WO1999043088A1

WO1999043088A1 - Method for automatically correcting a phase servo loop gain

Info

Publication number: WO1999043088A1
Application number: PCT/FR1999/000341
Authority: WO
Inventors: Daniel Bec
Original assignee: Atg
Priority date: 1998-02-18
Filing date: 1999-02-16
Publication date: 1999-08-26
Also published as: FR2775143A1; FR2775143B1

Abstract

The invention concerns a method for automatically correcting slow gain variations in a phase servo loop with a complex transfer function of the frequency represented in the Nyquist plane. In a preferred embodiment, the associated device essentially comprises a digital signal processing element (33) and annex circuit elements (adder (30), converters (31 and 34), memory (32), oscillator (35), logic circuit (36), reader-recorder controller (37)) for processing a signal (R) derived from reading the marks engraved on a digital optical disk and containing in particular a part for the synchronisation and phase adjustment of the reader-recorder clock (hp). The invention is particularly applicable to a phase servo loop of a digital optical disk reader-recorder reading clock.

Description

METHOD FOR AUTOMATICALLY CORRECTING THE LOOP OF A LOOP

OF CONTROL.

The present invention relates to a method for automatic correction of

5 slow variations of the gain of a control loop having a complex frequency transfer function, only a limited part of the transfer function of said control loop, represented in the Nyquist plane, being taken into account in the scope of the invention.

The present invention relates more particularly to the application of said method

IO to the automatic correction of the gain of a phase control loop of a reading clock, of a digital optical disc reader-recorder of said format sampled according to well-known terminology, synchronized with the signal read from pre-engraved marks on said disc.

The present invention further relates to a device for implementing the method for automatically correcting the gain of a phase control loop of a reading clock of a digital optical disc player-recorder.

The characteristics and advantages of the invention will appear on reading the description which follows, made with reference to the appended figures in which: FIG. 1 is the representation of a timing diagram of signals appearing in a control loop phase of a digital optical disc player-recorder; Figure 2 is the representation of a phase control loop of a digital optical disc player-recorder; FIG. 3 is the representation of a timing diagram of signals appearing in a phase control loop according to the invention; Figure 4 is a simplified representation of a control loop used in the context of the invention; FIG. 5 is the representation in the Nyquist plane of the transfer function of a control loop within the framework of the invention; FIG. 6 is the representation of the operations carried out in a method according to the invention; FIG. 7 is the representation in the Nyquist plane of the transfer function of a control loop in a first case of application of the invention; FIG. 8 is the representation in the Nyquist plane of the transfer function of a control loop in a second case of application of the invention; FIG. 9 is the representation of a device, applied to a phase control loop of the clock for reading the pre-etched marks of a digital optical disc player-recorder, implementing the method according to the invention. using a digital signal processing device; FIG. 10 is the representation of a chronogram, in the case of a phase control loop of a clock for reading the pre-etched marks of a digital optical disc player-recorder, of the signals of a following device the invention; FIG. 11 is the schematic representation of the calculation modules of a digital signal processing device forming part of a device according to the invention, in the case of a phase control loop of a clock of a digital optical disc player-recorder.

To facilitate understanding of the invention, we first place ourselves in the case of a digital optical disc reader-recorder of sampled format, the clock for reading the post-etched marks representing the user's data. , user data in the broad sense including user data proper and ancillary data including in particular error correction data, is generated from an hp clock for reading pre-engraved marks, said clock being itself synchronized and rephased on specific pre-engraved marks intended for said synchronization and rephasing functions and included in groups of pre-engraved marks arranged at regular intervals along a spiral track which extends over the entire surface of the disc. The post-engraved marks, representing the data of the user, are recorded in the zones located between the groups of pre-engraved marks, each zone containing the registered marks representing the data of the user immediately following the group of pre-engraved marks to which said zone is associated. Subsequently, we mainly consider only the pre-engraved marks and the hp clock which is derived therefrom.

In general, the analog signal r, resulting from the reading of specific pre-engraved marks intended for synchronization and rephasing of the clock hp and

2 a chronogram of which, as a function of time t, is shown in FIG. 1, is shaped, thanks to a fixed or controlled threshold, in a circuit 1, shown in FIG. 2, which supplies a signal s. The signal attacks one of the inputs of a phase comparator 2, the output of which attacks the input of a loop correction filter 3, itself followed by a variable frequency oscillator 4, the frequency of which is driven by the output voltage of the filter 3. The oscillator 4 supplies the clock signal hp. The output of oscillator 4 drives a logic circuit 5 generating for each period of occurrence of marks r and s a signal u of duration equal to an integer number of periods of oscillator 4 and substantially equal to the duration of signal s . The frequency f ^ of occurrence of the signals s and u is the frequency of occurrence of the groups of pre-etched marks, which appears when the spinning disc is read at its nominal speed of rotation; the frequency f _p of the signal h _p is for example equal to twice the inverse of the duration of the shortest of the pre-etched marks. The pre-engraved marks used for synchronization of the digital optical disc player-recorder are for example reduced to a single mark of duration equal to 4 / fp. Subsequently, to simplify the description of the invention and without reducing its scope, only the particular case of the single synchronization mark will be considered, even if it is common to use at least two pre-etched marks creating an interval of time which is prohibited by the modulation code used for recording user data.

Usually only the rising edges of the signals s and u are used in the phase comparator 2 so that a variation in duration of the signal s modifies the relative phase of the signal u with respect to that of the signal s. An improvement of this method could consist in using no longer the signal u, but a signal u 'consisting of two windows u' ι and u'2 both of the same duration and centered respectively on the rising edge of the signal u and on its falling edge. A phase loop error signal e is then obtained by performing the following operations, shown in FIG. 1: integration of a reference voltage between the instant of occurrence of the rising edge of the signal u 'i and the instant of occurrence of the rising edge of signal s, addition to the voltage obtained from the integration of a voltage opposite to said reference voltage between the instant of occurrence of the rising edge of signal s and the instant of occurrence of the falling edge of the signal u'i, maintaining the voltage obtained and adding to said maintained voltage the integration of said reference voltage between the instant of occurrence of the rising edge of the signal u'2 and the instant of occurrence of the falling edge of signal s, addition to the voltage obtained from the integration of a voltage

3 opposite to said reference voltage between the instant of occurrence of the falling edge of the signal s and the instant of occurrence of the falling edge of the signal u'2 and maintenance of this voltage. This method eliminates the phase error of the signal u 'with respect to the signal s due to a variation in the duration of the signal s, the control loop rephasing the middle of the signal u' with respect to the middle of the signal s. This method also makes it possible to eliminate the erroneous signals s, due to local defects of the disk, at least one edge of which is not in the corresponding window u '\ or u'2.

