Digital time converter and phase-locked loop
The digital time converter addresses precision and calibration issues in existing converters by using a sigma-delta architecture with integrator and feedback circuits, enhancing precision and reducing complexity in all-digital phase-locked loops.
Patent Information
- Application Number
- FR2021010080
- Authority / Receiving Office
- FR · FR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2021-09-24
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2041-09-24
AI Technical Summary
Existing digital time converters, particularly those used in all-digital phase-locked loops, suffer from limited precision in measuring duration due to stage propagation time, require calibration, and necessitate additional processing for normalization, often relying on unnecessary clock signals.
A digital time converter design incorporating a sigma-delta type architecture with integrator and feedback circuits that utilize a clock signal for pulse integration and quantization, achieving precise duration measurement without additional calibration and clock signals, and leveraging existing low-pass filtering in the phase-locked loop.
The solution provides enhanced precision, reduces complexity, eliminates the need for calibration, and improves signal-to-noise ratio by moving quantization noise to high frequencies, making it suitable for all-digital phase-locked loops.
Smart Images

Figure 00000047_0000 
Figure 00000047_0001 
Figure 00000048_0000
Abstract
Description
Title of the invention: Digital time converter and phase-locked loop Technical field
[0001] The present description relates generally to electronic circuits, and more particularly to converters of a duration into a digital signal representative of said duration, or digital time converters, and phase-locked loops comprising such digital time converters. Prior art
[0002] Digital time converters are known. These converters are configured to convert a duration into a digital signal representative of the value of this duration.
[0003] For example, a known digital time converter comprises a succession of stages forming a delay line. The state of each stage of the delay line is sampled at each active edge of a timing signal of the converter. Furthermore, the delay line receives as input an edge of a first signal, which propagates in the delay line. At the following active edge of the timing signal, the sampled state of the delay line is then representative of a duration separating the supply of the edge of the first signal to the delay line from the following active edge of the timing signal. A disadvantage of such a digital time converter, called a flash digital time converter, is that the measurement accuracy of the duration separating the edge of the first signal from the following active edge of the timing signal is limited by the propagation time of a stage of the converter.Such a converter also requires calibration of the propagation time in each stage of the converter, so that each stage introduces the same delay. Furthermore, when such a converter is implemented in an all-digital phase-locked loop (ADPLL), it is usually necessary to perform normalization of the converter output, which involves additional undesirable processing. Summary of the invention
[0004] There is a need to overcome all or part of the disadvantages of known digital time converters.
[0005] For example, there is a need to overcome all or part of the drawbacks of known digital time converters intended for implementation in an all-digital phase-locked loop.
[0006] For example, there is a need for a digital time converter which, when implemented in an all-digital phase loop, provides an output in the form of a normalized phase deviation.
[0007] For example, there is a need for a digital time converter which, when implemented in an all-digital phase-locked loop, does not require a clock signal other than the clock signals respectively received and supplied by the phase-locked loop.
[0008] For example, there is a need for a digital time converter that is simple and compact to implement.
[0009] For example, there is a need for a time-to-digital converter that, when implemented in an all-digital phase-locked loop, takes advantage of the low-pass filtering implemented in the loop.
[0010] For example, there is a need for a digital time converter that is particularly suitable for implementation in an all-digital phase-locked loop, and that is also usable in any other electronic application or system where digital time conversion is required.
[0011] One embodiment overcomes all or part of the drawbacks of known digital time converters, for example when they are implemented in an all-digital phase-locked loop.
[0012] One embodiment provides a digital time converter comprising: a first circuit configured to provide, at each active edge of a first signal, a first pulse of duration determined by a difference between said active edge of the first signal and an immediately successive active edge of a second signal; an integrator circuit configured, at each first pulse, to integrate: - the first impulse, - a second pulse starting after the first pulse and in synchronism with a clock signal, and - a third pulse starting after the second pulse and in synchronism with the clock signal, a second circuit configured to generate, for each first pulse, the corresponding second pulse; a third circuit configured to quantize an output signal of the integrator circuit on one bit and, at the start of each third pulse, store said binary quantization on a first bit; and a fourth circuit configured to generate, for each first pulse, the corresponding third pulse from the first bit.
[0013] According to one embodiment, the second circuit is configured to receive the clock signal, and to generate, at least in part from the clock signal, at least one third signal configured to control the memorizations by the third circuit and the generations of the third impulses by the fourth circuit.
[0014] According to one embodiment, the first circuit is configured to provide a fourth signal indicating a start and an end of each first pulse, the second circuit being configured to receive the fourth signal and to generate the third signal from the clock signal and the fourth signal.
[0015] According to one embodiment, the first and second circuits are configured so that the second pulses place a conversion zero of the converter in the middle of the dynamic range of the converter.
[0016] According to one embodiment, the fourth circuit is configured so that the integration of each third pulse implements a subtraction of the binary quantization stored at the start of said third pulse.
[0017] According to one embodiment, the first and second circuits are configured so that each second pulse causes, on the output signal of the integrator circuit, a variation of amplitude identical and of opposite sign to an amplitude variation that would cause a first pulse of duration determined by half the dynamic range of the converter.
[0018] According to one embodiment, the converter is of the sigma-delta type and the fourth circuit is configured so that each third pulse implements a negative sigma-delta feedback.
[0019] According to one embodiment: the first, second and third impulses are signed; the first and second circuits are configured so that a sign of the second pulses is opposite to a sign of the first pulses and a maximum duration of the first pulses that the converter can convert multiplied by an amplitude of the first pulses is equal to twice a duration of the second pulses multiplied by an amplitude of the second pulses; and the fourth circuit is configured so that a sign of each third pulse is determined by a state of the first of the bit and an amplitude of the third pulses multiplied by a duration of the third pulses is equal to the duration of the second pulses multiplied by the amplitude of the second pulses.
[0020] According to one embodiment, the second and third pulses each have a duration that is a multiple of a period of the clock signal, preferably equal to a period of the clock signal.
[0021] According to one embodiment, an output of said converter is at least partly determined by the first bit.
