Gear wheel fault detection by synchronous analysis of gait and / or amplitude

Abstract

Acoustic diagnostic method for detecting a non-conforming wheel of a clock movement, comprising the following steps: - detecting a plurality of sounds, each sound of the plurality of sounds corresponding to the passage of each tooth of the escape wheel over each of the pallets, forming alternations, - measuring, at each alternation, a characteristic quantity of the oscillator in operation, - choosing an alternation sequence associated with a rotation, at least partial, of the escape wheel or of at least one of the wheels of the plurality of wheels, on the basis of the predetermined number of teeth, - stacking, on the chosen alternation sequence, each measurement of the characteristic quantity for each alternation, so as to obtain a synchronous signal, - detecting a variability of the characteristic quantity, - deducing a non-conformity on one wheel of the plurality of wheels.

Description

[0001] The present invention relates generally to the field of watchmaking, more particularly to the field of quality control of watch components, especially those assembled in a watch mechanism.

[0002] Quality control methods and devices are known in the prior art.

[0003] There are methods for analyzing disturbances due to gear defects, these methods relying on visual analysis or FFT analysis ("Fast Fourier Transform" in English, or rapid Fourier transformation in French), but provide an imprecise view of the defects in the transmission.

[0004] Furthermore, the Witschi training document "Measurement Technique and Analysis of Watch Defects" is known in the prior art. However, it has at least the following drawbacks. In the prior art, it requires an operator to observe the signal live, without even having information on the location of the problem's origin. This can be useful and reasonably straightforward in the case of an escape wheel, but it becomes very complicated in the case of slow-rotating wheels. Moreover, this prior art method cannot be automated and is therefore expensive and time-consuming.

[0005] It is therefore desirable to improve their efficiency and capabilities, in order to better identify which component has a defect, and for example to notify the user about the component with the defect and / or the correction to be made.

[0006] The present invention aims to propose a new method of quality control in the field of watchmaking, in order to overcome the disadvantages of the prior art.

[0007] The objective of the analysis of the present invention is, among other things, to determine the impact of one of the elements of the gear train (preferably each element) in the perturbation of the torque transmitted to the escapement, which is to be as stable as possible in order to optimize the operation and chronometry of the watch movement and the watch.

[0008] The present invention enables a thorough analysis of finishing gear defects by measuring perturbations in amplitude in order to detect various quality defects, such as: Malroundness or concentricity defects, tooth cutting defects, ovality defects, location and center distance defects, existence of hard point(s) in the transmission possibly leading to a blockage (which can also be seen as a sum of defects).

[0009] All these defects, alone or cumulatively across the entire gear train, generate disturbances in the transmitted torque and prevent optimal operation, thus causing rework to analyze and reduce the size of the parts.

[0010] Thus, in a first aspect, the invention relates to an acoustic diagnostic method for detecting a non-conforming wheel in a watch movement, comprising the following steps: to provide the clockwork mechanism comprising an oscillator, an escape wheel, an anchor (preferably with pallets) and a plurality of wheels, a predetermined number of teeth per wheel; to detect a plurality of sounds, each sound of the plurality of sounds corresponding to the passage of each tooth of the escape wheel in cooperation with the anchor (or on each of the pallets), forming alternations (and / or of each wheel tooth of the plurality of wheels); to measure, at each alternation, a characteristic quantity of the oscillator in operation, such as an amplitude, a rate, a reference point, an actual lift angle; to choose an alternation sequence associated with a rotation, at least partial, of the escape wheel or of at least one of the wheels of the plurality of wheels, based on the predetermined number of teeth; to superimpose (or stack), on the chosen alternation sequence, each measurement of the characteristic quantity for each alternation.in order to obtain a synchronous signal, detect, on the basis of the synchronous signal, a variability of the characteristic quantity with respect to a predetermined tolerance, and deduce, if a defect is detected on the chosen alternation sequence, a non-conformity on one wheel out of the plurality of wheels.

[0011] This allows for improved quality control, as mentioned above, and helps determine which wheel in the clockwork mechanism (or watch movement) is non-compliant or defective, so that a corrective action can be proposed. This involves observing the temporal (or frequency) evolution of a signal calculated from a measurement, preferably sampled at the oscillator's frequency (e.g., 4 Hz), such as the rate and / or amplitude, or other parameters. Furthermore, having a signal or measurement per alternation (e.g., instead of a signal every two seconds) eliminates the need for a pre-calibrated component. Measuring each alternation (instead of every two seconds, for example) provides better resolution, allowing for tracing further up the gear train, towards the higher-speed components.It has been observed that state-of-the-art devices (for example, the Witschi mentioned above) typically only take one measurement every two seconds. Measuring each cycle with a good signal-to-noise ratio and a high acquisition rate allows for the identification of even the fastest-moving mechanisms.

