Correspondence

VACUUM ARCS: CATHODE-SPOT COMPONENTS AND but little detail is given. However, most observers have used
CURRENT DENSITY a low-resolution-power photographic technique, which did not
Indexing terms: Arcs, Cathodes allow studies of the cathode spot structure to be made.
Abstract There is evidence that the cathode-spot fragments consist of
smaller constituents representing the actual emission sites.
A series of high-speed frame and streak photographs These have been observed by Kesaev7 on mercury cathodes
of vacuum-arc-cathode spots, with magnifications and thin-film solid-metal cathodes (magnification x 250), and
of about x 20, has been used to investigate spot be Seeker and George4 when studying the erosion marks left
structure. The size of a cathode spot was deter- on solid-metal cathodes (magnification x 5000). However, no
mined with a microphotometer, and the current further division of the fragments could be detected in the
densities obtained were in the range 5xi0°-10 1] - A/m 2 . present studies when the photographs were examined visually
or analysed with a microphotometer.
A sample densitometer trace, with, the direction of the scann-
ing perpendicular to the streak, is shown in Fig. 2! Nearly
Measurements of the arc-cathode current density in low- all the fragment profiles observed2 can be approximated by a
pressure1 discharges have been reported several times. bell-shaped function exp {— (x/w) } , where w is the halfwidth
Froome *2 has used optical observations of the discharge- at the 0*608 level. Therefore integration due to streaking does
cathode region, and Wroe3 and Seeker and George4 have not alter the shape of the profile, and traces obtained from
examined the erosion marks left by the arc on the cathode both frame and streak photographs can be analysed in a simi-
surface. Recently, Sanger and Seeker5 have reviewed previous lar manner.
measurements of cold-cathode current densities.
Further, it is assumed that the flat top portion of a profile
This letter describes experimental results obtained in vacuum (Fig. 2) can be associated with the actual emitting area on the
arcs with a cathode consisting of a 2 mm-thick copper strip cathode surface. This region coincides with that of the high-
sandwiched between two 1 mm molybdenum sheets. This est density of the excited atoms, which is the most significant
arrangement limited the cathode-spot movement, which is source of light, and envelops the zone in which the primary
essential for high-magnification optical studies. The spot electrons experience inelastic collisions. This means that
usually remained on the copper part of the cathode. A con- the dimension of the densitometer trace may be equal to or
stant-current supply over a period of 10 ms was used, the greater than the dimension of the electron-emitting area.
discharge current being 25-100 A. The steeply sloping parts of the profile correspond probably
Photographs of the cathode surface at a magnification of to the constriction zone of the discharge, and the less steeply
about x 20, were obtained, using an image-convertor camera, sloping wings may be associated with the plasma column.
in both framing and streaking modes of operation. In the
former case, the exposure time was 10 \xs, and, in the latter
case, the writing speed was 0*25 mm/Vs. It was found that
the streak orientation with respect to the cathode did not
essentially influence the results.
Sample streak photographs of the cathode spot are illustrated
in Fig. \. The cathode spot is seen to experience continuous
rebuilding in the form of motion and variations in the size
and brightness of its constituent parts (or spot fragments)
as well as fluctuations in the number of fragments. The aver-
age number of fragments is approximately 1*2 for 25 A arcs
and approximately 2'5 for 50 A and higher currents.
The existence of spot fragments in low-pressure arcs with 6
metal electrodes has been reported recently by Zykovaet al.,

x1
0 2 4 6 8 °
x, m
Fig. 2
Sample densi/omeler trace
Magnification x 250

The diameter of an emitting area was estimated by extra-

polating the steeply sloping sides of the density profile and
finding the points of intersection with a horizontal line drawn
tangentially to the flat top of the profile.
