• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    In a data processing architecture comprising a control unit and converters CNj to be synchronized on an active edge of a common reference clock CLK, a synchronization method provides an arrangement of the converters in at least one serial chain, and a process synchronizing the converters by propagating a synchronization signal SYNC-m emitted by the control unit, said signal being retransmitted at the output OUT by each converter after resynchronization on an active clock edge, to a synchronization input IN d 'a next converter in the chain ,. Each converter comprises a configuration register REG of the synchronization, comprising at least one parameter of polarity Sel-edgej which sets the polarity of the reference clock edge for the reliable detection of a synchronization signal received at the input of the converter. A phase parameter Sel-shiftj also makes it possible to synchronize in phase the sampling clocks of n conversion cores of the converters working at a sampling frequency obtained by division by n of the reference clock frequency CLK.
    公开号:FR3043477A1
    申请号:FR1560739
    申请日:2015-11-10
    公开日:2017-05-12
    发明作者:Etienne Bouin；Remi Laube；Jerome Ligozat；Marc Stackler
    申请人:e2v Semiconductors SAS；
    IPC主号:
    专利说明:

    METHOD FOR SYNCHRONIZING DATA CONVERTERS BY A SIGNAL
    TRANSMIT FROM CLOSE TO NEAR
    Technical Field The invention relates to the synchronization of analog-digital and / or digital-to-analog data converters. It relates more particularly to fast converters whose working frequency is of the order of a hundred megahertz and more. The invention applies in particular to systems that require the synchronization of several of these fast converters, such as antenna arrays or I / Q modulated data communication systems.
    TECHNICAL PROBLEM
    Fast converters are generally formed of two or more converter cores, each of which operates at a slower (sampling) working frequency, and whose output signals (interleaving) are combined to arrive at the expected conversion frequency. The sampling clocks of these converters are thus usually generated internally by a clock generator which performs a frequency division, from the reference clock. If we have conversion cores per converter, the sampling frequency is obtained from the reference clock CLK by frequency division by n.
    In an architecture with several converters, it is necessary to be able to synchronize them, to start them in phase. In the case of fast converters, it is necessary to start the frequency dividers in phase, that is to say that it is not enough to say that they are reinitialized synchronously on an active edge of reference clock, it is still necessary to know which active edge, otherwise the sampling clocks obtained may not all be in phase depending on whether the frequency dividers start on an active clock edge or the next.
    One of the known solutions to this problem is to use a signal, which is usually called synchronization signal, which is a pulse distributed in phase on all the converters and its distribution is designed to obtain the propagation time adapted for each converter. , so that the active clock fronts on which the frequency dividers will be initialized correspond, that is to say that two by two these active edges will all be distant from a reference clock period number which is a multiple n: 0, n, 2xn, .... Everything is determined by design of the distribution paths of the reference clock and the synchronization signal on each of the converters.
    This technique thus requires a very careful distribution of the reference clock and the synchronization signal which is based on an accurate evaluation of the propagation delays in the electrical conduction paths bringing these signals to each of the converters in a given architecture. It is a question of taking into account all the parameters influencing the propagation: length and materials of the conductors, characteristics of the stages of entry and exit of the signals, characteristics of the welds, .... If one knows to conceive that rather well with the weak ones frequencies, in fast systems that occupy us with working frequencies of 100 megahertz and more, the task is more delicate. In particular, there is then a strong additional constraint concerning the acquisition of the synchronization signal which must occur in a time window much shorter than the clock period. All these difficulties are added and make expensive but also difficult synchronization by design that is sufficiently precise. This results at the system level with additional system complexity and degraded performance. Other known solutions do not use this synchronization signal. These are, for example, solutions using PLL phase-locked loops in the converters. But they pose other problems. In particular, they generally cause jitter problems on the sampling clocks, which also impacts the performance of the converters. These PLL solutions also pose temperature stability problems, not to mention the complexity they add to each converter.
    SUMMARY OF THE INVENTION The invention proposes to solve the technical problem relating to the distribution of the synchronization signal. We seek a simpler solution for implementation including when it is necessary to synchronize a large number of converters on a high frequency clock, for example 100 megahertz and beyond, with all the required precision.
    In a given architecture, the propagation delays of a signal from one point to another are fixed by the physical characteristics of the propagation path of this signal in the considered architecture. Rather than acting upstream, at design, to set this delay for each converter as in the prior art, the clever idea of the invention is to configure the converters in at least one serial chain, and replace the phase distribution of the synchronization signal on all the converters, by transmitting this signal from one converter to another to reach step by step all the converters of a chain, in combination with a parameterization of each converter of the chain , which reflects the propagation delays of the synchronization signal in this chain of converters and by which is transmitted at the output of each converter, a synchronization signal which is resynchronized, for the next converter. In each converter, the parameterization also makes it possible to choose the right active reference clock edge which retrieves the signal resetting the sampling clock generator, making it possible to synchronize in phase the sampling clocks of all the converters.
