Description Technical Field

[1] The invention relates to an apparatus for filtering out wire defects in metal wires preferably steel wires during or after wire drawing. It can be implemented as an add-on to an existing wire drawing bench or it can be put on a wire winder to detect flaws, weak spots or other wire aberrations during rewinding. A method to operate and adjust the apparatus is also given.

Background Art

[2] Metal wires and more in particular high strength (more than 2500 N/mm ² ), thin gauge (thinner than 0.30 mm) steel wires are increasingly being used in all kinds of applications. Their use is not longer limited to steel cord for the reinforcement of e.g. truck tyres where they give the tyre belt its stiffness or the tyre carcass it's strength. High strength, thin gauge wires have also found use in steel cord to reinforce belts used to lift elevator carts, in mechanical applications such as to reinforce synchronous belts, and even in the sawing of precious, hard and brittle materials as a sawing wire.

[3] Making steel wires in high strength and thin gauge brings particular

problems with it. Low strength, thick gauge wires (such as for making crane ropes) are much more forgiving to raw materials flaws. For example the effect of an undeformable inclusion that is present in the wire rod will be less for a thick, low strength wire than for a thin, high strength wire as the area taken up by the undeformable inclusion relative to the total cross section of the wire will be much less in the former than in the latter wire. While such an inclusion may lead to production loss due to wire fracture in the wire drawing process, it can even have larger consequences when it goes through unnoticed and is present in the final product. A broken wire in an elevator belt e.g. may bring a premature lay-off of the belt with it, as the wire end will start to wick out of the belt. In wire sawing, the unexpected breakage of the wire leads to complete stand-still of the process and loss of precious time and material.

[4] The applicant's motto is therefore "to keep the fracture within Bekaert" i.e. it is better that a weak spot in the wire is detected and eliminated during or shortly after production of the wire, rather than the customer or the end- user being confronted with losses or even safety issues.

[5] Systems are available that detect internal or surface flaws of the wire

(most of them being based on magnetic induction measurements) but these systems only mark if there is a problem but do not eliminate it.

Elimination is best done by just filtering all weak spots out by breakage at that spot. This can be done by submitting the wire over a certain test length to a minimal test tension by continuously running the entire length through a testing device. Every fracture that then occurs during testing, is a fracture the customer or end-user is saved from.

[6] Apparatus have been described wherein wire, shortly after drawing, is continuously subjected to an On-line tensile test'. See e.g. JP 2000 167618, wherein the wire is led over two sheaves, one fixed and one movable, much like a block-and-tackle system. A load is applied to the moveable sheave - by a dead weight, pneumatically, or hydraulically actuating systems - resulting in a tensile stress in the wire. The level of the tensile stress applied is between the tension applied on the wire during use (as this concerns a sawing wire this is the tension applied to the wire by the sawing machine) and up to about 70% of the breaking load of the wire.

[7] An alternative use of the same mechanical principle is described in JP

2007 1 18067. By increasing the force level to between 40 and 90% of the breaking load of the wire, it is claimed that one can reduce tensile residual stresses in the wire.

[8] The problem with this kind of sheave systems is that they require space behind the drawing machine or wire re-winder. Also the tensioning control system for controlling the amount of tension occupies more place than desirable. The inventors therefore sought solutions that occupy less space that can easily be retrofitted on existing wire drawing benches and/or winders and wherein the control of tension is simple. Disclosure of Invention

[9] The primary object of the invention is to offer an apparatus that filters out defects in drawn metal wires preferably steel wires by generating a fracture at that defect, such that the defect does not reach customers or end-users. A further object is to provide an apparatus that is compact and can easily be retrofitted on existing wire draw benches and/or winders. Another object is to have a simple to control system. A final object of the invention is to provide a method to operate the apparatus.

[10] According a first aspect of the invention an apparatus is claimed. It

comprises two capstans. Ά capstan' for the purpose of this application is a sheave with a flat surface - think of a cylinder - on which full or partial loops of wire are wound for transmitting forces to the wire running on it by friction between the wire and the surface. The flat surface of the capstan defines a certain diameter. Capstans can rotate on an axis to which they are fixedly or rotatable mounted. With 'fixedly' is meant that no relative rotation is possible between axis and capstan, with 'rotatable' is meant that relative rotation is possible between capstan and axis.

