FR2521051A1 - Automatic guidance system for welding burner - where proximity sensors adjacent to workpieces feed signals to computer driving servomotors aligning burner on seam - Google Patents

Abstract

Two workpieces are aligned at a specific angle w.r.t. each other, and are then joined by an external or an internal fillet weld. The process and appts. includes a carrier, which can be driven along the weld, and has two servomotors driving a vertical pillar in a horizontal direction (X), and a vertical direction (Y). On the pillar is a jack used to adjust the position of two lever arms, each of which carries a proximity sensor. The pillar also carries three more servomotors used to move a slide in the (X,Y) directions, and also to alter the angle of a welding burner mounted on the slide. The two proximity sensors are connected to a computer employed to adjust the position of the burner.

Description

 The present invention due to the work of MM.

Jean-Marie DETRICHE of the CERM Society and Paul
MARCHAL of the CEA relates to a self-adaptative method of positioning an element such as a welding torch with respect to two parts forming between them a known angle A, which can be constant or variable, as well as to a device for setting of this process.

 Preferably, although in a nonlimiting manner, the invention relates to the welding of flat sheets (at least locally, with respect to the dimensions of the welding head) forming between them an angle, measured in a plane perpendicular to the line of welding, between 500 and about 1400. Such an assembly will be called an angle in the following description. Of course, the welding of the two plates of an angle can be performed either from the inside or the outside. The method and the device according to the invention apply to these two types of welding.

 The assembly of sheets by welding to form an angle is a case often encountered in industry. This welding operation can be done either manually or automatically. However, when this operation is performed manually, it poses a large number of problems related in particular to the accuracy of the work to be performed and the need to preheat the parts when they are thick. As a result, it is virtually impossible to ensure intensive manual production and working conditions are often unhealthy and dangerous.

 For these different reasons, angle welding is often done automatically. For this purpose, special machines are currently used.

These machines must be able to adapt to the variations of the parts to be welded with respect to their theoretical form, for example manufacturing inaccuracies and especially the deformations resulting from the preheating and the heating during welding. These machines generally comprise a cross carriage that can be moved along two perpendicular axes and carrying both the welding torch and a mechanical probe consisting of a probe member (rigid rod) and a position transducer. Depending on whether the welding takes place from the inside or the outside of the angle, return means such as springs hold the probing member in abutment on the lips opposite the parts to be welded, or in the bottom of a chamfer arranged at the junction between the two parts so, in both cases, that it follows this junction.The transducer measures the position of the difference of the end of the probe relative to at its reference position and delivers two signals proportional to the deviations measured in two perpendicular directions. These signals are the position control commands of the carriage which is moved so as to permanently cancel these positional deviations, which brings the torch back to the reference position relative to the parts to be welded.

 These known machines mainly have the disadvantage of requiring perfect guidance of the probing member. In practice, such guidance is very rare because it becomes imperfect as soon as material is introduced into the groove that must follow the probing member. This is particularly the case when an imperfect realization of the chamfer, a local defect of the parts at their junction or the presence of weld points necessary to the prepositioning of the parts to be welded, which may lead to an escape of the probe member out of the groove.

This is also the case after the realization of a first weld bead during a multi-pass welding, since the groove is then filled and no longer exists.

As a result, the special machines used to date are therefore unsatisfactory.

 The present invention specifically relates to a method and a self-adaptive device for performing the welding of two corner pieces by means of a welding torch, while not having the disadvantages of automatic welding devices of the prior art and in allowing in particular to perform a precise welding in as many passes as necessary.

 However, the invention is not limited to the positioning of a welding torch and applies to the positioning of any other member or tool relative to an angle, before, during or after welding. Thus, the invention also applies to the positioning of a sensor for non-destructive testing of the weld bead, a tool for cutting a metal angle (for example for dismantling it), or a tool for machining, gluing or realizing a seal.

For this purpose and in accordance with the invention, there is provided a self-adapting method of positioning an organ with respect to two parts forming between them an angle measured in a plane perpendicular to their intersection, said member being able to move along this last characterized in that it consists: - to arrange before moving said body two cap
proximity probes mounted on a support of said orga
against said parts, - to measure with said sensors the distance between
parant said support and each of said parts,
when moving said organ along the inter
section, - to determine the corrections to be made to the posi
welding torch holder relative to
to the parts to be welded, according to the angle between
the pieces, measured distances and the orienta
the sensors relative to said clamps
and - to move said support in order to perform the horns
rections thus determined.

 According to another characteristic of the invention more particularly adapted to the realization of a multipass welding, positioning and orienting said member relative to said support before moving said member along the intersection.

Preferably, the two parts each having a flat portion near their intersection, said sensors then measuring the distance separating said support and each of said planar portions, the axes of the proximity sensors are arranged in a plane perpendicular to the intersection of the two parts and oriented relative to one of the axes of a reference mark associated with the support, so that the axis of each of the sensors defines with this
A reference axis has an angle B = Tt - - when the angle between the sheets, measured on the side of the torch is less than m.