[0022] According to one embodiment, the converter is of the multi-stage noise-shaping sigma-delta type, the converter comprising a first stage comprising the integrator circuit and the first, second, third and fourth circuits, and a second stage configured to receive an output signal from the first stage integrator circuit and to provide a second stage output digital signal from the first stage integrator circuit output signal, an output of said converter being determined at least in part by the first bit and the second stage output digital signal.
[0023] According to one embodiment, the second stage is configured to: generate, after each third pulse, a fourth pulse starting in synchronism with the clock signal; integrating, during each fourth pulse, a sum of the output signal of the integrator circuit of the first stage and a feedback signal, quantizing on one bit a result of said integration and storing on a second bit said binary quantization at the end of the fourth pulse; generating, during each third pulse, a fifth pulse from the second bit, and the feedback signal by integrating said fifth pulse; and storing the second bit on a third bit at each active edge of the first signal, the digital output signal of the second stage being generated from the second and third bits or corresponding to the second and third bits.
[0024] According to one embodiment, the second stage comprises: a fifth circuit configured to generate, after each third pulse, a fourth pulse starting in synchronism with the clock signal; an integrator circuit configured to integrate, during every fourth pulse, a sum of the output signal of the first stage integrator circuit and a feedback signal; a sixth circuit configured to quantize on one bit an output of the integrator circuit of the second stage and store said binary quantization on a second bit, at the end of each fourth pulse; a seventh circuit configured, at each third pulse, to provide said feedback signal by integrating, during said third pulse, a fifth pulse determined from the second bit; and a circuit configured to store the second bit on a third bit at each active edge of the first signal, the digital output signal of the second stage being generated from the second and third bits or corresponding to the second and third bits.
[0025] According to one embodiment, the converter comprises a first input configured to receive the first signal and a second input configured to receive the clock signal, and: - the second signal and the clock signal are confused, or - the converter comprises a frequency divider circuit, preferably by two, configured to receive the clock signal and to provide the second signal.
[0026] Another embodiment provides an all-digital phase-locked loop configured to receive a signal at a first frequency and to provide a signal at a second frequency equal to N / D times the first frequency, with N and D being two positive values, the phase-locked loop comprising a converter as described above, the first input of the converter being configured to receive the signal at the first frequency and the second input of the converter being configured to receive the signal at the second frequency.
[0027] According to one embodiment, the phase-locked loop comprises a circuit configured to provide a first digital signal incremented by the value N at the first frequency; a circuit configured to provide a second digital signal incremented by the value D at the second frequency and a third digital signal corresponding to a storage of the second digital signal clocked at the first frequency; a digital filter configured to receive a subtraction of an output of the converter and the third digital signal from the first digital signal and to control an oscillator configured to provide the signal at the second frequency, wherein the second signal from the converter is merged with the signal at the second frequency.
[0028] According to one embodiment, the phase-locked loop comprises: either a circuit configured to provide a first digital signal incremented by the value N at the first frequency, and a circuit configured for a second digital signal incremented by the value D / 2 at the second frequency, with D even, or a circuit configured to provide a first digital signal incremented by the value 2N at the first frequency and a circuit configured to provide a second digital signal incremented by the value D at the second frequency; a circuit configured to provide a third digital signal corresponding to a storage of the second digital signal clocked by the second signal from the converter; a digital filter configured to receive a subtraction of an output of the converter and the third digital signal from the first digital signal and to control an oscillator configured to provide the signal at the second frequency, the second signal from the converter being obtained by dividing by two the frequency of the signal at the second frequency. Brief description of the drawings
[0029] These and other features and advantages will be set forth in detail in the following description of particular embodiments made without limitation in relation to the attached figures among which:
[0030] [Fig.l] represents, very schematically and in the form of functional blocks, an embodiment of a digital time converter;
[0031] [Fig.2] schematically represents details of an exemplary embodiment of the digital time converter of [Fig.l];
[0032] [Fig.3] illustrates, by means of timing diagrams, the operation of the converter of [Fig.2];
[0033] [Fig.4] represents, schematically and in the form of blocks, an embodiment of a fully digital phase-locked loop;
[0034] [Fig.5] schematically represents details of an exemplary embodiment of the digital time converter of [Fig.l], in a case where the converter is implemented in a phase-locked loop of the type of that of [Fig.4];
[0035] [Fig.6] illustrates, by means of timing diagrams, the operation of the converter of [Fig.5];
[0036] [Fig.7] schematically represents details of another exemplary embodiment of the digital time converter of [Fig.1], in a case where the converter is implemented in a phase-locked loop of the type of that of [Fig.4];
[0037] [Fig.8] illustrates, by means of timing diagrams, the operation of the converter of [Fig.7];
[0038] [Fig.9] illustrates an alternative embodiment of the converter of [Fig.l];
[0039] [Fig. 10] illustrates an example of implementation of a part of the converter of the previous figures;
[0040] [Fig.l 1] illustrates an example of implementation of another part of the converter of the previous figures;
[0041] [Fig. 12] illustrates an example of implementation of yet another part of the converter of the preceding figures;
[0042] [Fig. 13] illustrates an example of implementation of yet another part of the converter of the preceding figures; and
[0043] [Fig. 14] illustrates a detailed embodiment of the variant embodiment of [Fig.9], Description of the embodiments
[0044] The same elements have been designated by the same references in the different figures. In particular, the structural and / or functional elements common to the different embodiments may have the same references and may have identical structural, dimensional and material properties.
[0045] For the sake of clarity, only the steps and elements useful for understanding the described embodiments have been shown and are detailed.
[0046] Unless otherwise specified, when referring to two elements connected to each other, this means directly connected without intermediate elements other than conductors, and when referring to two elements connected (in English "coupled") to each other, this means that these two elements can be connected or be connected by means of one or more other elements.
[0047] In the following description, when reference is made to absolute position qualifiers, such as the terms "front", "back", "top", "bottom", "left", "right", etc., or relative position qualifiers, such as the terms "above", "below", "upper", "lower", etc., or to orientation qualifiers, such as the terms "horizontal", "vertical", etc., reference is made unless otherwise specified to the orientation of the figures.
