ELECTROMAGNETIC FLEXURE
The present invention generally relates to electromagnetically operated devices,
or actuators of the type which use a magnetic field to move an armature or
flexure from one position to another. More specifically, the present invention
concerns the flexure for a binary actuated valve suitable for applications such as
fluid flow control.
There are many types of electromagnetically operated devices or actuators
which are currently used to control switch contacts and provide an open, closed,
or changeover functionality. Typical examples of such devices include those
which use the movement of the flexure or cantilever to open and/or close valves
controlling a flow of fluid. Many such valves require a continuous electric current
to hold the armature in one position or the other. This wastes energy and can
produce unwanted heat. To avoid using a continuous flow of electric current,
binary actuated valves have been developed such as that of the type disclosed
in US6935373.
Existing binary valves of the type disclosed in US6935373 operate bi-stably in
either the fully-open or fully-closed states, using permanent magnets to hold the
valve in each state. To change the state of the valve, a single short electrical
pulse is sent to the coil to reduce, remove, or reverse the attractive magnetic
force, causing the valve to switch states with the help of a mechanical spring.
Such a valve can be controlled using a pulse width modulation (PWM) transistortransistor
logic (TTL) signal, with an edge-detection circuit sending actuating
pulses to the coil in response to the edges of the PWM signal.
In many applications, it is desirable to have valves that can pass large flow rates
and switch with short time delays despite high pressure differentials across the
seal. One such application is pneumatic control of truck brakes. In this
application, it is desirable that valves have effective orifice diameters of up to
9 mm and switching times of 3 ms. Furthermore, pressure differentials across
the valve can be up to 2.5 bar. This combination of performance parameters is
not achievable with conventional valve technologies, which tend to have
switching times longer than 15 ms.
Typical existing binary valves, such as that shown in Figure 1, achieve switching
times shorter than 4 ms with a 9 bar pressure differential, but only for effective
orifice diameters smaller than 3.5 mm. Such performance is sufficient in
applications where fast and small pressure adjustments are required. However,
many applications require higher pressures, flow rates, and switching speeds.
The valve according to the present invention has been developed to overcome
the limitations of previous binary valves, such that it would achieve specifications
suitable for pneumatic brake actuation when placed directly on the brake
chamber. This requires changes in the pressure in the chamber greater than 0.5
bar at 12.5 Hz with a supply pressure of 12.5 bar.
According to the present invention, there is provided an electromagnetic valve
comprising:
a yoke;
a magnet having pole pieces defining a gap;
a flexure assembly having one end attached to the yoke, such that part of
the flexure assembly extends into the gap, the flexure assembly having at least
one resilient portion formed of a resilient material and at least one magnetisable
portion, wherein that part of the flexure assembly that extends into the gap is
movable between the pole pieces through an intermediate position towards
which it is resiliency biased such that a resilient mechanical force is generated
by deflecting the resilient portion from an undetected position; and
means for polarising the magnetisable portion of the flexure assembly so that
the part of the flexure assembly that is movable between the pole pieces is
attracted towards a pole piece by a magnetic force, thereby defining a valve
state;
wherein the magnetisable portion and the resilient portion of the flexure
assembly are configured such that the magnetic force defining the valve state is
greater than the resilient mechanical force;
wherein the magnetisable portion of the flexure assembly comprises
elements of magnetisable material,
wherein each element comprises regions of reduced permeability separating
the elements from the resilient portion, and
wherein the elements of magnetisable material do not contribute significantly
to the resilient mechanical force.
Designing the flexure assembly for a binary valve involves a tradeoff between
several variables. In order to handle high pressure, the flexure assembly
stiffness needs to be high. Although raising the stiffness also increases the
speed of response of the valve, a stiff flexure assembly experiences higher
stresses at the root of the cantilever. In addition, the attractive magnetic force,
or 'pull force,' must overcome the flexure stiffness as the flexure assembly
approaches its end stop during valve switching. This requires a large magnetic
flux to flow through the flexure assembly, implying a large cross-sectional area.
Ordinary flexure assemblies for binary valves tend to be rectangular in crosssection,
and are designed to balance the conflicting requirements mentioned in
the previous paragraph. However they can only pass a limited magnetic flux,
reducing the maximum switching pressure. By including a magnetisable portion
and the resilient portion as defined above, the present invention uniquely
separates the magnetic functionality of the flexure from the mechanical
functionality, creating more leeway in the design process and allowing higher
levels of performance to be achieved.
In preferred embodiments, the magnetisable portion comprises elements of
magnetisable material, for example, 'fingers', such that at least one end of each
element comprises regions of reduced permeability separating the elements from
the resilient portion. These elements of magnetisable material may or may not be
separate parts from the resilient portion. The separation of the elements of
magnetisable material allows the flexure assembly to bend without interference from
the elements, but also allows the magnetic flux to flow along the elements and
across small regions of reduced permeability at one or more ends of the elements.