AT
When the angle A between the sheets, measured on the side opposite the torch, is less than 2 #, the proximity sensors are also arranged in a plane perpendicular to the intersection of the two parts and oriented with respect to the axis of reference, so that the normal to the axis of each of the captures defines with this reference axis an angle B = 2.

Different definitions of wire angles and
B according to whether the welding takes place from the inside or outside of the angle allow to express the corrections aX and hY using the same relations.

Thus, in both cases, the corrections AX, #y to bring to the position of the support of the torch with respect to the parts to be welded are determined using the relations
aX = (dl - d2). sin B (1) d1 + d2
#Y = (2) .cos B (2)
This feature has the advantage of allowing the use of the same device and in particular identical calculation means, for performing internal and external welding. However, it will be noted that the definitions of the angles A and B and, consequently, the expressions of the corrections EX and hY to be carried out, can be modified without departing from the scope of the invention.

 The invention also relates to a self-adapting device for positioning a member relative to two parts forming between them a measured angle in a plane perpendicular to their intersection, said member being mounted on a support adapted to move to follow said intersection, characterized in that it comprises at least two proximity sensors mounted on said support, means for orienting the sensors vis-à-vis said parts, means associated with the proximity sensors to determine the distance separating the support and the surface of each of the parts, means for calculating the corrections to be made to the position of the support relative to the parts, as a function of the angle formed between the parts, the measured distances and the orientation of the sensors with respect to said parts , and means for moving the support in order to effect the corrections thus calculated.

 When used to perform multipass welds, such a device further comprises means for positioning and for orienting said member relative to said support.

According to a first embodiment of the invention, the proximity sensors each comprise a movable member held in contact with the surface of the flat part of the part vis-à-vis by elastic means, the means for determining the distance separating the support and the surface of each of the parts comprising, for each sensor, means for detecting the displacement of this movable member. According to two variants of this first embodiment, the movable member may comprise either a spherical portion or a wheel in contact with the surface of the part.
According to a second embodiment of the invention, the proximity sensors comprise eddy current sensors comprising at least one coil, the means for determining the distance separating the support and the surface of each of the parts to be welded comprising a means of processing of the signal delivered by each sensor, emitting a proximity signal that varies with the distance to be determined according to a non-linear function, and a proximity signal processing means processing the latter according to a function inverse to said function.

 According to a third embodiment of the invention, the proximity sensors each comprise an eddy current sensor comprising at least one coil, a signal processing means delivered by the sensor delivering a proximity signal representative of the distance separating the sensor of the surface of the part vis-à-vis, means for moving the sensor parallel to its axis and servo means acting on these means for moving the sensor in response to the proximity signal delivered by the processing means , in order to maintain this signal at a constant value, the means for determining the distance separating the support and the surface of each of the parts to be welded, comprising means for detecting the displacement of each sensor parallel to its axis.

 In order to allow the device according to the invention to perform the welding of the angles from the inside as well as from the outside, each sensor may be mounted at the end of a branch of the same length which is itself pivotally mounted by its another end on an axis parallel to the intersection of the two parts, the means for orienting the sensors acting on the branches so as to rotate about this axis by an equal and opposite angle.

Various embodiments of the invention will now be described by way of non-limiting example with reference to the accompanying drawings in which:
FIG. 1 is a sectional view in a plane xOy perpendicular to the weld line illustrating the principle of the welding process according to the invention applied to the welding of an angle from the inside,
FIG. 2 is a view comparable to FIG. 1 illustrating the principle of the welding method according to the invention applied to the welding of an angle from the outside,
FIG. 3 represents a self-adapting welding device according to the invention,
FIG. 4 shows two variants of a first embodiment of the proximity sensors used in the device of FIG. 3,
FIG. 5 schematically illustrates a second embodiment using eddy current proximity sensors and appropriate processing means for the signals delivered by these sensors;
FIG. 6 represents the first processing circuit which is associated with the eddy current sensors in the embodiment of FIG. 5,
FIG. 7 represents the logic phase shifting circuit used in the circuit of FIG. 6,
FIG. 8 represents the phase discriminator circuit used in the circuit of FIG. 6,
FIG. 9 represents the variations of the signal Vph delivered by the circuit of FIG. 6 as a function of the distance h separating the end of the eddy current sensor from the facing surface, and
FIG. 10 represents a third embodiment using current sensors.
Foucault who are enslaved to stay at a constant distance from the surface in vis-à-vis.

 FIG. 1 shows two solder plates 10 and 12 whose plane surfaces S1 and S2 intersect at a point P in the plane of the figure containing a reference reference xOy associated with the carrier of the welding torch T. The surfaces S1 and SZ form between them, in this same plane, an angle A less than 1800 in the case of Figure 1 which relates to the welding of the angles from the inside.