[1 1 ] The apparatus comprises a first capstan with a first capstan diameter D1 that is mounted to a first axis and a second capstan with second capstan diameter D2 mounted to a second axis. The terms 'first' and 'second' imply an order in the sense that the 'first capstan' should be the capstan where - during use - the wire first arrives before the wire further travels - possibly over pulleys, sheaves and other devices - to the second capstan after which it leaves the apparatus. The axes themselves are rotatable and can be driven or non-driven. With driven is meant that rotative motive force (i.e. torque) is applied to the axis (for example by a direct drive motor, a belt, worm-worm gear, a gear box or any other kind of torque transmission) from a frame of reference. With non-driven is meant that the axis can rotate freely at all times e.g. because the axis is mounted on a bearing relative to the reference frame.

[12] During use the axes turn (in case of a non-driven axis) or are made to turn (in case of a driven axis) at an angular speed (expressed in 'radians per second') of W1 for the first axis and an angular speed W2 for the second axis. Angular speeds and diameters are selected such that, without any wire being present, the circumferential speed of the second capstan is larger than the circumferential speed of the first capstan or that D2xW2/2 is larger than D1 xW1/2, hence D2xW2 is also larger than D1 xW1 . Angular speeds can be imposed on the axes separately e.g. both axes are driven by individual motors with defined angular speeds W1 and W2. More preferred is if the angular speeds of both axes are coupled to one another in a fixed gear ratio of W1 :W2. If the first axis is driven then at angular speed W1 , the second axis will turn at angular speed W2 or vice versa. A particular preferred embodiment is where none of the axes is driven, but both axes are coupled to one another in a fixed gearing ratio W1 :W2. In this way the apparatus can be introduced into an existing wire path as a stand-alone unit driven by the wire that is being pulled through.

[13] The filtering apparatus is characterised in that one of said first capstan or second capstan is fixedly mounted to the respective first axis or second axis and in that said second capstan is coupled to the first capstan by means of a torque generating coupling.

[14] The principle of operation of the apparatus is as follows: as the wire is entering the device it is held on the first capstan by loops of wire wound around it. The wire is held at sufficient tension (see further) that no slip between wire and first capstan occurs. Then the wire is led to the second capstan of which the circumferential speed (W2xD2/2) is at least larger than the circumferential speed of the first capstan (W1 xD1/2). Again enough loops are wound around the second capstan such that no slip occurs.

[15] If the coupling between the second axis and the second capstan would be rigid - which is a priori not excluded by the invention - this would result in an applied elongation to the wire £ _f i _Xe d of ((W2xD2/W1 xD1 )-1 ). If this elongation would be higher than the maximal elongation at break A _t of the wire, the wire would simply break. As A _t is in general not very large - typically less than 3% for the kind of steel wires envisioned - a very good control of the ratio W2xD2/W1 xD1 is then needed.

[16] There is also the problem that when applying a fixed elongation, the

application of which is purely friction based, this friction can vary depending on wire properties, surface and surface condition of the capstans and even environmental conditions such as temperature and humidity. Cooling of the wire on the capstans will lead for example to a heat shrink which will add to the applied elongation of the capstan. This is contrary to what happens in e.g. a wet wire drawing machine wherein friction is stable (wire is immersed in a fluid lubricant, capstans are cooled and at constant temperature) and the elongation is imposed by a drawing die.

[17] The inventors surprisingly found that by introducing a torque generating coupling between first capstan and second capstan greatly resolved the sensitivity to friction and the need for control of the elongation applied through the ratio W2xD2/W1 xD1 . The presence of this coupling induces a constant force to the wire travelling from the first to the second capstan and does not impose a constant elongation any more onto the wire.

Moreover, the simple adjustment of the torque generating device makes it possible to apply any tension to the wire larger than 0 and smaller than or equal to Ffixed, Ffixed being equal to AE £fi _Xe d, wherein A is the cross- sectional area and E is the modulus of the wire.