[0048] Unless otherwise specified, the expressions "about", "approximately", "substantially", and "of the order of" mean to within 10%, preferably to within 5%.
[0049] In the present application, unless otherwise indicated, the term "pulse" refers, for example, to a square wave variation in the level of a signal relative to an initial or resting state, also called zero level, of the signal. Unless otherwise indicated, the term "amplitude of a pulse" refers, for example, to the absolute value of the level of the signal during the pulse relative to its zero level, i.e., the absolute value of the level of the pulse, it being understood that a pulse may have a negative or positive level relative to the zero of the signal. A pulse with a negative level will be called a "negative pulse" and a pulse with a positive level will be called a "positive pulse". For example, a negative pulse of amplitude A corresponds to a pulse of level equal to -A and a positive pulse of amplitude A corresponds to a pulse of level equal to +A.
[0050] Unless otherwise indicated, the "dynamic" of a time-to-digital converter is, for example, the difference between the largest value and the smallest value that the converter can measure.
[0051] A sigma-delta type digital time converter is proposed here for measuring a duration between each active edge of a first signal and an active edge of a second signal following the active edge of the first signal.
[0052] According to one embodiment, at each active edge of the first signal, the converter is configured to successively integrate, with an integrator circuit, a first pulse representative of the duration to be measured, a second pulse configured to center an output dynamic of the integrator circuit, and a third pulse configured to implement the negative feedback of the sigma-delta loop, the converter being further configured to quantize and store on a bit a output of the integrator circuit, for example, at the start of every third pulse.
[0053] In other words, each second pulse is configured to cause a variation in the output signal of the integrator of identical amplitude and opposite sign to a variation in the output signal of the integrator caused by a first pulse corresponding to a duration to be measured equal to half the dynamic range of the converter. In other words, the second pulses are configured to place the conversion zero of the converter in the middle of the dynamic range of the converter.
[0054] For example, the converter is configured so that a sign of the first pulses is opposite that of the second pulses and so that a maximum duration of the first pulses that the converter can convert multiplied by an amplitude of the first pulses is equal to twice a duration of the second pulses multiplied by an amplitude of the second pulses.
[0055] Furthermore, in other words, each third pulse is for example configured to subtract at the input of the integrator circuit the stored result of the quantization of the output of the integrator circuit carried out at the start of the third pulse, so that after the integration of the third pulse, the output of the integrator circuit only comprises the residual quantization error.
[0056] For example, the converter is configured so that the sign of each third pulse is determined by the state of the bit corresponding to the storage of the quantization of the output of the integrator circuit, and so that an amplitude of the third pulses multiplied by a duration of the third pulses is equal to the amplitude of the second pulses multiplied by the duration of the second pulses.
[0057] In other words, a digital time converter of the sigma-delta type is proposed here, operating, at each conversion cycle, according to at least three phases successively comprising a first phase during which the duration to be converted is integrated, a second phase during which a conversion zero is centered in the middle of the dynamics of the converter, and a third phase during which the negative sigma-delta feedback is implemented.
[0058] [Fig.l] represents, very schematically and in the form of functional blocks, an embodiment of a digital time converter 1 of the sigma-delta type.
[0059] The converter 1 comprises a first circuit CL. The circuit Cl is configured to provide, at each active edge, for example at each rising edge, of a signal SI, a pulse PL. The duration of the pulse PI is determined by a difference between the active edge of the signal SI and an active edge, for example a rising edge, immediately following a signal S2. In other words, the circuit Cl is configured to provide, at each active edge of the signal SI, a pulse PI having a duration representative of the time separating this edge of the SI signal from the following active edge of the S2 signal. Each PI pulse is, for example, a pulse starting at an active edge of the SI signal and ending at the immediately following active edge of the S2 signal.
[0060] In the remainder of the description, Dmax is the maximum duration of the PI pulses that the converter can convert. For example, a PI pulse having a duration Dmax corresponds to a duration to be measured equal to the maximum duration that the converter can measure, i.e. to the maximum difference between an active edge of the signal SI and a following active edge of the signal S2 that the converter can measure.
[0061] In the remainder of the description, it is considered by way of example that the PI pulses all have a duration less than or equal to Dmax, or, in other words, that the durations to be measured with the converter 1 are all included in the dynamic range of the converter.
[0062] According to one embodiment, the converter 1 is configured to convert the duration between each active edge of the signal SI and an immediately successive active edge of the signal S2 into a digital output signal of the converter 1.
[0063] As an example, the circuit C1 is configured to receive the signals S1 and S2 and to provide the pulses P1.
[0064] According to one embodiment, at each active edge of the signal SI, the circuit C1 is configured to generate a signal or pulse CMD1. The duration of the pulse CMD1 is determined by the duration between the edge of the signal SI and the immediately following edge of the signal S2. The duration of the pulse CMD1 determines the duration of the pulse PI, or, in other words, the pulse PI is determined from the pulse CMD1. In other words, each pulse CMD1 indicates the start of a corresponding pulse PI and the end of this pulse PL. Each pulse CMD1 is, for example, a pulse starting at an active edge of the signal SI and ending at the immediately following active edge of the signal S2. Each pulse PI is, for example, a pulse starting at the same time as a corresponding pulse CMD1, and ending at the same time as this pulse CMD1.
[0065] By way of example, the circuit C1 comprises a flip-flop, not shown in [Fig.l], receiving the signals SI and S2 on two of its inputs and providing the pulses CMD1. Preferably, the flip-flop is an RS flip-flop of which an initialization input, i.e. the input S of the flip-flop, is configured to receive the signal SI, of which the reset input, i.e. the input R of the flip-flop, is configured to receive the signal S2, and of which an output, for example the output Q of the flip-flop, is configured to provide the signal CMD1. Preferably, the input S, respectively R, of the flip-flop is sensitive to the active edges of the signal SI, respectively S2. Of course, the person skilled in the art is able to replace this flip-flop with any other circuit making it possible to generate, from the signals S1 and S2, the CMD1 pulses. For example, the person skilled in the art is able to replace the RS flip-flop of the C1 circuit with a D-type flip-flop, or with a circuit other than a flip-flop.