The flexure assembly may consists of a single part as shown for example in Figures
10a and 10b described below. Examples of flexure assemblies consisting of more
parts will be given for example in Figures 10c and 10d described below. Hereafter,
the terms 'flexure' and 'flexure assembly' are used interchangeably.
Examples of devices according to the present invention will now be described
with reference to the accompanying drawings, in which:
Figures 1A and 1B respectively show a schematic of an existing binary
actuated valve with a 1.75 mm diameter orifice, and an overlaid diagram of the
relevant magnetic circuit model;
Figures 2A to 2D show four general stages that can be identified during
the switching of a binary actuated valve;
Figure 3 shows a binary actuated valve according to the present
invention;
Figure 4 lists the preferred specifications of a high-performance
pneumatic brake valve according to the present invention;
Figure 5 is a schematic drawing of a flexure in a binary valve;
Figure 6 schematically shows a valve configuration for pneumatic brakes;
Figure 7 shows a B-H curve for rolled cold steel;
Figure 8 is a graph showing the maximum switching pressure for different
coil configurations during static tests;
Figure 9 is a graph showing the maximum switching pressure for different
flexure thicknesses and coil configurations during static tests;
Figure 10 shows alternative flexure designs which may be used in a valve
according to the present invention;
Figure 11A shows a perspective view of a scalloped flexure design;
Figure 11B shows a frontal view of a scalloped flexure design, including a
frontal view of pole pieces above the flexure;
Figure 12 is a graph showing the maximum switching pressure for
different flexure and coil configurations during static tests;
Figure 13 shows a number of other possible flexure designs which may
be used in a valve according to the present invention;
Figure 14 shows a further possible design according to the invention; and
Figure 15 is another graph which shows the maximum switching pressure
for different flexure and coil configurations during static tests.
Figure 1 shows an existing binary actuated valve comprising a steel frame 1, a
coil 2 and strong permanent magnets 3. The only moving part of the valve is a
flexure 4, which is a cantilever that is held bi-stably against either one of two
pole-pieces 5a; 5b via the permanent magnets 3. The steel frame transfers the
magnetic flux from the permanent magnets 3 to the pole-pieces 5a; 5b.
Alternatively, the permanent magnets 3 may be placed in series with the polepieces
5a; 5b.
When the flexure 4 is held against the bottom pole-piece 5b, it blocks an orifice 6
and the valve is therefore closed. The diameter of the orifice 6 in this existing
valve is 1.75 mm. When the flexure 4 is held against the top pole-piece 5a, the
valve is fully open. To get the flexure 4 to switch states, a brief electrical pulse is
sent to the coil 2 to reduce, remove, or reverse the attractive magnetic force
which, combined with the mechanical stiffness of the flexure 4, pulls the flexure 4
to the opposite state. The flexure 4 stores some of the kinetic energy that would
ordinarily be lost in a conventional solenoid valve, converting the kinetic energy
to potential energy as the flexure 4 reaches the opposite pole-piece 5a or 5b.
This also softens the landing of the flexure 4, increasing the longevity of the
hardware.
Figure 2 schematically represents four general stages that can be identified
during the switching of a binary actuated valve:
A) Pulling the flexure to its seat;
B) Sealing;
C) Releasing the flexure from its seat; and
D) Flexure switching.
The four stages are described in more detail below. It is assumed that the valve
is submerged in a high pressure reservoir (not shown), and that a low pressure
reservoir (also not shown) is attached to its orifice 6. Therefore, the pressure
force tends to help seal the valve once it is closed. Note, however that it is
possible to configure the valve so that the pressure force tends to open the valve
rather than tending to close it.
A) The first stage of valve switching is shown in Figure 2A. A magnetic force,
Fm, opposes an elastic flexure force, Ff, to pull the flexure to its seat.
Considering a flexure of rectangular cross-section, can be approximated by:
where d is the cantilever deflection from its neutral point at the centre of its
throw, Ef is the Young's Modulus of the flexure material, bf is the width of the
flexure, tf is its thickness, and M s the cantilever length of the force application.
The magnetic force is often simplified to be
F_ (2)
2m o
where Ba is the magnetic flux density in the air between the flexure and the polepieces,
Aa is the characteristic area of the air gap, and m o is the permeability of
free space, which equals 4p x 10 7 N/A2.
B) The second stage of valve switching is shown in Figure 2B. Once the flexure
4 touches the orifice 6, the air pressure force, Fp, helps the magnetic force in
compressing the flexure against the seat, sealing off the orifice. The pressure
force is given by:
F p (3)
where PH and P are the pressures in the high- and low-pressure reservoirs
respectively, and d0 is the effective diameter of the orifice.