[0012] Furthermore, the present invention is preferably based on a tooth count, as this provides the necessary resolution (each event—tooth passage, revolution, or even half-revolution—is a finite multiple of alternations with a specific repetition profile). An exact signature of the event is obtained, independent of the part's adjustment (if the part is not adjusted at all, it is pointless to search for a repetition pattern every 5 seconds, for example, corresponding to the theoretical rotation time of the escape wheel). In other words, one of the advantages of the present invention is that it allows patterns to be grouped not according to a theory, but according to their autocorrelation, thus eliminating the need to know the precise durations of the various events.

[0013] Furthermore, the proposed method allows for the decomposition of the complete signal, highlighting the contribution of each gear to the signal. It should be noted that, in addition to detecting runout, the present invention makes it possible to detect stiff spots in the gear that could lead to the movement stopping. Moreover, the gear analysis proposed in the present invention is based, in one embodiment, on the combination of FFT (or another mathematical processing method) and the amplitude signal resynchronized over the period considered (i.e., an event) and averaged over all measured events, as well as on the detection of the last event before a stoppage.

[0014] The KPIs (for "Key Performance Indicator") of the analysis are established for each characteristic period of the gear train (rotation of the moving parts and passage of the teeth) and are, for example: value of the variation in amplitude (amplitude envelope) for the period considered (for example 4σ or any other value of σ, or any other value), the main periods having the most impact on the amplitude obtained by FFT, the sum of the impacts having to correspond approximately to the value of the envelope.

[0015] In watchmaking, a watch's rate characterizes its operation, evaluated according to its regularity. Instantaneous rate refers to the rate at the moment the watch is observed. Instantaneous rate is measured with a comparator chronometer, a device capable of precisely measuring the frequency of the watch's oscillator. Monthly rate refers to the difference between two states separated by a one-month interval (for example, a rate variation of ±20 seconds per month). Probable rate refers to the rate that the timekeeping instrument would likely have if it were placed or used under defined conditions similar to its intended use. It can refer, for example, to a duration: of 24 hours (probable daytime walking), of one month (probable monthly walking), of one year (probable annual walking).

[0016] In other words, rate is understood to be an isochronic deviation. Rate variability refers to a rate deviation in watchmaking.

[0017] The term "tooth passage" refers to a contact of that tooth with its counterpart, which may be a corresponding tooth on another wheel, a pallet of the escapement, or any other mechanism commonly used in watchmaking intended, or capable of cooperating with teeth, for example a finger, a Maltese cross, without being limited to these examples.

[0018] Superposition or stacking refers to observing and aligning all occurrences of an event in synchronization with its period (defined as the number of cycles). For example, superimposing occurrences can be used to average variations due to noise and obtain a clearer image of the signal. Note that synchronous stacking (or superposition) is possible when the superposition duration, and therefore the period of the analyzed event, is known.

[0019] In other words, the invention relates to an acoustic diagnostic method for detecting a non-conforming wheel in a watch movement, comprising the following steps: to provide the clockwork mechanism comprising an oscillator, an escape wheel and a plurality of wheels, a predetermined number of teeth per wheel; to measure a characteristic quantity of the oscillator in operation, such as an amplitude and / or a rate, a reference point, an actual lift angle; to detect a plurality of sounds, each sound of the plurality of sounds corresponding to the passage of each tooth of the escape wheel (and / or each tooth of the wheel in the plurality of wheels); to associate the sounds of the plurality of sounds with a rotation, at least partial, of the escape wheel or of at least one of the wheels in the plurality of wheels, based on the predetermined number of teeth; and preferably to stack the sounds to obtain a synchronous and averaged signal relative to each event; to detect, based on the associated sounds and / or on the measurement of the characteristic quantity, and over a predetermined period,If there is variability (or variation) in the characteristic quantity and / or rotation time of the escape wheel or at least one of the wheels in the plurality of wheels, deduce which wheel among the escape wheel and / or at least one of the wheels in the plurality of wheels is non-compliant, or deduce that the clockwork mechanism is compliant (i.e., no non-compliant wheels).

[0020] The invention can also be defined according to the following characteristics, taken individually or in combination.

[0021] Preferably, the characteristic quantity of the oscillator in operation is an amplitude and / or a step of the oscillator in operation.

[0022] Advantageously, the process includes the following step: to associate, or correlate, sounds from the plurality of sounds to one wheel from the plurality of wheels or the escape wheel, according to the predetermined number of teeth of said wheel.

[0023] This allows us to identify which wheel is causing the problem by associating or correlating sounds. For example, this could be done by detecting a repetitive pattern in the sounds and associating that (sound) pattern with a specific wheel. By detecting a deviation in the pattern from the expected pattern, it is thus possible to determine the non-compliant wheel.

[0024] Advantageously, the process includes the following step: detect a reproducible character of sounds from the plurality of sounds, preferably as a function of a period (or a frequency, the frequency being the inverse of the period) of rotation of the wheels, and / or a frequency of identified sounds, and / or a frequency of defects.

[0025] This allows us to determine the sound patterns (or reproducible characteristics of sounds) and to process them in order to conclude that there is a defect in the event of bias or deviation from the expected pattern.

