You are on page 1of 6

Remanent flux in current-transformer cores

R. G. Bruce, B.Sc, Graduate I.E.E., and A. Wright, Ph.D., M.Sc, A.M.I.E.E.

Synopsis
The presence of remanent flux in current-transformer cores can affect the behaviour of associated protective
schemes and metering equipment.
At the instant of interruption of a system fault, the core flux density in current transformers carrying
the fault current may be very high, perhaps even at saturation level, and remanent flux densities up to the
maximum possible for the particular core material may be left. The paj>,er considers the subsequent operation of such transformers, and it is shown that some residual flux may remain in a core indefinitely, since
the passage of normal load current may not bring about complete demagnetisation.
It is concluded that the large flux variations, which are normally present in the cores of metering
transformers, will cause such large amounts of any residual flux present to be eliminated that the accuracy
will only be affected to a negligible degree. In protective current transformers, however, where the normal
flux variations are small, any residual-flux reduction will tend to be small, and protective schemes must
be designed to allow for any effects which remanent flux may cause.
1

Introduction

The presence of remanent flux in a current transformer


would cause the core flux, in normal steady-state conditions,
to operate round a minor hysteresis loop displaced along the
B axis of the B/H diagram. This would, in general, reduce
the accuracy of the current transformation. It is therefore
important to know whether appreciable amounts of residual
flux may be left in a current-transformer core, and, if so,
whether it is ultimately removed by the passage of normal
load current. It is also important to know how much the
performance of transformers may be affected by any residual
flux which can remain in them.
In current transformers which are used to supply protective
equipment, the presence of remanence would reduce the
available flux swing in one direction; this would make the
avoidance of saturation during fault conditions more difficult.
In those unit schemes of protection where the output of two
identical current transformers is balanced, saturation is not
important provided it occurs simultaneously in each. However, an internal fault, say, may leave different amounts of
remanent flux in the two cores, and maloperation may occur
during a subsequent through fault, since the transformers
would now tend to enter saturation at different times.
These effects were discussed by Gray and Wright,1 and
Seeley2 showed that maloperation of balanced protective
schemes may be caused by residual magnetism in one or more
of the associated current-transformer cores. He showed,
further, that a significant proportion of the residual magnetism
in a core may still be present after 24 h under normal operating
conditions. No investigation was made, however, to determine
how the residual-flux-density reduction depends on the magnitude of the alternating flux density which is subsequently
applied, nor was a study made of how rapidly the reduction
occurs.
It was felt by the authors that a more complete investigation
of the phenomena was warranted.
Ways in which residual magnetism may
be left in current-transformer cores
There are a number of conditions which may leave
residual magnetism in the cores of current transformers
connected to a power system; e.g. the magnetising-current
inrush of a power transformer. In general, however, the

Paper 4984 J, first received 28th July and in final form 20th December
196S
Mr. Bruce and Dr. Wright are with the Department of Electrical and
Electronic Engineering, University of Nottingham, Nottingham,
England

PROC. IEE, Vol. 113, No. 5, MA Y 1966

greatestfluxvariations imposed on a transformer will be under


fault conditions; the magnitude of residual flux which may be
left by such conditions is examined in Section 2.1.
2.1

Fault clearance during steady-state conditions

For simplicity it will be assumed that, under a certain


fault condition, the flux in a current-transformer core is
operating round the hysteresis loop shown in Fig. 1. Since
circuit breakers normally interrupt a fault at a current zero,

magnetising
force H

Fig. 1
Hysteresis loop of core material

the operating conditions of the iron at this instant will tend


to be at one,of two fixed points on the hysteresis loop, their
position depending on the burden power factor. With a
purely resistive burden the flux will be at a maximum, i.e. at
A or B, and with purely reactive burden it will be at P or Q.
For the working point to remain at A, say, a direct magnetising force Hx must be available, and can only be provided
by a direct current in the secondary circuit. To support this,
the core flux must decrease to induce the necessary direct
e.m.f. The operating point of the iron will thus progress
along the hysteresis loop until equilibrium is reached at
point C, at such a speed that the rate of flux change at every
point is sufficient to supply the magnetising force maintaining
the iron at that point. The time constant of this decay will
clearly be dependent on the losses in the transformer core
and secondary circuit.
To maintain a negative magnetising force when a fault is
interrupted at point P, a positive flux movement must occur,
and equilibrium will be attained at R. Residual fluxes up to
915