    In this way, we release the constraints upstream, on the design and implementation of application circuits, so we reduce costs. And a powerful synchronization of the converters is carried out, at the cost of a step of learning synchronization configuration parameters, of easy implementation, which is shown to be done only once, and of the addition of sequential and combinational logic elements in each converter, making it possible to determine and apply the parameterization. The invention thus relates to a method for synchronizing data converters of the digital / analog and / or analog / digital type on an active edge of a common reference clock CLK, characterized in that the converters form at least one transmission chain. series of a synchronization signal transmitted by a control unit of the converters such that: the converter of rank 1 in the chain has a synchronization input connected to a synchronization control output of the control unit; each converter of rank j greater than 1 in the chain has a synchronization input connected to the synchronization output of the converter of rank j-1 in the chain; the synchronization method being characterized in that a synchronization process of the converters of the chain on a CLK reference clock active edge activated by the control unit comprises the following steps controlled by the control unit: - initialization in each converter of a synchronization configuration register having at least one polarity parameter defining as a polarity of a reference clock edge for the signal detection at the synchronization input of the converter, the polarity of the active edge of reference clock or reverse polarity; then - sending on the synchronization control output of the control unit a synchronization signal, which is a pulse width at least equal to a reference clock period; the synchronization method being further characterized in that each converter in the chain is configured to perform the following steps: a) detecting a sync input timing signal on a reference clock edge CLK which has the polarity defined by the value of said polarity parameter configured in the converter, - b) alignment on the next reference clock active edge of the signal detected in step a) to provide a synchronization signal aligned with an active edge of reference clock and applying said signal to the synchronization output of the converter.
    In one implementation, each converter in the chain is formed of n conversion cores at a sampling frequency which is provided by a divider by n of the reference clock frequency, where n is a non-zero integer, and configured to perform following step b), the following steps: - c) offset of said aligned synchronization signal obtained in step b) by an integer of reference clock period (s) defined by the value of a phase parameter (Sel-shiftj) provided by said converter configuration register; and -d) applying the shifted signal obtained as a reset signal of said frequency divider by n, wherein and said integer defined by the phase parameter has a value between 0 and n-1 inclusive.
    Each converter in the chain is configured to perform a step e) of checking the stability of an active level of signal received at the synchronization input acquired on the reference clock edge defined in the detection step a) by the value of said polarity parameter, by comparison of the first value acquired on said detection edge, with at least a second value of said acquired signal on an edge in advance, by a predetermined time, on said detection edge and at least one third value of said signal acquired on a delay front, of a determined delay, on said detection edge, and if the values are not all identical, activation of a corresponding flag bit in said converter configuration register.
    The method comprises a learning phase for configuring the polarity parameter of each converter successively from the first input converter to the last converter of the chain, by means of the flag bit of the converters configuration register, then configuring a value p of the phase parameter of each converter, from a determination of the integer number M of reference clock periods separating the sync pulse edge at the output of the converter from the sync pulse edge. output of a converter upstream in the chain, and the value p of the phase parameter is such that the sum M + p is a multiple of n. The invention also relates to an analog / digital or digital / analog data converter which comprises an input for receiving a synchronization signal and an associated synchronization circuit for synchronizing the converter on an active edge of a reference clock signal. according to the method of the invention. Other features and advantages of the invention are presented in the following description, with reference to the accompanying drawings in which: FIG. 1 illustrates a data architecture in which the converters are configured to form a serial propagation chain of a synchronization signal according to the invention; FIG. 2 is a simplified block diagram of a converter comprising the circuit elements for the implementation of a synchronization method in a converter chain according to the invention; FIGS. 3 and 4 illustrate timing diagrams of the signals generated in each converter, as a function of the configuration parameters of the synchronization according to the invention; FIG. 5 is a timing diagram which more particularly illustrates the phase alignment of the sampling clocks in a chain of converters according to the method of the invention; FIGS. 6 and 7 are a timing diagram of the signals and a block diagram of the steps of a learning process according to the invention, for configuring the polarity parameter of each converter, using a synchronization signal transmitted by the control unit which is synchronous with the reference clock; - Figures 8 and 9 illustrate a variant of the learning method using a synchronization signal issued by the control unit which is asynchronous; and FIGS. 10 and 11 schematically show different possibilities of constituting the chain (s) of converters compatible with a synchronization method according to the invention.
    Detailed Description The invention applies to data processing systems which comprise a control unit which controls a set of data converters of the analog / digital and / or digital / analog type, which must work synchronously. In these architectures, the control unit, generally produced by a programmable logic circuit of FPGA (Field-Programmable Gate Array) type, is designed to drive the converters in a master-slave communication scheme, by means of a bus peripheral interface, such as for example the bus called SPI bus (for Serial Peripheral Interface) which is a synchronous serial data bus widely used. The peripheral interface bus enables the exchange of data between the control unit and the converters, in particular DATA data which is the data to be converted (digital-to-analog converter) or which is the result of conversions (digital-to-analog converter) . These aspects will not be detailed further. They are well known to those skilled in the art.
    FIG. 1 illustrates a processing architecture comprising K converters controlled by a control unit UC, which are arranged in a series chain which allows their synchronization according to the invention on a common reference clock CLK. In practice the electrical distribution of this clock on the control unit and the converters is carried out using the techniques of the state of the art, to ensure synchronous distribution on all components.
    According to the invention, each converter in the chain is configured to receive a synchronization signal on an input pin IN; and transmitting a synchronization signal to a next converter on an output pin OUT. The serial chain of distribution of the synchronization signal on the converters according to the invention can thus be constituted as follows: a synchronization signal SYNC-m is emitted by the control unit UC of the converters; it is applied to the first converter in the chain and transmitted after resynchronization to the next converter and so on until the last converter in the chain. SYNC-inj is the signal received on the input pin IN of the converter of rank j in the chain; and SYNC-outj the signal transmitted by the row converter j on its output pin OUT, after detection of the SYNC-inj signal and alignment on a reference clock active edge as will be explained in the following. Finally, for the synchronization method according to the invention, the synchronization pulse has a duration (or width) which is at least equal to a clock period CLK, allowing acquisition of the active level of the synchronization signal by each converter on at least one of the rising and falling edges of a clock pulse CLK.