However, in the case of certain rewritable discs using the so-called phase change technique or in the case of certain recordable discs called WORM only once, the determination of a threshold is delicate, since said discs provide a read signal which does not does not have a zero average value and said average value can vary considerably when moving from an area of the user data recorded to an area not yet recorded. This determination of the threshold can be particularly delicate, or sometimes even impossible, during the initialization of the digital optical disc reader-recorder, after it is put into operation and before the synchronization and the rephasing of the clock h _p are acquired, and consequently requires a significant time to perform said acquisition of synchronization and rephasing and finish the initialization of said digital optical disc reader-writer. It is therefore preferred to seek, within the framework of the invention, a method of synchronization and rephasing of the clock hp with respect to the signal r not requiring the use of a threshold for shaping said signal r. For this purpose, in the context of the invention, a signal r is taken from the pre-etched marks of the disc intended for the synchronization of the reader-recorder using a pair v of sampling signals, synchronous with the clock hp, the timing diagram of which is shown in FIG. 3, the centers of the two signals of said pair v of sampling signals being separated by a time interval τ substantially constant and equal to the duration of the signal r, duration measured by example at the points of inflection of the curve representative of the signal r, ie approximately at the mid-height of said signal r. If τ \ and x are the respective amplitudes of the first sample and the second sample of the signal r, a phase comparator is produced between the clock hp which is the source of the couple v of sampling signals and the signal r , by making the difference Δr = r \ - τ_: indeed, if said couple v is in phase advance on the signal r, as shown in FIG. 3 a, r \ <and Δr <0; if said couple v is in phase with

4 the signal r, as shown in FIG. 3b, ri = and Δr = 0; if said pair v is lagging behind the signal r, as shown in FIG. 3c, ri> n and Δr> 0. The convergence of the synchronization and the rephasing of the pair v of sampling signals with respect to the signals from reading the pre-etched marks is fast. Said convergence is all the more rapid as the time interval τ of the couple v - therefore the duration of the signal r - does not exist in the signal read coming from the reading of the marks engraved on the disc, pre-engraved marks other than the marks whose reading provides the signal r and post-etched marks when the user data are recorded, or at least the spectrum of the signal r appears only very weakly in the signal coming from the reading of all the marks engraved on the disc other than those whose reading provides the signal r. Another solution, allowing very rapid convergence of said synchronization and of said rephasing, consists in the execution of the synchronization and rephasing operations in a specific zone of the disc containing only the pre-etched marks; however, this choice lengthens the overall initialization time of the reader-recorder and is therefore not retained. The fact that the slaving of the hp clock in frequency and in phase is carried out in a single phase comparator explains the fact that we are talking about the stabilization of a phase slaving loop.

Concerning the constancy of the time interval τ, it is useful to recall some notions relating to the different types of optical discs to which the invention applies. If the disc, driven at constant angular speed, has homologous engraved marks of constant angular length over its entire extent - disc known by the term CAV -, the time interval τ remains constant over the entire extent of the disc. If the disc, driven at constant angular speed, is composed of concentric annular zones in each of which the homologous engraved marks have constant angular lengths, and if, in its innermost part, each annular zone has homologous engraved marks whose length linear is roughly the same in all annular zones - disc known by the term ZCAV -, the time interval τ remains constant in each annular zone, but varies from one annular zone to another (at the same time as the clocks player-recorder). If the disc is of the CAV type for its pre-etched marks and of the ZCAV type for its recorded marks representing the data of the user, the time interval τ remains constant over the entire extent of the disc. If the disc has holomogue engraved marks of

5 constant linear length over its entire extent and if it is driven at an angular speed inversely proportional to the distance from its center of the area being read or written - disc known as CLV -, the time interval τ remains constant over the entire extent of the disc. In the following we are not concerned with the type of disc, CAV. ZCAV or CLV, and it is considered that the time interval τ is constant, or at least constant with respect to the clock of the reader-recorder which makes it possible to generate it.

The gain of the phase control loop and therefore its stability depend on the constancy of the value that takes the magnitude Δr / τ around Δr = 0; the constancy of said quantity depends in particular directly on the repeatability of the shape of the sides of the pre-etched marks, the reading of which provides the signal r, which can vary from one disc to another and during the same disc. To ensure good stability of the phase control loop, it is necessary to be able to correct its gain to take account of variations in the quantity Δr / τ and in the variation of any other quantity having the same effect. If we again consider the difference Δr, substantially proportional to the phase error between the middle of the couple v and the middle of the signal r, it appears that even for a fixed duration τ of the time interval between the signals sampling of the couple v, the amplitude of said difference varies when the duration, the shape or the amplitude of the signal r vary, whereas there is no phase error when said difference is zero. In a phase control loop of a digital optical disc player-recorder, various components of said control loop or various elements located upstream of said control loop can cause such variations; these are for example: the size and shape of the pre-etched marks of the disc intended for synchronization and rephasing of the clock h _p , as already mentioned, the distance from the center of the disc of said pre-etched marks intended for synchronization and in rephasing, the diameter of the focused spot of the digital optical disc reader-recorder, the power of the laser beam before its interaction with the disc, the local reflection coefficient of the disc, the gain of the pre-amplifier stages according to the photodetector cells receiving the light beam after its interaction with the disc or the gain of amplifier stages located in the phase control loop. The phase comparator being included in a phase control loop, any change in amplitude of the signal it delivers, or in the amplification of said signal

6 in said control loop, modifies the operation of said control loop by changing its gain and simultaneously its bandwidth: a decrease in gain results in an increase in the residual phase error of the control loop and an increase in gain results in a risk of oscillation of the control loop. To optimize the operation of the servo loop, it is necessary to stabilize it and therefore keep its gain constant. It is important to note that only the correction of slow variations in gain should be considered here, due for example, using some of the previous examples, to the variation of the duration of the pre-engraved marks intended for synchronization and rephasing when passing from 'from one disc to another, or to the variation in the diameter of the focused spot during the rise in temperature of the digital optical disc player-recorder or during its aging; more generally considered here is any slow variation of elements constituting the phase control loop or of elements situated upstream of said control loop resulting in a change in gain of said phase control loop. On the other hand, any rapid variation of some of the said elements should not be taken into account, but on the contrary eliminated as far as possible; for example, any significant variation in shape and size of a pre-engraved mark intended for synchronization and rephasing should be eliminated in relation to the shape and dimension which they should have with reference to the preceding corresponding pre-engraved mark, related variation in particular to local defects of the disc and which can be taken into account by eliminating the contribution of the incriminated mark using a criterion of plausibility. A major characteristic of the invention is therefore to compensate for the slow variations in the elements constituting the phase control loop or in elements located upstream of said control loop resulting in a gain variation of said control loop.