[0026] Advantageously, the process includes the following step: count the number of teeth per wheel based on the plurality of sounds detected.

[0027] This allows us to associate a number of sounds with a specific tooth. For example, a particular tooth should rotate at a certain speed: consequently, this tooth can be associated with a sound of a given intensity striking at the frequency corresponding to the rotational speed. Indeed, with each tooth engagement, the wheel will rotate through an angle corresponding to its number of teeth and will produce a sound. This sound will be repeated according to the number of teeth. This applies particularly well to the escape wheel. Even more so, this applies to the sounds emitted by the contact between the escape wheel and the pallet fork levers.

[0028] Advantageously, the process includes the following step: indexing teeth of the escape wheel or plurality of wheels according to the measured characteristic quantity.

[0029] This allows each tooth of the wheel to be numbered in order to know which tooth is engaged or meshed in the gear.

[0030] Advantageously, the process includes the following step: identify deviations, patterns (in particular continuous or discontinuous error patterns) or anomalies from predetermined thresholds, in the characteristic quantity, (or in the amplitude of the oscillator, of the rate), and / or of a rotation time of the escape wheel or of at least one of the wheels of the plurality of wheels, so as to detect which wheel among the escape wheel and / or at least one of the wheels of the plurality of wheels, is non-compliant.

[0031] For example, a round tooth produces a continuous error signal or pattern, while a damaged tooth produces a discrete error signal or pattern. In both examples, industrial applications are relevant and useful.

[0032] Advantageously, the process includes the following step: represent, on a graph, the characteristic quantity, the amplitude of the oscillator, of the rate, and / or of a rotation time of the escape wheel or of at least one of the wheels of the plurality of wheels, so as to detect which wheel among the escape wheel and / or at least one of the wheels of the plurality of wheels, is non-compliant.

[0033] Advantageously, the process includes the following step: propose a correction after deducing which wheel among the exhaust wheel and / or at least one of the wheels from the plurality of wheels, is non-compliant.

[0034] Advantageously, the correction can be chosen from among the replacement of the escape wheel, all or part of the plurality of wheels, reworking one or more wheel teeth, and / or correcting a wheel shaft, and / or reworking the runout of one or more wheels.

[0035] Advantageously, the process includes the following step: deduce that the non-conforming wheel has a defect among a defect of roundness or concentricity, a defect in cutting the teeth, a defect in ovality, a defect in location, a defect in center distance, a defect in normal wear or a defect in premature wear.

[0036] This makes it possible to identify a plurality of defects in the watchmaking mechanism and to have precise knowledge of which element is causing the problem.

[0037] Advantageously, the process includes the following step: perform at least one Fourier transform on the synchronous signal.

[0038] Advantageously, the process includes the following step: stacking patterns (of sounds) by synchronous analysis to deduce a representative pattern, such as average pattern or median pattern.

[0039] Advantageously, the process includes the following step: to perform a statistical analysis on the data (i.e. measurement of characteristic quantity, sounds), in order to define a mean, a standard deviation, and / or extreme points of the statistical distribution.

[0040] This allows for the application of mathematical, computer, and / or automatic processing to the detected sounds in order to better identify the defect. In particular, the Fourier transform can be performed on the synchronous signal of a predetermined moving part, rather than on the entire sound data. This allows for tooth-level accuracy, limited to the predetermined moving part (i.e., a specific wheel). A Fourier transform on the entire time-domain signal is significantly impacted, even distorted, by the decrease in amplitude over time due to the discharge of the drum, which highlights the advantage of using a synchronous signal. The present invention also makes it possible to detect singular events (non-continuous in the rotation of a moving part) due to the synchronous analysis approach.The analysis is also more robust than that of the prior art because it is able to rule out isolated aberrant elements and confirm the presence of the defect over the entire signal between the beginning and end of the power reserve. Advantageously, the method includes the following step: identify a natural frequency corresponding to a defect, such as a natural frequency corresponding to a revolution time for a roundness defect, a natural frequency corresponding to half a revolution time for an ovality defect, or a natural frequency corresponding to a hard spot comprising a series of harmonics.

[0041] Advantageously, the process includes the following step: establish an average value of the characteristic quantity (for example, the amplitude of the oscillator, the rate), and / or a rotation time of the escape wheel or of at least one of the wheels of the plurality of wheels, and detect, by comparing to the corresponding average value, the variability of the characteristic quantity, and / or the rotation time of the escape wheel or of at least one of the wheels of the plurality of wheels.

[0042] In other words, the invention relates to an analysis method for establishing the characteristic signal (or amplitude) associated with the rotation of each component, in order to identify its various defects and propose an appropriate correction. This involves identifying the time of each event and counting a repetition pattern (for example, a multiple of an integer number of escapement teeth related to an event). The average event is then constructed, and the shape of the average curve is analyzed to deduce the presence of the various possible defects. It is thus possible to confirm a potential binding point by identifying the stopping point (moving part and tooth of the binding and / or stopping point).