C) The third stage of valve switching is shown in Figure 2C. A solenoid coil has
to be actuated to modify the magnetic force sufficiently such that Fp + Fm < Ff +
Fc, where Fc is the "coil force", which is used as an illustrative force at this point
in the explanation. Preferably, the actuated coil completely negates the
magnetic force, allowing the entire flexure force to overcome the pressure force.
The influence of the coil can be described approximately by:
where Nc is the number of coil turns, ic is the current C (indicated by arrows) in
the coil, and f(5) is represents a function of the flexure position.
D) The final stage of valve switching is shown in Figure 2D. It is assumed that
once the flexure 4 is lifted from the orifice, a pressure balance occurs such that
there is no pressure force on the flexure. In addition, it is assumed that the pull
force from the magnets is completely eliminated by the influence of the coil. In
this case, the flexure's motion is dominated by free vibration at its fundamental
resonant frequency, fn, which for a rectangular cantilever is given by:
where p f is the density of the flexure material. It is acknowledged that the actual
flexure motion is influenced by imbalances between the magnetic and coil
forces, by second order effects of the pressure, and by variation of the geometry
of the flexure.
An embodiment of a binary valve according to the present invention (except for
the flexure) is shown in Figure 3. A summary of the preferred specifications for
the binary actuated valve according to the present invention is given in the table
of Figure 4. The reasoning behind these specifications is explained in more
detailed below. It will be appreciated that the parameters given in Figure 4
represent the optimal values determined by the inventors, and that the present
invention is not restricted to the nominal parameters given in this example.
The minimum diameter for a binary actuated valve for a pneumatic ABS system,
according to the present invention, is 8 mm, with a preferred diameter of 9 mm.
Such diameters are large enough to pass air flows that are sufficient to track the
demand pressure variations during a slip-controlled stop on a rough surface.
Typical frequencies of body-bounce and wheel-hop for heavy vehicles tend to be
in the range of 3 Hz and 12.5 Hz respectively. The PWM frequency requirement
may be determined based on the Nyquist sampling theorem known in the art.
With this theorem, assuming a safety factor of 2, the PWM frequency of the
pneumatic valves is required to be 50 Hz in order to follow oscillating pressure
demands to reject both wheel-hop and body-bounce.
Existing pneumatic brake systems can achieve pressure control accuracies of
0.1-0.2 bar. To improve on this, the target accuracy of the new valves is set as
0.05 bar. It is further specified that this accuracy should be achieved within a
settling time of 200 ms. Using gain values of at least 3, assuming a PWM
frequency of 50 Hz, and constraining the range of mark-space ratios to reach as
low as 15%, a valve according to the present invention would have to take less
than 3 ms to change states. Mechanical motion was previously observed to take
up half the switching time of the valve, and, since the mechanical motion
constitutes half a period of free vibration, then the required 3 ms reaction time of
the valve means that the flexure must have a natural frequency greater than
1/(0.003 s) = 333 Hz.
The binary actuated valve according to the present invention was designed to
fulfil the criteria tabulated in Figure 4 and described above. It would be
appreciated, however, that the present invention is not limited to the nominal
values, but that these values are the preferred ones for this application of the
valve technology.
Figure 5 is a schematic drawing of a 'flat', rectangular flexure 40 in a binary
valve having an orifice 60 of 9 mm in diameter. When the flexure is deflected, a
cylinder of space 65 is created between the orifice 60 and the flexure 4. The
surface area of the side of the cylinder needs to be greater than the orifice
surface area so as not to impede the flow. In other words,
d h > —^- (6)
where h is the height of the cylinder.
The valve according to the present invention was also designed to have a
theoretically infinite service life, based on an endurance limit of 40% of the
ultimate strength of the material, su. The stress will be largest at the root of the
flexure, and is calculated for an example rectangular flexure 40 using:
3SE t
2L
Returning to Figure 3, the valve comprises a mild-steel C-frame, or yoke 10,
which has two parts so that different metals could be used for the flexure 40 and
the yoke 10. It will be appreciated that the yoke 10 may come in different
shapes, some of which are described below. However, the valve could also be
made as one piece, multiple pieces, or variants of the number of pieces shown
in Figure 3. Preferably, the 'neck' of the yoke 10 (the portion of the yoke
between the section within which the flexure 40 is clamped and the section that
is connected to a magnet and a pole-piece) is large enough to prevent
bottlenecking the flux flow, and ample space is made available around the
flexure 40 for a solenoid coil 20. Strong magnets 30a; 30b, such as neodymiumiron-
boron (NdFeB) magnets, are placed next to bright mild steel pole-pieces
50a; 50b. However, the magnets could be placed in alternative locations that
are in series with the pole-pieces 50a; 50b as well , such as at the back of the
valve or on the necks of the valve.