[18] As a further bonus the ratio (W2xD2/W1 xD1 )-1 can be chosen

substantially higher than the total elongation A _t of the wire. By keeping the torque generated by the coupling below that of a fixed coupling, there is no risk that a too high elongation will be imposed on the wire.

[19] Due the presence of the torque generating device, the linear velocity of the wire on the second capstan will be higher than the linear velocity of the first capstan. This difference will depend on the ratio W2xD2/W1 xD1 . The linear velocity of the wire V2 on the second capstan will only be slightly higher than the linear velocity of the wire on the first capstan V1 . The ratio in linear velocity V2 V1 is equal to the elongation plus one (ε+1 ), the elongation being the consequence of the tensioning of the wire. So whether or not wire is present during use, the second linear velocity is larger than the first linear velocity.

[20] The torque generating coupling can be situated in different places along the force path that is formed by the axes, belts or gears connecting mechanically the first capstan to the second capstan. This force path is in balance with the force path formed by the wire during use. Preferred positions for the torque generating coupling are:

^■ the torque generating coupling is situated between second axis and second capstan, or;

^■ the torque generating coupling is situated between first axis and first capstan, or;

^■ the torque generating coupling is situated between first and second axis or;

^■ the torque generating coupling is situated between first and second capstan.

[21 ] By preference the torque generating coupling is adjustable. Adjustment can be in discrete steps or can be continuous.

[22] Possible torque generating couplings are simple friction couplings where a friction body (e.g. a brake pad in the form of a ring) is pushed with a normal, controlled force on a brake disk. Problems here are the wear of the brake pad, the heat generated and the difficulty to control the torque generated. Other torque generating couplings are powder couplings where torque is transferred over a powder - usually a metallic powder - between discs that are pressed together with a controlled normal force. If the powder is ferromagnetic, the apparent viscosity of the powder can be controlled through a magnetic field e.g. from an electromagnetic coil (electromagnetic powder coupling). Also fluid couplings can be used wherein fluid between several pairs of discs (e.g. even discs connected to the capstan, odd discs to the second axis) transmits the torque. This can be either due to a change in viscosity (viscous fluid coupling) or by exchange of momentum through an impellor - runner turbine combination.

[23] However, by far the most preferred coupling is a magnetic coupling. In a magnetic coupling a ring of alternating pole permanent magnets - currently high performance magnets such as neodymium-iron or samarium-cobalt based magnets - fixed to e.g. an axis is separated by a gap from a ring of alternating pole magnets fixed to the corresponding capstan drive hole. When torque acts on either the axis or the capstan, this torque will be transmitted over the magnetic field to respectively the capstan and the axis. The number of magnets will determine the smoothness of the transmission (the more magnets the smoother). The amount of torque transmitted will depend on the width of the gap as the magnetic field strength of the permanent magnets decreases rapidly with distance. The adjustment of the torque generated is therefore achieved by means of the simple adjustment of the gap. Hence no control of a normal force is needed which makes a magnetic coupling the most preferred coupling. In the gap there may be a vacuum, or air, or a fluid, or separator discs or bushings.

[24] Basically there are two designs: there is the axial design wherein the

magnetic field lines run parallel to the axis of rotation (in that case magnets are arranged on discs) or there is the radial design wherein the field lines of the magnets run radial. In that case the magnet rings are mounted one into the other. The radial design is most preferred as it allows for easy mounting of the coupling in between the axis and the capstan.

[25] By preference the axes are situated in planes parallel to one another.

More preferred is if the axes are parallel to one another. Alternatively or additionally it is preferred if axes and capstans are organised such that wire arriving on the surface of a capstan and wire departing from it are in the plane perpendicular to the axis. When the axes are parallel this implies that both capstans are situated in the same plane provided no deflectors (such as sheaves or rolls) are present in the wire path. This makes the fixed gearing between both axes easier as gears can be placed between the axes in a plane parallel to the plane of the capstans. The axes can also be co-axial i.e. one axis is inside the other axis, where the other axis takes the form of a hollow shaft.