    It is proposed in the following to first describe the synchronization process of the converters according to the invention, and then the processes for determining the configuration parameters used by this process.
    Previously, it is necessary to specify the conventions that have been chosen: the reference clock active edge CLK on which synchronization is performed is the rising edge; the default binary value of the synchronization configuration register parameters is the null value (0). The synchronization signal is a synchronization pulse, which is a positive logic pulse. The detection of such a signal corresponds to the detection or acquisition of an active level, which is with the chosen convention, the high level. It will be possible to make the transpositions and adaptations necessary for systems using different conventions.
    FIG. 2 schematically illustrates the elements of the synchronization circuit and the synchronization configuration register provided in each converter, for the implementation of the synchronization of the converters of the chain according to the method of the invention.
    The synchronization circuit comprises sequential and combinational logic circuits. A first circuit LS1 makes it possible to provide a function for detecting and resynchronizing a synchronization signal SYNCJnj received at the input. A second circuit LS2 is provided which makes it possible to ensure an offset function of this signal, to enable phase starting of the frequency dividers of all the converters in the chain, when these converters consist of conversion cores with a working frequency. slower than the clock frequency CLK. The functions of these first two circuits LS1 and LS2 are performed in connection with the parameters defined in the REGj configuration register of the synchronization. A third circuit LS3 makes it possible to detect whether the clock edge CLK positioned by the polarity parameter is the correct edge. If it is not, it makes it possible to activate a Flagj flag bit provided in the configuration register, which is notably used in the learning phase, to precisely determine the value of the polarity parameter for the converter in question.
    LS1 circuit and parameterized polarity
    It has been seen that the parameter of polarity Sel-edgej is that which makes it possible to fix in each converter the adequate polarity of the clock edge CLK used to reliably acquire the active level of the synchronization signal.
    This parameter is used by the first circuit LS1 sequential logic and combinatorial converter. This circuit LS1 receives as input the synchronization signal Sync-inj received on the input pin IN. It outputs the synchronization signal Sync_outj which is transmitted on the output pin. This circuit LS1 is designed for the conventions on the signals indicated above, for - detecting a synchronization pulse received at the input, on a clock edge CLK which may be the rising edge or the falling edge depending on the value of the polarity parameter Salt -edgej; and - outputting a synchronization pulse which is synchronous with the reference clock, that is to say aligned with the active clock edge, which for the selected conventions is the rising edge.
    Figure 2 provides an exemplary embodiment of this circuit, by way of illustration. In this example, the circuit LS1 comprises two pairs of D flip-flops controlled in master slave mode, in series.
    The first pair performs the detection function. It is sequenced by a clock signal H1 which is generated by a logic gate, in the example an exclusive OR gate which receives on one input, the clock signal CLK, and on the other, the inverse of the parameter of polarity Sel-edge ,. When Sel-edgei = 0, we have H1 = / CLK (Figure 3); and when Sel-edgej = 1, we have H1 = CLK. Thus, the output Qm1 of the master flip-flop of the first pair takes the value of the signal SYNC-inj applied as input to (during) the low level of the clock H1; and keep its previous state on the high level; for the Qe1 output of the slave flip-flop of the first pair, it is the opposite: it takes the value of the signal SYNC-inj applied as input to (during) the high level of the clock H1 and keeps its previous state on the level low.
    The second pair performs the alignment function on a next CLK clock-active edge. In this second pair, the master flip-flop is sequenced by a clock signal H2 which is generated by a logic gate, in the example an AND gate which receives on one input, the clock signal CLK, and on the other , the Sel-edge polarity parameter ,. When Sel-edgei = 0, we thus have H2 = 0 (FIG. 3) and the Qm2 output of this flip-flop recopies its input all the time, that is to say the signal Qe1; and when Sel-edgei = 1 (Figure 4), we have H2 = CLK. The slave latch of the second pair is sequenced by the reference clock. Its Qe2 output provides the SYNC-outj synchronous synchronization signal to be outputted OUT.
    The timing diagram of FIG. 3 corresponds to the case where Sel-edge has been set to 0 for this converter, indicating that the active level (1) of the signal SYNCJnj which will be (is) received is stable at the time of a falling edge of the CLK clock.
    The timing diagram of FIG. 4 corresponds to the inverse case where Sel-edge has been set to 1: that is to say that the active level (1) of the signal SYNC-inj which will be received is stable at moment of a rising edge of the CLK clock.
    Note that in both cases detection and alignment are performed over a reference clock period.
    LS2 circuit and parameterized phase
    We have seen that the phase parameter Sel-shiftj is the parameter which makes it possible to start the frequency dividers of the converters in phase. If the converters are each formed with a single conversion core whose sampling frequency is the clock frequency CLK, this parameter is set to its default value (zero) in all the converters. When the converters are each formed with n conversion cores, n integer at least equal to 2, the sampling clock of the conversion cores is provided by a divider by a factor n of the clock frequency CLK.
    During the synchronization process, the frequency dividers of the converters will not be reset on the same active clock edge CLK, but in a delayed manner in connection with the propagation of the synchronization signal from one converter to the other. But the phase parameter ensures that they are reset on active clock edges that are considered two by two, are distant from a number that is multiple of n clock periods CLK. In this way, the sampling clocks all start in phase.