 Preferably, the plane defined by reference numeral xOy is perpendicular to the surfaces and S2 and, therefore, to the weld line between parts 10 and 12. The Ox axis has been chosen as being horizontal and the vertical axis Oy, but it goes without saying that this choice is arbitrary and that any other definition of the reference reference xOy could be used.

In the rest of the presentation, we will assume:
AT
a) that the angle A defined between the surfaces S1 and
does not vary significantly during penny
dage, and
(b) the direction of the bisector of that angle A
does not vary significantly at
during a welding operation.

 These two hypotheses are not limiting because, assuming that at least one of them is not realized, it is possible by means of an auxiliary device not forming part of the invention, on the one hand, to determine at any moment the value of the angle  formed between the two surfaces and, secondly, to determine at any moment the variations in the direction of the bisector of the angle #A.

 The information thus collected can be used without difficulty to correct at any time the positions of the various parts of the device according to the invention, so that everything happens then as if the assumptions a) and b) above were still carried out.

 In accordance with the invention, the xOy coordinate system associated with the wearer of the welding torch T is defined as a fictitious point F of intersection of fictitious surfaces S'1 and S'2, which corresponds to the place occupied by the point P. intersecting surfaces S1 and S2 when the positioning of the carrier has been achieved. Taking into account the hypotheses a) and b) above, and assuming that the OH1 and OH2 distances separating the center O from the reference xOy of the fictitious surfaces St1 and S'2, perpendicularly to the latter, are equal, we place the point F on the reference axis Oy and the normals OH1 and OH2 are each with the vertical OF the same angle B.

FIG. 1 shows that the fictitious surfaces S'1 and Sss2 are respectively parallel to the real surfaces S1 and S2 of the parts 10 and 12. If, according to the invention, proximity sensors are
C1 and C2 are oriented so that their axes coincide with the directions OH1 and OH2, it will therefore be possible to measure the distances OH'1 and OH'2 between the point O and each of the surfaces S1 and
This orientation can be easily achieved before any measurement by tilting each of the heads C1 and C2 of an angle B with respect to the vertical
OF, this angle B being determined very simply from the angle by the relation: B - W ~ A (3)
2
If the welding torch is to weld plates 10 and 12 in several passes; it will be readily understood, as shown in FIG. 1, that the coordinates xt and Yt of the projection on the plane xOy the end t of the welding torch T, in the xoy coordinate system, as well as the orientation a of the axis tt 'of this torch relative to the reference axis Oy must be modified for each weld pass.

Thus, FIG. 1 shows the position and the orientation of the welding torch relative to the xOy mark after a first pass p1 has been made and before the completion of the second pass p2 which requires the arrival of the end t of the torch in t2. For a given angle and for a given pass, and considering the assumptions a) and b) made previously, the coordinates xt, Yt and the orientation at are preprogrammed constants. According to the invention, the position and orientation of the torch will therefore be determined before each welding pass.

During the actual welding operation, the proximity sensors C1 and C2 and the torch
T being positioned correctly, it remains to determine at any time the corrections AX and #Y to be made along the directions Ox and Oy to bring the point F in P and, consequently, the end t of the torch in t2. These corrections are determined from the knowledge of the angle B, calculated as seen from the angle A, and from the measurement of the distances d1 and d2 which respectively correspond to the segments OH'1 and OH ' 2 in FIG. 1, using the sensors C1 and C2.In fact, these corrections are determined by the relations (1) and (2) already mentioned, which we recall below.
AX = (dl -d2). sin B (1)
#Y = (dl + d2). cos B (2)
2
As illustrated in Figure 2, the principle of welding an angle from the outside is virtually the same as that just exposed for welding an angle from the inside with reference to Figure 1 .

Thus, if the angle A is defined as the angle less than 1800 formed by the surfaces S1 and the sheets to be welded 10 and 12 on the opposite side to the welding torch, and if the angle B is defined as being angle formed between the extension OF of the vertical axis Oy of the reference mark xOy and two segments O '1 and O' 2 which are equal and symmetrical with respect to the axis Oy and perpendicular to the normals' 1 H11 and A
O'2 H'2 to the surfaces S1 and S2 the angle B is related to the angle A by the relation
B = 7 (4)
If proximity sensors C1 and C2 are placed along the segments 0 '1 H' 1 and '2' 2, these sensors will thus detect the distances between the points' 1 and 2 and the surfaces S1 and S2 facing each other. when relation (4) is satisfied.

 As in the case of the inner angles described with reference to FIG. 1, the end t of the welding torch T is then positioned and the axis tt 'of this torch is oriented according to the pass to be made, on the basis of the data xt, Yt and at corresponding to this pass.

 Finally, one determines the corrections AX and hY to bring during welding so that the fictitious point F comes to be placed on the point P, intersection of the surfaces S1 and S2 in the plane xOy. Taking into account the definitions of the angles A and B, these corrections AX and AY are given by the same relations (1) and (2) that in the case where one carries out the welding of the angle by the interior.