[26] A most preferred embodiment is where the first and second axes

coalesce, are one and the same i.e. there is only one axis. A first advantage of this is of course that it saves an axis. A second advantage is that the gearing ratio W1 :W2 is automatically fixed to 1 :1 . A third advantage is that space is saved. A fourth advantage is that in this way it becomes possible to refurbish existing machines, on which already a capstan is present such as a wire drawing bench and/or a wire winder, with the filtering apparatus. This single axis can be driven or not-driven. A driven axis can e.g. be the drive axis of the wire drawing capstan or the winder capstan. A particularly preferred embodiment is when this single axis is not driven. The apparatus can then be introduced in a wire path as a stand-alone unit. The apparatus is then driven by the wire that is pulled through it and still remains functioning as a defect filter.

[27] In order to guide the wire from the first to the second capstan in this single axis embodiment, one or more reversal rolls can be introduced in the wire path. One reversal roll suffices in principle. Over the reversal roll, the wire is led from the first capstan to the second capstan. By preference the reversal roll is introduced such that no reverse bending is induced in the wire. Hence when following a wire on its path the bends are always in the same direction. Reverse bending may introduce torsions in the wire.

[28] An additional tension control can be introduced by braking or driving the reversal roll. If the reversal roll is driven at a linear speed larger than W1 xD1/2, the wire will additionally be tensioned between first capstan and reversal roll and the tension between reversal roll and second capstan will diminish. Alternatively, the reversal roll can be braked in which case the tension between reversal roll and second capstan is increased, while the tension between first capstan and reversal roll diminishes.

[29] A good alternative embodiment is when there are two rolls present. The first roll is associated with the first capstan, the second roll is associated with the second capstan. Both rolls can turn independently from one another. The function of the rolls is to prevent that subsequent loops on the capstan would interfere with one another. By threading the loops of wire over the rolls and the associated capstan, the subsequent loops on a single capstan spread and do not disturb each other during running. This spreading can further be influenced by putting the reversal roll axis under a small angle to the single axis (but both still in the same plane).

[30] In a further preferred embodiment, a straightening device is introduced in the wire path of the apparatus. A straightening device or 'straightener' is a sequence of grooved rollers in a substantially single plane where repeated reverse bending in that plane induces desirable residual stresses on the wire. Goals of using a straightener can be diverse: they can be introduced to give the wire a certain cast (cast is the general curvature adopted by a wire when freely suspended) or - just the opposite - to make the wire straight. They can also be used to influence residual internal stresses on the wire. Compressive stresses at the surface for example are known to improve fatigue resistance of wires. See US 4,612,792 in that respect. Another use is to induce torsions on wire or even cords by putting the grooved rollers slightly above or below the plane of reference.

Straighteners are usually combined: different straighteners are put in series with an angle between the planes of reference (e.g. perpendicular) while the wire is aligned substantially along the intersection line of those planes.

[31 ] The functioning of a straightener is intricate and complex. However, an important parameter for the proper functioning of a straightener is that the tension of the wire running through it must be constant and preferably controllable. This is achieved with the apparatus described. By putting the straightener in the wire path between first and second capstan its' functioning is greatly improved and less prone to variation.

[32] The wire path can be divided into a number of zones. There is the entry zone which are the wire loops on the first capstan possibly extended over a reversal roll (if present). The wire in the 'entry zone' enters at the entry tension Ti of the wire (i.e. the tension before the capstan) and the tension rises to the tension T ₂ of the 'tensioning zone'. The 'tensioning zone' is where the wire leaves the first capstan and arrives at the second capstan thereby possibly passing a reversal roll. In the 'tensioning zone' the tension is at the tension induced by the torque generating coupling and is controlled. Finally the wire exits through the loops of the 'exit zone'. In the 'exit zone' the wire enters at tension T ₂ and exits at exit tension T ₃ which may be higher but is preferably lower than T ₂ . The exit zone starts where the wire enters the second capstan possibly extended over a reversal roll and exits form that capstan.

