An electromagnetic flow meter is disclosed which can be used to
measure flow in low conductivity fluids, for example alcohol, turpentine,
oil or other organic solvents. In one embodiment, the electromagnetic
flow meter includes an elongate flow conduit having a direction
of elongation corresponding to a direction of fluid flow, magnetic
field generating means for generating a magnetic field across the
flow conduit and potential sensing electrodes for sensing a potential
generated by the magnetic field in a fluid flowing through the conduit.
Both the field generating means and the potential sensing electrodes
of the flow meter are elongate in the direction of flow.
1. An electromagnetic flow meter comprising: an elongate flow conduit
having a direction of elongation corresponding to a direction of
fluid flow; magnetic field generating means for generating a magnetic
field across the flow conduit; potential sensing electrodes for
sensing a potential generated by the magnetic field in a fluid flowing
through the conduit; characterised in that both the field generating
means and the potential sensing electrodes are elongate in the direction
2. A flow meter according to claim 1 wherein the electrodes are
longer than twice the mean conduit cross-sectional dimension, preferably
at least three times, more preferably at least five times said dimension.
3. A flow meter according to claim 1 wherein the elongate electrodes
have a length in the direction of flow substantially greater than
the electrode dimension in a perpendicular direction.
4. A flow meter according to claim 3 wherein the electrode length
in the direction of flow is at least twice the perpendicular dimension,
preferably at least five times, more preferably at least ten times
the perpendicular dimension.
5. A flow meter according to claim 1 wherein the electrodes have
a length in the direction of elongation of the fluid conduit at
least equal to the mean conduit cross-sectional dimension.
6. A flow meter according to claim 1 wherein the length of the
electrodes is at least half the length of the field generating means.
7. A flow meter according to claim 1 arranged to measure flow in
a fluid having conductivity of the order of 10.sup.-4 mhos/m or
8. A flow meter according to claim 1 further comprising an input
amplifier connected to the potential sensing electrodes having an
input impedance of at least 10.sup.12 ohms, more preferably of the
order of 10.sup.15 ohms.
9. An electromagnetic flow meter comprising an elongate flow conduit
having a length along a first direction corresponding to a direction
of flow through the conduit, a width in a second direction substantially
perpendicular to the first direction and a height in a third direction
substantially orthogonal to the first and second directions; at
least a first field generating coil positioned above the flow conduit
for generating a magnetic field substantially along said third direction;
first and second potential sensing electrodes positioned at opposite
sides across the conduit for sensing potential developed across
the conduit width; wherein the field generating coil has a length
in the first direction at least as long as the geometric mean of
the height and width of the conduit and wherein the first and second
electrodes also have a length in the first direction substantially
greater than the height of the electrodes in the third direction
and preferably at least as long as the geometric mean of the height
and width of the conduit.
10. A flow meter according to claim 9 having a second field generating
coil opposite the first field generating coil.
11. A flow meter according to claim 9 wherein the conduit is substantially
circular in cross-section.
12. An electromagnetic flow meter adapted and arranged to obtain
a measurement of flow of a fluid having a conductivity of the order
of 10.sup.-4 mhos/m or less.
13. An electromagnetic flow meter adapted and arranged to obtain
a measure of flow in an oil-based fluid.
14. An oil, turpentine or organic fluid flow meter comprising an
electromagnetic flow meter having an elongate field generating means
and elongate potential sensing electrodes and a high impedance input
amplifier for measuring a potential generated across the fluid to
derive the measure of flow therefrom.
15. An electromagnetic flow meter comprising means deriving a measure
of flow from a quantity of oil flowing through an elongate metering
conduit of the flowmeter.
16. A method of deriving a measure of flow of a fluid having a
conductivity of the order of 10.sup.-4 mhos/m and/or an oil-based
fluid, the method comprising passing the fluid through an electromagnetic
flow meter having an elongate field generating coil and elongate
potential sensing electrodes and deriving a measure of flow from
a potential sensed by the potential sensing electrodes using a high
 The present invention relates to electromagnetic flow meters.