[0043] It is possible to produce mobile-by-mobile analysis.

[0044] Firstly, it is necessary to carry out a synchronous analysis of the rate and / or amplitude over part or all of the duration of a power reserve for each constituent element of the gear train and identify a defect and propose a correction. IlIt is possible to limit the period to a sufficiently long one to achieve enough repetitions without necessarily utilizing the entire power reserve, particularly for a caliber with a long power reserve (for example, five days). Thus, it is not necessary to require the full power reserve, but one to several rotations (preferably a few, such as two, three, four, five, or more than ten) are needed for the components in question. Therefore, for a component in the gear train, the sufficiently long period corresponds preferably to three complete rotations of the component and, even more preferably, to the equivalent of one complete rotation of the mainspring barrel.On the other hand, it's interesting to note that the analysis depends on the power reserve indicator: indeed, critical points generally appear at lower winding levels. Therefore, to detect them, it's useful to perform an analysis over a full rotation of the mainspring barrel at a low winding level and compare it to the analysis at a high winding level. Furthermore, an example of synchronous analysis is to stack (that is, to divide by time modulo) each pattern of an event to determine a pattern, such as an average pattern, a median pattern, a minimum pattern, a maximum pattern, and so on.

[0045] Secondly, it is necessary to use a tool that provides sensitivity to oscillation.

[0046] Thus, the invention relates to a method for detecting a defect in a gear of a clockwork movement with a known oscillator oscillation frequency and comprising an escape wheel, the number of teeth of the movement's gears being known, the method comprising the following steps: to start the clockwork mechanism, to measure the amplitude and / or rate of the oscillator, to identify the time of each rotation of a gear and to count a whole number of escapement teeth relative to a rotation of a gear (for example, one revolution of a seconds wheel takes 60 seconds and corresponds to 240 escapement teeth, or an average / second gearing takes 4 seconds and corresponds to 16 escapement teeth), index each escapement tooth passage relative to the duration of the event (for example tooth no. 1 / 16 in the event), calculate the variation of the measured quantities by synchronous analysis, deduce a defect on a gear.

[0047] The invention relates, in a second aspect, to a device arranged to implement the process of the first aspect.

[0048] Advantageously, the device includes an acoustic sensor.

[0049] Advantageously, the device includes an optical sensor.

[0050] Advantageously, the device includes a computing unit or processor.

[0051] Advantageously, the device includes an association unit arranged to associate sounds with movement components, for example, one wheel or another wheel.

[0052] The invention relates, in a third aspect, to a method for controlling a watch movement comprising the steps of: to place a watch movement in an acoustic analysis device, such as the device according to the second aspect, capable of measuring a characteristic quantity of the oscillator in operation, such as an amplitude, a rate, a reference point, and / or an actual lift angle, to identify a conformity or non-conformity, preferably automatically, of a wheel of the watch movement by means of, or through, the diagnostic process according to the first aspect.

[0053] The invention relates, in a fourth aspect, to a method for adjusting or maintaining a watch movement comprising the following steps: carry out the diagnostic process according to the first aspect, issue a setting instruction to a user.

[0054] Advantageously, the setting instruction is displayed on a connected device, such as a connected magnifying glass.

[0055] Advantageously, the adjustment instruction to a user corresponds to a wheel change, a change in positioning of at least one ruby, a bridge change, a plate change and / or adjusting the play of one or more wheels.

[0056] Advantageously, calculate the efficiency loss and / or the efficiency associated with the gear train or a subset of the gear train.

[0057] Il It is indeed possible to calculate an overall yield loss (for example, x%). We observe the impact of the defect, for example 10°, and we reduce it to the average amplitude, for example 250°. We can then deduce that the defect impacts the amplitude by 4% (which can be converted into yield points, modulo a few parameters of the gauge).

[0058] This helps the operator to identify the most important correction to make.

[0059] Other features and advantages of the present invention will become more apparent upon reading the following detailed description, of embodiments of the invention given by way of non-limiting example and illustrated by the accompanying drawings, in which: there figure 1 represents a measure of power reserve and amplitude for a watch movement, with amplitude as a function of running time, in the method according to the present invention, the figure 2 represents the amplitude as a function of time over a limited period of time of the measurement of the figure 1 , there figure 3 represents the amplitude and its Fourier transforms in frequency and period, the figure 4 represents the characteristic frequencies and characteristic periods after Fourier transforms, the figure 5 represents variations in amplitude over periods corresponding to wheels of the clockwork mechanism, the figure 6represents the average amplitude over one wheel rotation of the first wheel among the wheels of the clockwork mechanism, the figure 7 represents the average amplitude over one wheel rotation of another example of the first wheel among the wheels of the clockwork mechanism but on a different movement, the figure 8 represents the average amplitude over one wheel revolution of yet another example of the first wheel among the wheels of the clockwork mechanism on yet another movement, the figure 9 represents the amplitude over a given measurement in tickets, a ticket being a measurement taken in the reference horological positions with integration and stabilization times, the Figure 10 represents variations in amplitude over periods corresponding to wheels of the clockwork mechanism, the figure 11 represents the amplitude on another given ticket measurement, the figure 12represents variations in amplitude over periods corresponding to wheels of the clockwork mechanism, the figure 13 represents an acquisition of the temporal amplitude signal with temporal superposition, the figure 14 represents a signature or signal synchronous with the oscillation, the figure 15a represents a sinusoidal regression on the synchronous signal, the figure 15b represents a tooth engaged at rest on the synchronous signal, the figure 15c represents a singular pinion or a singular tooth on the synchronous signal, the figure 16 represents a synchronous signal to identify a rounding defect.