the magnitude at C or D may be left, depending on the burden


power factor, although the maximum remanence will be left
by the range of lagging power factors, where interruption
occurs between A and C or B and D. Because metering current
transformers usually operate over an appreciable flux-density
range under normal load conditions, it is highly likely that
saturation of the core will occur during a fault. On clearance
of the fault, a residual magnetism of the maximum possible
for the particular core material may remain. On the other
hand, protective current transformers, because of their need
to function satisfactorily under fault conditions, are often
designed to operate at very small flux levels under normal
load conditions, and thus the steady-state flux variations
during a fault may not leave a significant residual magnetism.
2.2

Interruption of faults while transients are present

The effect of an exponentially decaying unidirectional


component of fault current is well known.3'4 To support a
direct component of secondary current requires a continuous
unidirectional flux change, which results in a high flux level
during the transient period. It can be shown that, with a
resistive burden, the resulting flux variation is of the form
shown in Fig. 2. The corresponding exciting-current waveform

time

Fig. 2

Flux variation following a system fault


Total flux
Steady-state component
Transient component

will be similar; a d.c. component of exciting current must


eventually decay to zero, the time constant depending upon
the losses in the transformer core and secondary circuit. This
decay will be associated with a corresponding flux decay,
however, owing to the nonlinear nature of the core material,
equilibrium may be obtained about a residual magnetism.
The minor hysteresis loop so formed is centred on the H axis,
so that no d.c. component of magnetising current is present.
Thus, if a fault containing a d.c. transient is interrupted
within the first few cycles, the core may be at a high flux
density, perhaps even saturated, and will decay from this
level, leaving considerable residual magnetism. Interruption
will again occur at fixed points on the flux waveform, and
maximum remanence will be left in a current transformer with
a resistive burden, provided the fault is cleared after the
correct halfcycle.
The presence of inductance in the secondary circuit reduces
the flux transient given by a particular condition, and also
the flux density at the instant of interruption. Nevertheless,
appreciable residual flux may be left unless the burden power
factor is very low.
3

Behaviour obtained on energising


current transformers which have
remanent flux in their cores
Because of the nonlinear nature of the magnetisation
characteristics of core materials, it is not possible to determine
easily, analytically, the positions ;to which hysteresis loops
will move for given operating conditions. The behaviour was
916

therefore studied experimentally on current transformers


incorporating cores of each of ths three basic types of
material: hot-rolled silicon-iron alloy (Stalloy); cold-reduced
grain-oriented silicon-iron alloy (Crystalloy); and nickeliron alloy (Mumetal). In each case the windings were continuously wound on ring-shaped cores.
3.1

Measurement of core flux

The investigation required that measurement be made


of the residual flux density existing in a continuous ring core.
For this the more usual methods of flux measurement, which
rely on the presence of an air gap, were not practicable.
A method which has been used in the past relies on the
variation of incremental permeability with residual magnetism.2 This method, however, was thought to be unsuitable,
because with low magnetising forces the permeability of many
ferromagnetic materials, particularly silicon-iron alloys,
varies considerably with time, following a magnetic or
mechanical disturbance of the material.5
The method which was finally adopted involved the
destruction of the residual flux present. However, rather
than obtaining a direct measurement by reducing the
remanence to zero, it was found more convenient to drive
the core well into the saturation region with a suitably large
direct magnetising force. The resulting flux change was
measured using a fluxmeter. The final state of the iron was
thus almost the same, irrespective of the initial core flux, and
this resulted in a near linear relationship between fluxmeter
deflection and the residual flux density.
An obvious disadvantage of this method was that it
destroyed the quantity being measured, necessitating the
initial core flux condition to be restored after each measurement. Apart from this, the method proved to be very
satisfactory.
In addition to the above measurement, it was interesting
to observe the waveforms of both B and H which were
obtained on energising a current transformer containing a
remanent flux. A current transformer having a 1 : 1 turns
ratio was used with its primary and secondary circuits interconnected, so that the difference between the currents of the
two circuits passed through a small linear resistor. Under
this condition, the difference current must equal the transformer exciting current. The waveform of flux density was
simultaneously obtained by integrating the e.m.f. induced in
an additional winding. An operational feedback amplifier was
used for this.
3.2