Preferably, the pole-pieces 50a; 50b are conical to 'focus' the flux, thereby
increasing the magnetic hold force. An orifice 60 may be housed in non
magnetic through-tubes 61 that are screwed and fixed by lock nuts 7 1 into the
pole-pieces 50a; 50b to hold them against the magnets 30 and yoke 10. A
packer 5 1 may also be used so that the flat face of a standard toroidal magnet
with a rectangular cross-section can be mated to a yoke whose face is not flat,
facilitating angular adjustments of the pole-piece assembly such that the flexure
sits flat on a pole-piece face when deflected .
Intuitively, one would think that the flexure stiffness should be driven to its
maximum to achieve the maximum F f and fn through equations ( 1) and (5).
However, this would raise the stress at the root of the cantilever, as illustrated by
equation (7). Moreover, although equations (2) and (4) suggest the magnetic
circuit in the valve is independent of the valve mechanics, these equations are
only first order approximations. In reality, only a finite amount of magnetic
permeability is available in a material, constraining the amount of flux that can be
transmitted by the flexure. This attractive magnetic force must overcome the
flexure stiffness during the first stage of valve switching, when the flexure 40
deflects to the opposite pole-piece to provide the appropriate opening area
according to equation (6). It follows that the flexure stiffness must be small
enough that the limited magnetic attraction available will hold the flexure against
either pole-piece.
The flexure 40 in the example embodiment according to the present invention
shown in Figure 3 has a thickness of 1.4 mm and a width of 30 mm. A coil 20
may be wrapped around the flexure 40 using a plastic coil former that features
an inner cut-out large enough to permit free motion of the flexure 40 throughout
its travel. Rubber O-rings 80 may be used to create a seal between the flexure
40 and the pole-pieces 50a; 50b. The example embodiment of a valve shown in
Figure 3 preferably accommodates an O-ring 80 with ds = 1.6 mm, Ds = 12.1
mm, and a compression of up to 0.24 mm. However other sealing arrangements
are possible, with the flexible surface located on the flexure 40, or with the
flexible surface located some distance away from the pole piece using a linkage
connected to the flexure.
Two valves V according to Figure 3 were fabricated for the use in an
experimental airbrake system according to the diagram shown schematically in
Figure 6. The system comprises a disk brake DB and a load cell LC. The
arrows indicate the direction of airflow supplied by an air compressor AC,
through an inlet I and an outlet O of the system. The valves V were used to
control the compressor pressure at the inlet I and the chamber pressure at the
outlet O. Finite element simulations suggested that the flexure 40 would have a
natural frequency of 575 Hz and a stress at the root of the cantilever of
497 MPa, which is less than the endurance limit of the design material. In
addition, the simulations suggested that the flexure 40 hold force would be 140
N. Attempts were made to observe the valve switching under an applied
upstream pressure. However, it was discovered that for a flat flexure 40 made of
EN42 spring steel, the maximum upstream pressure under which the valve could
switch was not 12 bar, but 3 bar. In other words, despite meeting all of the
mechanical requirements, a conventional, flat flexure 40 made from EN42 spring
steel could not meet the magnetic requirements for the valve. Magnetic
stainless steel materials were also tried for the flexure, but resulted in lower
switching pressures. This was despite the valve having adequate hold force and
flexure stiffness when tested.
Considering the reasonable hold force and flexure stiffness displayed by the
valve, it was theorized that the coil was not altering the magnetic hold force as
much as was originally expected. However, this did not explain what aspect of
the magnetic design would have to be modified to improve switching. To resolve
the problem, an experimental design optimization of the valve was performed in
concert with a theoretical analysis based on magnetic circuit theory.
The mechanism behind the magnetization of materials is well known in the art.
As the strength of the magnetic field applied to the material is increased, small
magnetic domains within the material that initially have random orientations
become aligned with the applied field. Eventually, all of the domains become
aligned with the applied field. At this point, the material is saturated and
responds like free space to any further increase in the applied magnetic field
strength. Since the core is now indistinguishable from the outside air, most of
the additional magnetic flux leaks through the air.
The saturation phenomenon is commonly illustrated through the use of B-H
curves, which plots flux density vs. magnetization. The B-H curve used in the
theoretical analysis of the flexure is shown in Figure 7. The relative permeability
of the material, m , is the local slope of the B-H curve at a given magnetic field
strength, and gives an indication of how much flux can be passed through the
material at a given level of magnetization. It is commonly assumed in the art
that mG is a constant value, with the value taken near the origin of the B-H plot.
This is approximately valid at low values of flux density in the material, but not
for high values of the flux density, where the local slope is reduced. Simulations
of the valve indicated that the rectangular flexure shown in Figure 3 was
saturated when it was close to the pole-pieces.