 Electromagnetic flow meters have been used for many years
for measuring the flow of a conductive fluid. Electromagnetic flow
meters have a number of advantages over mechanical meters as they
can be made relatively compact and without moving parts to wear
or become damaged by inclusions within the fluid. Electromagnetic
flow meters work on the principle that a magnetic field (which may
be static or time-varying) induces an EMF across a moving fluid.
As a practical matter however, the EMF is typically small and measuring
it with practical instruments without undue sensitivity to noise
is a concern.
 It conveniently so happens that water-based fluids, which
constitute a usefully large proportion of process fluids to be monitored,
tend to have a conductivity that can be measured using an electromagnetic
flowmeter of convenient dimensions. In fact, pure (demineralised)
water has a relatively low conductivity and this may be problematic
to measure as the impedance of a practically sized meter would be
very high. However, measurement of pure water is rarely called for
as practical fluids have much higher conductivities, tap water having
a conductivity approximately 100 times greater than pure water and
seawater having a conductivity approximately 10000 times greater.
Typical water-based process fluids have conductivities falling between
the latter two. Thus, for (impure) aqueous fluids, electromagnetic
flow meters can conveniently be manufactured with practical dimensions.
 Non-polar, "insulating", fluids such as oil-based
fluids are not conventionally considered suitable for use with electromagnetic
flow meters and there are a wide variety of other metering techniques
adapted for such fluids.
 The reason for this can best be understood by considering
typical conductivity values. Fluids such as turpentine have conductivities
of the order of 2500 times lower than that of tap water and 200000
times lower than seawater. Alcohols typically have similar conductivities
(or slightly higher) and fluids such as kerosene and oils have still
 Accordingly, whilst electromagnetic flowmeters have become
widely used for measuring the flow of impure aqueous fluids, other
techniques are indicated for "non-conductive" fluids,
e.g. organic solvents, turpentine, alcohol or oil-based fluids.
 The present invention stems from an attempt by the inventors
to overcome the conventional wisdom in the art that electromagnetic
flow meters are fundamentally unsuited to measurement of "insulating"
 The geometry of an electromagnetic flow meter is usually
tightly constrained by the physical requirements of the associated
pipework, which typically means that the flow meter is formed into
a cylindrical pipe of the same diameter as the surrounding pipework.
The measuring electrodes are typically small discs. It has been
considered that changing the geometry of the pipe into a non-circular
section so that, for example, the proposed electrodes are closer
together might reduce the impedance. However, only a small reduction
can be gained (and a significant reduction would be needed) and
in any event the change in shape may cause turbulence and/or other
flow problems (such as increased flow resistance) and so may not
be possible or may be undesirable in many applications.
 Increasing the diameter of the disc electrodes may also
decrease the impedance but again only by a relatively small amount.
Furthermore, as the diameter of the electrodes increases, contamination
of a portion of an electrode may lead to a change in the measured
potential. For this reason, it is generally considered highly desirable
to keep the electrodes small in diameter. Such electrodes effectively
provide a point sample of potential, which is substantially unaffected
by partial contamination.
 Thus, minor changes in the geometry of the meter might lead
to small changes in impedance but are likely to introduce problems
and so have not appeared to provide a promising route.
 Acknowledging this, some alternative attempts have been
made to extend the use of electromagnetic flow meters to lower conductivity
fluids e.g. demineralised water. In particular, it has been suggested
[V. Cushing, "Induction flowmeter", Review of Scientific
Instruments (1958) Volume 29 pp 692-697] that using very high frequencies
may enable readings to be taken with fluids of lower conductivity.
Theoretical considerations for this approach seem encouraging. However,
due to various practical problems associated with such techniques
these proposals have not hitherto developed into successful practical
products. Nonetheless, this remains an avenue which seems worthy
of exploration and conceivably, if the practical problems could
be overcome, use of high frequencies might provide a solution to
the problem of measuring low conductivity fluids.