Change in residual flux produced by the passage


of a sinusoidal alternating current

In this Section, variations in remanent flux brought


about by steady-state alternating fluxes only will be considered, i.e. resulting from gradual changes in primary current.
The behaviour obtained with currents containing transient
components will be discussed later.
The conditions which would be experienced when normal
load current is resumed, following the clearance of a fault
which has left residual magnetism in the core of a current
transformer, is now examined. As has been explained, the
minor hysteresis loop required by the operating condition
must take up a position centred on the i/axis so that there
is no direct component in the exciting current. The negative
halfcycle of the required magnetising current will result in a
demagnetisation of the core which depends on the peak value
of that current, as shown in Fig. 3; i.e. if the load current
were gradually increased from zero, the negative tips of the
minor loops generated would trace out a path corresponding
to the demagnetisation curve for the iron from the initial
value of residual flux density.
If, after reaching a particular maximum primary current,
the latter is gradually decreased, the minor hysteresis loops
generated will form loops inside the previous larger loop.
PROC. IEE, Vol. 113, No. 5, MA Y 1966

Reduction of this current to zero would cause the working


point of the iron to be cycled towards the centre of these
loops, in exactly the same manner as the complete demagnetisation of a specimen.
flux
density
B

71

could only be of significance with silicon-iron core materials.


It is known that both the above effects gradually reach an
equilibrium condition, and would not finally result in the
complete removal of a sufficiently large residual magnetism.
The extent of the reductions in residual magnetism brought
about by impressing various amplitudes of alternating flux
were measured in current transformers having each of the
three basic core materials. Reductions from three initial
magnitudes of residual flux density were determined, the
greatest in each case being close to the maximum possible
in that particular material. Fig. 4 shows the immediate
reductions in residual flux which were obtained.
0-8

magnetising
force
H

Fig. 3

Demagnetisation produced by the positioning of minor hysteresis loops

S 0-4

It is important to note that subsequent increase and


decrease of current through the transformer will cause no
further change in the residual magnetism left in the core,
provided that the current does not exceed its previous maximum value. Increasing the current beyond this limit would
produce further demagnetisation in the same manner as
before.
3.2.1

Experimental study

Since, under steady-state conditions, the flux variation


in a current transformer is affected only by the burden
impedance and not its power factor, it was convenient during
the following tests to operate the transformers with resistive
burdens.
The following procedure was adopted to examine the
extent by which various amplitudes of alternating flux affect
a given remanent flux. A known residual flux density was
left in the core of the transformer under test, by passing a
suitable direct current through a winding. The primary current
was then slowly increased to a value giving the required
magnitude of alternating flux density; on gradual removal
of this current, after the desired time interval, the residual
flux density remaining in the core was measured.
With each of the core materials tested, the passage of a
given primary current produced a very rapid initial reduction
of residual magnetism; on leaving this current flowing for a
considerable period of time, only a slight further reduction
was given. For example, with an initial remanence of
0-61 Wb/m2 in a current transformer having a Stalloy core,
the passage of primary current, corresponding to a peak flux
density of 0-11 Wb/m2, immediately reduced the residual
magnetism present to 0-33 Wb/m2. After a number of hours
the additional reduction obtained was of the order of
0-02 Wb/m2. Beyond this it was not possible to detect any
further decrease in magnetism, thus indicating that an equilibrium position is finally reached at which the transformer
will continue to operate indefinitely.
This small additional decrease in remanence probably
results mainly from causes not directly associated with the
preceding theory. Firstly, it is well known that mechanical
vibration causes a reduction in magnetisation; thus, even if
external vibration of the core were eliminated, it is conceivable that magnetostriction, particularly with large amplitudes of alternating flux, would be sufficient to cause a small
reduction in residual magnetism. Also, because of the
phenomenon known as the 'time decrease in permeability',5
it would be expected that at low alternating flux levels, the
magnetising current required for a particular operating condition would gradually increase with time, so producing a
further reduction in residual magnetism. This, however,
PROC. IEE, Vol. 113, No. 5, MA Y 1966

0-2
04
06
peak value of alternating flux density, Wb/m2

(0
1-6

0-8

03
0-6
0-9
peak value of alternating flux density,Wb/m2

(ii)
0-4

0-2

0-4
peak value of alternating flux density, Wb/m2

Fig. 4

0-6

(iii)