Before describing this major characteristic of the invention, it is necessary to define some quantities used in the context of the invention. In general, a control loop, represented in FIG. 4, which is both the phase control loop considered until now and any other control loop, receives an input signal E, which is transmitted to a subtractor element 10. The direct branch of the control loop receives the output of the subtractor element 10; said direct branch symbolically consists of a single element 1 1 of

7 transfer function A, A being a complex function of the frequency; element 11 can usually include a correction filter for the servo loop. At the output of element 11, the signal S output from the control loop is supplied. The signal S is transmitted to the reaction branch, which symbolically consists of a single element 12 of transfer function B, B being a complex function of the frequency. The signal S x B = S \ output from said reaction branch is transmitted to the subtractor element 10. The subtractor element 10 provides the main branch with the loop error signal ε = E - S i. The product A x B = T is the open loop transfer function of the servo loop. In the case of the phase control loop of a digital optical disc player-recorder, the signal E is the middle phase of the signal r. the quantity Si is the middle phase of the couple v of sampling signals, the loop error signal ε is the difference Δr of the samples ri and x_ taken from the signal r. said difference being substantially proportional to the phase difference between the middle of the signal r and the middle of the pair v of sampling signals and the signal S is the clock hp.

To allow the description of the major characteristic of the invention, it is necessary to describe more precisely the transfer function T of the control loop which has just been described. We consider the representative curve, shown in Figure 5, of the transfer function T of the servo loop usually represented in the Nyquist plane by plotting its real component Re (T) on the abscissa and its imaginary component Im ( T) on the ordinate; we consider the point O (0; 0) origin of the coordinates and the point C of coordinates (-1; 0), point corresponding to the oscillation condition of the servo loop. We consider more particularly the portion of the curve representative of the transfer function T, located in the quadrant corresponding to Re (T) <0 and Im (T) <0 near point C and passing at a distance greater than a distance point C, for example outside the circle of radius 1/2 centered on point C. to provide a margin of stability to the control loop. We then consider a half-line D from C, of negative slope and located in said quadrant; it cuts said portion of the curve representative of said transfer function T at a point M with coordinates (-1 + a; -b), a and b being two positive real numbers and a being less than 1; T being a function of the frequency, the point M corresponds to a given frequency f. Insofar as the phase control loop is satisfactorily corrected, the considered portion of the curve representative of the transfer function T crosses the quadrant considered with an average slope roughly of the order of 1 and passes at a minimum imposed distance from point C, said minimum distance characterizing the stability of the control loop . The portion considered of the curve representative of the transfer function T, in said quadrant where the phase correction of the servo loop is preferably carried out, is tangent to a half-line Δ from O. half-line whose slope is roughly of the order of 1: in this zone the curve representative of the transfer function T is to the left and above said half-line Δ. The vectors ÇO, OM and CM = ÇO + OM respectively represent the complex numbers 1. T and 1 + T. In the control loop, which has just been described and which is shown in FIG. 4, we have: E = ε x (1 + B x A) = ε x (1 + T), ie E / ε = 1 + T and consequently the vector CM represents the complex number E / ε. If we know the complex component of E of given frequency f and if we measure the complex component of ε of given frequency f, we know the complex number (complex component of frequency f of E / complex component of frequency f of ε), therefore point M; the component of E being able to be taken as phase reference, said component of frequency f of E can then be taken real.

When the constituent elements of the control loop or elements located upstream of said control loop cause by their variation, taking all of them globally into account, a variation of the transfer function T of said control loop which is limited to a multiplication of said transfer function by a positive real number k, the curve representative of said transfer function T in the Nyquist plane undergoes a homothety of center O and of ratio k which transforms it into the curve representative of the transfer function T 'and in particular the point M is transformed into a point M', with OM '/ OM = k; simultaneously ε varies. Stabilizing the operation of the control loop - keeping the gain of said control loop constant - amounts to making the transfer function T and invariant. since its representative curve in the Nyquist plane is supposed to undergo a simple homothety. amounts to making one of the points of its representative curve invariant, in particular to bringing the point M ', corresponding to the given frequency f, to its nominal position M. Knowledge of the complex component of frequency f of E and the measurement of the complex component of frequency f of ε allow at any time to know the point M '

9 and, given M, to know the ratio k. If we multiply the transfer function A of the main branch of the control loop - at any point on the main branch - by the positive real correction factor 1 / k, the transfer function is kept invariant and the operation of the control loop is stabilized. In order to know the components of given frequency f, respectively of E and of ε, with a sufficient signal-to-noise ratio, it is usually necessary to introduce an external signal of known amplitude at the given frequency f and possibly to make this introduction as well as the measurement of the given frequency component f of ε at an instant when the loop is not used, even under its normal operating conditions. The frequency f is of course chosen apart from any possible discrete line linked to the normal operation of the servo loop.

In the case where the gain variation of the loop is only substantially real positive, a case which could be encountered, for example, when a slight saturation of an amplifier appears and which would correspond to a number k having a real component positive and an imaginary component small compared to its real component, and keeping a correction which would remain for example a multiplication of the gain of the loop by a correction factor l / (Re k) or l / | k | this would lead to an imperfect correction of the gain of the loop, M 'not being completely brought back to its nominal position M, correction which would nevertheless remain an optimization. Thereafter only a purely real number k will be considered.