[0060] There figure 1 represents a power reserve and amplitude measurement for a watch movement, with the amplitude as a function of running time, in the method according to the present invention, used as a starting point. The resolution required for the invention is preferably an oscillation resolution.

[0061] The acoustic diagnostic method for detecting a non-conforming wheel in a clockwork movement of the present invention comprises the following steps: to provide the clockwork mechanism comprising an oscillator, an escape wheel, an anchor (with pallets), and a plurality of wheels, the number of teeth per wheel being predetermined; to detect a plurality of sounds, each sound of the plurality of sounds corresponding to the passage of each tooth of the escape wheel over each of the pallets, forming alternations, or of each tooth of the wheel in the plurality of wheels; to measure, at each alternation, a characteristic quantity of the oscillator in operation, such as an amplitude, a rate, a reference point, an actual lift angle; to choose an alternation sequence associated with a rotation, at least partial, of the escape wheel or of at least one of the wheels in the plurality of wheels, based on the predetermined number of teeth; to superimpose or stack, on the chosen alternation sequence, each measurement of the characteristic quantity for each alternation, so as to obtain a synchronous signal; to detect,Based on the synchronous signal, a variability of the characteristic quantity relative to a predetermined tolerance, deduce if a fault is detected on the chosen alternation sequence, a non-conformity on one wheel out of the plurality of wheels.

[0062] In other words, within the framework of the present invention, it is possible to capture information relating to each tooth of the escapement board (step, amplitude, sounds, duration of rotation, etc.).

[0063] To this end, it is necessary to identify the time of each event and count a whole number of escape teeth relative to that event (for example, one rotation of a second gear takes one revolution in 60 seconds, corresponding to 240 escape teeth, or an average / second gear engagement takes (approximately) 4 seconds, or 16 escape teeth). Note that the given durations are non-exhaustive examples.

[0064] It is then necessary to index each escape tooth passage with respect to the duration of the event (tooth no. 1 / 16 in the event),

[0065] Next, the pattern of repetition of the event must be identified (for example: an event of 60 seconds every 60 seconds for one rotation of a second mobile or an event of (approximately) 4 seconds every 60 seconds for one tooth of this mobile).

[0066] It is necessary to identify the sequence of repetition of the event (revolution no. 52 / 156 for example)

[0067] It is necessary to calculate the mean, median, or statistically representative event.

[0068] It is necessary to plot and / or calculate the variation of the measured quantities over the average event. This processing can be visual or performed automatically by a processor. For example, it is possible to identify the tooth number that generates a minimum amplitude, which may correspond to the presence of a hard spot, and then compare this to the last tooth number observed in the event of a stoppage (as indicated below).

[0069] Next, a mathematical treatment such as an FFT circumscribed to the event should be performed, for: identify a natural frequency corresponding to the revolution time (malrond defect), and / or identify a natural frequency corresponding to half the revolution time (ovalisation defect), and / or identify a natural frequency corresponding to a hard point including a series of harmonics, identify the emergence of natural frequency relative to others at the end of the power reserve, also symptomatic of hard points. This is the mathematical analysis stage on the average event. Indeed, the FFT is optional; it provides confirmation for the analysis.

[0070] In the event of a stoppage, it is possible to identify the last event that occurred.

[0071] Preferably, the sequence is chosen and / or defined beforehand based on the wheel on which a defect is being sought. For example, 40 oscillations on a 20-tooth escape wheel or 20,000 oscillations on a large average rotation. The evolution for an escape wheel can be performed over 20 oscillations (40 oscillations), for a seconds wheel 240 oscillations (480 oscillations), and for a small average wheel 2,400 oscillations (4,800 oscillations). Note that these examples are specific to a given caliber with a given gear ratio. They are not exhaustive.

[0072] Thus, with reference to the figure 1 It is possible to represent the power reserve measurement for the watch movement, with the amplitude as a function of running time.

[0073] The power reserve here, for example, lasts approximately 42 hours. We observe that the amplitude decreases overall as the power reserve time passes. Furthermore, a zone with a disturbed amplitude can be identified towards the end of the power reserve.

[0074] However, by zooming in on a period of 300 seconds, it is possible to obtain additional information, as explained below.

