting amplifier 342. Filter 25
may be similarly split in or for the embodiment of Figs. 4 and
26 5 to provide separate filter paths from the input terminal 26 to
27 detectors 412 and 413 or multiplier or frequency converter inputs
28 127 and 227.
29 By way of example, the mixer 442 may be composed of
conventional components, such as of two diodes interconnected in
31 an OR-element configuration between input terminals at 421 and
32 422 and the previously mentioned output terminal 43. Within the
33 scope of that embodiment, the component 442 may, however, include
34 a modulator, such as the above mentioned modulator 142. In this
respect and in geaeral, the circuitry in Fig. 6 between the
36 filtered arc signature portion or segment at 27 and the terminal
37 43 may correspond to the non-linear processor 42 shown in Fig.
*rB
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
1 2. The output of such processor 42 or 242, 342, and 442 at 43
2 may be filtered or otherwise processed at 244. For example, the
3 circuit 244 may include a baadpass filter corresponding to the
4 bandpass filter 4426 or 144 mentioned above with respect to
5 Figs. 2 to 5. Alternatively or additionally, the component 244
6 in Fig. 6 may include a standard IF amplifier, such as the
7 commercially available 440 Hz IF amplifier to name an example.
8 In method terms, the embodiment of Fig. 6 processes a
9 fractal segment, such as shown at 16, or a higher frequency
10 segment of the arc signature in the two paths 327 and 427 out of
11 phase with each other, and detects the electric arc 22 from such
12 out-of-phase segments or portions of the arc signature.
13 In F:pparatus terms, the embodiment shown in Fig. 6 has an
14 input 327 of a first phase processor, herein called "inverting
15 amplifier" 242 connected to the electric filter output 27 in a
16 first signal path 327 - 421, and has an input 427 of a second
17 phase processor, herein called "non-inverting amplifier" 342,
18 connected to the electric filter output 27 in a second signal
19 path 427 ~~ 422; with such first and second phase processors being
20 180° or otherwise out of phase with each other.
21 The more the outputs 421 and 422 are out of phase with each
22 other, the greater is the common-mode or similar rejection of
23 extraneor~s narrow-band signals 52 in the embodiment of Fig. 6.
24 The embo~',imeats of Figs. 2 to 5 thus share with each other a
25 feature according to which the narrow-band extraneous signals,
26 such a.s mentioned above or shown in Fig. 3 at 52 in a fractal
27 subset 1G or other fractal of the arc signature are diminished
28 in energ3~ relative to a remainder of such fractal subset portion
29 before detection of an electric arc 22 from such fractal subset.
30 At least in the embodiments shown in Figs. 3 and 5, the
31 fractal subset 16 is subjected to a frequency transformation,
32 such as shown in'Fig. 3, and an electric arc 22 is detected from
33 such fractal subset after such frequency transformation. By way
34 of example, a fractal subset 16 may be cross-modulated or may be
35 added to itself, such as disclosed above with respect to Fig. 5
36 or as mentioned with respect to Fig. 6, and the electric arc is
37 detected from such crass-modulated or added-to-itself fractal
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
36
1 subset. '.~'he embodiment of Fig. 4 adds the variant of parallel
2 different kind detection or demodulation, and the embodiment of
3 Fig. 6 adds the variant of out-of-phase prr~cessing.
4 In either case, a fractal subset o~: the arc signature
portion may be treated as a modulated carrier having a modulation
6 indicative of any electric arc 22, and such electric arc may be
7 detected from such modulated carrier for further rejections of
8 extraneous signals. The demodulator system disca.osed above with
9 respect tc~ Fig. 2 can also be used in the embodiment of Fig. 6,
the terminal 47 of which may be the same as the input terminal
11 47 of the time sad level sensing and alarm circuitry 48 - 50 in
12 Fig. 2, With respect to which various systems of modulation have
13 been mentioned above.
14 Of course, a wideband signal is not an arc signature unless
it disvla~rs chaotic frequency changes. ~~~ccordingly, a stage
16 including comparator 55, may be provided such as shown in Fig.
17 7 to detect and to display a pickup of a disturbance or signal
18 that is not only wideband in the region of interest, but that is
19 also c~ixaotic in nature as an arc signature is, and that is
sustained for a period of time, such as determined by the RC
21 component 58.
22 Fig. 7 is a circuit diagram of an optical indicator of
23 possible c:c actual arcing that can be used at various stages in
24 the systems of Figs. 2, 4, 5 and 6, according to a further
embodiment of the invention.
26 By waif of example, indications of the progress of the signal
27 through ax~~ monitoring circuitry may be accomplished with three
28 similar functional blocks or differential indicators of which a
29 prototype is shown in Fig. 7 at 54.
Such circuit 54 includes an operationa:!_ amplifier 55 having
31 its nou-inverting input 56 connected to a circuit input 57
32 through a lowpass filter and RC timing component 58 to prevent
33 response to short-term transients. The inverting input 60 of
34 that op amj~ is connected to comparator level resistors 61 and 62.