[0066] The converter 1 further comprises an integrator circuit INT. The circuit INT is configured to integrate pulses that it receives, and to provide an output signal RES1 corresponding to the result of this integration. More particularly, at each first pulse PI, the circuit INT is configured to integrate the pulse PI, then a pulse P2 corresponding to, or associated with, this pulse PI, then a pulse P3 corresponding to, or associated with, this pulse PL.
[0067] For each PI pulse, the associated P2 pulse starts after the PI pulse, in synchronism with a clock signal clk, and the associated P3 pulse starts after the P2 pulse, also in synchronism with the clk signal. For example, the P2 and P3 pulses start in synchronism with the rising edges of the clk signal. According to another example, the P2 and P3 pulses start in synchronism with the falling edges of the clk signal. According to one embodiment, the P2 and P3 pulses each have multiple durations, preferably integers, of a period of the clk signal, for example durations equal to a period of the clk signal.
[0068] According to one embodiment, the frequency of the signal clk is sufficiently large so that the conversion by the converter of a duration triggered by an active edge of the signal S1 is completed before the start of the following conversion caused by the following edge of the signal SL. For example, a multiplicative factor between the frequency of the signal clk and that of the signal S1 is determined so that, following each active edge of the signal SI, the converter 1 performs an update of its output before a new active edge of the signal S1 occurs. The determination of this multiplicative factor is within the scope of the person skilled in the art.
[0069] A circuit C2 of the converter 1 is configured to generate, for each pulse PI, the corresponding pulse P2. For example, the circuit C2 is configured to receive the signals clk and CMD1, and to provide the corresponding pulses P2. For example, the circuit C2 is configured to generate the pulses P2 from the signals clk and CMD1.
[0070] The circuit C2 is configured so that the pulses P2 center an output dynamic of the integrator INT. In other words, the circuits C2 and Cl are configured so that each pulse P2 causes a variation of the signal RES 1 of identical amplitude and opposite sign to a variation of the signal RES1 caused by a pulse PI corresponding to a duration to be measured equal to half the dynamic of the converter. In other words, the circuits C2 and Cl are configured so that the second pulses place the conversion zero of the converter in the middle of the dynamic of the converter, that is to say so that the second pulses P2 produce, during their integrations by the INT circuit, an offset or bias which places the zero of the conversion dynamics in the middle of the maximum duration that the converter 1 can measure.
[0071] We call Al the amplitude of the PI pulses, A2 the amplitude of the P2 pulses and K2*Tclk the duration of the P2 pulses, with Al and A2 strictly positive real values, K2 a positive and non-zero integer, and Tclk the duration of a period of the clk signal.
[0072] According to one embodiment, the PI and P2 pulses are signed. The circuits C1 and C2 are then configured so that the PI and P2 pulses are of opposite sign and that the product of the maximum duration Dmax of the PI pulses by the amplitude Al of the PI pulses is equal to twice the product of the duration K2*Tclk of the P2 pulses by the amplitude A2 of the P2 pulses. In other words, the circuits C1 and C2 are configured so that the PI and P2 pulses are of opposite sign and that Dmax*Al = 2*K2*Tclk*A2. For example, the PI pulses are positive and therefore each have a level equal to Al and the P2 pulses are negative and therefore each have a level equal to -A2. For example, when Al = 2*A2 and K2 = 1, then the duration Dmax of the PI pulses is equal to Tclk.In such an example, if each PI pulse starts at an active edge of signal SI and ends at the next active edge of signal S2, then the maximum time that the converter can measure between each active edge of signal SI and the next active edge of signal S2 is equal to Dmax.
[0073] The present description is not limited to the above example, and the person skilled in the art is able to provide other values of K2, A1 and A2 allowing the operation described above.
[0074] A circuit C3 of the converter 1 is configured to quantize on one bit the output signal RES1 of the integrator circuit INT, and to memorize, at the start of each pulse P3, this binary quantization of the signal RES1 on one bit OUT1.
[0075] According to one embodiment, the circuit C3 is configured to provide a bit compl corresponding to the binary quantization of the signal RES1, and to update the bit OUT1 from the bit compl at the start of each pulse P3. More particularly, the circuit C3 is configured to compare the signal RES 1 to a level ref, preferably equal to half the output dynamic of the circuit INT, and to provide the bit compl to a first binary state when the signal RES1 is greater than the reference level ref, and to a second binary state when the signal RES 1 is less than the reference level ref. For example, the first binary state of the signal compl corresponds to a high level of the signal compl, and the second binary state of the signal compl corresponds to a low level of the signal compl.For example, the ref level or potential is equal to half the difference between a high supply potential Vdd and a low supply potential, for example ground or Vss=-Vdd, of converter 1.
[0076] According to one embodiment, the memorizations implemented by the circuit C3, that is to say the updates of the state of the output bit OUT1 of the circuit C3, are controlled by a signal, or pulses, CMD3. The signal CMD3 has active edges synchronized, or coincident, with the beginnings of the pulses P3. For example, each pulse P3 begins at the same time as an active edge of the signal CMD3, or, in other words, at the same time as a pulse CMD3. Preferably, the durations of the pulses CMD3 and the pulses P3 are equal. For example, the pulses P3 are determined from the signal CMD3, or, in other words, the signal CMD3 controls the generation or supply of the pulses P3.For example, the circuit C3 comprises a flip-flop, for example of type D, which samples the result of the complete binary quantization of the signal RES1 with respect to the level ref at each start of a pulse P3, for example at each start of a pulse CMD3, and which provides the bit OUT1 as output.
[0077] According to one embodiment, the circuit C2 is configured to provide the signal CMD3. According to one embodiment, the circuit C2 is configured to provide the signal CMD3 at least in part from the signal clk. According to one embodiment, the circuit C2 is configured to provide the signal CMD3 from the signal clk and the signal CMD1.