[0075] There figure 2 represents the amplitude as a function of time, within the context of a zoom of approximately 300 seconds.

[0076] The zoom over the 300-second period corresponds to a small average rotation, also including several repeating patterns: the passage of the barrel tooth / GM (for "Large Average" or center wheel) every 150 seconds, the rotation of the intermediate wheel every 40 seconds, the passage of the GM tooth / PM (for "Small Average" or middle wheel) every 30 seconds, the rotation of the escapement every 5 seconds, the passage of the PM tooth / INTER (for intermediate wheel) every 4 seconds.

[0077] There figure 3 represents the amplitude and its Fourier transforms in frequency and period.

[0078] The FFT is preferably performed only on the first 120,000 samples corresponding to amplitude values ​​greater than 200° so as not to include outliers, if any.

[0079] The graph at the top of the figure 3 represents the amplitude as a function of the passing time of the power reserve, for the 120,000 samples.

[0080] The graph in the middle of the figure 3represents the characteristic frequencies after FFT. We can identify 8 frequencies of interest.

[0081] The graph at the bottom of the figure 3 represents the characteristic periods. We can identify periods of interest 9.

[0082] There figure 4 represents the characteristic frequencies and characteristic periods after Fourier transforms.

[0083] Results at very low frequencies are preferably filtered to observe the frequencies of interest associated with the gear (between 0 seconds and 300 seconds).

[0084] The numerical values ​​of the ten main peaks having a period of less than 350 seconds make it possible to obtain the main disturbances of the gear on the amplitude signal. Table 1 index impact (°) frequency period(s) 1 4.706859589 0.033334375 29.9990625 2 3.619336605 0.003333438 299.990625 3 2.705308676 0.006666875 149.9953125 4 1.103452325 0.01333375 74.99765625 5 0.957385421 0.20000625 4.99984375 6 0.884805799 0.020000625 49.9984375 7 0.830545425 0.06666875 14.99953125 8 0.747223258 0.023334063 42.85580357 9 0.725010157 0.036667813 27.271875 10 0.716868281 0.016667188 59.998125

[0085] Disadvantages of the FFT: Very low frequencies are predominant in an FFT performed on a full power reserve and are of no particular interest, except perhaps for analyzing the barrel unwinding. The filtering step of the values ​​obtained at the end of the power reserve (PR) is important for the consistency of the results. The impact value corresponds to the amplitude of the cosine function; it must be multiplied by two to make it comparable to the impact visible on the time amplitude graph. This necessary interpretation hinders the reading of the FFT. Harmonics can be integrated into the interpretation (for example, the perturbation associated with the rotation of the average rotation is the sum of the perturbation every 300 seconds, but also partially every 150 seconds).

[0086] Performing the analysis over the duration of the average event avoids interference from low frequencies, the lowest frequency being that of the event itself. The average event allows for a more robust identification of the individual component (i.e., which tooth of which gear) that generates the minimum amplitude. This individual component generating the minimum amplitude would not be visible in FFT analysis, which would result in a series of harmonics that would "fly under the radar" of fault detection. It is precisely the combination of a frequency analysis limited to the duration of the studied event and a temporal analysis by individual component that allows for the identification of faults.

[0087] THE figures 1, 2 , 3, 4 (and their corresponding descriptions) are derived from the prior art, but have been used and modified during the implementation of the process of the present invention, as a starting point.

[0088] There figure 5represents variations in amplitude over periods corresponding to wheels of the clockwork mechanism.

[0089] With the present invention, a more synthetic view of the disturbances in the gear is obtained, as illustrated in figure 5 .

[0090] The rotation of one small average revolution over 300 seconds generates an amplitude variation of 21.5° (visible at the PM 300.0 s rotation on the figure 5 ).

[0091] This variation is mainly made up of a (peak-to-peak) variation of 12.1° every 30 seconds. This 30-second period corresponds to a shift from the GM board tooth to the small to medium sprocket (for example, a 10-tooth sprocket).

[0092] We also detect the presence of a dent on the mobile and especially the presence of a hard spot on the rotation of the mobile.

[0093] The transmission between the large and small middle gears is the most disruptive element in the mechanism (for example, due to cutting defects, roundness, or ovalization of the plate). A correction to this movement could be made by replacing this moving part.

[0094] There figure 6 represents the average amplitude over one wheel rotation of a first wheel among the wheels of the clockwork mechanism.

[0095] Observations may be different for other RM measurements.

[0096] On the first wheel, a rotation of one revolution presents two hard points 11, which are spaced 31 teeth apart. However, the moving part under study transmits its energy to two moving parts whose angle of inclination also corresponds to 31 teeth. The transmission is poor, due to a defect in location and center distance. The transmission efficiency is reduced.

[0097] There figure 7represents the average amplitude over one wheel rotation of another example of the first wheel among the wheels of the clockwork mechanism but on a different movement.

[0098] On the first wheel of this other example, we can also observe the same behavior with two hard points 12 separated by the same number of teeth.