The invention is first of all a method for automatically correcting slow variations in the gain of a control loop, having a complex frequency transfer function, said transfer function T being represented in the Nyquist plane with its part. real Re (T) on the abscissa and its imaginary part Im (T) on the ordinate, a portion of said transfer function being located in the quadrant corresponding to the abscissas and negative ordinates and passing at a distance greater than a minimum distance from point C of coordinates (-1; 0) of said Nyquist plane, the intersection of a half-straight line coming from said point C of coordinates (-1; 0) and of negative slope and of said portion of the curve representative of the function of transfer T being a point M corresponding to a frequency f, the transfer function T being subject to the variation of elements constituting said control loop or elements located upstream of said bo ucle of enslavement to variations resulting in the multiplication of said transfer function T by a number

10 positive real k and consequently resulting in homothety, center the point of coordinates (0; 0) and report said positive real number k, of the curve representative of said transfer function and in particular of said point M corresponding to the frequency f . characterized in that said point M corresponding to the frequency f having been chosen and therefore the frequency f being given, a signal of known amplitude is introduced at said given frequency f at the input of said servo loop, the component at said given frequency f of the error signal ε 'of said servo loop, the inverse of said real number k is calculated from said signal of known amplitude at said given frequency f of the component measured at said given frequency f of the error signal of said control loop and of the known nominal position of said point M corresponding to said given frequency f, said point M is brought back to its nominal position by multiplying by a correction factor 1 / k equal to l inverse of said positive real number k to the transfer function of the main branch of said control loop and said factor is kept constant until the next correction of said gain. A control loop according to the method of the invention, shown in Figure 6. consists of: an addition module 20, a subtraction module 21, a main branch consisting of a module 22 with a complex transfer function A ', which can usually include a correction filter for the servo loop, and with a multiplication module 23, of a reaction branch consisting of a module 24 with a transfer function complex B ′, of a source module 25 supplying two sinusoidal signals Ep and Eq of known amplitude and of given frequency f in phase quadrature with respect to each other, Eq being late with respect to E _p , two selective detection modules at the frequency f 26 and 27 and a calculation and maintenance module 28; calculation here means obtaining a result signal from an original signal both with the aid of an electronic device, either analogical or logical, as with the aid of a digital signal processing device after digitization of the signal origin, as will be specified later in an exemplary embodiment. The input signal of the servo loop E 'and the signal Ep are supplied to the addition module 20; the output signal from the addition module 20, the signal E '+ Ep = E]', is supplied to the subtraction module 21; the output signal of the subtraction module 21, the loop error signal ε 'of the servo loop, is supplied to module 22 whose output signal, the signal ε' x A '= SQ', is supplied to the multiplication module 23: the output signal of the module 23, that is to say the output signal S 'of the

11 servo loop, is supplied to module 24; the output signal from module 24, the signal S 'x B' = S \ ', is supplied to the subtractor module 21, which supplies the loop error signal ε' = Ei '- Si'. The signals ε 'and E _p are supplied to module 26, which supplies the signal c, real part of the contribution εp' of E _p to the loop error signal ε '; the signals ε 'and E _q are supplied to module 27. which supplies the signal of the imaginary part of the contribution εp' of Ep to the loop error signal ε '. The signals c and d are supplied to module 28 which calculates the inverse 1 / k - positive real number - of the variation factor - increase or decrease factor - of the transfer function of the servo loop from the signal Ep, said signals c and d and knowledge of the nominal operating state of the control loop and maintains it until the next calculation; said factor 1 / k, maintained by the module 28, is supplied to the multiplication module 23, thus closing a control loop for correcting the gain of the control loop; the product A 'x B' / k is the transfer function T.

In summary, the method according to the invention for correcting the gain of a servo loop is characterized by the execution of the following successive operations: addition of a known sinusoidal signal Ep of given frequency f to the input signal E 'of the control loop, sampling of the loop error signal ε' of the control loop and selective detection at the given frequency f of the real component c and of the imaginary component d of the contribution of said sinusoidal signal known of frequency given f to the loop error signal ε ', calculation of a correction factor of the gain of the servo loop equal to the inverse of the real number k, multiplication of the gain of the direct branch of said loop enslavement by said factor and maintenance of said factor.

It is possible, during the calculation period of said correction factor 1 / k of the gain of the control loop, to put the control loop in the case of operation without gain correction, that is to say of give the output of the calculation and maintenance module the value 1, calculate under these conditions a new value of k and carry out the multiplication by 1 / k of the gain of the main branch of the servo loop. It is also possible, during the calculation period of said correction factor 1 / k, to keep the previous correction; it is the latter case that we preferentially consider from now on. In the latter case, if one is in step i + 1 of correction of the gain of the control loop, a differential correction factor 1 / + \ is calculated in the presence of the correction of factor 1 / kj, then we calculate the

12 correction factor 1 / kj + j = (1 / k'j + i) x (1 / k); the loop gain is multiplied by the correction factor 1 / kj + i and the said correction factor is maintained.

The calculation and maintenance module 28 then performs the following operations during the i + lth correction phase of the control loop: maintenance of the previous correction factor 1 / kj, calculation of the differential correction factor 1 / k'j + , development of the new calculation factor 1 / kj + i, replacement of the previous correction factor 1 / kj, by the new correction factor 1 / kj + j which is applied to the multiplication module 23 and maintenance of said new factor of correction.

Of course the signal Ep is chosen as small as possible so as not to disturb the output signal S 'notably, but large enough to allow detection of the two components c and d of the contribution εp' of the known sinusoidal signal E _p of given frequency f to the loop error signal ε 'with a sufficient signal-to-noise ratio. The calculation module 28 can, in particular in the case of the phase control loop of a digital optical disc player-recorder, include significant filtering to eliminate possible accidental disturbances due to abnormal rapid variations in the constituent elements the control loop or elements located upstream of said control loop; in this case it is possible to envisage a time constant of the low-pass filter of the order of 0.1 seconds or even of the order of 1 second or more; the choice of the value of said time constant depending very particularly on the nature and the distribution of the local defects of the disc which disturb the pre-etched marks and more particularly the marks intended for the synchronization and the rephasing of the clock hp of the reader- recorder. Still in the case of the phase control loop of a digital optical disc player-recorder, it is possible possibly to carry out the gain correction of the control loop only from time to time, in particular during periods of the digital optical disc player-recorder during which there is no data transfer with the outside - neither writing nor reading. The value of the gain correction of the servo loop is then maintained at its last state until the preparation of the next correction; the value of said correction is in any case maintained in principle at least during the time interval between the times of occurrence of two successive groups of pre-etched marks. In applications other than the gain correction of the phase control loop of a digital optical disc player-recorder,

13 abnormal rapid variations of the constituent elements of the control loop or of elements located upstream of said control loop could be weaker or even nonexistent and consequently require only a more reduced filtering or even no filtering of the calculation module 28 If we again consider the two components c and d extracted from the loop error signal ε 'and corresponding to the contribution εp' of the known sinusoidal signal Ep of given frequency f to said loop error signal ε ', the vector CM 'represents the complex number Ep / (c + id), ie Ep x (c - id) / (c ^ + d ^) and the vector OM' the complex number -1 + Ep x (c - id) / (<y + d ^); in the absence of variations of the constituent elements of the control loop or of elements located upstream of said control loop the point M 'is merged with the point M and the vector OM represents the complex number -1 + a - ib. As we only consider the variations of the transfer function T of the servo loop which result in the multiplication of said transfer function by a positive real number k, the inverse of said real number k is obtained by calculation (-1 + a) / [-1 + Ep xc / (c ^ + d ^)] or by the calculation b / [E _p xd / (c ^ + d ^)], these two values being equal.