Decrease in residual magnetism produced by normal operating


conditions in three core materials
(i) Stalloy
Initial residual flux density a 0-74 Wb/m*
6 0-61 Wb/m*
c 0-37 Wb/m*
(ii) Crystalloy
Initial residual flux density a 1 -43 Wb/m*
b 1-12 Wb/m*
c 0-68 Wb/m*
(iii) Mumetal
Initial residual flux density a 0- 33 Wb/m2
* 0-22 Wb/m*
c 0-13 Wb/m*

917

The shapes of these graphs are dependent on the magnetisation characteristics and hysteresis loops of the particular
materials considered. In each case, relatively largefluxdensities
are required to remove all the residual magnetism present.
Metering current transformers usually operate at large
flux densities, and a large proportion of any residual magnetism
left in them will be removed. The effect on their accuracy
will normally be negligible. To demonstrate this, a Mumetalcored current transformer was tested while operating with a
peak alternating flux density of 0-3Wb/m2, first, after the
core had been demagnetised, and then with a residual magnetism of 0-33Wb/m2. It was found that, in the latter case,
the exciting current was only of the ord* of 3 % greater than
that required under normal conditions. Taking as an example
a metering current transformer of class AM, which has a
maximum ratio error limit of 1 0 % , and a phase error
limit of 30', a 3 % increase in exciting current on such a
transformer operating at these limits would cause a change
in the respective errors of approximately 0 03% and 0-9'.
The flux variation in protective current transformers may
be very small during normal load conditions, and, consequently, a considerable amount of residual flux may remain
in them. This flux will only become significant when the
associated protective equipment is required to deal with the
next power-system fault. If this next fault is within the protected zone, the effect of the residual flux on the protectiveequipment fault setting will be the factor which is of interest.
The magnitude of this effect will depend on the type of
protective equipment in use. To give an indication of the
possible changes in setting, two extreme cases are considered:
If low-impedance relays are used in the scheme, the flux
swings in the current-transformer cores, at the fault-setting
current level, will tend to be small, and the reduction in the
residual flux will be negligible. The incremental permeability
will be low and the exciting current will be increased above
the expected value. It was found that a Stalloy-cored transformer operating at a peak alternating flux density of
0 03Wb/m2, after a remanence of 0-74Wb/m2had been left
in the core, required an exciting current 58% greater than
that required to support the same flux with the core demagnetised. Although this increase is large, the effect on the
transformer secondary current will be small, as the exciting
current is not likely to exceed 2-3 % of the primary current.
The protective setting would thus rise by not more than about
2% of its nominal value, a change which would be of no
practical importance.
If high-impedance relays were used in the protective
equipment, the transformer-core flux swings would be large
at the fault setting current, and most of the residual flux
would be destroyed. The incremental permeability would not,
in these circumstances, be appreciably below that obtained
with zero residual flux; e.g. the above Stalloy-cored transformer operating with a peak alternating-flux density of
0-3Wb/m2, after a residual flux density of 0-74Wb/m2 had
been left in its core, required an exciting current 7-5%
greater than normal. Since most of the primary current acts
as exciting current when high-impedance relays are in use,
the protective fault setting rises by almost the same amount
as the exciting current. Increases up to 7 5 % of the nominal
setting may thus be expected. Again, this increase would
normally be of little importance.
If the next fault, after residual flux has been left in a core,
is external to the protected zone, and the required flux swing
is in the same direction as the residual flux, the available
swing before saturation is reached will be much reduced. For
those schemes where saturation must not occur, the transformers must be made large to overcome this effect; e.g. if a
residual flux of 60% of the saturation level can be retained,
a core would have to be provided of 2-$- times the size
which would be needed if residual flux could not be left,
since only 40% of the flux swing can be regarded as always
being available.
918

It is not possible to discuss the exact significance of residual


flux in those schemes where some saturation is possible
during external faults. In general, however, the presence of
remanent flux leads to larger unbalance currents under
adverse conditions, as explained by Gray and Wright.1
Greater attention must be paid to biasing features in the
protective equipment used to overcome this effect.
The positioning of successive minor hysteresis loops resulting from a gradual increase and decrease of primary current,
through a current transformer containing remanence, are
shown in Fig. 5. These were obtained by simultaneously displaying the waveforms of B and H on an oscilloscope. It was
demonstrated that further increases and decreases of primary
current, up to this previous maximum, gave no further
decrease in residual magnetism; the loops generated by this
were merely a similar pattern to those indicated in Fig. 5b.