    The phase parameter is applied to the second circuit LS2 of the synchronization circuit of the converter. It receives as input the signal SYNC-outj provided by the first circuit LS1; it outputs the signal SYNC-Corej applied to reset the sampling clock generator Fej (frequency divider) used in the heart or cores of the converter.
    The circuit LS2 comprises in practice n-1 delay circuits, for example flip-flops D sequenced by the clock signal CLK, each circuit delaying a clock period CLK the signal received as input; and an input channel multiplexer and an output channel controlled by the phase parameter Sel-shiftj. This multiplexer receives as inputs the signal SYNC-Corej and the output of each of the n-1 delay circuits, and selects the input channel corresponding to the offset set by the parameter Sel-shiftj.
    The timing diagram of FIG. 5 illustrates the effects of the circuit LS2 and the phase parameter. It represents the signals of synchronization of input (SYNC-inj), output (SYNC-outj) and core (SYNC-Corej) of three successive converters of the chain, which are at n = 2 cores working at half frequency of the clock frequency CLK. In this timing diagram, the output synchronization signal SYNC-outj of the upstream converter CNj is taken as the reference. For the two converters CNj and CNj + -i, the active edges of their output signals SYNC-outj and SYNC-outj + i are separated by 2 clock periods CLK: these two signals can be applied without any offset (Sel-shiftj and Sel- shiftj + i to 0): their sampling clocks Fej will start in phase. For the two converters CNj and CNj + 2, the active edges of their output signals SYNC-outj and SYNC-outj + 2 are separated by 5 clock periods CLK, which is not multiple of 2. In the example, then adds an offset of 1 clock period CLK to the signal SYNC-outj + 2, to arrive at a multiple number of 2 (by setting the phase parameter Sel-shiftj + 2 of the converter CNj + 2 to the value 1). Thus, the clocks of these three converters Fe i, Fe i + i and Fe i + 2 will all start in phase, as illustrated in FIG. 5. The value of the phase parameter of each converter with respect to an upstream converter in the chain is thus determined.
    LS3 circuit and flag bit
    The Flagj flag bit of the configuration register of a converter makes it possible to signal to the control unit that the acquisition of the SYNC-inj input synchronization signal is not reliable, that is to say that the CLK clock edge defined by the polarity parameter for the detection of the active level of this signal (LS1 circuit) falls in an area where this signal is unstable, and therefore must be modified. FIG. 6 illustrates this situation: the active pulse front of the signal SYNC-in-i is concomitant with the clock edge CLK fixed by the parameter Sel-edgei for the detection, which at that moment is the falling edge. We are in a zone of instability ZI of the signal and the detection at this moment is not reliable. The third circuit LS3 of the synchronization circuit of the converter makes it possible to detect this situation, and to activate if necessary the Flagj flag bit.
    This circuit LS3 comprises in practice three detection stages which each receive the input synchronization signal SYNC-inj and which are sequenced from the reference clock CLK. A first stage is configured to detect an active level of the SYNC-inj signal on the CLK clock edge determined by the default value of the Sel-edgej polarity parameter. The other two stages perform this detection one on a front slightly ahead on this clock face (-Δ), the other slightly late (+ Δ). For the high frequencies that occupy us, this slight positive or negative delay is typically of the order of ten picoseconds (10'12s). If the three stages provide the same logical output value corresponding to the active level of the signal, it is because this level is well established at the time of the detection edge. If at least one provided a different logical value, it is that the signal SYNC-inj was being established at the high value (or low): one is in a zone of instability ZI. The circuit LS3 then activates the flag Flagj bit in the configuration register, in the example by setting it to the value 1. In practice the different detection stages are designed substantially like the detection stage of the circuit LS1, including in in addition to delay circuits for generating the detection fronts in advance and late on the reference clock edge. This flag bit is used in particular learning phase to change the value of the polarity parameter of the converter. Another use is described below in the synchronization process of the converter chain.
    Process of synchronization of the converter chain
    During the learning phase that will be described later, the control unit UC stores the polarity and phase parameters determined for each converter, for example in non-volatile memory. At each new power-up, the control unit programs the configuration registers of each of the converters. It then triggers the synchronization process of the converters.
    In one embodiment, this process is triggered by the transmission by the control unit of a synchronization signal SYNC-m which is synchronous with the clock CLK, i.e. the pulse is emitted on an active CLK clock face.
    By virtue of the polarity and phase parameters configured for each converter, this synchronization signal will propagate from the first converter to the last of each converter chain constituted according to the invention, by generating, in passing through each converter, a signal of synchronization of the conversion heart or cores correctly positioned so that at the end of the process, all the sampling clocks will be in phase, as illustrated for example by the timing diagram of Figure 5.
    In a variant, it can be provided that the synchronization process is triggered by the transmission by the control unit of a synchronization signal SYNC-m which is asynchronous. In this case, there is an uncertainty on the detection of the synchronization pulse by the first converter CN-ι at the input of the chain. This uncertainty is removed by providing that the control unit checks the state of the flag bit of this converter CNi: If it is not activated, it is because the active pulse level was steadily established at moment of the detection edge positioned by the polarity parameter. If enabled, the controller issues a new asynchronous sync pulse. As soon as the first converter has correctly detected the input synchronization pulse, as it outputs a synchronization signal SYNC-outi which is by synchronous construction, the synchronization process can continue in the following converters, as described supra with synchronous SYNC-m signal.
    It is advantageously provided that the synchronization output pin of the last converter of the chain is looped back to the control unit: the control unit thus has an indication of the end of the synchronization phase of the converters of each chain.