[0078] A circuit C4 of the converter 1 is configured to generate, for each pulse PI, the corresponding pulse P3 from the bit OUT1.
[0079] According to one embodiment, the circuit C4 is configured so that the integration of each pulse P3 implements a subtraction of the binary quantization stored at the start of this third pulse P3. In more detail, the circuit C4 is configured so that, for each pulse PI, the integration of the associated pulse P3 by the circuit INT corresponds to a subtraction, on the signal RES1, of the binary quantization stored at the start of the pulse P3 by the circuit C3. In other words, the circuit C4 is configured so that each pulse P3 implements the sigma-delta negative feedback, that is to say the sigma-delta type negative feedback implemented in the converter 1.
[0080] According to one embodiment, the circuit C4 is configured to generate the pulses P3 from the bit OUT1 and the signal CMD3. In other words, the supply of the pulses P3 is controlled by the signal CMD3.
[0081] According to one embodiment, the P3 pulses are signed like the P1 and P2 pulses. The sign of each P3 pulse is then determined by the value (or state) of the OUT2 bit. For example, when the OUT2 bit is in a first binary state, the P3 pulse then has a negative level equal to -A3, and, when the OUT2 bit is in a second binary state, the P3 pulse then has a positive level equal to +A3. A3 is a strictly positive real value corresponding to the amplitude of each P3 pulse and being identical for all the P3 pulses. Because each P3 pulse has a amplitude equal to A3, the signal P3 has a peak-to-peak amplitude equal to 2*A3.
[0082] K3*Tclk is the duration of the P3 pulses, with K3 a positive and non-zero integer. According to one embodiment, the circuit C4 is configured so that the product of the duration K3*Tclk of the P3 pulses by the amplitude A3 of the P3 pulses is equal to the product of the duration K2*Tclk of the P2 pulses by the amplitude A2 of the P2 pulses. In other words, the circuit C4 is configured so that K3*A3 = K2*A2.
[0083] The converter 1 described above is particularly simple and compact to implement.
[0084] Furthermore, converter 1 does not require a calibration phase unlike many known digital time converters.
[0085] In the sigma-delta converter 1, the quantization is implemented on a single bit. This results in better conversion linearity than if the quantization had been implemented on several bits. Although a quantization on a single bit generates a large quantization error in the form of strong quantization noise, the converter 1 is of the sigma-delta type and therefore has the property of shifting this quantization noise towards high frequencies, in such a way that the low-frequency noise density will be reduced while the high-frequency noise density will be increased. Thus, by applying digital low-pass filtering to the output of the converter 1, a large part of the quantization noise, for example more than 99% of the quantization noise, is eliminated and the signal-to-noise ratio of the useful signal is improved, for example by 40 dB. In the example illustrated in [Fig.l], the OUT1 signal corresponds to the output signal of the converter. The implementation of a digital low-pass filtering, for example clocked by the frequency of the SI signal, of this OUT1 output signal then provides a binary word indicating the value of the measured duration with less quantization noise than the OUT1 bit.
[0086] According to an embodiment in which the converter 1 is implemented in an all-digital phase-locked loop, the low-pass digital filtering of the signal OUT1 is implemented by the digital filter of the phase-locked loop, it being understood that the output signal of the converter 1 can be combined with other digital signals before being supplied to the filter. In such an embodiment, the converter 1 takes advantage of the low-pass digital filter already present in the all-digital phase-locked loop.
[0087] According to an embodiment in which the converter 1 is implemented in an all-digital phase-locked loop, the signal clk corresponds to the periodic output signal of the phase-locked loop, the signal S1 corresponds to the periodic input signal of the phase-locked loop, and the signal S2 is obtained from the signal clk. Thus, the converter does not require any other clock or timing signals than those available at the input and output of the phase-locked loop.
[0088] According to an embodiment in which the converter 1 is implemented in an all-digital phase-locked loop, the converter 1 directly provides a normalized phase difference, without requiring additional calculation.
[0089] [Fig.2] schematically represents details of an exemplary embodiment of the converter 1 of [Fig.l].
[0090] In this example, the active edges of the signals S1 and S2 are the rising edges of these signals, it being understood that the person skilled in the art is able to adapt the description given below to the case where the active edges of one and / or the other of the signals S1 and S2 are falling edges.
[0091] Furthermore, in this example, the level of the PI pulses is equal to +A1, the level of the P2 pulses is equal to -A2, and K2 and K3 are each equal to 1. Thus, the amplitude A3 of the P3 pulses is equal to the amplitude A2 of the P2 pulses, and, furthermore, Dmax = 2*A2 / Al*Tclk.
[0092] In the example of [Fig.2], the circuit C1 comprises a flip-flop 200. The flip-flop 200 is configured so that a rising edge of the signal SI causes the signal CMD1 to be set to the high state and the following edge of the signal S2 causes the signal CMD1 to be reset to the low state. In other words, the flip-flop 200 is configured so that a rising edge of the signal SI causes the start of a corresponding pulse CMD1 and the following edge of the signal S2 causes the end of this pulse CMD1. For example, the flip-flop 200 is an RS type flip-flop having an input S (initialization input) receiving the signal SI, an input R (reset input) receiving the signal S2 and an output Q providing the pulses CMD1.
[0093] In the example of [Fig.2], the circuit C1 also comprises a circuit 202 (block "A1" in [Fig.2]). The circuit 202 is configured to receive the pulses CMD1 and provide the corresponding pulses PI. In other words, the circuit C1 is configured to provide the pulses PI of amplitude A1 from the signal CMD1. By way of example, the circuit 202 comprises one or more resistive elements and / or a switch controlled by the signal CMD1.
[0094] In the example of [Fig.2], the circuit C2 is configured to generate, after each pulse CMD1, a pulse P2 starting at the same time as an active edge, for example rising, of the signal clk, and to generate, after each pulse P2, a pulse CMD3 starting at the same time as an active edge, for example rising, of the signal clk. The circuit C2 is configured so that the duration of each of the pulses P2 is equal to K2*Tclk and the duration of each of the pulses CMD3 is equal to K3*Tclk, with K2 and K3 each equal to 1 in this example.