That op ar~.p 55 has a feedback circuit 64 which may include a
36 feedbaclc c.3pacitor or other impedance 65 and a unidirectional
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
37
1 current. conducting device, such as shown at 66, for such purposes
2 as noise reduction, prevention of premature or excessive
3 switching. Hy way of example, the op amp 55 may be of the type
4 LM35BAN.
The indicator circuit 54 includes light-emitting diodes or
6 LEDs 68 and 69 switched by transistors 71 and 72 biased through
7 resistors. including series resistors 73 and 74 and a pair of
8 resistc~ars 75 and 76. Transistors 71 and 72 may, for instance,
9 be of the type 2N2222.
A re:;istor 78 connects transistor 72 to the output of the
11 comparatow op amp 55. Transistor 71 is normally biased ON
12 through tt.se series-connected resistors 75 and 76. This turns ON
13 the fixst LED 68 which, for instance, may be a green LED.
14 Conversel~r, the second LED 69 may be a red LED. However, the
second trxnsiator 72 and thus the red LED 69 are biased off at
16 that point.
17 As signals having frequencies in the monitored fractal
18 subset or other band of interest occur at the output 43 of the
19 non-linear processor 42 or modulator 142 or mixer 442 in the
embodiments of Figs. 2, 4 and 5, and thereby at the input 57 of
21 the circuit 54 connected thereto, the output of the comparator
22 55 goes positive, turning the transistor 72 ON, and shutting the
23 transi:~tor~ 71 OFF. This turns the red LED 69 ON and turns the
24 green LED 68 OFF, thereby indicating to an observer that
frequenciE:s in the band of interest for arc detection are
26 occurring, such as through a disturbance that may be, but not
27 necessarily is indicative of an electric arc 22. The gain of the
28 circuit 5~ may be adjusted to avoid sharp transitions in
29 switching states. This helps a user gain a 'qualitative feel'
for the amplitude of the signal at that point by gauging the
31 mixture of red and green LED colors. The unidirectional current
32 conducting device 66 may be omitted, if the circuit of Fig. 7 is
33 so used in. any of the circuits of Figs. 2, 4, 5 and 6.
34 Alternatively, the terminal 57 of the display circuit 54 may
be connected to the terminal 45 in Figs. 2, 5 or 6, or a
36 duplicate of the circuit shown in Fig. 7 may be so connected to
37 that terminal 45 and thereby to the output of the bandpass filter
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
38
1 or IF amplifier 44, 144 or 244.
2 Such display stage 54 then indicates through its red LED an
3 occurrence of wideband signals in a bandwidth of interest, such
4 as in the monitored fractal subset; a well-known criterium of arc
signatures. Gain adjustments in such circuit 54 again may give
6 a user a 'qualitative feel' With respect to picked-up wideband
7 signals at that point by gauging the mixture the red and green
8 LSD colors.
9 However, a wideband signal is not an arc signature unless
it displays chaotic frequency changes. Accordingly, the
11 circuitry shown in Fig. 7 may be used as a final display stage
12 is the monitoring circuits shown in Figs. 2, 4, 5 and 6. For
13 instance, the input terminal 57 of the circuit 54 may be
14 connected to the alarm output terminal 49 shown in Fig. 2. In
fact, the circuitry of Fig. 7 from terminal 57 through op amp 55
16 may be used as liming circuit 58 and as comparator 55 in the
17 above mentioned timing and level sensing circuitry 48 shown in
18 Fig. 2.
19 The circuitry 54 may thus be used to detect and to display
a pickup of a disturbance or signal that is not only wideband in
21 the region of interest, but that is also chaotic in nature as an
22 arc signature is,. and that is sustained for a period of time,
23 such as determined by the RC component 58.
24 As disturbances or signals at wideband frequencies in the
range of interest occur and vary chaotically, the output of the
26 comparator 55 at circuitry 48 and terminal 49 goes positive,
27 turning ON the transistor 72, and shutting OFF the transistor 71.
2 8 Thi s turns ON the red LED 6 9 and turns OFF the green LED 6 8 ,
29 thereby indicating to an observer the occurrence of an arc 22 in
line 20.
31 The embodiment shown with the aid of Fig. 7 thus provides
32 a prewarning of a possible electric arcs preferably in two or
33 three stages, culminating in a display of an occurrence of a
34 chaotic wideband signal in a bandwidth of the monitored fractal
subset, or otherwise in a bandwidth of interest such as described
36 above in connection with these Figs. 2, 4, 5, 6 and 7.
37 Principles and circuitry herein disclosed may be employed
CA 02298789 2000-O1-28
WO 99/09424 PCT/US97/14497
39
1 in various arc monitoring functions, such as mentioned above.
2 In the case of such uses as research, development, and
3 maintenance in such areas as internal combustion engines or
4 electric ignition, electric welding, and electric lightiag, such
as mentioned above, circuitry of the type shown in Fig. 2 may be
6 employed up to the termiaal 47, with and without circuitry of the
7 type shown in Figs. 4, 5 and 6. The kind of signal display
8 stages shown in Fig. 7 aad more sophisticated signal display and
9 evaluation stages may be used in such cases.
This extensive disclosure will render apparent or suggest
11 to those skilled in the art various modifications and variations
12 within the spirit and scope of the invention.