 FIG. 3 shows an exemplary embodiment of a device for implementing the method whose principle has just been described with reference to FIGS. 1 and 2.

 The mechanical part of the device represented in FIG. 3 comprises a support plate 14 intended to be mounted on a mobile carrier (not shown) enabling the welding torch T to move along the joint J formed between the surfaces S1 and S2 of the sheets 10 and 12 to be welded. The plate 14 carries, via a correction mechanism 16, a vertical support arm 18 whose lower end carries, on the one hand, the proximity sensors C1 and C2 by means of a mechanism 20 and, on the other hand, the welding torch T via a positioning and orientation mechanism 22.

 The correction mechanism 16 shown in FIG. 3 comprises two horizontal columns 24 and two vertical columns 26 integral with the plate 14 and parallel respectively to the horizontal directions Ox and vertical Oy of the reference xOy, perpendicular to the joint J, which is associated with fictitiously to the support arm 18. A piece 28 is slidably mounted on each of the columns 24. The two parts 28 are connected by a vertical worm 30 which is received in a tapped hole formed at the upper end of the vertical arm 18 The worm 30 is rotated by a motor My allowing the hY corrections to be made as will be seen later.

 In a comparable manner, a piece 32 is slidably mounted on each of the columns 26, the pieces 32 being interconnected by a horizontal worm 34 received in a threaded hole also formed at the upper end of the support arm 18. A motor M control the rotation of the screw 34 in order to make the AX corrections as will be seen later.

In the embodiment shown, the mechanism 20 for orienting the sensors C1 and
C2 perpendicularly to the surfaces S1 and S2, that is to say according to the angle B defined above with reference to FIGS. 1 and 2, may be used both for welding an internal angle and for producing the weld. an external angle. For this purpose, we see in Figure 3 that each of the sensors
C1 and C2 is mounted at the end of an arm or branch 36, 38 respectively, the other end of each arm 36 and 38 being pivotally mounted about a common axis 40 parallel to the joint J to be welded. More specifically, the axes of each of the sensors C1 and C2 are perpendicular to the arm 36 and 38 corresponding and the latter are of the same length and maintained symmetrically with respect to the reference axis Oy.The set consisting of C1 sensors and C2 and the arms 36 and 38 is disposed in the same plane perpendicular to the joint J. The arms 36 and 38 thus materialize the segments O '1 and O' 2 in Figure 2. It should be noted that segments of the same type could be plotted in FIG. 1 without modifying the relations obtained from this figure.

 The holding of the arms 36 and 38 in position and the adjustment of their orientation are carried out by means of an electric jack V fixed close to the lower end of the support arm 18 and rotating a vertical worm 84 on which is mounted a nut 86 connected to the arms 36 and 38 by hinged rods 88 and 90 respectively.

 The positioning and orientation mechanism 22 of the torch T comprises two columns 92 parallel to the horizontal axis Ox and two columns 94 parallel to the vertical axis Oy. These columns are integral with the lower part of the support arm 18.

 A piece 96 is slidably mounted on each of the columns 92 and these parts are interconnected by a vertical worm 98 which enters a tapped hole formed in a movable support plate 100. A motor My controlling the rotation of the worm 98 thus ensures the positioning Yt of the end of the torch T.

 In a comparable manner, a part 102 is slidably mounted on each of the columns 94, the parts 102 being interconnected by a horizontal worm 104 received in a threaded hole also formed in the support plate 100. A motor Mx controls the rotation of the screw 104 and thus ensures the xt positioning of the end of the torch T.

Finally, the support plate 100 carries an engine
My controlling the inclination of the torch T relative to the plate 100 and thus ensuring the orientation <Lt of the torch relative to the reference axis Oy.

The angular deflection of the motor Ma is at most e600 approximately. t
The device according to the invention comprises, in addition to the mechanism which has just been described, a means 106, 108 respectively associated with each of the sensors
C1 and C2 for delivering, in association with the latter, signals d1, d2 respectively representative of the distances OH'1 and OH'2 in FIG. 1 and distances O'1H'1 and 0'2H'2 in FIG. The device further comprises a calculation means 109 on
h which is displayed beforehand the value of the angle B determined according to the angle AA formed between the sheets (using relations (3) or (4) as appropriate) and which receives during welding the signals d and d2 in order to calculate at any time the values of the correction signals EX and hY.

The values of the correction signals EX and ssY are calculated in the circuit 68 using the relations (1) and (2) set forth above with reference to FIGS. 1 and 2. They are used to control the motors Mx and My to correct at all times the position of the support arm 18 and the torch assembly
T, detectors C1 and CL, which it supports.