Several parameters were investigated to optimise the flow of flux through the
valve, but only the coil and the flexure are discussed here for brevity, since
changing other parameters did not produce significant results. A suite of tests
was performed to evaluate the performance of different coil configurations. The
configurations examined included winding coils in series around the flexure;
around the pole-pieces; around the neck of the yoke; and combinations of these
arrangements. Each configuration was tested for increasing currents, capturing
the effect of increasing the magnetomotive force on the hold force. Results of
the tests are shown in Figure 8. The currents were derived from the voltage
input to the coil and the resistance of the coil, which was measured before and
after each test.
Two distinct trends are seen in the results of Figure 8: one for configurations that
included a coil wound around the flexure, and one for configurations that
included a coil wound around the pole-pieces or inline with the magnet. When
the coil was wound around the flexure, for example the curves labelled '300
Turns Around Flexure Only,' higher switching pressures were typically
encountered at lower magnetomotive forces compared to when there was no coil
around the flexure. However, winding the coil around the pole-pieces appears to
have facilitated higher switching pressures at higher magnetomotive forces, for
example the curve labelled '300 Turns Around Pole-Pieces Only.' In this case, a
clear saturation of the maximum pressure occurs for a magnetomotive force of
approximately 2500 Ampere-turns.
The magnetic circuit simulation suggested that winding the coil around the
flexure is the most efficient arrangement with respect to re-routing the flow of flux
to facilitate a change of state of the valve. This is because winding the coil
around the flexure effectively short-circuits the flow of flux between the two
permanent magnets. But, Figure 8 indicates that the flexure saturates at low
magnetomotive forces with the coil wound around it (a dotted line is drawn on
Figure 8 showing the expected point of saturation for these coil configurations).
According to the magnetic circuit simulation, winding the coil around the polepieces
is less efficient than winding it around the flexure. However, with the coil
located in series with the permanent magnets, it is postulated that the coil is able
to oppose the flow of flux generated by the magnets directly. Consequently, flux
leakage is less of a factor and more of the energy from the coil may be directed
to overcoming the permanent magnets, explaining the higher switching
pressures found in Figure 8 for coils in series with the magnets at higher
magnetomotive forces. Some of this flux is still routed through the flexure,
though, which eventually saturates.
By increasing the MMF and changing the coil locations, the switching pressure
was increased to 6 bar from the initially attained value of 3 bar. The 6 bar
maximum switching pressure achieved with the best coil configuration was still
well below the target design pressure of 12 bar, though. Moreover, the switching
pressure was achieved for a magnetomotive force of 4500 A-t, which translates
into a current of 22.5 A for a 200 turn coil. Given the 0.5 mm diameter of the
wire used, the currents could not be raised much further to achieve higher
switching pressures without melting the coil. Changes to other parameters were
therefore investigated to improve the valve's performance. These investigations
are described below.
To increase the amount of flux that can be channelled through the flexure, its
cross-sectional area would have to be increased. Looking at equation (1), the
flexure stiffness relates linearly to its width and cubically to its thickness. It
follows that increasing the width of the flexure would have a much smaller effect
on the stiffness than increasing the thickness. This smaller effect would in turn
minimise the amount of extra magnetic force needed to fight the stiffness.
However, the width of the flexure is more than 20 times the thickness in the
embodiment shown in Figure 3. Consequently, increasing the width of the
flexure by just a small amount to raise the cross-sectional area would have a
significant effect on the physical envelope of the valve when compared to
increasing the thickness. Thicker flexures increase the magnetic flux, but also
increase the mechanical stiffness and hence the necessary magnetic hold force.
The research was therefore directed towards inventing a flexure that is
effectively thicker in strategic areas, substantially separating its mechanical
functionality from the primary magnetic functionality responsible for switching the
flexure from one state to the other.
The investigation was performed by taking a normal flexure and attaching
1.1 mm thick 'slivers' of mild steel to it using tape. The slivers were short
enough that they sat between the pole-pieces and the back of the yoke, and the
bonding was flexible enough that the slivers did not significantly affect the
flexure's stiffness. The tests evaluating the maximum switching pressure that
were described previously were then re-run with approximately 15 A sent to 100
turn and 200 turn coils wound around the new flexure configurations.
Results from the tests are plotted over results from the tests for the different coil
configurations in Figure 9. It is clear that thickening the flexure between the
pole-pieces and the back of the yoke had a far more dramatic effect on the
maximum switching pressure than optimizing the coil configuration, with the
pressure increasing by 100% over the best coils. A peculiar result is that
thickening the flexure on the opposite side of the orifice had a greater effect than
thickening the flexure on the same side of the orifice. This is counterintuitive,
since the opposite pole-piece is over 1 mm away from the flexure, and one
would expect the majority of the flux to try to travel through the adjacent polepiece
to the slivers on the flexure. But, thickening the side opposite the orifice
brings the slivers within range of the magnetic attraction of the other pole-piece,
encouraging the flexure to switch to the other state.