[0095] As an example, as illustrated in [Fig.2], the circuit C2 comprises three flip-flops 206, 208 and 210. The flip-flop 206 is configured to provide, from the signal CMD1, a run signal indicating for each CMD1 pulse whether the pulse is finished or not. Flip-flop 208 is configured to provide a CMD2 signal determining the start and end of each P2 pulse from the run signal and the clk signal. Flip-flop 210 is configured to provide the CMD3 signal from the CMD2 signal and the clk signal. Circuit C2 further comprises a circuit 212 (block "-A2") configured to provide the P2 pulses from the CMD2 signal, each P2 pulse having an amplitude A2.
[0096] As an example, the flip-flop 206 is a D type flip-flop having a data input D receiving a high level ('1' in [Fig.2]), for example the supply potential Vdd of the converter 1, a clock (or synchronization) input C active on the falling edge and receiving the signal CMD1, a reset input R active on the high level and receiving the signal CMD2 and an output Q providing the run signal.
[0097] For example, flip-flop 208 is a D flip-flop having a data input D receiving the run signal, a clock input C receiving the clk signal and being sensitive, or active, on the active edges of the clk signal, and an output Q providing the CMD2 signal.
[0098] For example, the flip-flop 210 is a D type flip-flop having a data input D receiving the signal CMD2, a clock input C receiving the signal clk and being sensitive, or active, on the active edges of the signal clk, and an output Q providing the signal CMD3.
[0099] The circuit 212 is configured to receive the CMD2 pulses and to provide the corresponding P2 pulses. In this example, the PI pulses are positive and each have a level equal to A1, the P2 pulses then being negative and each having a level equal to -A2. By way of example, the circuit 212 comprises one or more resistive elements and / or a switch controlled by the CMD2 signal.
[0100] In the example of [Fig.2], the circuit INT comprises an operational amplifier 214 mounted as an integrator, the integrator assembly of the operational amplifier 214 not being detailed in [Fig.2]. The amplifier 214 comprises a first input coupled, preferably connected, to a node 216 configured to receive the pulses PI, P2 and P3. An output of the amplifier 214 provides the signal RES1.
[0101] By way of example, as illustrated in [Fig.2], the circuit C3 may comprise a comparator 218, for example implemented from an operational amplifier. The comparator 218 has an input receiving the signal RES1, another input receiving the ref level to which the signal RES1 is compared, and an output providing a binary compl signal in a first state, for example the high state, when the signal RES1 is greater than the ref level and in a second state, for example the low state, when the signal RES1 is less than the ref level. The compl signal corresponds to the quantization on a bit, or binary quantization, of the signal RES1. The circuit C3 can then also comprise, as illustrated in [Fig.2], a storage element 220 clocked by the signal CMD3. The element 220 is configured to provide the bit OUT1 from the bit compl, for example to update the state of the bit OUT1 from the state of the bit compl at each start of a CMD3 pulse. In other words, the element 220 is configured to sample the bit compl, for example at each rising or falling edge of the signal CMD3, and store on the bit OUT1 the sampled level of the bit compl. For example, the element 220 is a D flip-flop having a data input D receiving the signal compl, a clock input C receiving the signal CMD3 and being active on the rising edge of the signal CMD3, and an output Q providing the bit OUTL
[0102] In the example of [Fig.2], circuit C4 corresponds to a circuit 222 (block A3 / +A3" in [Fig.2]). Circuit 222 receives signal OUT1 and signal CMD3 and provides the corresponding pulses P3. For example, circuit 222 comprises one or more resistive elements and / or a switch controlled by signal CMD3.
[0103] As an example, we consider the case where a positive, respectively negative, pulse PI, P2 or P3 at the input of the integrator 214 causes an increase, respectively a decrease, of the signal RES1. In this example, after each update of the bit OUT1, the circuit 222 provides a negative pulse P3 of amplitude A3 when the bit OUT1 is in a binary state indicating that the signal RES1 is greater than the ref level and a positive pulse P3 of amplitude A3 when the bit OUT1 is in a binary state indicating that the signal RES1 is less than the ref level. In other words, after each update of the bit OUT1, the circuit 222 provides a pulse P3 of level equal to -A3 when the bit OUT1 indicates that the signal RES1 is greater than the ref level, and of level equal to +A3 when the bit OUT1 indicates that the signal RES1 is less than the ref level.
[0104] The person skilled in the art will be able to deduce from the example above, the polarity of the pulses P3 as a function of the state of the bit OUT1 in the case where a positive, respectively negative, pulse PI, P2 or P3 at the input of the integrator 214 causes a decrease, respectively an increase, of the signal RES1.
[0105] [Fig. 3] illustrates, by means of timing diagrams, an example of operation of the converter 1 of [Fig. 1] when it is implemented in the manner described in relation to [Fig. 2]. In the example of [Fig. 3], K2 and K3 are each equal to 1, and the amplitudes A2 and A3 are equal to 0.5 times the amplitude A1, although the person skilled in the art will be able to adapt the description given below to the case where K2, K2, A1, A2 and A3 have other values. In this example, a positive, respectively negative, pulse PI, P2 or P3 at the input of the integrator 214 causes an increase, respectively a decrease, of the signal RES1, and, in addition, that the bit OUT1 is set to update to high, respectively low, when at the time of this update the RES1 signal is higher, respectively lower, than the ref level.
[0106] For example, at a time t0, the output RES1 of the circuit INT and the signals PI, P2 and P3 are at their zero level, the signals SI and S2, run and OUT being in the low state.
[0107] [Fig.3] illustrates a single operating cycle of converter 1, i.e. the succession of steps repeated after each active front of signal SI.
[0108] At a time t1 after time t0, the signal SI switches to the high state. The rising edge of the signal SI causes the signal CMD1 (not shown in [Fig.3]) to switch to the high state and the signal PI to switch to the level +A1. In other words, the rising edge of the signal S1 causes the start of a pulse CMD1 and a pulse PL
[0109] From time t1, because the circuit INT receives the positive PI pulse, the signal RES 1 varies, in this example increases, with a slope proportional to the level of the PI pulse, each PI pulse here having a level equal to +A1.