For a given angle, the orientation of the sensors C1 and C2 is performed only once before welding by means of the jack V according to the value of the angle AA. On the other hand, the positioning and the orientation of the torch T by means of the motors Mx, My and Ma are carried out between each successive welding pass. Finally, the setting of the torch T on the joint J is carried out at any moment during welding by the engines
Mx and My as a function of the distances d1 and d2 determined using the sensors C1 and C2.

 In the device which has just been described with reference to FIG. 3, the sensors C1 and C2 as well as the means 106 and 108 associated with them may be of different types.

 Thus, FIG. 4 shows two variants of a first embodiment of the sensors C1 and C2 and means 106 and 108, according to which there is a mechanical contact between the sensor and the surface S1, S2 respectively to face.

In the first embodiment shown in FIG. 4a, each of the sensors C1 and
C2 comprises a first part 110 fixed to the end of the corresponding arm and on which is slidably mounted a second part 112 which extends towards the surface S1, S2 respectively by a rod 114 terminated by a contact sphere 116 A spring of compression 118 placed between the parts 110 and 112 permanently urges the sphere 116 in contact with the surface S1, S2 respectively when the distance between the part 110 and this surface is less than the maximum operating distance for which a collar 120 formed on the moving part 112 comes into contact with a shoulder 122 formed on the part 110.

 In this embodiment of FIG. 4a, the means 106, 108 for delivering a signal d1, d2 representative of the distance O'1H'1 and O12H'2 respectively (FIG. 2) comprise a potentiometer 124 housed in the piece 110 and with which is in contact a slider 126 carried by an extension of the rod 114 inside the workpiece 110.

 In the embodiment of Figure 4b, there is the piece 110 attached to the arm supporting the sensor, the workpiece 112 sliding on the workpiece 110, the rod 114 secured to the workpiece 112, the compression spring 118 placed between the parts 110 and 112 to bias the latter towards the surface S1, S2 respectively, and the assembly formed by the flange 120 and the shoulder 122 limiting the stroke of the sensor.

This variant differs mainly from the variant of FIG. 4a in that the contact sphere 116 is replaced by a roller 116 'kept in contact with the surface S1, S2 by the spring 118. Moreover, it can be seen from FIG. 4b that the slider 126 'is pivotally mounted on the piece 110 and on the end of an extension of the rod 114 inside the piece 110, instead of being directly attached to the end of this extension as in the variant of Figure 4a. As a result, the potentiometer 124 'has the shape of a circular arc centered on the point of articulation of the slider 126' on the part 110.

 If we compare the embodiments of Figures 4a and 4b, we see that the first of them is simpler to implement and that the friction between the contact sphere 116 and the surface of the sheet vis-à-vis on-line may result in dangerous jolts to a smooth control. This disadvantage is eliminated in the variant of Figure 4b where the contact is made via the wheel 116 '. However, the latter variant has the disadvantage of defining a preferred direction of movement perpendicular to the axis of the wheel.

 FIG. 5 schematically shows a second embodiment of the sensors C1 and C2 and means 106 and 108 associated therewith.

In this embodiment, each sensor
is an eddy current sensor that consists of
in the simplest case shown in Figure 5
of a single coil B.

Recall that the principle of such a sensor
is to feed a coil by a current elec
high frequency sinusoidal receiver, so as to create
an alternating field that induces in a piece of maté
Electrically conductive material disposed opposite the eddy current sensor, in turn creating a field that opposes the original field and modifies
the impedance of the coil. The variations in the impedance of the coil thus give an indication of the disposition of the latter with respect to the part facing each other.

In order to deduce from the signal delivered by the coil B an output signal representative of the distance h separating the coil from the surface opposite, that is to say, indirectly, distances d1 and d2 to be determined, each coil is associated with the processing means 106, 108 (FIG. 3). In the case of Figure 5, these means 64, 66 consist of a first processing circuit 128 which will be described in more detail with reference to Figures 6 to 8 and a second processing circuit 130. Note from now on that the presence of this last circuit 130 is justified by the non-linear character of the function defined by the response curve of the signal Vph delivered by the circuit
128, depending on the distance h.The circuit 130 has indeed only function to make linear the va
riations of the output signal V of the means 106, 108
(Figure 3) depending on the distance h. For this purpose,
the circuit 130 processes the signal Vph delivered by the
circuit 128 by applying the function V'ph = f (h)
opposite of the one that is defined by the latter cir
cooked.

 FIG. 5 shows in greater detail the electronic processing circuit 128 of the signals delivered by the eddy current sensor.

This circuit makes it possible to carry out a differential phase measurement between the signal coming from the coil B and the signal emitted by an oscillator 44 taken as a reference. It therefore makes it possible to carry out a measurement of proximity, that is to say the measurement of the distance h separating the lower end of the coil B from the surface S1, S2 facing each other.