New flexure arrangements were designed that would 'thicken' the flexure in a
magnetic sense without making it thicker in a mechanical sense (i.e. without
making it stiffer). Four such designs are shown in Figure 10.
The first design, in Figure 10A, involves adding 'fingers' to the flexure F 1 such
that the mechanical stiffness, which is predominantly dictated by the thickness of
the flexure at its root, remains similar to the original flexure design. The fingers
are shown in Figure 10A as being an integral part to the original, flat, rectangular
portion of the flexure F 1, but they may also be separate parts attached to the
flexure, to simplify manufacture or to allow the use of different materials. The
basic section of the central flexure element has a thickness of 1.30 mm and a
width of 30 mm. The fingers in this embodiment each have thicknesses of
1.075 mm, with a 1 mm air gap separating the fingers from the main body of the
flexure and a small air gap (<0.5 mm) separating the fingers from the back of the
yoke when the flexure is deflected. Because this gap is small, it can still be
crossed easily by the magnetic flux. The thickness of the fingers was limited by
the need to fit the entire valve in the previously built pressure chambers. It will
be appreciated that these are exemplary dimensions of this embodiment of the
flexure and that the dimensions for other embodiments of this design may vary.
FEA simulations suggested that the deflection force of the flexure would be 120
N when touching an orifice (decreased from 150 N for the flat, completely
rectangular flexure), the associated stress at the root would be 530 MPa, and
the natural frequency would be 356 Hz. Although the natural frequency was
reduced, it was still above the minimum 333 Hz required by the specifications
shown in Figure 4.
The second design, shown in Figure 10B, adds a scallop to the flexure S such
that the flexure S 'hugs' a pole-piece and better directs the flux. Once again, the
fingers and the scallop are shown to be integral with the original, flat portion of
the flexure in Figure 10B, but may also be made as separate parts attached to
the flexure. The thickness of the clamped section was also reduced to that of
the 1.30 mm thick central flexure element. Since the valve would now easily fit
in the pressure chambers, the fingers were thickened to 2.1 mm each, retaining
their original 1 mm spacing from the central flexure element and 0.5 mm spacing
from the back of the yoke when deflected. It will be appreciated that these are
exemplary dimensions of this embodiment of the flexure and that the dimensions
for other embodiments of this design may vary. The deflection force of the
flexure S was predicted to be 120 N when touching the O-ring, while the stress
at the root and the natural frequency were simulated to be 432 MPa and 550 Hz
respectively.
The third design, shown in Figure 10C, decouples the fingers F2 from the flexure
entirely. In this design, 2.1 mm thick, static mild steel fingers F2 were clamped
between the flexure and yoke, with scallops at the end of the fingers so they
hugged the pole-pieces. It will be appreciated that these are exemplary
dimensions of this embodiment of the flexure and fingers, and that the
dimensions for other embodiments of this design may vary. Shims may be
placed between the flexure and the fingers to separate the two, allowing for free
motion at the root of the flexure, thereby minimally affecting its stiffness.
Moreover, the fingers may be carefully bent such that they barely touched the
flexure along its length when it was deflected.
The fourth design, shown in Figure 10D, has additional fingers oriented in the
opposite direction to the ones in Figures 10A and 10B. A small air gap between
the fingers and the thickened end of the flexure F3 allows the flow of magnetic
flux, but the mechanical stiffness is close to that of a flexure without the fingers.
The fingers may be integral parts of the original, flat portion of the flexure, as
shown in Figure 10D, or they may be made as separate parts attached to the
flexure.
The second design as shown in Figure 10B, the flexure S with a scallop, looked
the most promising from a theoretical point of view, since the scallop would
provide a large amount of area where flux could be transferred from a pole-piece
to the flexure. It follows that it was decided to fabricate that design while running
preliminary tests with the third design. A detailed representation of the scalloped
design shown in Figure 10B is shown in Figures 11A (in perspective view) and
11B (frontal view). The frontal view of Figure 11B shows the scalloped flexure S
placed below a polepiece P 1. The pole-piece P 1 is formed of one part, and is
shown in section view so that the orifice can be seen. The preferred material for
making the scalloped flexure may be EN42 spring steel, however other materials
may be used.
Speed tests run with the scalloped flexure shown in Figures 10B and 11 showed
that the valve took 2.5-3 ms to switch states. Tests to determine the maximum
switching pressures were run with both the flexure S featuring a scallop (Figure
10B), and the design featuring fingers F2 clamped between the flexure and the
yoke (Figure 10C). The results are plotted on top of the results for different coil
configurations in Figure 12.