 The present invention, however, is based on a different
 According to a first aspect, the present invention provides
an electromagnetic flow meter comprising:
 an elongate flow conduit having a direction of elongation
corresponding to a direction of fluid flow;
 magnetic field generating means for generating a magnetic
field across the flow conduit;
 potential sensing electrodes for sensing a potential generated
by the magnetic field in a fluid flowing through the flow conduit;
 characterised in that both the field generating means and
the potential sensing electrodes are elongate in the direction of
 It should be noted that this contradicts conventional wisdom
concerning flow meter design. Conventionally, it is accepted practice
that the field generating means should extend axially either side
of the potential sensing electrodes in a flow meter by a sufficient
amount that electrical currents primarily generated as the fluid
flows into and out of the field are not generated directly in the
vicinity of the potential sensing electrodes. This implies that
the electrodes should be significantly shorter than the coil. Furthermore,
as the length of the field generating coil increases, its resistance,
weight and the cost increase, as well as the overall size of the
meter, so it is generally desirable to keep the coil as small as
possible and effectively use point electrodes, as discussed above.
It has been shown that the optimum length for a coil is slightly
less than the diameter of the conduit, typically at 0.9 times the
diameter of the conduit [M. K. Bevir, "Induced Voltage Electromagnetic
Flowmeters" Phd Thesis, University of Warwick, 1969]. As further
background, it is noted that, before this assessment of optimum
coil length was commonplace, some meters may have been constructed
with slightly longer coils, but still with short electrodes.
 In the present invention, as will be explained further below,
elongate coils are used which are deliberately beyond the conventionally
accepted "optimum" length, typically longer than twice,
three times, four times or in preferred embodiments even greater
than five times the diameter of the conduit. References herein to
conduit diameter are intended to encompass the appropriate cross-sectional
distance (or geometric mean of height and width) of a non-circular
 Pursuant to the invention, it has been appreciated that
increasing the length of the coils has the important effect of decreasing
the effective source impedance of the source of the EMF and this
can reduce the effective source impedance for a low conductivity
fluid to a usable level. This can surprisingly outweigh the expected
disadvantage of having a longer coil than is conventionally considered
 In the case of the electrodes, because of the conventional
concern about electrical currents at the edges of the field and
the desire to keep the field generating coil as small as conveniently
possible, prior art electrodes have deliberately been kept short
in the direction of flow. Although in certain cases some larger
plate electrodes have been used, in particularly in non-circular
conduits, it is generally preferred to use circular or square or
deliberately shortened electrodes in which the length of the electrode
in the direction of flow is substantially the same as or less than
the electrode dimension in a direction perpendicular to the direction
 In the present invention, in contrast, elongate electrodes
are used which preferably have a length in the direction of flow
substantially greater than the electrode dimension in a perpendicular
direction, preferably at least twice and in embodiments typically
as much as at least five, preferably at least ten and more preferably
of the order of twenty times the perpendicular dimension.
 Pursuant to the invention, it has been appreciated that
increasing the length of the electrodes can significantly decrease
the effective electrode impedance, by an order of magnitude, without
becoming susceptible to the known problems of contamination of a
portion of the electrodes associated with prior art "larger"
electrodes. Preferably, the electrodes have a length in the direction
of elongation of the fluid conduit at least equal to the conduit
diameter (or separation of electrodes or equivalent cross-sectional
dimension of a non-circular conduit). The electrodes preferably
have a length similar to the length of the field generating means,
preferably at least half the length of the field generating means,
preferably at least three quarters, typically at least 80%, ideally
of the order of 90% or more of the length of the field generating
 Such a flow meter may be reliably capable of measuring flow
in a fluid having conductivity of the order of 10.sup.-4 mhos/m
or less without the requirement for specialist instrumentation and
without being unduly susceptible to noise. Preferably, the flow
meter further comprises an input amplifier connected to the potential
sensing electrodes having an input impedance of at least 10.sup.12
ohms, more preferably of the order of 10.sup.15 ohms or more.