 In order to detect variations in the impedance of the coil B, it is placed with a resistor R in a symmetrical measuring bridge whose two branches R1 and R2 serve to ensure equilibrium. The measuring bridge thus formed is fed between the high point defined by the junction of the coil B and the resistor R and the low point defined by the junction of the branches R1 and R2 by a high frequency sinusoidal signal (for example 240 kHz). from the HF oscillator 44. The sinusoidal output signals Vb1 and Vb2 are in accordance with the common terminals of the branches B and R1 and R1 and R2 are respectively transmitted to the + terminal and to the - terminal of a differential amplifier 46 which delivers a sinusoidal signal Vb. The signal Vb is injected into an adder 48 with a signal Vo which corresponds to the signal injected into the measuring bridge by the oscillator 44 and phase shifted by 75 by a phase-shifter 50. The sinusoidal signal V c delivered by the summator 48 is injected at the + terminal of a comparator 52 whose terminal - is connected to the ground. Similarly, the sinusoidal signal delivered by the oscillator 44 is injected at the + terminal of another comparator 54 whose terminal - is also connected to the earth. The comparators 52 and 54 deliver positive logic signals for the positive half-waves of the sine-wave signals fed to them. The logic signal delivered by the comparator 54 is representative of the alternations of the sinusoidal signal delivered by the oscillator 44. It is injected into a logic phase-shifter 56 delivering a phase-shifted signal of a given value with respect to the input signal. This phase-shifted signal is in turn injected into a phase discriminator 58 at the same time as the signal delivered by the comparator 52, the latter being representative of the value of the signal at the terminals of the coil B. The phase discriminator 58 delivers a signal Vph representative of the phase difference between the signal injected into the bridge and that across the coil B.

As shown in FIG. 7, the logic phase shifter 56 comprises an inverter 60 and two circuits
RC for delaying the signal injected at E in the phase shifter and the output signal of the inverter 60.

The signal injected into the phase shifter and delayed by one of the RC circuits, as well as the output signal of the inverter 60 are injected into a first gate
NO / OR 62. Similarly, the signal injected at E into the phase shifter 56 as well as the signal coming out of the invertor 60 and delayed by the second circuit RC are injected into a second NAND / OR gate 64. The signals delivered by each of the NO / OR gates 62 and 64 are injected into an OR gate 66 which delivers at S a logic signal, each pulse of which corresponds to the beginning and the end of a slot of the signal injected at the input.
E of the phase shifter 56. This logic signal is injected at the input of a monostable 68, at the input J of a flip-flop JK 70 and, via an inverter 72, at the input K of this last. The output signal of the monostable 68, which corresponds to a series of fixed time slots starting with a determined delay with respect to each pulse of the signal delivered at S, is injected at the input H of the flip-flop JK whose output signal delivered in Q is identical to the injected signal E input of the phase shifter but out of phase according to a value determined by the delay imposed by the monostable 68. This output signal is injected into the phase discriminator 58.

 As shown in FIG. 8, the two inputs of the phase discriminator 58 are connected, on the one hand, to an EXCLUSIVE OR gate 74 and, on the other hand, to an EXCLUSIVE OR gate 76 after passing through an inverter 18 of the signal. from the logic phase shifter 56.

The output of each of the EXCLUSIVE OR gates 74, 76 is connected to an integrator 78, 80 whose output signals, representing the DC component of each of the signals injected at the input of the phase discriminator, are injected into a subtractor 82 delivering the DC output signal Vph whose value corresponds to the phase difference between the signal delivered by the phase-shifter 56 and the signal coming from the comparator 52.

 The variations of the signal Vph delivered by the processing circuit 128 which has just been described with reference to FIGS. 6 to 8 as a function of the distance h between the end of the sensor C1, C2 and the surface S1, S2 are represented on Figure 9.

 As this figure clearly shows, there is a limit huma, at the distance h beyond which the variations of the output signal Vph become practically zero. For a given piece, this distance can be about 25 mm for a 10 mm diameter sensor. Of course, the distance h can be adjusted at will by acting on the carrier (not shown) supporting the mechanical part of the device shown in FIG.

 A first solution to the problem posed by the non-linear character of the response of the processing circuit 128, shown in FIG. 9, has just been described with reference to FIG.

 A second solution to this problem will now be described with reference to FIG. 10 which represents another embodiment of the means 106, 108 (FIG. 3), when the sensors C1 and C2 are eddy current sensors.

 The principle of this embodiment consists in slaving the coil B of each of the eddy current sensors at a constant distance h 0 from the surface S 1, S 2 facing each other and detecting the displacements of the sensor resulting from this slaving.

 For this purpose, we see in Figure 10 that the coil B is secured to a part 132 slidably mounted by means of a guide 134 on a support 136 fixed to the end of the arm 36, 38 (Figure 3) bearing the sensor. The direction of the sliding thus defined is perpendicular to the facing surface when the arms 36, 38 are oriented correctly by means of the jack V. The displacement of the piece 132 in the guide 134 is obtained by means of a fixed motor 138 the support 136 and controlling the rotation of a threaded rod 140 parallel to the direction of movement of the workpiece 132 and being screwed into a threaded hole formed in this piece.