[0110] At a time t2 after time t1, the signal S2 switches to the high state. The rising edge of the signal S2 causes the signal CMD1 (not shown in [Fig.3]) to switch to the low state and the signal PI to return to its zero level. In other words, the rising edge of the signal S2 causes the end of the pulses CMD1 and PL II. As a result, the signal RES1 stops varying and stores its value at time t2.
[0111] Furthermore, the end of the CMD1 pulse (not shown in [Fig.3]), i.e. the falling edge of the CMD1 signal received by the flip-flop 206, causes the run signal to switch to the high state.
[0112] A time t3 subsequent to time t2 corresponds to the first rising edge of the signal clk following the end of the pulse PL. This rising edge of the signal clk causes the signal CMD2 (not shown in [Fig.3]) to switch to the high state and the signal P2 to the level -A2. In other words, this edge of the signal clk causes the start of a pulse CMD2 and a pulse P2. In addition, the start of the pulse CMD2 causes the signal run to switch to the low state.
[0113] From time t3, because the circuit INT receives the negative pulse P2, the signal RES 1 varies, in this example decreases, with a slope proportional to the level of the pulse P2, each pulse P2 here having a level equal to -A2.
[0114] A time t4 after time t3 corresponds to the next rising edge of the signal clk. This rising edge of the signal clk causes the signal CMD2 (not shown in [Fig.3]) to switch to the low state and the signal P2 to return to its zero level. In other words, the rising edge of the signal clk causes the end of the pulses CMD2 and P2. In addition, this rising edge of the signal clk causes the signal CMD3 (not shown in [Fig.3]) to switch to the high state, thus updating the bit OUT1 and the start of a pulse P3.
[0115] In this example, when it is updated at time t4, the OUT1 bit switches to the high state because the RES1 signal is higher than the ref level at time t4.
[0116] In this example, the high state of the OUT1 bit therefore results in a negative pulse P3 of amplitude A3, that is to say a pulse of level equal to -A3, and the low state of the OUT1 bit results in a positive pulse P3 of amplitude A3, that is to say a pulse of level equal to +A3. Thus, in this example, the pulse P3 starting at time t4 has a level equal to -A3 because the OUT1 bit is in the high state.
[0117] From time t4, because the circuit INT receives the pulse P3, the signal RES1 varies with a slope proportional to the level of the pulse P3. More particularly, in this example, the signal RES1 decreases with a slope proportional to the negative level -A3 of the pulse P3.
[0118] A time t5 after time t4 corresponds to the next rising edge of the signal clk. This rising edge of the signal clk causes the signal CMD3 to switch to the low state, therefore the end of the pulse P3, that is to say the return to the zero level of the signal P3.
[0119] Furthermore, from time t5, as the INT circuit no longer receives a PI, P2 or P3 pulse, the RES1 signal memorizes, or keeps, its value at time t5.
[0120] After time t5, each new rising edge of signal SI causes a new operating cycle of converter 1, i.e. a new succession of pulses PI, P2 and P3, a modification of signal RES1 from its current value, and an update of bit OUT1 on the basis of signal RES1.
[0121] [Fig.4] illustrates an example of an all-digital phase-locked loop 4, according to one embodiment, where the phase-locked loop 4 comprises the converter 1.
[0122] The phase-locked loop 4 is configured to receive a clkref signal at a first frequency and to provide the clk signal at a second frequency equal to N / D times the first frequency, with N and D being two positive values. Preferably, D is greater than or equal to 1, and N is greater than 1 and generally greater than D.
[0123] The phase-locked loop 4 comprises the converter 1. The converter 1 comprises an input configured to receive the signal clkref, which then corresponds to the signal S1 described in relation to Figures 1 and 2, and a second input configured to receive the signal clk. In the exemplary embodiment illustrated by [Fig.4], the signals S2 (see Figures 1 and 2) and clk of the converter 1 are merged.
[0124] Converter 1 provides an output signal TDCOUT. When converter 1 is of the type described in relation to [Fig.l] or [Fig.2], the signal TDCOUT is for example equal to the bit OUT.
[0125] In [Fig.4], the phase-locked loop 4 comprises a circuit accN, a circuit accD, a register reg, a digital arithmetic circuit 400, a digital low-pass filter 402 and a digitally controlled oscillator 404.
[0126] The circuit accN is configured to provide a digital signal accNOUT incremented by the value N at the frequency of the signal clkref. In other words, at each active edge of the signal clkref, the signal accNOUT is incremented by the value N. In other words, the signal accNOUT represents the accumulated phase of the signal clkref.
[0127] The circuit accD is configured to provide a digital signal accDOUT incremented by the value D at the frequency of the signal clk. In other words, at each active edge of the signal clk, the signal accDOUT is incremented by the value D.
[0128] The reg circuit is configured to sample the accDOUT signal in synchronism with the clkref signal, for example in synchronism with the active edges of the clkref signal, and to provide an accDOUT' signal corresponding to the storage of this sampling. In other words, the reg circuit copies the accDOUT signal to its accDOUT' output at each active edge of the clkref signal. In other words, the accDOUT' signal represents the accumulated phase of the clk signal sampled by clkref.
[0129] The circuits accD and reg therefore form a circuit configured to provide the signal accDOUT incremented by the value D at the frequency of the signal clk and the signal accDOUT' corresponding to the sampling by the signal clkref of the signal accDOUT. In other words, the circuits accD and reg form a circuit configured to provide the signal accDOUT' corresponding to the sampling, by the signal clkref, of the signal accDOUT incremented by the value D at the frequency of the signal clk.
[0130] The digital filter 402, in practice a low-pass filter, is configured to receive and filter a result FIN of a subtraction of the output signal TDCOUT of the converter 1 and the signal accDOUT' from the signal accNOUT, and to control the oscillator 404. Thus, the circuit 400 is configured to receive the signals TDCOUT, accNOUT and accDOUT' and to provide, at the input of the filter 402, the digital signal FIN corresponding to the digital signal accNOUT from which the digital signals TDCOUT and accDOUT' have been subtracted.