[33] The straightener can be placed in the entry zone, the tensioning zone or the exit zone. By preference it is placed in the tensioning zone as there the tension is stable and controllable. Within the entry zone the tension can vary between Ti and T ₂ and depends on the position of the loop and the friction of the wire to the capstan. Loops close to the end of the entry zone will be nearer to T ₂ , loops at the start of the entry zone have a tension close to Ti . The same occurs mutatis mutandis in the exit zone: the tension can vary between T ₂ and T ₃ depending on the position of the loop and the friction of the wire to the capstan.

[34] When the straightener is placed in the tension zone, it may be necessary to add a reversal roll to easily thread the wire through the straightener.

[35] According a second aspect of the invention a wire drawing bench is

provided that comprises the filtering apparatus as described above. Such a wire drawing bench can be a dry drawing bench (making use of powdery soap to lubricate the wire when pulled through the drawing dies) or it can be a wet drawing bench (when dies in die holders are submerged in a liquid lubricant). In either case the filtering apparatus is placed after the last drawing die (for the purpose of this application the 'last drawing die' is the die with the smallest diameter. A synonym is 'head die') and outside any lubrication, as a sufficient amount of friction is important for the proper functioning of the apparatus.

[36] A particularly preferred embodiment is when the first capstan corresponds to the drawing capstan that follows the last die i.e. the head capstan that pulls the wire at final diameter through the last die or head die. The first axis corresponds then to the axis of the head capstan. The second capstan may be mounted on a second axis that is turning in a fixed gear ratio to the first axis. Or even more preferred is if the second capstan is also mounted on the axis of the first capstan that is the axis of the head capstan.

[37] According a third aspect of the invention, a winder comprising the filter apparatus according the first aspect of the invention is claimed. A winder in general has a pay-off section (for delivering wire) and a take-up unit (for winding the wire on a carrier). The filter apparatus can easily be retrofit to an existing winder. In a particularly preferred embodiment no drive is needed to make the equipment work. The capstans are made to turn by means of the wire that is being pulled through. Of course the final take-up unit must be able to deliver enough power to make the capstans turning and to overcome the torque generated by the torque generating coupling. [38] According a fourth aspect of the invention a method to filter out defect from steel wires by use of the apparatus according the first aspect of the invention is described. First the wire under test is fed to the apparatus (from a pay-off spool, from a drawing bench or any other device known for wire generation or treatment), say at a tension T The wire is laced around the first capstan with first one or more loops. One loop is complete when the wire arrives back at its point of departure on the capstan. If there is a straightener in the wire path, the wire can be led through that straightener. The wire then pursues its path by lacing it over the second capstan with second one or more loops. The wire is extracted from the second capstan at a tension T ₃ . In the wire path between the first and second capstan (the 'tensioning zone') the wire is under a tension of T ₂ . This is the 'test tension' and its value can be adjusted by simply adjusting the torque generating coupling.

[39] When a straightener device is present, this test level will be slightly larger after the wire has passed the straightener than before entering the straightener as some force is needed to pull the wire through. In any case such deviation is rather small.

[40] Depending on the test tension level induced on the wire different effects can be achieved. If the test tension is larger than the wire tension that will be applied on it during its further use, the wire be filtered from defects to a degree sufficient for its use: a first advantageous use. An example is the 25 N tension that is applied on a sawing wire in a multi wire saw machine. Setting the test tension on 30 N already results in a substantial filtering of defects. For example the test tension can be set to larger than 20% or 30% or even 40% of the breaking load of the wire. As long as the wire remains in its elastic region that extends for most wires up to about 70%, and for some wires up to 80% and for high tensile wires up to 90% of their breaking load, no particular changes will be induced on the wire except for the filtering effect.

[41 ] A second advantageous use occurs when the test tension is in the plastic region of the wire (i.e. above 70% for most wires, above 80% for some wires and above 90% for special wires) in addition to the filtering effect, changes in the wire can be induced. For example the cast of the wire can be altered. Another advantageous use is to change the residual internal stresses in the wire. When combined with a straightener this cast effect or residual internal stress effect can be achieved easier as the straightener may lift parts of the wire cross section into plasticity by adding bending stresses to the imposed tensional stresses.