The figures for the current density deduced from the densito-
meter traces give, therefore, an estimate of some minimal
value of the current density at the cathode surface. The
experiments reported here do not resolve fragments into
smaller emission sites [or microspots (cells), according to
Kesaev's terminology]. If microspots do exist in these ex-
perimental conditions, the real current density must be even
Fig.l higher than the values deduced from the present observations.
The current density within an individual fragment of a spot
Streak photographs showing cathode-spot fragments cannot be directly estimated. It was assumed, therefore, that
I = 50 A, magnification * 17, writing speed 0'25 mm/\is, the current density was constant and equal over the estimated
streak perpendicular to the cathode emitting areas of all the fragments within a spot. More than
PROC. IEE, Vol. 119, No. 6, JUNE 1972 773
1000 densitometer traces were examined, and the current specimens of different thicknesses to check whether the
density in the spot was found to fluctuate within the range method of Zakrzewski and Pietras is useful in anomaly in-
5 x 109-10 11 A/m 2 . vestigations. Also,note should be taken of Fig. 5 of their
paper, which shows that the amplitude of the actual flux is
These results are in reasonable agreement with data reported about 7 or 8% below that calculated.
previously by Froome, 1>2, Wroe3 and Seeker and George.4
The instantaneous values for the measured current density The method depends on the assumption that the hysteresis
are also reasonably within the region of the current densities loop for a thick specimen magnetised slowly is applicable
required for the interpretation of electron emission within to all elements of the specimen, an assumption which ignores
the temperature-field-emission mechanism,as was discus- inhomogeneities due to domain size. For the method to be
sed in a previous publication.8 well founded, in theory, not the thickness of the specimen
but rather the mesh dimension of the grid spacing used for
21st January 1972 B.E.DJAKOV the computer analysis has to be large compared with do-
R. HOLMES main thickness. However, the mesh dimension of 0-105 mm
used by the authors is of the same order as the spacing
Department of Electrical Engineering & Electronics between domain walls in electrical sheet steel.
University of Liverpool
Brownlow Hill, Also, there are limitations on the use of the computing tech-
PO Box 147 nique, which requires a specified applied-field waveform.
Liverpool L69 3BX, England Such a method could not easily be applied to a familiar prac-
tical problem (a transformer application) in which the back
e.m.f. in the magnetising circuit has the determined wave-
form.
References
1 FROOME, K.D.: 'The behaviour of the cathode spot on an The authors are right to question calculations in which
undisturbed mercury surface1, Proc.Phys.Soc.B, 1949, 'elliptical hysteresis loops are used which do not reproduce
62,pp.805-812 the physical phenomena properly'. On the other hand, one
2 FROOME, K.D.: 'The behaviour of the cathode spot on an must be practical and keep in mind the object of the analy-
undisturbed liquid surface of low work function', ibid., sis. The computer technique presented may serve as a tool
1950,63, pp. 377-380 to help in design work to estimate losses in thick magne-
3 WROE,H.: 'The current density on the solid cathodes in tisable sections of a machine. It cannot be relied on in a
the vacuum arc', Nature, 1958,182, pp. 338-344 rigorous research investigation of anomalous magnetisation
4 SECKER, P. E., and GEORGE, I. A.: 'Preliminary mea- losses. However, equally, the use of elliptical hysteresis
surements of arc cathode current density', J.Phys.D, 1969, loops, with the mathematical convenience of a complex per-
2, pp.918-920 meability containing two parameters which can each be
5 SANGER,C.C.,and SECKER, P.E.: 'Arc-cathode current- varied with frequency and field amplitude, may yet provide
density measurements', J. Phys. D., 1971,4, pp. 1940-1945 a more effective practical basis of computation.