This process can be implemented using a purely electronic device and in this case each module described in the process can be replaced by a set of elementary electronic components making it possible to carry out the operations charged to said module or else all the operations to be carried out by the various modules described can be carried out by one or more specific electronic components designed for this purpose. For example, the selective frequency detection of the two modules 26 and 27 can be carried out by a usual synchronous detection device produced in the form of elementary electronic components or else be carried out by a part of a more specific specific electronic component defined for this purpose. Similarly, the calculation module 28 can be produced using a specific electronic component, or a portion of a specific electronic component, designed for this purpose.

However, the operations corresponding to the various calculation modules described above can be carried out by calculation using a digital signal processing device; this solution is interesting in particular because such a digital signal processing device makes it possible to carry out in addition other useful operations than those corresponding to said modules described and has great flexibility

14 adaptation. This solution is acceptable insofar as the speed at which said operations must be executed is not too high and therefore compatible with a digital signal processing device. This is the case for gain correction of a phase control loop of a digital optical disc player-recorder. Indeed with the case chosen as an example of a single pre-engraved mark intended for the synchronization and rephasing of the clock hp of duration equal to 4 / f _p and if we consider a period of occurrence of the groups of pre-engraved marks equal for example to 256 / fp, the gain correction operation need only be carried out in a time interval less than 256 / f and not in a time interval of 4 / f _p or of the same order of magnitude, which, taking into account the order of magnitude of the frequency f _p of a conventional digital optical disc player-recorder, makes it possible to use conventional digital signal processing circuits and to perform not only the operations according to the invention, but also operations other than those forming the subject of the invention.

The calculation of the correction factor 1 / k nevertheless requires a large number of steps of the clock of the digital signal processing device. It is preferable with a digital device to perform an approximate, but rapid gain correction calculation, and even to apply it only partially; we apply a correction which is only a fraction f (l / k) of the calculated correction: if 1 / k = 1 + e, e being a real number, we apply for example only a correction 1 + e / 2. A fast calculation makes it possible to take a larger number of samples and makes it possible to avoid the error which would arise from a sub-sampling of the correction signal. In addition, in the case of digital signal processing, the filtering, mentioned during the description of the calculation module 28, can include the calculation of an average m (l / k), for example either an arithmetic average on a determined number of the last correction factors 1 / k or f (l / k) calculated successively and memorized, that is to say a weighted average exponentially carried out on all the correction factors calculated while favoring the last correction factors calculated. This procedure is not a hindrance insofar as the correction does not need to be carried out quickly; on the contrary, it has the advantage of minimizing the effect of rapid disturbances which could be due to elements constituting the control loop or to elements located upstream of said control loop.

We again consider point M of the curve representative of the transfer function T of the servo loop in the Nyquist plane. The slope of the curve

15 representative is roughly equal to 1. If we now consider, as shown in figure 7. a point M such that the direction CM makes an angle of 45 ° with the axis Re (T). the directions CM and OM are roughly perpendicular. Taking point C as the origin of the coordinates and as the coordinate axes the two axes CX and CY respectively parallel to the axis Re (T) and to the axis Im (T), the point O has the coordinates (1; 0 ) and point M has coordinates (a; -a). When there are variations of elements constituting the control loop or of elements located upstream of said control loop, the transfer function T is replaced by the transfer function T 'and therefore in the plane from Nyquist point M is replaced by point M '. We have algebraically OM '= OM + ε _e , OM and OM' being positive and ε _e being a real positive or negative number of absolute value much less than OM. The coordinates of M ', taking point C as the origin of the coordinates and the axes CX and CY as the coordinate axes, are then respectively Re (CM') = a - ε _e cos θ and Im (CM ') = -a - ε _e sin θ. θ being the acute angle, measured arithmetically, made by the direction OM with the axis CX. If OM '> OM, let ε _e > 0, | Re (CM') | <| Re (CM) | = | Im (CM) | <| Im (CM ') | ; if OM '<OM, let ε _e <0, | Re (CM') | > | Re (CM) | = | Im (CM) | > | Im (CM ') | ; if OM '= OM, let ε _e = 0, | Re (CM') | = | Re (CM) | = | Im (CM) | = | Im (CM ') |. To correct the gain of the servo loop, if | Re (CM ') | <| Im (CM ') |, the gain of the control loop must be reduced; if | Re (CM ') | > | Im (CM ') |, increase the gain of the control loop.

If we add the sinusoidal signal of known amplitude E _p of given frequency f. when the point corresponding to the frequency f is in M ', the contribution εp' of the signal E _p to the loop error signal ε 'has a real component c and an imaginary component d and the coordinates of said point M', with C as the origin of the coordinates and CX and CY as the axes of the coordinates, are respectively Re (CM ') = Re (E _p / ε _p ) = E _p xc / (c ² + d ² ) and Im (CM') = Im ( E _p / ε _p ) = -E _p xd / (c ² + d ² ). Consequently, to correct the gain of the servo loop, if | c | <| d |, the gain of the control loop must be reduced and if | c | > | d |, increase the gain of the control loop. Taking into account the preferential choice to bring a series of differential corrections - at the i + lth step of the gain correction of the control loop we replace the correction 1 / kj by the correction 1 / k [+ \ = (1 / kj) x (1 / k'j + i) - and to apply only a partial correction, the differential correction factor remains close to 1 and consequently can be taken in form

16 1 / k'i + = 1 + nx (I + l - | dj + ι |), where CJ + I and dj + i are respectively the real component and the imaginary component of Ep / εp 'during the i + lth correction gain of the control loop and where n is a positive real number chosen according to the characteristics of the control loop and taking into account the types of faults that the control loop presents and the extreme values that the factor of correction is likely to take; to simplify the calculations n can advantageously, as far as possible, be an integer power of 2 negative or positive. At the initialization of the gain correction device of the servo loop the value 1 is given to the coefficient 1 / k. Considering that the correction factor 1 / kj remains for example between 1/3 and 3, we can even content ourselves with replacing the multiplication (1 / k _j ) x (1 / k'j + i) by the addition 1 / kj + nx (| CJ + I | - | dj + ι |), having suitably chosen n so as not to over-correct when 1 / kj is small. The correction factor 1 / k _j is then the addition to 1 of a sum of terms nx (| c; | - | di |), with 1 <j <i.