[42] Important for the proper working of the apparatus or the method is that the number of said first one or more loops and the number of said second one or more loops should be sufficient that no slip occurs between wire and first capstan and wire and second capstan during use. Slip on a capstan is generally modelled through the Euler friction formula: if the exit tension at a capstan T _ou t is held larger than Τί _η ·β ^"μθ wherein T _in is the tension at which the wire enters the capstan, μ is the angular friction coefficient (in rad ^"1 ) and Θ is the total contact angle (in rad, one loop corresponds to 2ΊΤ,

'contact' is between wire and capstan) no slip will occur. Applying this criterion on the second capstan (T ₃ relative to T ₂ ) and on the first capstan (T ₂ relative to Ti) will result in the required number of loops. As a general rule the number of second one or more loops will be higher than the number of first one or more loops as the exit tension is usually the lowest.

[43] As the friction coefficient 'μ' should be sufficiently high, the use of friction reducing agents (lubricants, oils, waxes or other) on the capstans should be avoided at all means.

[44] Possibly the first one or more loops can be split between the first capstan and the at least one reversal roll. As the contact angle between wire and capstan then reduces, the number of first one or more loops must be concurrently adapted. Alternatively or concurrently the second one or more loops can be split between the second capstan and the at least one reversal roll. The number of second one or more loops must be

concomitantly adapted.

[45] A method for using the apparatus as described above comprises the

following steps:

• feeding a metal wire, such as a steel wire, to be tested

• lacing said wire around the first capstan with first one or more loops • optionally said wire is led through a straightener

• lacing said wire around said second capstan with second one or more loops,

• extracting said wire from said apparatus

characterised in that

the torque generating coupling is adjusted such that a constant test tension on said wire is induced when going from said first to said second capstan.

[46] The method can further be complemented by the feature that said torque generating device is adjusted to a torque that induces a test tension that is at least 20 percent of the breaking load of said wire.

[47] The above two methods wherein the number of the first one or more loops and the number of the second one or more loops are sufficient that no slip occurs between the wire and the first capstan and between the wire and the second capstan.

[48] The above three methods to use the apparatus wherein first and second axis coalesce further having the feature that said first one or more loops are shared between said first capstan and said at least one reversal roll, and wherein said second one or more loops are shared between said at least one reversal roll and said second capstan.

[49] A particularly favoured method is when the capstans are driven by means of the wire being pulled through.

Brief Description of Figures in the Drawings

[50] Figure 1 describes a prior-art head capstan of a wire drawing machine.

[51 ] Figure 2 describes a first preferred embodiment of the invention.

[52] Figure 3 describes a second preferred embodiment of the invention

inclusive a straightener.

[53] Figure 4 shows a third preferred embodiment of the invention with a

different mounting of the torque generating coupling.

[54] Figure 5 shows a general working principle of the invention. Mode(s) for Carrying Out the Invention

[55] Figure 1 depicts schematically a head capstan on a wire drawing bench 100. The wire 102 exiting from the head die 104 is guided in loops over the head capstan 106 and the reversal roll 108. The head capstan is fixedly mounted on the driven axis 124. The reversal roll 108 can also have other functions such a length counter wheel. After some loops the wire leaves the machine over a sheave 1 10. The number of loops is sufficient to overcome the force needed to draw the wire through the head die.

[56] Figure 5 shows the apparatus 500 in an embodiment with two axes. Wire 518 enters the apparatus on the first capstan 504. The wire 518 is laced around the first capstan with first one or more loops. The first capstan 504 is fixedly connected to first axis 502. The first capstan 504 has a diameter D1 that is equal to 2xRi . In this example the first axis 502 is driven. The wire 518 pursues its route to second capstan 508. Again the wire is looped on the capstan with second one or more loops. The second capstan 508 has diameter D2 equal to 2xRi .

[57] The second axis 506 to which the second capstan 508 is coupled is driven by a gearwheel 516 to which the axis 506 is fixedly connected. Gearwheel 516 meshes with reversal wheel 512 that on its turn meshes with gearwheel 514 that is fixedly connected to axis 502 and hence also capstan 504. The reversal wheel 512 is introduced such that both capstans turn in equal direction. In case gearwheels 514 and 516 would directly mesh (no reversal wheel) the threading of the loops would lead to a reverse bending of the wire which is less desired.