[0099] There figure 8 represents the average amplitude over one wheel revolution of yet another example of the first wheel among the wheels of the clockwork mechanism, on yet another movement.

[0100] Finally, on the first wheel taken in yet another example, only a slight roundness is visible, the transmission disturbance is minor and acceptable.

[0101] There figure 9 represents the amplitude on a given ticket measurement. Example ticket 1:

[0102] On a ticket measured over 60 seconds across six positions, i.e. 360 seconds (six zones 6a, 6b, 6c, 6d, 6e, 6f to the figure 9 ): this movement is distinguished by a noise on the acoustic amplitude signal that is greater than that of the rest of the population.

[0103] There Figure 10 represents variations in amplitude over periods corresponding to wheels of the clockwork mechanism. Analysis of ticket 1 using this invention:

[0104] We obtain a view limited to the seconds and escapement mechanism of the disturbances in the gear train. The analysis, covering only a few revolutions (here, a 60-second ticket only), is by definition imprecise, particularly the lowest frequency term associated with the rotation of the seconds mechanism, which recovers some of the effects of the upstream gear train.

[0105] However, it is possible to deduce the following, even with few turns.

[0106] The seconds hand is a disruptive element in this mechanism due to the following defects: Malrond since each turn generates a variation of approximately 4.5°, ovalization of the plate or pinion since each half-turn generates a variation of 4.2°, of cutting since each tooth of the second plate with the escape pinion generates a variation of approximately 2.9°.

[0107] A correction to this movement could be made by replacing this moving part.

[0108] The movement also features a malrond of the escape wheel generating a variation of 1.5° every 5 seconds.

[0109] There figure 11 represents the amplitude on another given ticket measurement. Example ticket 2 :

[0110] On a ticket measured over 60 seconds: this one shows a low noise.

[0111] There figure 12 represents variations in amplitude over periods corresponding to wheels of the clockwork mechanism. Analysis of ticket 2 using this invention:

[0112] In this movement, the seconds hand generates no defects; the main disturbances in the gear train are due to: The transmission between the middle gear and the seconds pinion (every 6.6 seconds) generates a variation of approximately 1.5°, and the transmission between the seconds gear and the escapement pinion (every 0.62 seconds) generates a variation of approximately 1.3°. The effects observed every 1.25 seconds and 2.5 seconds are also consequences of this meshing.

[0113] This movement sets a benchmark in terms of transmission quality.

[0114] Regarding the predetermined tolerance (also called the predetermined tolerance threshold), the following can be stipulated, by way of non-limiting example: declaring a non-conforming tooth as being at a distance from the mean in terms of the number of standard deviations, for a pinion tooth to be non-conforming, its distance must be at 1.64 sigma for example, for a board tooth to be non-conforming, its distance must be at 2.5 sigma for example.

[0115] This is justified by considering that a sprocket has approximately 10 teeth (1 tooth = 10%), and in a normal distribution, 90% of the points lie within an interval of approximately ±1.64 standard deviations around the mean. A board with approximately 80 teeth (1 tooth = 1.25%) would have 98.75% of its points lying within an interval of approximately ±2.5 standard deviations around the mean.

[0116] There figure 13 represents the acquisition of the temporal amplitude signal with temporal superposition.

[0117] There figure 14 represents a signature or a signal synchronous with the oscillation.

[0118] Taking the example of the amplitude signal associated with a moving object with an average revolution time of 300s, i.e., 1200 oscillations (regardless of the movement's operating speed setting), the different identification steps are: synchronous signature to oscillation, identification 101 of the area engaged at the end of the power reserve (the part is non-compliant, the nominal power reserve has not been reached, tooth on stop), synchronous signal resampled with one point per board tooth (which allows to smooth the noise), statistical descriptors [min, max and mean], 102, 103, 104, sinusoidal regression 105, identification 106 of a singular tooth (according to the criterion of the number of standard deviations).

[0119] In this example, the amplitude value is more than 1.64 sigma from the average, but the singular pinion tooth corresponds precisely to the tooth engaged when stopping at the end of the power reserve: the moving part has a hard point and this is the cause of the stop.

[0120] This is further explained in the figures 15a to 15c detailed below.

[0121] There figure 15a represents the sinusoidal regression 105 on the synchronous signal.

[0122] There figure 15b represents the tooth engaged at standstill at the end of the RM power reserve (identification of the tooth engaged at standstill 101 on the synchronous signal).

[0123] There figure 15c represents a singular pinion (or a singular tooth) on the synchronous signal (identification of the singular pinion or the singular tooth 106).

[0124] Furthermore, to be complete, the temporal superposition of all successive sequences of 1200 oscillations allows the synchronous signature to be constructed.

[0125] Furthermore, regarding the tolerance on all periodic defects (out-of-roundness, cutting, etc.), we can note the following.

[0126] The tolerance becomes a percentage of permissible loss relative to the nominal value. For example, up to 1% amplitude loss associated with the defect is allowed.