    Learning process of configuration parameters
    For each converter chain constituted by the invention, it is necessary to have a learning phase, to configure in each converter, the polarity and phase parameters for the synchronization process. After powering on the system, the converters are initialized: in particular the converter configuration registers are initialized to a default value, which is generally the null value. In all the converters we thus have: Sel-edgej = 0 and Sel-shiftj = 0. With the conventions adopted, this means that in the circuits LS1 of all the converters, the clock edge CLK for the detection is the falling edge; and the LS2 circuits of all the converters select the synchronization input channel without offset. The control unit is configured to activate a learning sequence P1, to stepwise determine the polarity parameters of the converters, as illustrated by the timing diagram of FIG. 6 and the step diagram of FIG. 7.
    The phase begins with the transmission of a synchronization signal SYNC-m (a pulse) synchronized on a clock active edge CLK. The control unit then listens to the converters to detect if a flag bit is activated.
    At the level of the converters, the sequence is as follows: The synchronization pulse sent by the control unit arrives at the input of the chain on the first CNL converter after a delay (which one does not need to know) which is fixed and depends only on the characteristics of the signal conduction path from the OUT-sync output of the control unit to the IN input of this converter (Figure 1). The converter performs the detection of the signal SYNC-ΐητ received at the input (circuit LS1) and checks (circuit LS3) if this detection is made in an instability zone ZI of this signal, (step ST-a). If this is not the case, the signal propagates to the next converter, CN2.
    But if, as illustrated in FIG. 6, the propagation delay between the output OUT-sync of the control unit and the input IN of this converter is such that the high level of the signal SYNC-i ^ is established around or at the moment of the detection edge, the falling edge in the example, which is the clock edge CLK which has the polarity defined by the default value of the polarity parameter Sel-edge-ι, the converter (by its circuit LS3) activates its flag Flagi bit (step ST-b). The control unit will then detect the activation of the Flagi flag bit of the configuration register of the converter CNi and will: reset it (step A); - set the Sel-edgei polarity parameter to 1 (step B); and - launch a new synchronization sequence P1 (step C).
    A new SYNC-m synchronous synchronization signal is then transmitted to the first converter of the chain as shown in ® in FIG. 6. This time, the polarity parameter of the first converter is appropriately positioned: the pulse of synchronization is reliably detected and the converter outputs a resynchronized pulse on the active clock edge CLK, SYNC-outi, as illustrated in sur in FIG.
    As before, the synchronization pulse arrives at the input of the second converter CN2 after a determined delay which depends only on the characteristics of the conduction path of the signal from the output of the previous converter to the input of this converter. In turn, this converter CN2 carries out the detection of the signal SYNC-in2 received in input (circuit LS1) and checks (circuit LS3) if this detection is made in a zone of instability ZI of this signal (step ST-a). If this is not the case, the signal propagates to the next converter, CN3.
    In the example, as illustrated in ® in FIG. 6, the converter detects a zone ZI of instability for the polarity defined by the parameter SEL-edge2 (circuit LS3) and activates its Flag flag bit (step ST-b).
    As seen previously, the control unit detects this, resets this Flag2 flag bit (step A); sets the value of the Sel-edge2 polarity parameter of the CN2 converter (step B) to 1; and activates a new synchronization sequence P1 (step C).
    A new SYNC-m synchronous sync signal is output, as shown in ® in Figure 6.
    This time, the synchronization signal will be correctly detected and transmitted by the first two converters already set. The delay of propagation of the synchronization pulse from the transmission to the input of the third converter is thus well determined, fixed. It is then possible to determine the appropriate polarity parameter for this third converter as previously described. And thus determines the correct polarity parameter for each successive converter in the chain, to the last.
    It is noted that this learning process is compatible with a synchronization process initiated by the control unit by a synchronization signal SYNC-m which is synchronous or asynchronous, as described above.
    Figures 8 and 9 illustrate a variant of the training sequence in which the control unit transmits an asynchronous SYNC-m synchronization signal. In this case, the control unit does not modify the polarity parameter of the first converter CNi, if its flag bit is activated; it resets the flag bit and transmits a new SYNC-m asynchronous synchronization signal and repeats this until the first converter successfully detects the signal it receives as input. The sequence is unchanged for the other converters in the chain. It should be noted that this learning process is compatible only with a synchronization process initiated by the control unit by a synchronization signal SYNC-m that is also asynchronous.
    Once all the polarity parameters have been configured, it is then possible to configure the phase parameters Sel-shiftj, the converters of the chain. Indeed, all the propagation delays of the synchronization signal in the chain become deterministic: the detections are reliable and the synchronization signals at the output OUT of the converters are all synchronized on an active clock edge CLK. It is therefore possible to determine the number of clock periods CLK separating the sync pulse edge at the output of each converter from a synchronization pulse edge at the output of another converter, and this for the entire chain. The principle is, as illustrated in FIG. 5, to measure the difference between the synchronization pulse fronts at the output of two converters, and when this difference is equal to an integer M of clock periods CLK which is not multiple of n, the value of the phase parameter of the downstream converter is set to the value p such that M + p is a multiple of n. p can therefore be set to 0, 1, ... n-1.
    This can be done in different ways for example by observing, on the oscilloscope, output signals or synchronization signals. Or by precise calculation of the propagation delay between the output OUT of a converter and the input IN of the following converter, taking into account the conductor material, its length, the characteristics of the welds, the characteristics of the input and output stages , .... or even using test sequences generated by the converters. This determination is not performed by an automatic process managed by the control unit, as for the polarity parameter. But when this determination is made, the value of the phase parameter thus determined for each converter is memorized in the control unit. These values are used to configure the converters each time the system is powered up.