Figure 12 shows that the design with clamped fingers F2 increased the switching
pressure by 4.3 bar over the best coil configurations, while the scalloped flexure
S performed even better, increasing the switching pressure by 5.8 bar over the
best coil configurations. These test results further confirmed the previous
deduction that the steeper gradient of the switching pressure vs. magnetomotive
force curves occurs when there is no saturation of the flexure. Although Figure
12 only shows test results up to 9 bar, it is conceivable that the 12 bar design
pressure initially specified for the valve could be achieved by the scalloped
flexure S with a higher magnetomotive force.
The valve according to the present invention may be used in many applications
requiring rapid switching on and off of the flow of a liquid or gas. These include
air brakes, pneumatic and hydraulic actuators, and other applications where
rapid switching of a fluid flow (and optionally pulse-width modulation) can be
used to control mechanical systems.
Figure 13 shows a number of possible flexure designs that may be used, in
addition to those shown in Figures 10 and 11, in a valve according to the current
invention. The design shown in Figure 13G includes slivers 100 of magnetic
material connected to the central flexure such that they do not constrain the
central flexure from bending.
Figure 14 shows a further flexure design in accordance with the present
invention. Figure 14C shows a section view of the flexure and pole-piece. The
flexure Z has a clamped portion A; a resilient portion B; a flat portion C; and a
seat D. In addition, the flexure Z has fingers E; and a magnetic coupling to the
fingers F. Similar to Figure 10 D, the fingers E protrude from the clamped region
of the flexure. This reduces the mass of moving part of the flexure compared to
flexure S, therefore increasing its natural frequency according to equation (5).
The portions of the fingers E adjacent to the clamped portion A of the flexure are
also clamped, and the portions of the fingers E adjacent to the resilient portion B
of the flexure are not clamped. All components A-E may be manufactured
integrally, but may also be made as separate parts attached to each other.
Making components A-E as separate parts would allow individual consideration
for magnetic permeability, mass, corrosion resistance, strength, and ease of
manufacture of each of the parts, as well as the flexure as a whole.
In the design shown in Figure 14, the flat portion and seat are separated by
annular protruding element e. This element may be attached, or integrally
formed with either the flat portion or the seat. At least one protruding element e
may be used on either side of the flat portion. In addition, the pole-piece P 1 in
this design is formed of a magnetic portion G; and a non-magnetic portion H.
The non-magnetic portion H includes another protruding element f , which may
be attached to, or integrally formed with the non-magnetic portion H of the polepiece.
Protruding element f may press into the rubber element, creating a seal.
Advantageously, the diameter of this seal would only be as large as the orifice
diameter, and consequently smaller than the diameter of a seal created by a
rubber O-ring integrated into the pole-piece (for example, Figure 11B). The
smaller diameter seal would lower the pressure force according to equation (3),
and as Figure 2 c shows, a lower pressure force would require lower
magnetomotive force from the coil to cause the flexure to switch states.
The height of the protruding element e on the flat portion C of the flexure may be
dimensioned to provide stress relief on the rubber seal, by limiting the amount
the protruding element f on the pole-piece presses into the rubber. The height of
protruding element e on the flexure may be used to adjust the air-gap between
the flat portion C of the flexure and the pole-piece face when the valve is in a
given state. Preventing metal-to-metal contact between the flat portion and the
pole-piece face, by including a small air-gap, can significantly reduce the
magnetic hold force, and therefore the magnetomotive force required by the coil
to switch states. It should be noted that in this embodiment of the valve, the
protruding element e does not make a magnetic circuit with the non-magnetic
portion of the pole-piece H when they touch each other.
The protruding element e may include one or more 'slots' g, as shown in Figure
14B, in order to vent air in the small annular volume created between the
protruding element e on the flexure and the protruding element f on the polepiece
when the valve is closed.
The thickness of the flat portion D of the flexure in Figure 14 is preferably
dimensioned so that it carries the required flux without saturating, and without
compromising the mass and dynamic properties of the flexure. The thickness
was 3.5 mm in this embodiment of the flexure. The thickness of the resilient
portion B of the flexure is designed to achieve a given resilient mechanical force,
and was 1.16 mm in this embodiment of the flexure. The thickness of the fingers
E is designed to prevent magnetic saturation of the flexure, and was 2.35 mm in
this embodiment of the flexure. The thickness of the clamped portion A of the
flexure is designed to provide clearance between the main resilient portion B and
the fingers E. This thickness was 3.4 mm in this embodiment of the flexure. It
will be appreciated that these are exemplary dimensions of this embodiment of
the flexure and that the dimensions for other embodiments of this design may
vary. Suitable values may be selected using equations (1-7).