[0127] Example 1: Malrond of escape wheel (one revolution = 20 oscillations or 5 s).

[0128] The roundness causes a variation of 4.8° and therefore a loss of 2.4° on average over a sufficiently long observation time (for example at least 30s), i.e. a loss of 1% of amplitude considering a nominal amplitude of 240°: this part is declared non-compliant with regard to the tolerance.

[0129] If the signal is perfect, we are horizontal at the maximum value.

[0130] There figure 16 represents a synchronous signal to identify a rounding defect.

[0131] To the figure 16 This is an example of an escape wheel misalignment with one revolution corresponding to 20 oscillations or 5 seconds.

[0132] In addition, reference is made to example 2: cutting profile at the level of the great mean (one tooth passage -968 oscillations or 242 s).

[0133] The cutting profile produced results in a 5° variation with each tooth pass, corresponding to a 1% loss. The cutting is not within tolerance.

[0134] It will be understood that various modifications and / or improvements obvious to a person skilled in the art can be made to the different embodiments of the invention described in this description.

[0135] Finally, it should be noted that it is possible to combine the methods of implementation as much as possible or necessary.

Claims

1. An acoustic diagnostic method for detecting a non-conforming wheel in a clock movement, comprising the following steps: - providing the clock movement comprising an oscillator, an escape wheel, an anchor, and a plurality of wheels, the number of teeth per wheel being predetermined, - detecting a plurality of sounds, each sound of the plurality of sounds corresponding to the passage of each tooth of the escape wheel in cooperation with the anchor, forming alternations, - measuring, at each alternation, a characteristic quantity of the oscillator in operation, such as an amplitude, a step, a reference point, an actual lift angle, - choosing an alternation sequence associated with a rotation, at least partial, of the escape wheel or of at least one of the wheels of the plurality of wheels, based on the predetermined number of teeth, - superimposing, on the chosen alternation sequence, each measurement of the characteristic quantity for each alternation,in order to obtain a synchronous signal, - detect, on the basis of the synchronous signal, a variability of the characteristic quantity with respect to a predetermined tolerance, - deduce, if a defect is detected on the chosen alternation sequence, a non-conformity on one wheel out of the plurality of wheels.

2. Acoustic diagnostic method according to the preceding claim, comprising the following step: - indexing teeth of the exhaust wheel or plurality of wheels according to the measured characteristic quantity.

3. Acoustic diagnostic method according to any one of the preceding claims, comprising the following step: - identifying deviations, patterns or anomalies from predetermined thresholds, in the characteristic quantity, so as to detect which wheel among the exhaust wheel and / or at least one of the wheels in the plurality of wheels, is non-compliant.

4. Acoustic diagnostic method according to any one of the preceding claims, comprising the following step: - proposing a correction after deducing which wheel among the exhaust wheel and / or at least one of the wheels in the plurality of wheels, is non-compliant.

5. Acoustic diagnostic method according to any one of the preceding claims, comprising the following step: - deduce that the non-conforming wheel has a defect among a defect of roundness or concentricity, a defect in cutting the teeth, a defect in ovality, a defect in location, a defect in center distance, a defect in normal wear or a defect in premature wear.

6. Acoustic diagnostic method according to any one of the preceding claims, comprising the following step: - perform at least one Fourier transform on the synchronous signal.

7. Acoustic diagnostic method according to any one of the preceding claims, comprising the following step: - identify a natural frequency corresponding to a defect, such as a natural frequency corresponding to a revolution time for a roundness defect, a natural frequency corresponding to half a revolution time for an ovality defect, or a natural frequency corresponding to a hard spot comprising a series of harmonics.

8. Acoustic diagnostic method according to any one of the preceding claims, comprising the following step: - establishing an average value of the characteristic quantity, and / or a rotation time of the exhaust wheel or of at least one of the wheels of the plurality of wheels, and - detecting, by comparing to the corresponding average value, the variability of the characteristic quantity, and / or the rotation time of the exhaust wheel or of at least one of the wheels of the plurality of wheels.

9. Method for checking a watch movement comprising the steps of: - placing a watch movement in an acoustic analysis device capable of measuring a characteristic quantity of the oscillator in operation, such as an amplitude, a rate, a reference point, and / or an actual lift angle, - identifying a conformity or non-conformity, preferably automatically, of a wheel of the watch movement by the diagnostic method according to one of claims 1 to 8.

10. Method for adjusting or maintaining a watch movement comprising the following steps: - carrying out the diagnostic procedure according to one of claims 1 to 8, - issuing an adjustment instruction to a user.

11. Adjustment method according to the preceding claim, wherein the adjustment instruction is displayed on a connected device, such as a connected magnifying glass.

12. Adjustment method according to any one of claims 10 to 11, wherein the adjustment instruction to a user corresponds to a wheel change, a change in positioning of at least one ruby, a bridge change, a plate change and / or retouching the runout of one or more wheels.

Images