 In a preferred embodiment, the invention provides an electromagnetic
flow meter comprising an elongate flow conduit having a length along
a first direction corresponding to a direction of flow through the
conduit, a width in a second direction substantially perpendicular
to the first direction and a height in a third direction substantially
orthogonal to the first and second directions;
 at least a first field generating coil positioned above
the flow conduit for generating a magnetic field substantially along
said third direction;
 first and second potential sensing electrodes positioned
at opposite sides across the conduit for sensing potential developed
across the conduit width;
 wherein both the field generating coil and the first and
second electrodes are elongate along said first direction.
 Preferably the field generating coil has a length in the
first direction at least 3 times as long as the geometric mean of
the height and width of the conduit. Preferably the first and second
electrodes have a length in the first which is similar to, but slightly
less than, the length of the field generating coil.
 References to height, width, above and across are for convenience
and are not intended to imply any spatial orientation.
 Preferably the flow meter has a second field generating
coil opposite the first field generating coil. The conduit may be
circular in cross-section in which case, the height and width (and
geometric mean thereof) are all equal to the diameter of the conduit.
The electromagnetic flow meter preferably has an input amplifier
for measuring the potential across the first and second electrodes
having an input impedance of at least 10.sup.12 ohms and preferably
10.sup.15 ohms or more.
 The invention further provides an electromagnetic flow meter
adapted and arranged to obtain a measurement of flow of a fluid
having a conductivity of the order of 10.sup.-4 mhos/m or less.
The invention further provides an electromagnetic flow meter adapted
and arranged to obtain a measure of flow in an oil-based fluid.
 The invention further provides an oil, alcohol, turpentine
or other organic fluid (preferably oil) flow meter comprising an
electromagnetic flow meter having an elongate field generating means
and elongate potential sensing electrodes and a high impedance input
amplifier for measuring a potential generated across the fluid to
derive a measure of flow therefrom.
 The invention further provides a method of deriving a measure
of flow of a fluid having a conductivity of the order of 1 0.sup.4mhos/m
and/or an oil-based fluid, the method comprising passing the fluid
through an electromagnetic flowmeter having an elongate field generating
coil and elongate potential sensing electrodes and deriving a measure
of flow from a potential sensed by the potential sensing electrodes
using a high impedance amplifier.
 An embodiment of the invention will now be described, by
way of example, with reference to the accompanying drawings in which:
 FIG. 1 is a schematic diagram of a rectangular section meter
for use in explaining the theory underlying the invention;
 FIG. 2 is a schematic diagram of the current flow in a fluid
in the meter of FIG. 1;
 FIG. 3 is a graph of potential and current density against
position in the meter of FIG. 1;
 FIG. 4 schematically depicts an equivalent electrical circuit
to the meter of FIG. 1;
 FIG. 5 is a graph of resistance against the ratio l/a of
the meter of FIG. 1;
 FIG. 6 schematically depicts an equivalent electrical circuit
to a meter including measurement electronics;
 FIG. 7 shows the electric potential and the current distribution
for a meter where l/a is 3;
 FIG. 8 shows schematically a meter with long electrodes
and a cylindrical conduit in accordance with an embodiment of the
 FIG. 9 is an illustration of the coil assembly of a practical
meter of a first embodiment;
 FIG. 10 is an illustration of the first embodiment showing
 FIG. 11 is a schematic cross section of the practical meter
according to the first embodiment.
 To assist in understanding the present invention, a theoretical
explanation will first be given.
 A magnetic flow meter is in essence a type of M.H.D. (magneto-hydrodynamic)
generator. A simple form consists of a duct, lined with an electrically
insulating lining through which the fluid to be measured flows,
and in operation the fluid is made to pass through a transverse
magnetic field within the duct. The moving fluid creates an electric
field, E in a direction perpendicular to the motion, and to the
magnetic field. The strength of the electric field depends upon
the velocity V, and upon the intensity of the field B. Appropriately
placed electrodes are connected to electronic apparatus that measures
the potential difference between them and converts this into a form
that indicates the mean flow through the meter. The details of the
electronic apparatus are not germane and many conventional schemes
are known which may be used herein.