    The configuration register of each converter thus comprises in practice 1 bit reserved for the flag bit, 1 bit reserved for the polarity parameter and r bits, with n = 2r, reserved for the phase parameter. The invention that has just been described easily accommodates all the forms of arrangement of converters that can be found in data processing architectures in the various application domains: serial arrangement, parallel, multi-branched tree ..., or combinations of these arrangements. Figures 10 and 11 give such examples of tree configuration (Figure 10) or mixed (Figure 11). The synchronization method and the associated learning method which have been explained apply in the same way to each of the chains of one or more converters constituted in these systems.
    It is not limited to architectures using an SPI bus as slave master interface bus between the control unit and the converters.
    权利要求:
    Claims (10)
    [1" id="c-fr-0001]
    A method of synchronizing data converters of digital / analog and / or analog / digital type on an active edge of a common reference clock CLK, characterized in that the converters form at least one serial transmission line of a synchronization signal (SYNC-m) transmitted by a control unit (UC) of the converters such that: - the converter (CNi) of rank 1 in the chain has a synchronization input (IN) connected to a synchronization control output (OUT-sync) of the control unit; each converter (CNj) of rank j greater than 1 in the chain has a synchronization input (IN) connected to the synchronization output (OUT) of the row converter j-1 (CNj-ι) in the chain; the synchronization method being characterized in that a synchronization process of the converters of the chain on a reference clock active edge activated by the control unit comprises the following steps controlled by the control unit: - initialization in each converter of a synchronization configuration register (REGj) having at least one polarity parameter (Sel-Edgej) defining as a polarity a reference clock edge (CLK) for the signal detection at the synchronization input (IN) of the converter, the polarity of the active reference clock edge or the reverse polarity; then - transmitting on the synchronization control output (SYNC-out) of the control unit of a synchronization signal (SYNC-m), which is a pulse of width at least equal to a reference clock period; the synchronization method being further characterized in that each converter in the chain is configured to perform the following steps: - a) detecting a synchronization signal (SYNCjnj) at the synchronization input (IN), on a clock edge reference signal CLK which has the polarity defined by the value of said polarity parameter (Sel-Edgej) configured in the converter, - b) alignment on the next reference clock active edge of the signal detected in step a) to provide a synchronization signal aligned with an active reference clock edge and applying said signal (SYNC-outj) to the synchronization output (OUT) of the converter.
    [2" id="c-fr-0002]
    The method of claim 1, wherein each converter (CNj) in the chain is formed of n conversion cores at a sampling frequency (Fej) which is provided by a divider by n of the reference clock frequency. , where n is a non-zero integer, and configured to carry out following step b) the following steps: c) shifting of said aligned synchronization signal obtained in step b) by a whole number of periods ( s) reference clock (CLK) defined by the value of a phase parameter (Sel-shiftj) provided by said converter configuration register; and -d) applying the obtained shifted signal (SYNC-Corej) as a reset signal of said frequency divider by n, where and said integer defined by the phase parameter has a value between 0 and n-1 inclusive.
    [3" id="c-fr-0003]
    The method of claim 1 or 2, wherein each converter in the chain is further configured to perform the following step: e) checking the stability of an active signal level (SYNC-inj) received at the input of synchronization (IN) acquired on the reference clock edge defined in step a) of detection by the value of said polarity parameter, by comparison of the first value acquired on said detection edge, with at least a second value of said signal acquired on an edge in advance, of a determined delay, on said detection edge and at least a third value of said signal acquired on a delay front, by a determined delay, on said detection edge, and if the values are not all identical, activation of a corresponding flag bit (Flagj) in said converter configuration register.
    [4" id="c-fr-0004]
    4. Method according to claim 3, characterized in that the flag bit of the converter configuration register is used by the control unit in a learning phase (P1) to change the value of the polarity parameter in each converter. a string, from a first binary value initialized by default to a second binary value, and in that in said learning phase, the control unit transmits a new synchronization signal each time a flag bit is activated in the configuration register of a converter of the chain, said learning phase allowing the configuration of the polarity parameter of each converter successively from the first input converter to the last converter of the chain.
    [5" id="c-fr-0005]
    5. Method according to claim 4, characterized in that after said learning step of the polarity parameters of the converters, a p value of the phase parameter of each converter is determined from a determination of the integer number M of reference clock periods separating the synchronization pulse edge at the output (OUT) of the converter from the sync pulse edge at the output of an upstream converter in the chain, and the p value of the phase parameter is such that the sum M + p is a multiple of n.
    [6" id="c-fr-0006]
    6. Method according to any one of the preceding claims, characterized in that the synchronization signal (SYNC-m) transmitted by the control unit is a signal synchronized on an active edge of the reference clock CLK.
    [7" id="c-fr-0007]
    7. Method according to any one of the preceding claims, characterized in that the synchronization process of the converters of the chain comprises the transmission by the control unit of an asynchronous synchronization signal (SYNC-m) and in that that the control unit transmits a new asynchronous synchronization signal each time the flag bit is activated in the configuration register of the converter (CN-ι) of rank 1 at the input of the string.