Fig. 5
BjH diagram showing successive minor hysteresis loops when primary
current is gradually (a) increased, (b) decreased
Initial remanence = 0-61 Wb/m2
Peak values of alternating flux density in the minor loops = 0,0 022,0 044,0 066,
0-11 Wb/m*

magnetising

force
H

Fig. 6
Minor hysteresis loops generated on energising the primary circuit
flux
density
B

initial magnetisation
curve y - "

TR

magnetising
force
H

Fig. 7
Additional magnetisation produced when the primary circuit is
energised
PROC. 1EE, Vol. 113, No. 5, MA Y 1966

3.3

Change in residual flux produced by primary


current transients

To simplify this examination initially, the primary


circuit will be assumed to have unity power factor; i.e. no
d.c. transient will be present when the circuit is closed.
'"" -"I1

I p ,5A r.m.s.
H,At/n

300
100
0-9

"m:,\mi mm'

Referring to Fig. 6, consider that residual magnetism represented by OR has been left in a current transformer core,
and that the primary circuit is then closed at a time such that
the initial flux swing required is from peak to peak of its
steady-state value. In the case of an upward flux swing, a

vvTvvTvJ
1

,!

AAAAAAAftA WWAMM WAAAAAAA W W M i W W W W

M WWWW\A wmvww W\,V\A

0 6

/ *
B,Wb/m2 0-3
0
-0-3
-06

iD,5A rms.

f-5' ' , " ' . '

... .

, +100r
[
H,At/m
0-6
, 0-3
B,Wb/m
0
-0-3
-O6

i
I 'V

I
.V.VAV//J

,,,

....

'

I p ,5A nm.s.r

wvwvw
0-6
03
B,Wb/m2 0
-0-3
-06

^ W W f WWWWt! VWyWAV WVWWW WWMW WWV^^.V A/WfW

Fig. 8

Oscillograms and B\H diagrams showing the transients produced on energising the primary circuit
Core material: Stalloy
Initial remanence <= 0 _ _ ..
Primary circuit X\R = 5-4
Supply frequency = 50c/s
a, b Resistive burden
Peak value of steady-state alternating flux density = 0Transient flux of same sense as remanence
c, d Conditions as for (a) and (6), except transient flux is of opposite sense to the initial remanence
e, f Inductive burden
Peak value of steady-state alternating flux density = 0-08 Wb/m2
Transient flux of opposite sense to remanence

PROC. IEE, Vol. 113, No. 5, MAY 1966

919

minor hysteresis loop, as indicated by RS, will be followed.


It is clear, however, that the iron cannot continue to operate
around such a loop, owing to its offset nature, and, to reach
equilibrium, it will move to the left, the negative tip following
the demagnetisation curve from the point R. The speed of
this motion will be such that at every point the rate of change
of remanent flux is sufficient to maintain the direct component
of magnetising force at that point, and it will therefore be
dependent upon the losses in the transformer core and
secondary circuit.
On the other hand, if the initial flux swing had been in
the opposite direction, then, during this swing, the iron would
be driven down the demagnetisation curve RT, say, and then
during the next cycle would operate around a minor loop,
such as TU. The position of this loop would be dependent
upon the magnetic characteristics of the core material; it is
possible that the movement necessary to reach equilibrium
may be in either direction, i.e. to the left and downwards,
as before, or to the right and upwards. For switching at any
other point on the flux waveform, the first minor loop
generated would have its negative tip lying between R and T.
It should be realised that this behaviour will be modified
if the initial upward flux swing is large enough to cause
further magnetisation of the core. In Fig. 7 a loop, such as
ST, will be generated which will move towards an equilibrium
position in exactly the same manner as before.
The performance obtained when the primary current contains an exponentially decaying direct component is now
considered. The flux variation required by this condition, as
has been mentioned in Section 2.2, is an exponential transient
flux change with the steady-state variation superimposed.
Thus, in the case of a transient flux in the same sense as an
initial remanent flux, the working point of the iron, represented in the B\H diagram, will be driven upwards and to
the right in a series of loops. As before, the offset nature of
these loops will cause a subsequent movement to the left
and downwards, until an equilibrium position is reached.
A transient flux in the opposite direction would drive the
core in a series of loops down the demagnetisation curve.
In fact, for a sufficiently large transient flux, the final equilibrium loop could be centred about a remanent flux of the
opposite sense to that initially present.
3.3.1