 To the mechanism which has just been described is associated an electronic part which comprises a processing circuit 128 of the signals delivered by the coil B identical to the circuit 128 of the embodiment of FIG. 5. The signal Vph delivered by this circuit 128 is injected in a summator 142 on which a reference value ho is displayed corresponding to the value that it is desired to permanently give to the distance separating the end of the coil B from the surface facing it. the curve represented in the rectangle schematizing the summator 142 in FIG. 10, the latter makes it possible to add a reference voltage to the characteristic Vp # = f (h) i so that it vanishes for the value ho ~ The signal delivered by the summator 142 is a signal representative of the difference between the actual distance h separating the end of the coil B from the surface opposite and the distance ho on which it is desired to wedge.

 In order to transform it into a correction signal, the signal delivered by the adder 142 is inverted in an inverter 144, then amplified through a preamplifier 146 and an amplifier 148. The signal output from the amplifier 148 is injected into the motor 138. which performs the desired correction in order to maintain the coil B at the distance ho from the facing surface.

 In the embodiment of FIG. 10, the means 106 and 108 making it possible to deliver the signals d1 and d2 as illustrated in FIG. 3 consist, for each of the proximity sensors C1, C2, by a position sensor 150 such that a potentiometer which is associated with the motor 138 and thus detects the movements of the coil B.

 If we compare the embodiments of Figures 5 and 10 which both implement eddy current proximity sensors, we see that the first of them is particularly simple from the mechanical point of view but requires circuits processing signals delivered by the relatively complex eddy current sensor.

On the other hand, the embodiment of FIG. 10 has the disadvantage of requiring motorization of the sensor but is characterized from the electronic point of view by very simple processing circuits.

 Of course, the invention is not limited to the embodiments which have just been described by way of example but covers all variants thereof.

This remark applies in particular to the mechanical embodiments which have been described as well with reference to FIG. 3 as with reference to FIGS. 4 and 10. It is possible to equip an industrial robot with such a head which would be its terminal organ. In addition, the mechanism shown in Figure 3 is such that the torch T is movable relative to the support arm 18 on which are directly articulated sensors C1 and C2, but it will be understood that a system where the torch is directly connected to the arm 18 and movable sensors relative to this arm would be equally suitable.

 In a comparable manner, each of the mechanisms 16, 20 and 22 of FIG. 3 can be replaced by an equivalent mechanism. For example, each of the arms 36, 38 of the mechanism 20 could be integral with a pinion, the two gears being meshing with one another and one of them being rotated by a motor. The latter can then be a step-by-step motor if the accuracy is sufficient, or a motor associated with a position sensor to angularly wedge arms 36 and 38 precisely in the opposite case. Finally, the jack V of the mechanism 20 shown in FIG. 3 can be electric, pneumatic or other, and a stop system can be provided to ensure precise positioning of the arms 36 and 38.

 In addition, we have seen that the invention is not limited to the positioning of a welding torch and can also be applied to the positioning of any other member or tool. From this point of view, it is clear that the mechanism 22 which relates more specifically to the application of the invention to multipass welding can be eliminated in other applications.

 Finally, it should be noted that the axes of the proximity sensors are not necessarily arranged in a plane perpendicular to the weld line nor perpendicular to the surfaces of the parts. The implementation of the invention simply assumes that the orientations of these axes are known so that the calculation means can take into account to calculate the values of the corrections aX and hY to perform.

For example, if the sensors are not perpendicular to the surfaces facing each other, it is possible to calculate aX and hY, provided that the angle between the axis of the torch and the bisector of the angle of the parts to be welded. This angle must be constant all along the joint, or measurable at every moment. If 6 is this angle, we can write
(5) ax (dl-d2) sin B
cos #
(6) #Y = d1 + d2 cos B
2 cos #
In another example, if the angle A varies, it is also possible to calculate AX and hY; if the two pieces deviate from the theoretical positions of the angles al and a2 respectively, and if the intersections of the axes of the sensors with the pieces are respectively at distances 11 and 12 from the line of seam, one can write (7) 1 = (d1 + l1sin a; - d2-12 sin a2) sin B
d1 + l 1sin al + d2 + 12 sin a2 cos B
(8) by = cos 6
2
Finally, the positioning method of 1'invention also applies to the case where the parts are not flat, from the moment when their geometry is known along the weld joint considered. It is then sufficient for this geometry to be stored in the form of the correspondence table between each pair of measurements d1 and d2 and each pair of displacements aX and hY to be performed.