    [8" id="c-fr-0008]
    An analog / digital or digital / analog data converter comprising an input (IN) for receiving a synchronization signal and an associated synchronization circuit for synchronizing the converter to an active edge of a reference clock signal (CLK), characterized in that the converter comprises a configuration register (REGj) accessible read / write via an interface bus by a control unit (UC) of the converter, said register comprising at least one polarity parameter ( SEL-edgej) for defining a reference clock edge polarity for the detection of said synchronization signal (SYNC-inj) received at the input, and in that the synchronization circuit comprises a detection circuit (LS1) which comprises: a) a detection stage of an active level of said synchronization signal received on a reference clock edge (CLK) which has the polarity defined by the binary value of said polarity ameter (Sel_edgej), and - b) an alignment stage of the synchronization signal detected by the detection stage which outputs (Qe2) a synchronization signal aligned with a reference clock active edge (CLK ), and said aligned synchronization signal (SYNC-outj) is applied to a synchronization output (OUT) of the converter.
    [9" id="c-fr-0009]
    A converter according to claim 8, which is formed of n conversion cores at a sampling frequency (Fej) which is provided by an n-divider of the reference clock frequency, where n is a non-zero integer, characterized in that the configuration register comprises another phase parameter (Sel-shiftj) whose value is equal to an integer between 0 and n-1, including terminals, and in that the synchronization circuit comprises a circuit shifter (LS2) configured to apply to the output signal (Qe2) of said alignment stage an offset of an integer number of reference clock periods (CLK), which is given by the value of said phase parameter , and apply the obtained shifted signal (SYNC-Corej) as a reset signal of the frequency divider by n.
    [10" id="c-fr-0010]
    The converter according to claim 8 or 9, wherein the synchronization circuit comprises an input level stability control (SYNC-inj) circuit (LS3), which comprises a first stage of acquiring a level of said synchronization signal (SYNC-inj) on a reference clock edge having the polarity defined by the value of said polarity parameter, at least a second acquisition stage of a level of said synchronization signal (SYNC-inj) on an edge in advance, by a determined delay, on said reference clock edge defined in the first stage and at least a third stage of acquisition of a level of said signal synchronization (SYNC-inj) on a delay front, by a determined delay, on said reference clock edge defined in the first stage, and if the values provided by the different stages are not all identical, the output of said circuit checking activates a flag bit (Flagj) in said converter configuration register.
    类似技术:
    公开号 | 公开日 | 专利标题
    EP3375092B1|2019-09-04|Method for the synchronisation of data converters by a signal transmitted in close proximity
    EP0645888B1|1998-03-04|Digital delay line
    EP0260632A1|1988-03-23|Device for the resynchronization of one or more binary data trains at an identical or sub-multiple rate on a synchronous reference clock signal
    EP1841066B1|2010-06-09|Estimating the jitter of a clock signal
    FR2590425A1|1987-05-22|APPARATUS FOR ADJUSTING THE PHASES OF DATA SIGNALS APPLIED TO A DATA USE CIRCUIT
    FR2589585A1|1987-05-07|CIRCUIT FOR SYNCHRONIZING A MULTI-CHANNEL CIRCUIT CONTROLLER
    EP0887653A1|1998-12-30|Functional test method and circuit containing means for implementing said method
    WO2013030504A1|2013-03-07|Method for synchronising a server cluster and server cluster implementing said method
    EP1114377A1|2001-07-11|Modular data collection system
    EP0249930A1|1987-12-23|Method for synchronizing two binary streams
    FR2858434A1|2005-02-04|Programmable clock synchronizer for computer system, configures synchronization controllers operating in different clock domain responsive to sync pulse that is sampled in clock domains
    EP0064923B1|1986-04-09|System for the phase synchronization of digital data streams, and its application to the commutation of said data streams
    EP1445865A1|2004-08-11|Frequency divider with funnel structure
    EP0939372B1|2003-01-15|Device for production testing of the dynamic characteristics of components using serial transmission
    FR2669751A1|1992-05-29|DIGITAL PULSE GENERATOR AND RELATED METHODS.
    FR2905040A1|2008-02-22|METHOD FOR PRODUCING A DIGITAL WORD REPRESENTATIVE OF A NON-WHOLE REPORT BETWEEN THE RESPECTIVE PERIODS OF TWO SIGNALS, AND CORRESPONDING DEVICE
    EP0696116A1|1996-02-07|Method for sampling a digital signal
    EP0174880B1|1989-09-20|Arrangement for the localisation of the transitions of a data signal in respect of a clock signal, and synchronisation mechanism using such an arrangement
    FR2820254A1|2002-08-02|FRACTIONAL FREQUENCY DIVISION OF A DIGITAL SIGNAL
    FR2770704A1|1999-05-07|Phase lock loop method
    EP3716523A1|2020-09-30|Method for synchronising digital data sent in series
    FR3025901A1|2016-03-18|DEVICE FOR GENERATING A CLOCK SIGNAL BY FREQUENCY MULTIPLICATION
    FR3111249A1|2021-12-10|Method for synchronizing analog-to-digital or digital-to-analog converters, and corresponding system.