We again consider point M of the curve representative of the transfer function T of the servo loop in the Nyquist plane. If we now consider, as shown in Figure 8, a point M such that the direction CM makes an angle of 90 ° ^~ with the axis Re (T), i.e. the direction CM coincides with the axis CY, the directions CM and OM roughly make an angle of 45 ° between them. Taking point C as the origin of the coordinates as above and as the coordinate axes the two axes CX and CY respectively parallel to the axis Re (T) and to the axis Im (T), the point O has the coordinates (1 ; 0) and point M has coordinates (0; -b). When there are variations of elements constituting the control loop or of elements located upstream of said control loop, the transfer function T is replaced by the transfer function T 'and therefore in the plane from Nyquist point M is replaced by point M '. We have algebraically OM '= OM + ε _e , OM and OM' being positive and ε _e being a real positive or negative number of absolute value much less than OM. The coordinates of M ', taking point C as the origin of the coordinates and the axes CX and CY as the coordinate axes, are then respectively Re (CM') = - ε _e cos θ and Im (CM ') = -b - ε _e sin θ, θ being the acute angle, measured arithmetically, made by the direction OM with the axis CX. If OM '> OM, let ε _e > 0, Re (CM') <0; if OM '<OM, let ε _e <0. Re (CM')>0; if OM '= OM, let ε _e = 0, Re (CM') - 0. To correct the gain of the

17 control loop, if Re (CM ') <0, the gain of the control loop must be reduced; if Re (CM ')> 0, the gain of the control loop must be increased. If we add the sinusoidal signal of known amplitude E _p of given frequency f, when the point corresponding to the frequency f is in M ', the contribution εp' of the signal Ep to the loop error signal ε 'has a real component c and an imaginary component d and the coordinates of said point M ', with C as the origin of the coordinates and CX and CY as the axes of the coordinates, are respectively Re (CM') = Re (E _p / ε _p ) = E _p xc / (c ² + d ² ) and Im (CM ') = Im (Ep / ε _p ) = -Ep xd / (c ² + d ² ). Consequently, to correct the gain of the control loop, if c <0, it is necessary to decrease the gain of the control loop and if c> 0, it is necessary to increase gain of the control loop. Given the preferential choice of bringing a series of differential corrections - at the i + lth step of the gain correction of the control loop, we replace the correction 1 / k _j by the correction 1 / k _j + i = ( 1 / kj) x (1 / k'j + j) - and to apply only a partial correction, the differential correction factor remains close to 1 and therefore can be taken as 1 / k'j + i = 1 + nx CJ + I, where CJ + I is the real component of Ep / εp 'during the i + lth correction of the gain of the control loop and where n is a positive real number chosen according to the characteristics of the control loop and taking into account the types of faults that the control loop has and the extreme values that the correction factor is likely to take; as already mentioned, to simplify the calculations n can advantageously, as far as possible, be an integer power of 2 negative or positive. At the initialization of the gain correction device of the servo loop the value 1 is given to the coefficient 1 / k. Considering that the correction factor 1 / kj remains for example between 1/3 and 3, we can even content ourselves with replacing the multiplication (1 / kj) x (1 / k'j ₊ ι) by the addition 1 / k _j + nx CJ + I. having suitably chosen n so as not to over-correct when 1 / k _j is small. The correction factor 1 / kj is then the addition to 1 of a sum of terms nxc; , with l ≤j ≤ i.

An example of a device, used in the case of a phase control loop of the hp clock of a digital optical disc player-recorder, implementing a method according to the invention and performing by digital calculation the operations according to said method, comprises a digital signal processing device 33 and ancillary circuit elements, as shown in FIG. 9. A signal R. of which a

18 chronogram is shown in Figure 10, is derived from the reading of the marks engraved on the digital optical disc - pre-engraved marks of the groups of pre-engraved marks and marks, post-engraved during the recording of the user data, which each follow groups of pre-engraved brands. Part of the signal R, resulting from the reading of the groups of the pre-engraved marks, contains not only the signal r intended for the synchronization and the rephasing of the clock hp of the digital optical disc player-recorder, but also signals intended for the follow of the runway - to the radial tracking of the runway and to the focusing in the plane containing the runway -, runway address signals intended to allow access to the runway, said runway usually being in the form of a spiral and bearing in generally both said groups of pre-etched marks and post-etched marks representing user data, and the track address marks intended to allow access to the track and in particular to the areas containing the post- engraved representing user data. An analog-digital converter 31 receives the output signal Ri = R + Ep from an adder 30, Ep being said signal of known amplitude of given frequency; said analog-digital converter 31 is controlled by a sampling signal, the clock signal hp of frequency fp, which makes it possible to sample the signal R \, for example on the rising edges of said signal hp, and to digitize it. The clock signal hp comes from an oscillator 35 controllable in frequency by a voltage coming from a digital-analog converter 34, which itself receives the output signal from the digital signal processing device 33; one could just as well consider, instead of the clock 35 driven in frequency by a voltage and the digital-analog converter 34, a frequency synthesizer driven directly by the digital signal of the digital signal processing device 33. The part of the signal R, resulting from the reading of the pre-etched marks of a group of pre-etched marks begins, as shown in FIG. 10, at an instant t ^ and ends at an instant tβ. The analog-digital converter 31 is permanently controlled by the clock signal hp of frequency fp and continuously generates digital samples. The output signal from the analog-digital converter 31, the sampled signal R2, contains in particular the signal resulting from the reading of the pre-etched mark from which two samples are taken intended for the generation of the error signal of the servo control loop. phase; the output signal from the analog-digital converter 31 is supplied to a memory 32, for example of the so-called FIFO type. Among the samples generated, only are