[58] The number of teeth on second gearwheel 516 is less than the number of teeth on the first gearwheel 514, which makes the angular speed W2 of the second axis larger than the angular speed W1 of the first axis. So even if D1 would be equal to D2, the condition that W1 xD1 is smaller than W2xD2 would still be met. In this embodiment Ri has deliberately been chosen somewhat smaller than R ₂ which may increase the ratio

W2xD2/W1 xD1 above the elongation at break A _t of the wire 518 under test. [59] The coupling of the first capstan 504 to the second capstan 508 is through torque generating coupling 510 which is e.g. a friction disk coupling. In this embodiment the coupling is situated between the second capstan and the second axis. The coupling is adjustable through increase of the normal friction force. The stretch of wire spanning from first capstan to second capstan is subjected to a test tension T ₂ that is controllable by the coupling. Any defect of the wire with local breaking load lower than T ₂ will be eliminated. The test tension can be measured for example by a wire tension meter (e.g. Hans-Schmidt). When gauged for different tensions, the adjustment of the friction coupling can be used to set the test tension.

[60] The number n ₂ of second one or more loops is chosen such that no slip occurs i.e. T ₃ (the exit tension) is larger than T ₂ xexp(- μ·η ₂ ·2ττ). Likewise the number ni of first one or more loops is chosen such that no slip occurs on the first capstan i.e. T ₂ is larger than Tixexp(- μ·ηι·2ττ). Ti is the tension of the wire at entry.

[61 ] Figure 2 shows a more practical embodiment of the filtering apparatus 200 as implemented on an existing wire drawing bench. A first capstan 206 - which is also the head capstan following the head die 204 - is fixedly mounted on drive axis 224 that is driven by the motor of the drawing bench. The first one or more loops of wire 202 are threaded over reversal role 208 with 4 loops shared between capstan 206 and reversal roll 208. At the fifth loop the wire travels to the second capstan 212 mounted on the same drive axis 224. The second capstan is mounted to axis 224 with bearings 230 in between. Hence W1 is equal to W2 as there is only one axis. From there the wire is divided over second capstan 212 and second reversal roll 208' for about 12 loops. The second reversal roll 208' rotates independently from first reversal roll 208. The second capstan 212 is coupled to the first capstan 206 through torque generating coupling 214 which in this case is an easy adjustable, radial magnetic coupling with an indicator scale so that torque levels can be set reliably. The torque generating device is in this embodiment situated between the second axis - that is the same as the first axis - and the second capstan. [62] In a further embodiment depicted in Figure 3, the embodiment of Figure 2 is completed with a straightener device 318 that is present in the tensioning zone - i.e. where the wire travels from first to second capstan - of the wire path. In the figures, like parts are identified with like unit and tens numerals. An additional reversal roll 316 is introduced in order to allow the straightener 318 to be mounted conveniently. The wire 302 passing through the straightener is kept at the constant test tension. In addition, the repeated reverse bending of the straightener induces additional bending stresses in the wire that will help even more to filter out defective spots on the wire.

[63] In Figure 4, an alternative positioning of the torque generating coupling is illustrated. The apparatus is again a single axis arrangement wherein both first 406 and second capstan 412 share the same axis 424. Now the second capstan 412 is fixedly connected to the axis 424 while the first capstan 406 is rotatably connected to the axis 424 by means of bearings 430. In this embodiment the torque generating coupling 414 is situated between the first and second capstan.

[64] In the different embodiments described the capstans can also be driven by the wire that is being pulled off from the second capstan with tension T ₃ rather than by driving the first axis. In practise, the tension T ₃ that drives the apparatus, will be lower than the induced test tension T ₂ - set by the magnetic coupling 214, 314, 414 - provided enough loops are present on the second capstan to prevent slipping of the wire. The higher the torque generation coupling is set, the higher the tension T ₂ and the more loops must be laid around the second capstan 412. Particularly the embodiment according to Figure 4 is suitable for being used as a stand-alone device in that the motive force enters through the second capstan and is transferred over the torque generating coupling to the first capstan.