Claims (13)

 1. Self-adaptative method for positioning a member (T) relative to two pieces (10, 12) forming between them an angle (A) measured in a plane perpendicular to their intersection, said member being able to move along the latter , characterized in that it consists - to arrange before moving said organ two cap  proximity probes (C1, C2) mounted on a support  (18) of said member vis-à-vis said parts, - to measure with the aid of said sensors (C1, C2) said  tance (dl, d2) separating said support and each  said parts, during the displacement of said organ  (T) along the intersection, - to determine the corrections (aX, hY) to be made to  the position of the support (18) of the welding torch  relative to the parts to be welded (10, 12),  angle) between the parts, distances  (dl, d2) measured and the orientation of the sensors  relative to said parts and - to move said support to perform the horns  rections (X, aY) thus determined.  2. Method according to claim 1, characterized in that one positions and in that one orients said member (T) relative to said support (18), before moving said member along the intersection (J). ).  3. Method according to any one of claims 1 and 2, characterized in that the axes of the proximity sensors (C1, C2) are arranged in a plane perpendicular to the intersection (J) and oriented relative to one another. (oy) axes of a reference mark (xOy) associated with the support (18), so that the axis of each of the sensors defines with this reference axis (y) an angle  A when the angle A between the sheets, measured on the side of the torch, is less than 2w.  4. Method according to any one of claims 1 and 2, characterized in that the axes of the proximity sensors (C1, C2) are arranged in a plane perpendicular to the intersection (J) and oriented relative to the one ( Oy) axes of a reference mark (xOy) associated with the support (18), so that the normal to the axis of each of the sensors defines with this reference axis (Oy) an angle  2 "" when the angle (A) between the sheets, measured on the side opposite the torch, is less than 2w.  5. Method according to any one of claims 3 or 4, characterized in that the corrections (#X, dY) to be made to the position of the support relative to the parts to be welded are determined using the relations  #X = (d1 + d2) .sin B  d1 + d2  #Y = () .cos B  2  6.A self-adapting device for positioning a member # (T) relative to two pieces (10, 12) forming between them an angle & measured in a plane perpendicular to their intersection (J), said member (T) being mounted on a support (18) able to move to follow said intersection, characterized in that it comprises at least two proximity sensors (C1, C2) mounted on said support, means (20) for orienting the sensors in contact with each other; said means (106, 108) associated with the proximity sensors (C1, C2) for determining the distance (d1, d2) between the support (18) and the surface of each of the parts (10, 12). , calculating means (109) of the corrections (aX, hY) to be made to the position of the support (18) with respect to the parts, as a function of the angle #) formed between the parts, distances (dl, d2) measured and the orientation of the sensors relative to said parts, and means (16) for moving the support (18) to perform the corrections (AX, hY) thus calculated.  7. Device according to claim 6, characterized in that it further comprises means (Mxt, Myt, Mat, 92 to 104) for positioning and for directing said member (T) relative to said support (18).  8. Device according to any one of claims 6 and 7, characterized in that the proximity sensors (C1, C2) each comprise a movable member (112, 114, 116-116 ') maintained in contact with the surface (S1 , S2) of the flat portion of the piece facing each other by elastic means (118), the means (106, 108) for determining the distance (dl, d2) between the support (18) and the surface of each of the parts comprising, for each sensor, means (124-124 ', 126-126') for detecting the displacement of this member.  9. Device according to claim 8, characterized in that the movable member comprises a spherical portion (116) in contact with said surface tS1, S2) -  10. Device according to claim 8, characterized in that the movable member comprises a roulette (116 ') in contact with said surface (S1, S2).  11. Device according to any one of claims 6 and 7, characterized in that the proximity sensors (C1, C2) comprise eddy current sensors comprising at least one coil (B), the means for determining the distance between the support (18) and the surface of each of the parts comprising means for processing (128) the signal delivered by each sensor, emitting a proximity signal (Vph) which varies with the distance (dl, d2) to be determined according to a function nonlinear and a processing means (130) of the proximity signal (Vph) processing the latter in a function inverse to said function.  12. Device according to any one of claims 6 and 7, characterized in that the proximity sensors (C1, C2) each comprise an eddy current sensor comprising at least one coil (B), a processing means (128). ) of the signal delivered by the sensor, delivering a proximity signal (Vph) representative of the distance (dl, d2) separating the sensor from the surface (S1, S2) of the piece facing each other, means (132) , 134, 138, 140) for moving the sensor parallel to its axis and servo means (142 to 148) acting on these means for moving the sensor in response to the proximity signal (Vph) delivered by the processing means ( 128), in order to maintain this signal at a constant value, the means (106, 108) for determining the distance (d1, d2) separating the support (18) and the surface of each of the parts (10, 12) comprising, for each sensor, means (150) for detecting the displacement of the sensor parallel to its axis.  13. Device according to any one of claims 6 to 12, characterized in that each sensor (C1, C2) is mounted at the end of a branch (C1, C2) which is pivotally mounted by its other end on a axis (40) parallel to the intersection (J) of the parts (10, 12) and in that the means (20) for orienting the sensors act on said branches so as to rotate them about said axis (40) of an equal and opposite angle.