19 selected to be memorized the samples which are useful for the various functions - synchronization and rephasing, radial track tracking, focusing in the plane containing the track and access to the areas where the user's data are recorded - which the processing device fills digital signal 33. To enable said selection to be made, a selection signal w is applied to memory 32; said selection signal w comes from the processing by a logic circuit 36 of the frequency signal fp of the clock hp generated by the oscillator 35. To simplify the description of the invention, since we are only interested here when synchronizing and rephasing the clock hp, only the part w ′ of the selection signal w is considered, part which performs the selection of the two samples taken from the signal r and intended for the synchronization and rephasing of the hp clock. In the case of the example provided above of a single pre-etched mark, of duration 4 / fp, intended for the synchronization and the rephasing of the clock hp, said part w ′ of the signal w, shown in FIG. 10, is consisting for example of two pulses, the time interval between the centers of said two pulses being substantially equal to the duration of the signal r, said two pulses being generated only once during each of the time intervals, extending from an instant t ^ to the instant tg, which immediately follows it, of reading a group of pre-engraved marks; the signal w 'is the pair v of sampling signals described above. The logic circuit 36, which selects samples of ranks given in a predetermined series of samples starting at time t ^, is reset to the initial position and the memory 32 is reset to zero by a signal z from the signal processing device digital 33, before each instant t ^ at the start of reading a group of pre-etched marks. At the rate of its calculation - and not at the rate of the clock signal hp of frequency fp - the digital signal processing device 33 takes the signal R3 using the control signal m the digital samples stored in memory 32 . The digital signal processing device 33 supplies as output a digital signal R4, result of the calculation of the operations according to the method of the invention, intended for the phase correction of the clock hp generated by the oscillator 35. The signal R4 is converted into an analog signal R5 by the digital-analog converter 34, said signal R5 controlling said frequency-controlled oscillator 35. The digital signal processing device 33 is in connection with the controller 37 of the reader-recorder, which manages and controls the execution of the various functions fulfilled by the reader-recorder and in particular those fulfilled by said digital signal processing device 33.

20 Before continuing with a more detailed description of the example device according to the invention, we consider a digital optical disc and a digital optical disc player-recorder using such a disc and comprising a phase control loop provided with a digital signal processing device as described in the invention: the digital signal processing device makes it possible to advantageously fulfill functions annexed to the phase control and to the correction of the gain of the phase control loop, functions that we will clarify now.

Due to the finite diameter of the focused reading beam and the width of the pre-etched marks proportional to the radius of the area of the disc where said marks are located, the signal edges from said pre-etched marks vary according to the radius and therefore the gain in the phase control loop varies depending on the radius; in the case of rapid access through part of the useful range of the disc, this variation in gain could not be corrected in real time; the variation of the gain of the phase control loop being known as a function of the radius, a correction factor can be stored - a table of discrete values corresponding to successive concentric bands of the disc - and applied in open loop, in real time , by the digital signal processing device 33 to the phase control loop.

Another phenomenon is the existence of local disc faults: because of these local faults, the signal R, resulting from the reading of the groups of pre-etched marks of the disc, and in particular the signal r, resulting from the reading of the marks intended when synchronizing and rephasing the hp clock, there may be gaps or, more generally, significant disturbances; as a result, the error signal of the phase control loop, which is the difference of the two samples r and r2 selected by the part w 'of the selection signal w and stored in memory 32 during the reading of each group of pre-engraved marks, may be aberrant; two thresholds of opposite values are memorized and make it possible to eliminate the loop error signal whenever it is not included between these two thresholds.

The digital signal processing device 33, which on the one hand is an element of the phase control loop of a digital optical disc player-recorder and on the other hand makes it possible to carry out the correction according to the invention slow variations in the gain of said loop, performs calculations and logic operations shown schematically in Figure 11. The signal processing device comprises a unit

21 control unit 50 which performs all the internal operations and enables dialogue with the external elements already described: the analog-digital converter 31, the memory 32, the digital-analog converter 34, the selection logic circuit 36 and the reader controller 37 digital optical disc recorder and supplying the signal of known amplitude E _p and of given frequency f to the adder 30; the calculations and logical operations are represented schematically by modules. A module 51 performs the difference r ₃ of the two samples ri and r2 supplied in the signal R3 extracted from the memory 32; the difference signal r ₃ is supplied to a module 52, which performs an open loop correction, in real time, as a function of the radius, of the gain of the phase control loop by multiplying the difference signal r ₃ by the one of said correction factors which are functions of the radius and stored in a table and provide a result signal X4 as an output; the signal rj. is supplied to a module 53, which performs a comparison of the signal rj. with said two memorized thresholds: if the signal X4 is in the interval of said two memorized thresholds the module 53 outputs the signal rj., otherwise it provides a zero correction, ie 4 = 0; a variant may also consist in supplying the value of the signal rj. obtained during the reading of the group of pre-etched marks immediately preceding the reading of the group of pre-etched marks having supplied the value out of tolerance of the signal r When, when reading of two groups of pre-etched marks immediately following one another, the two values obtained for the signal X4 are out of tolerance, the logic sector containing said two groups of pre-etched marks immediately following one another can advantageously be eliminated by raising an alarm to the controller 37 of the optical disc player-recorder digital. A module 54 detects in the signal 4 the real and imaginary components at the given frequency f, contribution of the sinusoidal signal Ep of known amplitude and given frequency f, deduces therefrom the differential correction nx (| CJ + I | - | dj + ιQ, which must be added during the i + lth correction step of the gain of the phase control loop to the previous correction factor 1 / k _j to obtain the i + lth correction factor 1 / kj + j, performs as has already been specified an average m (l / k) of the previous calculated correction factors; said average m (l / k) is supplied to a multiplier module 55 which supplies the corrected signal R4 =. xm (l / k); the signal R4 is supplied to the digital-analog converter 34. In summary, the digital signal processing device performs both the operations useful for the normal operation of the phase control loop and the gain correction operations of

22 said loop. It can also perform other operations of access, radial runway tracking and focusing as already mentioned.