 FIG. 1 shows the layout of a simplified instrument. It has
a rectangular duct of height 2a and width w. The region over which
there is a transverse magnetic field extends for a length of 2l,
in the direction of the flow. That is to say there is a magnetic
field from z=-l to z=l. The fluid flows in the z direction with
a uniform velocity V.sub.z. Electrodes would normally be placed
at z =0 on the top and bottom surfaces of the channel.
 Currents will circulate at each end of the device where
the magnetic field, and thence the induced e.m.f. die away. This
is shown in FIG. 2.
 Due to the circulating currents, the sensing electrodes
are conventionally kept small and far from the ends of the coil,
to minimize the unpredictable effects of these currents. This simple
expedient simplifies conventional analysis and is normally sufficient
for practical meters. However, pursuant to the invention, further
analysis has been carried out, and this will now be detailed.
 The equation that governs the operation of such a meter
is the Ohm's law equation. In its' vector form this is 1 J _ = E
_ + V _ .times. B _ .
 By manipulating Ohm's law and making use of Maxwell's fourth
equation we find that U, the electric potential, can be found by
solving 2 2 U = - R m a U z ,
 where R.sub.m is the magnetic Reynolds a az number. The
magnetic Reynolds number is a dimensionless group that is a measure
of the size of the magnetic field produced by the circulating eddy
currents relative to the applied field. For most liquids (with the
possible exception of molten metals) and importantly for the low
conductivity liquids we are considering, it is negligibly small
so that we can take .gradient..sup.2U=0. This can be solved, subject
to the appropriate boundary conditions.
 This allows us to calculate the potential at y=.+-.a, the
top and bottom edges of the device. By differentiating these equations
we can find the electric field and thence, by using Ohm's law, the
current density may be found.
 FIG. 3 shows the electrical potential on the top edge of
a duct where both land a are both 0.05 m, from z=0 to z=0.15 m.
The current density at y=0 (that is to say on the axis), is also
shown. This case is similar to the one shown pictorially in FIG.
2. Thus our more detailed analysis is consistent with the conventional
simplification. It can be seen that current flows in a positive
direction throughout the field region and returns in the other direction
at the ends of the meter.
 Pursuant to the present analysis, we have treated the device
as an M.H.D. generator having a source impedance R.sub.s associated
with the field region and a built in load R.sub.e attributable to
the circulating currents at each end. Note that, in a conventional
analysis, the circulating currents are not normally considered as
the electrodes are deliberately sized and placed so that the effect
of the currents is minimal. The equivalent circuit is shown in FIG.
4. The input to this circuit would be V.sub.zB.sub.xa whilst the
output from the electrodes would be S.multidot.V.sub.zB.sub.xa where
the sensitivity S is 3 R e R e + R s .
 We have used equation (1) to derive the current density
within the duct and then, by integration, calculate the total current
that circulates. Since we know that the maximum driving voltage
is V.sub.zB.sub.xa we have calculated the total resistance(R.sub.e+R.sub.s)i-
n the circuit.
 FIG. 5 shows the total resistance plotted against l/a. For
short meters we anticipated that R.sub.s would be 4 1 a wl .
 This resistance (R.sub.s based on geometry) is also shown
on the graph. It can be seen that for very short meters the total
resistance in the circuit tends towards this geometric value whereas
as l/a becomes large the resistance becomes constant and is about
5 2.691 w .
 From this we have concluded that for long meters the resistance
will, unlike a conventional practical meter, be solely determined
by the currents that circulate at the ends of the meter (which currents
do not normally feature in a conventional meter). From consideration
of the equivalent circuit shown in FIG. 4 it can be seen that as
R.sub.e becomes large compared with R.sub.s the sensitivity tends
 The output impedance of a conventionally arranged meter
fitted with small circular "point contact" electrodes
is 6 1 d
 (but we have found that 7 2 d
 may be more accurate), where d is the diameter of the electrode.
This resistance, R.sub.o, appears in the equivalent circuit as shown
in FIG. 6. For point contact electrodes it is much greater than
R.sub.e (which is not normally considered) so that the common practice
of making the input impedance of the electronics much greater than
R.sub.o is satisfactory.