    FR2789247A1|2000-08-04|Modular electronic circuit having improved synchronization for use in processing and transmission of electronic data at very high frequencies has modules comprising cells with means for coordinating clock signals
    FR2703534A1|1994-10-07|Device for digital control of a variable digital oscillator
    同族专利:
    公开号 | 公开日
    CN108352829A|2018-07-31|
    KR20180079339A|2018-07-10|
    CN108352829B|2022-02-25|
    AU2016354402B2|2020-11-19|
    FR3043477B1|2017-11-24|
    WO2017080925A1|2017-05-18|
    TW201740683A|2017-11-16|
    JP2018537031A|2018-12-13|
    SG11201803662SA|2018-06-28|
    TWI690161B|2020-04-01|
    JP6898319B2|2021-07-07|
    US10320406B2|2019-06-11|
    CA3004791A1|2017-05-18|
    AU2016354402A1|2018-06-21|
    US20180323794A1|2018-11-08|
    EP3375092B1|2019-09-04|
    EP3375092A1|2018-09-19|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    EP0798863A1|1996-03-29|1997-10-01|Koninklijke Philips Electronics N.V.|Analogue/digital converter having a high sampling rate|
    DE102004050648A1|2004-10-18|2006-04-20|Rohde & Schwarz Gmbh & Co. Kg|Arrangement for the synchronous output of analog signals generated in two or more digital / analog converters|
    US5436628A|1993-09-13|1995-07-25|Intel Corporation|Programmable frequency timing generator with phase adjust|
    JP2985957B1|1998-09-11|1999-12-06|松下電器産業株式会社|Phase comparator and digital phase locked loop|
    JP3657188B2|2000-10-31|2005-06-08|Ｎｅｃエレクトロニクス株式会社|Device and its operating method|
    JP2002232402A|2001-02-07|2002-08-16|Toyo Commun Equip Co Ltd|Automatic phase inversion circuit for clock signal|
    US7328299B2|2004-11-23|2008-02-05|Atmel Corporation|Interface for compressed data transfer between host system and parallel data processing system|
    CN101098220B|2006-06-29|2010-08-18|中兴通讯股份有限公司|Digital phase-locked loop based clock synchronization method and system thereof|
    CN101546207B|2008-03-28|2011-08-10|盛群半导体股份有限公司|Clock signal switching circuit|
    WO2009148457A1|2008-06-06|2009-12-10|Lsi Corporation|Systems and methods for synchronous, retimed analog to digital conversion|
    US7728753B2|2008-10-13|2010-06-01|National Semiconductor Corporation|Continuous synchronization for multiple ADCs|
    CN102495912B|2011-10-26|2013-11-20|电子科技大学|Multi-channel high-speed data acquisition system with synchronous correction function|
    JP2014017590A|2012-07-06|2014-01-30|Fujitsu Ltd|Electronic circuit and communication method|
    US9037893B2|2013-03-15|2015-05-19|Analog Devices, Inc.|Synchronizing data transfer from a core to a physical interface|
    WO2015162763A1|2014-04-24|2015-10-29|三菱電機株式会社|Network system|
    JP6213538B2|2015-09-24|2017-10-18|横河電機株式会社|Signal processing circuit|FR3094593B1|2019-03-29|2021-02-19|Teledyne E2V Semiconductors Sas|Method of synchronizing digital data sent in series|
    FR3100068B1|2019-08-20|2021-07-16|Teledyne E2V Semiconductors Sas|Method for synchronizing analog data at the output of a plurality of digital / analog converters|
    FR3111249A1|2020-06-09|2021-12-10|Teledyne E2V Semiconductors Sas|Method for synchronizing analog-to-digital or digital-to-analog converters, and corresponding system.|
    法律状态:
    2016-10-28| PLFP| Fee payment|Year of fee payment: 2 |
    2017-05-12| PLSC| Publication of the preliminary search report|Effective date: 20170512 |
    2017-10-26| PLFP| Fee payment|Year of fee payment: 3 |
    2018-06-22| CD| Change of name or company name|Owner name: TELEDYNE E2V SEMICONDUCTORS SAS, FR Effective date: 20180523 |
    2019-11-25| PLFP| Fee payment|Year of fee payment: 5 |
    2021-08-06| ST| Notification of lapse|Effective date: 20210705 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1560739A|FR3043477B1|2015-11-10|2015-11-10|METHOD FOR SYNCHRONIZING DATA CONVERTERS BY A SIGNAL TRANSMITTED FROM CLOSE TO NEAR|FR1560739A| FR3043477B1|2015-11-10|2015-11-10|METHOD FOR SYNCHRONIZING DATA CONVERTERS BY A SIGNAL TRANSMITTED FROM CLOSE TO NEAR|
    AU2016354402A| AU2016354402B2|2015-11-10|2016-11-04|Method for synchronising data converters by means of a signal transmitted from one to the next|
    US15/774,455| US10320406B2|2015-11-10|2016-11-04|Method for synchronizing data converters by means of a signal transmitted from one to the next|
    CA3004791A| CA3004791A1|2015-11-10|2016-11-04|Procede de synchronisation de convertisseurs de donnees par un signal transmis de proche en proche|
    JP2018523437A| JP6898319B2|2015-11-10|2016-11-04|A method of synchronizing data converters with signals transmitted from one data converter to the next.|
    KR1020187013192A| KR20180079339A|2015-11-10|2016-11-04|How to Synchronize Data Converters with Sequentially Transmitted Signals|
    CN201680065692.5A| CN108352829B|2015-11-10|2016-11-04|Method for synchronizing analog-to-digital data converters and/or digital-to-analog data converters|
    EP16793826.5A| EP3375092B1|2015-11-10|2016-11-04|Method for the synchronisation of data converters by a signal transmitted in close proximity|
    SG11201803662SA| SG11201803662SA|2015-11-10|2016-11-04|Method for synchronising data converters by means of a signal transmitted from one to the next|
    PCT/EP2016/076689| WO2017080925A1|2015-11-10|2016-11-04|Method for synchronising data converters by means of a signal transmitted from one to the next|
    TW105136296A| TWI690161B|2015-11-10|2016-11-08|Method for synchronizing data converters with a signal transmitted from converter to converter|
    [返回顶部]