In the device according to the invention which has just been described and which is applied to the correction of the gain of the phase control loop of the hp clock of a digital optical disc player-recorder, the conversion is carried out analog / digital of two samples ri and r2 of the signal R using the clock signal hp; if we refer to the description of the process according to the invention, shown in FIG. 6, the analog / digital conversion of two samples of the signal Ei 'is carried out using the signal S', the relative phase of the signals hp and R, ie the relative phase of the signals S'i and Ei ', in this case being the quantity to be canceled in the servo loop. In the case of a control loop where the magnitudes at stake are the amplitudes of the signals Si 'and E \ respectively, it is the difference between the amplitudes of the signals Si' and Ei 'which must be canceled; it is then necessary to carry out the analog / digital conversion of a sample of the signal Ej 'and of a sample of the signal Si' or else only of a sample of the signal ε 'using a clock coming for example from the device digital signal processing used in the control loop, without changing the spirit of the invention.

23

Claims

1 - Method for automatically correcting slow variations in the gain of a control loop having a complex frequency transfer function, said transfer function (T) being represented in the Nyquist plane with its real part (Re (T )) on the abscissa and its imaginary part (Im (T)) on the ordinate, a portion of said transfer function being located in the quadrant corresponding to the abscissas and the negative ordinates and passing at a distance greater than a minimum distance from the point (C ) of coordinates (-1; 0) of said Nyquist plane, the intersection of a half-straight line coming from said point (C) of coordinates (-1; 0) and of negative slope and of said portion of the curve representative of the transfer function (T) being a point (M) corresponding to a frequency (f), the transfer function (T) being subject to the variation of elements constituting the said servo loop or of elements located upstream of said loop of slaving to variations resulting in the multiplication of said transfer function (T) by a positive real number k and consequently resulting in homothety, from center to the point of coordinates (0; 0) and to report said positive real number (k), of the curve representative of said transfer function and in particular of said point (M) corresponding to the frequency (f), characterized in that said point (M) corresponding to the frequency (f) having been chosen and therefore the frequency (f) being given, a signal of known amplitude (Ep) is introduced at said given frequency (f) at the input of said servo loop, the component is measured at said given frequency (f) of the error signal (ε ') of said servo loop, the inverse of said real number (k) is calculated from said signal of known amplitude (Ep) at said given frequency ( f) of the component measured at said given frequency (f) of the error signal of said servo loop and from the known nominal position of said point (M) corresponding to said given frequency (f), said point (M ) at its nominal position by multiplying by a correction factor (1 / k, f ( l / k), m (l / k)) equal to the inverse of said positive real number (k) or derived from said inverse the transfer function of the main branch of said servo loop and said factor is kept constant from said multiplication until the next correction of said gain.

2 - Method according to claim 1, characterized in that said half-straight line originating from said point (C) of coordinates (-1; 0) and of negative slope makes an angle of 45 ° with the real axis (CX).

24 3 - Method according to claim 2, characterized in that the calculation of said correction factor (1 / k) is performed approximately by performing the sum of the correction factor calculated during the previous correction of the gain of the servo loop and of a term proportional to the difference between the absolute value of the real component (c) and the absolute value of the imaginary component (d) of the contribution of the signal of known amplitude (Ep) and given frequency (f) the error signal (ε ¹ ) of the servo loop.

4 - Method according to claim 1, characterized in that said half-straight line from said point (C) of coordinates (-1; 0) and of negative slope makes an angle of 90 ° with the real axis (CX).

5 - Method according to claim 4, characterized in that the calculation of said correction factor (1 / k) is performed approximately by performing the sum of the correction factor calculated during the previous correction of the gain of the servo loop and of a term proportional to the real component (c) of the contribution of the signal of known amplitude (Ep) and of given frequency (f) to the error signal of the control loop.

6 - Method according to one of claims 1 to 5, characterized in that the correction actually applied is only a fraction (f (l / k)) of said correction factor (1 / k) calculated. 7 - Method according to one of claims 1 to 6, characterized in that the correction actually applied is obtained by performing an average (m (l / k)) of the correction factor which has just been calculated and of the correction factors which have been calculated previously.

8 - Method according to one of claims 1 to 7, characterized in that the control loop is a phase control loop of a reading clock of a digital optical disc player-recorder, using as signal input the signal (r) for reading the pre-engraved marks of said disc intended for the synchronization and the rephasing of said clock of the digital optical disc player-recorder.

9 - Method according to claim 8, characterized in that said phase control loop comprises as input a phase comparison obtained by making the difference in the amplitudes of the samples of said signal from said pre-etched marks intended for the synchronization of the clocks of said reader- disc recorder

25 digital optics, said samples being obtained using a pair (v) of two signals, the centers of said two signals being separated by a substantially constant duration (τ) equal to an integer number of periods of the reading clock pre-etched marks of said digital optical disc player-recorder and substantially equal to the duration of the signal from the reading of pre-etched marks intended for synchronizing the clocks of said digital optical disc player-recorder.

10 - Process according to claim 9, characterized in that the signal from the reading of the pre-engraved marks intended for the synchronization of the clocks of said digital optical disc reader-recorder comes from the reading of a single pre-engraved mark giving a reading signal of duration substantially equal to four times the period of the clock for reading pre-etched marks.

1 1 - Device for correcting the gain of the phase control loop of a reading clock of a digital optical disc player-recorder using two digitized samples of the signal read from the pre-etched marks of said disc, comprising a device for digital signal processing (33), characterized in that said digital signal processing device performs both the signal processing allowing the operation of the phase control loop and the signal processing allowing the gain stabilization of said loop, said signal processing allowing the stabilization of said gain comprising the injection of a signal of known amplitude (Ep) of given frequency (f) upstream of said loop, the sampling of the error signal (ε ¹ , r ₃ ) from said loop, the extraction of said error signal (ε ', r ₃ ) from the real (c) and imaginary (d) components of given frequency (f) from the contribution of said amplitude signal known (Ep) and given frequency (f), calculating a correction factor ((1 / k), m (l / k)) and multiplying said loop error signal by said correction factor. 12 - Device according to claim 11, characterized in that the digital signal processing device performs, in addition to the basic operations allowing the operation of the phase control loop and the stabilization of the gain of said loop, the additional operations testing the likelihood of the input signal of said loop and correcting in open loop the gain of said loop as a function of the radius of the disc.

13 - Device according to one of claims 1 1 or 12, characterized in that said device performs, in addition to all the operations allowing the operation of the

26 control loop ^"phase and the stabilization of the gain of said loop, operations for at least one of the radial tracking functions track, focusing and access of the optical disc reader-recorder.

27