In summary, the flexure, Z, in Figure 14 has an improved seat design, benefits
from unnecessary material being removed, and is formed of multiple parts, which
results in improved manufacturability, reduced size and reduced weight, while
providing a larger hold force but smaller switching current (magnetomotive
force). This design gives substantially improved switching performance.
Figure 15 shows test results (switching pressure vs. magnetomotive force
curves) for three type of flexures: the flexure Z as shown in Figure 14, a 'flat',
standard flexure 40, a scalloped flexure S 1 (of the type shown in Figures 10B
and 11) . It may be seen from Figure 15 that, at a magnetomotive force of 700,
flexure Z increased the switching pressure by approximately 10 bar compared to
the scalloped flexure S and the 'flat', standard flexure 40.
Although Figures 10, 11, 13 and 14 give a sample of possible flexure shapes
that can be used, it will be appreciated that the flexures according to the current
invention are not limited to the physical forms shown in those figures.
CLAIMS
An electromagnetic valve comprising:
a yoke;
a magnet having pole pieces defining a gap;
a flexure assembly having one end attached to the yoke, such that part
of the flexure assembly extends into the gap, the flexure assembly having at
least one resilient portion formed of a resilient material and at least one
magnetisable portion, wherein that part of the flexure assembly that extends
into the gap is movable between the pole pieces through an intermediate
position towards which it is resiliency biased such that a resilient mechanical
force is generated by deflecting the resilient portion from an undeflected
position; and
means for polarising the magnetisable portion of the flexure assembly so
that the part of the flexure assembly that is movable between the pole pieces
is attracted towards a pole piece by a magnetic force, thereby defining a
valve state;
wherein the magnetisable portion and the resilient portion of the flexure
assembly are configured such that the magnetic force defining the valve
state is greater than the resilient mechanical force;
wherein the magnetisable portion of the flexure assembly comprises
elements of magnetisable material,
wherein each element comprises regions of reduced permeability
separating the elements from the resilient portion, and
wherein the elements of magnetisable material do not contribute
significantly to the resilient mechanical force.
The electromagnetic valve according to claim 1, wherein the regions of
reduced permeability are located at the ends of the elements of
magnetisable material.
The electromagnetic valve according to claim 1 or claim 2, wherein the
elements of magnetisable material are attached to the resilient portion.
The electromagnetic valve according to any preceding claim, wherein the
elements of magnetisable material extend from an end of the flexure
assembly opposite from the end of the flexure assembly attached to the yoke
and moveable between the pole-piece, such that the regions of reduced
permeability are located at opposite ends of the elements of magnetisable
material and separate the elements from the yoke.
The electromagnetic valve according to claims 1 to 3, wherein the elements
of magnetisable material extend from the end of the flexure assembly
attached to the yoke such that the regions of reduced permeability separate
the elements of magnetisable material from an end of the flexure assembly
opposite from the end of the flexure assembly that is attached to the yoke
and moveable between the pole-pieces.
The electromagnetic valve according to any preceding claim, wherein the
end of the main flexure assembly movable between the pole-pieces is of
greater thickness than the resilient portion.
The electromagnetic valve according to any preceding claim, wherein at least
one resilient element is attached to or integrally formed with the portion of
the flexure movable between the pole-pieces.
The electromagnetic valve according to any preceding claim, wherein the
resilient element is contained inside at least one protruding element on the
magnetisable portion of the flexure that is movable between the pole-pieces.
9. The electromagnetic valve according to any preceding claim, wherein the
protruding element on the magnetisable portion of the flexure includes at
least one slot.
10. The electromagnetic valve according to any preceding claim, wherein a polepiece
assembly includes a protruding element that can press into the
resilient element, creating a seal.
11. The electromagnetic valve according to any preceding claim, wherein a polepiece
is made from separate parts that may or may not be of different
magnetic permeabilities.
12. The electromagnetic valve according any preceding claim, wherein the orifice
is housed in a through-tube.
13. The electromagnetic valve according to any of the preceding claims, wherein
the magnetisable portion comprises a scallop.
14. The electromagnetic valve according to any preceding claim, wherein the
magnetisable portion comprises at least one protruding element and wherein
at least one pole piece comprises at least one recess for receiving the at
least one protruding element.
15. The electromagnetic valve according to any preceding claim, wherein the
protruding elements on the magnetisable portion movable between the polepieces
can be designed to create regions of reduced permeability between
the magnetisable portion and the pole-piece assemblies.
16. The electromagnetic valve according to any preceding claim, wherein the
protruding elements on the magnetisable portion movable between the polepieces
can be designed to limit the amount the protruding element on the
pole-piece assembly presses into the rubber element.
17. The electromagnetic valve according to any of the preceding claims, wherein
the flexure assembly consists of a single part.
18. The electromagnetic valve according to any of the preceding claims, wherein
the flexure assembly consists of more than one part.