6 ZYKOVA, N. M., KANTZEL, V. V., RAKHOVSKII, V. I.,
SELIVERSTVOVA, I. F., and USTIMEZ, A. P.: 'Dynamics It is of prime importance to realise that, even in thick speci-
of development of the cathode and anode regions in elec- mens, the hysteresis loop itself may be distorted and rounded
tric arcs', Sov. Phys. -Tech. Phys., 1971,15, pp. 1844-1849 off into a rough elliptical form owingtothe microscopic eddy-
current effects related to the domain structure. In my own
7 KESAEV, I.G.: 'Cathode processes in electric arcs' expermentsA on thick specimens using sinusoidal magnetis-
(Nauka, Moscow, 1968) ing field waveforms at 100-400 Hz, I found that the anoma-
8 DJAKOV, B. E., and HOLMES, R.: 'Cathode spot division lous eddy-type effect was smaller by an order of a factor of
in vacuum arcs with solid metal cathodes', J.Phys.D., ten than was usually evident at 50 Hz in thin sheet-steel
1971,4,pp.504-510 specimens. However, there was a definite effect as if the
hysteresis loop was opening out as frequency increased.
The loss angle attributable to hysteresis increased with
frequency. This implies a time-lag effect in the change of
flux density resulting from field change. As frequency in-
creases, a sinusoidal field of fixed amplitude would then
result in a smaller and smaller flux-density amplitude,
METHOD OF CALCULATING THE ELECTROMAGNETIC consistent with the discrepancy between the calculated and
FIELD AND POWER LOSSES IN FERROMAGNETIC MATERI- measured flux amplitudes of Zakrzewski and Pietras's Fig. 5.
ALS, TAKING INTO ACCOUNT MAGNETIC HYSTERESIS
14th December 1971 ' H.ASPDEN
Zakrzewski and Pietras suggest in their paper [Proc. IEE,
1971,118, (11), pp. 1679-1685] that computer analysis of wave- Acres High
form distortion due to hysteresis may explain the problem Hadrian Way
of 'anomalous losses which are observed in the cold-rolled Chilworth
steel sheets'. Then, applying their analysis to an actual Southampton, England
specimen, they obtain a theoretical loss which is only 'some
5-5% greater than the calculated losses', stated to be 'good
agreement by any standards'. The basis of the paper is that the static hysteresis loops
are sufficient to describe the magnetic characteristics of
The reader should be cautious about applying this method, the steel specimen when determining its electromagnetic
because the assumptions implicit in the analysis may not behaviour under different sorts of alternating excitation.
apply, particularly in thin low-loss materials. The experi- The numerical method adopted does not require simplifica-
mental result may be misleading, because only one test tions as do analytical methods, so that more detailed physical
figure on a sheet specimen is reported using sinusoidal field features may be incorporated in the calculations.
excitation at 50 Hz. The specimen was 2-1 mm thick, many
times thicker than the sheet steels in which anomalous loss- I did not intend to claim that a calculation of the kind given
es are evident. The test involved an unusually high degree in the paper did eliminate anomalous loss, but suggested
of skin effect compared with the normal magnetisation con- that it might do so. Future work is to be directed at find-
ditions in power transformers. If, for example, there is ing the discrepancy in thin cold-rolled steel sheet between
significant anomalous loss, and it is truly due to a normal- the loss as calculated by the method described in the paper
type eddy-current effect, the penetration of the field will be and the loss as measured. If the two losses are equal, the
less than calculated because the screening effect is under- anomalous loss will indeed have been eliminated; if they
estimated. Thus, because the controlled parameter is the are not equal, one will have a more realistic estimate of
sinusoidal surface field, the average magnetic flux will be the magnitude of the anomalous losses. It must be agreed
less, and so the local hysteresis loss, which depends on the
maximum flux, must be less than calculated for no-anomaly that a substantial number of experiments in different
conditions. It follows that experimental agreement of total materials under different conditions would be needed to
loss (hysteresis plus eddy) with a theoretical figure is no confirm the elimination of the anomalous losses.
warranty that the analysis is valid. It needs a comprehensive The paper illustrated the solution of the problem in which
research investigation at different frequencies and with the magnetic field strength Ho is varying sinusoidally. I .
774 PROC. IEE, Vol. 119, No. 6, JUNE 1972