 FIG. 7 shows the electric potential and the current distribution
for a meter where l/a is 3. For this instrument, and for longer
meters, there is effectively no current flow at z=0 so that in the
plane of the electrodes 8 J _ = 0
 and therefore E=-V.times.B. The sensitivity S is now unity.
Eddy currents still circulate at the ends of the instrument but
they no longer have any effect upon the potential at the electrodes.
The equivalent circuits shown in FIGS. 4 and 6 which effectively
model a conventional meter no longer have any meaning. This is because
the source impedance is not meaningfully associated with the length
of the field region within the meter. A surprising result we have
found is that, for long meters, the output impedance is therefore
dictated by the electrode geometry and the eddy currents do not
cause the anticipated problems.
 For meters where l/a.gtoreq.3 we have therefore considered
the possibility of lowering the output impedance by using a different
electrode design. Whereas point contact electrodes have been in
use for many years and have conventionally accepted advantages,
we have considered alternatives. In a conventionally optimal meter
where l/a is about 0.5 the resistance to current flow through the
liquid between two such electrodes is 9 2 d .
 If we consider the impedance between two electrodes which
are semi-circular in section but are long in the direction of the
flow (see FIG. 8) then we find that the resistance between them
is 10 2 L log e 2 s b
 where L is half of the length of the electrode in the direction
of the flow, s is the separation between the electrodes (in the
case of a meter with a circular duct this is its diameter), and
b is the width of the semi circular section. We have concluded that,
surprisingly, we can advantageously make a meter where the field
region is long and where the electrodes are about as long as the
field, or L=l (approximately).
 We have determined the ratio of the source impedance of
this device to that of a similar one having point contact electrodes
is 11 2 l d 2 log e 2 s b
 which is 12 d l log e 2 s b .
 The value of 13 log e 2 s b
 is likely to be about 3 (for practical reasons we have
explored, it cannot be less than about 2). This means that the ratio
is likely to be about 14 3 d l .
 If d had been 8 mm and l is 0.24 m this means that 15 3
d l = 0.1
 or in other words the source impedance would have been reduced
by an order of magnitude.
 We now consider a practical example where the fluid to be
measured is an alcohol with an electrical conductivity of 0.13e-6
mhos/m (0.0013 uS/cm).
 Firstly, we consider a conventional instrument having point
contact electrodes 8 mm in diameter and a magnetic field that extends
over a length equivalent to 0.9 times the tube diameter.
 The electrode impedance for a point contact electrode is
16 2 a
 where a is the electrode diameter and .sigma. is the electrical
conductivity of the fluid, so the electrode impedance will be 17
2 3.142 .times. 0.13 - 6 .times. 8 - 3 = 612 M .
 Since the field extends over a length of 0.9 times the tube
diameter, which in this case is 50 mm, the length of the field region
will be 0.045 m and the source impedance of the magneto-hydrodynamic
(MHD) generator within the meter will be about 18 ( 1 l )
 which is 54.7 M.OMEGA..
 Thus the total impedance is of the order of 1000 M.OMEGA..
It is difficult to measure signals with such an impedance without
electrical noise causing problems. A further problem with low conductivity
liquids is that minor static charges do not dissipate rapidly and
will cause triboelectric noise as they pass the electrodes and,
with such high impedance, this will be significant.
 We now consider a novel meter in accordance with an embodiment,
in which the electrodes are still 8 mm wide, but in this case the
elongate electrodes are 240 mm long (30 times the height) and the
magnetic field extends over a (similar) length of 270 mm. As in
the conventional meter case the meter bore is 50 mm, so the electrodes
and coil are about 5 diameters long.
 For our novel meter the impedance between the two measuring
electrodes is 19 2 l log e 2 S a
 where S is 50 mm, a is 8 mm and l is 240 mm so that 20 R
e = 2 3.142 .times. 0.13 - 6 .times. 0.24 log e 100 8 = 51.7 M .
 This is approximately 12 times lower.