Experimental study

Observations were made of the transients of B and


H produced on closure of the primary circuit of a current
transformer in which a remanent flux had been left. These
were recorded both as B\H diagrams and also as separate
oscillograms of the two quantities.
In order that significant transient effects could be obtained,
it was necessary to use values of primary-circuit X/R considerably greater than those which would normally be
encountered under healthy operating conditions, although
such values could be obtained under fault conditions. Since
the flux transient produced by this condition is dependent
upon the secondary-circuit power factor, the tests were
carried out with both purely resistive and highly inductive
burdens.
Typical results are shown in Fig. 8. The upper trace of the
oscillograms is that of primary current, and shows the extent
of the exponential current transient. In each case, switching,
was made near to a peak in the steady-state component of
current; the resulting transients were therefore close to the
maximum in the corresponding direction. The B/H diagrams
show the transient loops over only the first few cycles; this
was necessary owing to the very slow movement of the loops
as equilibrium was approached. The total transient can be
seen from the separate oscillograms.

920

Conclusions
The presence of remanent flux in current-transformer
cores can significantly affect the performance of balanced
forms of power-system protective equipment, and, while
present, would affect the accuracy of metering transformers as it would cause the permeability of the core to be
reduced.
It was shown in Section 2 that residual flux densities up
to the maximum possible for the particular core material
may be left in a current transformer, following the clearance
of a system fault. Higher values of remanence will, in general,
be left when burdens are of high power factor, since
greater flux densities are likely at the instant of fault
interruption.
It was found that the most satisfactory method of measuring
the residual flux in a core, and thereby studying the way in
which it decays in normal operation, was to use the destructive
method described in Section 3.1. Although this was not an
ideal method, it was found to be much more accurate than
the incremental permeability method which has been used for
this type of work in the past.
The remanent flux in a core is reduced when an alternating
current is passed through the windings of a transformer. The
degree of reduction is dependent on the magnitude of the
alternating flux produced; it is not affected by the secondarycircuit power factor, provided that the current is increased
gradually. Once equilibrium has been established, subsequent
variation of the current does not produce any further reduction in the amount of residual flux, provided that it does not
exceed the value at which equilibrium was reached.
When the primary circuit of a current transformer, in
which residual flux has been left, is energised, the final
equilibrium position of the minor hysteresis loop will depend
on the flux variation required during the transient period.
If the remanence is small or the initial flux swing large, the
equilibrium condition may be centred about a remanent flux
greater, or even of the opposite sense, than the original value.
In general, however, unless the initial remanence is very
small or the primary-circuit X/R is large, a reduction in
residual magnetism will be produced, although the reduction
may be dependent on the point on wave of switching.
The speed at which thefluxreduction occurs is affected by
the core material, but equilibrium, for any given set of conditions, is virtually achieved in a few seconds, the longer-term
small reductions probably being due to factors such as
magnetostriction.
Large alternating flux variations are needed to effect complete removal of the residual flux in a core, and, as such
variations are not likely in protective transformer cores,
protective schemes must allow for the possibility of large
residual fluxes being present. As, however, thefluxvariations
in metering current transformers are usually large, the amount
of residual flux which may be present in them will be very
small, and measurements show that the decrease in accuracy
due to this will generally be negligible.
5

References

1 GRAY, w., and WRIGHT A.: 'Voltage transformers and current


transformers associated with switchgear', Proc. IEE, 1953, 100,
Pt. II, p. 223
2 SEELEY, H. T. : 'The effect of current-transformer residual magnetism
on balanced-current or differential relays', Trans. Amer. Inst. Elect.
Engrs., 1943, 62, p. 164
3 WRIGHT, A.: 'An investigation into the performance of current
transformers and relay circuits employed in the earth fault protection of power systems', M.Sc. Thesis, Durham University, 1954
4 MATHEWS, P. : 'Protective current transformers and circuits' (Chapman and Hall, 1955)
5

WEBB, c. E., and FORD, L. H.: 'The time-decrease of permeability

at low magnetising forces', / . IEE, 1934, 75, p. 787

PROC. IEE, Vol. 113, No. 5, MA Y1966

You might also like