 The source impedance of this circular MHD generator 21 (
2 L )
 where L=2 l will, in this case be 20.4 M.OMEGA.. This is
over 6 times lower. 22 N . B . 612 M 20.4 M = 30
 The total impedance is of the order of 80 M.OMEGA., an order
of magnitude lower.
 Thus the effect of using a long meter with elongate electrodes
is that the electrode impedance and the source impedance are both
significantly reduced. A further advantage we have found is that
the use of long electrodes is an effective way of reducing both
the flow induced noise and the electrical noise. In particular,
we have concluded that small static charges may still remain in
low conductivity fluids but, because the electrodes are long in
the direction of flow, noise from any charges tends to average out.
In a practical meter, we have found that magnetic flow meters having
a field region several diameters long and which are equipped with
elongate electrodes, are significantly better at measuring the flow
of liquids whose electrical conductivity is low.
 Referring to FIGS. 9 to 11 a practical embodiment will now
be described. A housing and magnetic return circuit 10 typically
formed from steel, houses pole pieces 12 upper coil 14 and lower
coil 16. The housing has an insulating lining 22 here formed from
nitrile rubber, defining a bore (here 50 mm diameter) in which are
provided opposed sensing electrodes 18. As shown, the coils are
oriented about a vertical axis and the sensing electrodes are positioned
across the horizontal diagonal although of course the spatial orientation
is not critical.
 Optional earthing electrodes 20 are positioned along the
coil axis (where they will not pick up an induced emf due to the
coil). The electrodes 1820 and coils 1416 are connected to metering
electronics in housing 24. The metering electronics can be largely
conventional and will not be described or depicted in detail. However,
it is important to note that a very high impedance input amplifier
should be used, preferably 10.sup.15 ohms, which can be provided
by a commercially available FET input operational amplifier. As
explained above, the elongate electrodes are long along the direction
of flow (here 480 mm long) but of typically relatively small dimensions
(here 8 mm) high (in the perpendicular dimension). The thickness
of the electrodes is not critical but here they are formed as an
approximately semi-circular section for structural rigidity.
 Although the construction is novel, the operation of the
meter is essentially conventional and will be well understood by
one skilled in the art. In essence, an excitation field, typically
time varying, e.g. AC or pulsed DC or a more complex scheme is applied
by applying a current to the coils 1416 and the potential induced
in the moving fluid (here alcohol) is measured using a high impedance
differential amplifier connected to sensing electrodes 18. After
appropriate scaling and calibration, a measure of flow is obtained
from the sensed potential.
 The above is only an example and it can be seen that the
impedance can be reduced further proportionately by extending the
meter. The meter can be made very long, installed into several meters
of pipe. Making the electrodes long is not problematic and a long
liner can be extruded. However, making a long coil introduces some
practical concerns. One solution is to extrude the pole pieces.
Winding the coil around such a long pole piece cannot be done with
unmodified conventional coil winding equipment but is nonetheless
susceptible to automation using a moving shuttle and can of course
be done manually. Alternatively, a modular coil construction may
be adopted, particularly if the coil has a small number of turns
which plug together--this may be acceptable if a relatively large
current is passed through relatively few turns. As shown above,
the electrode impedance is inversely proportional to the length
of the electrode, and the source impedance of the MHD generator
is inversely proportional to the length of the coil.
 In preferred embodiments the length of the electrode is
approximately equal to the length of the coil, and thus the total
impedance is approximately inversely proportional to the length
of the coil. The total impedance is also inversely proportional
to the conductivity of the fluid. Hence an appropriate length for
the coil can be determined based on the conductivity of the fluid,
in order to obtain a total impedance below a desired threshold and
allow a successful flow measurement to be made. In general, a fluid
with a lower conductivity will require a longer coil in a practical
device. Very long coils and field regions can be produced using
the techniques suggested above if required.
 Although described in the context of particular examples,
it will be appreciated that modifications of detail may be made
and the invention is not limited to the specific embodiments. All
features disclosed herein may be provided independently or in alternative
combinations unless otherwise stated.