[0131] The filter 402 provides a digital control signal DCOCTRL to the oscillator 404. The signal DCOCTRL corresponds to the result of the low-pass filtering of the signal FIN by the filter 402. For example, the transfer function in z of the filter 402 can be of the type a+(b / (lz *)).
[0132] The oscillator 404 is configured to provide the signal clk at a frequency determined by the output value of the filter 402, i.e. by the value of the signal DCOCTRL.
[0133] In the phase-locked loop 4, the subtraction of the signal accDOUT' from the signal accNOUT represents the phase error of the loop, tainted by a residual phase error PhiError between the frequency of the signal clkref and that of the signal clk. This error PhiError is in practice equal to the ratio of a value DeltaT to a period Tclk of the signal clk, with DeltaT the duration separating each edge of the signal clkref from the following edge of the signal clk. The prediction of the converter 1 and the subtraction of the TDCOUT signal to the result of subtracting the accDOUT signal from the accNOUT signal allows to remove the residual phase error PhiError in the FIN signal representing the phase error of loop 4.
[0134] Although not illustrated here, converter 1 may be implemented in all-digital phase-locked loops different from that of [Fig. 4]. For example, converter 1 may be implemented in an all-digital phase-locked loop not including circuits accN, accD, reg and 400, in which the output TDCOUT of converter 1 is directly supplied to filter 402, converter 1 receives signal clkref or a signal obtained by dividing the frequency of signal clk by D and signal clk or a signal obtained by dividing the frequency of signal clk by N.
[0135] [Fig.5] schematically represents details of an exemplary embodiment of the converter 1 of [Fig.l], in a case where the converter is implemented in a phase-locked loop of the type of that of [Fig.4].
[0136] In this exemplary embodiment, the signals S2 (Figures 1 and 2) and clk are merged, and the signals SI (Figures 1 and 2) and clkref are merged. Furthermore, in this example, the duration of the pulses P2 and P3 is equal to one period Tclk of the signal clk (K2=K3=1), and the amplitudes A2 and A3 are equal to 0.5 times the amplitude A1. Thus, the maximum duration Dmax of the PI pulses is equal to Tclk, and, in this example where the PI pulses start with the rising edges of the signal SI and end with the rising edges of the signal S2, the dynamic range of the converter 1 is also equal to Dmax=Tclk.
[0137] Converter 1 of [Fig.5] is similar to converter 1 of [Fig.3], and only the differences between these converters are highlighted here.
[0138] In particular, since the signals S2 and clk are merged, the flip-flop 206 ( [Fig.2]) of the circuit Cl is omitted, and the data input D of the flip-flop 208 of the circuit Cl directly receives the signal CMD1.
[0139] [Fig.6] illustrates, by means of timing diagrams, the operation of converter 1 of [Fig.5].
[0140] The timing diagrams of [Fig.6] are similar to those of [Fig.3], and only the differences between the latter will be highlighted here. In particular, [Fig.6] does not include the run signal because the converter 1 of [Fig.5] does not include this signal. In the example of [Fig.6], as in the example of [Fig.3], a positive, respectively negative, PI, P2 or P3 pulse at the input of the integrator 214 causes an increase, respectively a decrease, of the signal RES1, and the bit OUT1 is updated to the high, respectively low state, when at the time of this update the signal RES1 is higher, respectively lower, than the ref level.
[0141] In [Fig.6], an instant t10 corresponds to the instant t0 described in relation to [Fig.3].
[0142] In [Fig.6], an instant tl 1 subsequent to the instant tl0 corresponds to the instant tl described in relation to [Fig.3], that is to say to an active edge (in this example a rising edge) of the signal clkref.
[0143] In [Fig.6], an instant tl2 subsequent to instant tl 1 corresponds to the first active edge (in this example a rising edge) of the signal clk following the rising edge of the signal clkref at instant tl 1. This rising edge of the signal clk causes the end of the pulses PI and CMD1, the signal CMD1 not being represented in [Fig.6]. Furthermore, this rising edge of the signal clk causes the start of the pulses CMD2 and P2, the signal CMD2 not being represented in [Fig.6].
[0144] In other words, in [Fig.6], the instants t2 (end of the PI pulse) and t3 (start of the P2 pulse) described in relation to [Fig.3] are merged and correspond to instant 112.
[0145] In [Fig.6], an instant tl3 subsequent to the instant tl2 corresponds to the instant t4 described in relation to [Fig.3] (rising edge of the signal clk, end of the pulse P2 and start of the pulse P3), and an instant tl4 subsequent to the instant tl3 corresponds to the instant t5 described in relation to [Fig.3] (rising edge of the signal clk and end of the pulse P3).
[0146] After time tl4, each new rising edge of the signal clkref causes a new operating cycle of the converter 1, i.e. a new succession of pulses PI, P2 and P3, a modification of the signal RES1 from its current value, and an update of the bit OUT1 on the basis of the signal RES1.
[0147] The person skilled in the art is able to understand that, in the example illustrated by [Fig.6], the frequency of the signal clk is at least three times greater than that of the signal clkref so that the converter 1 has the time to generate and integrate three successive pulses PI, P2 and P3 after each active edge of the signal clkref and before the following active edge of the signal clkref.
[0148] In converter 1 of [Fig.5], because the peak-to-peak amplitude of signal P3 is equal to 2*A3 and is equal to that of signal PI, namely A1, the converter in fact normalizes the value of bit OUT1 over a period of signal clk. In other words, bit OUT1 represents the value of the residual phase error PhiError and can be supplied as is at the input of circuit 400 of phase-locked loop 4 described in relation to [Fig.4].
[0149] [Fig.7] schematically shows details of another exemplary embodiment of the digital time converter of [Fig.l], in a case where the converter is implemented in a phase-locked loop of the type of [Fig.4].
[0150] In particular, the embodiment of [Fig.7] differs from that of [Fig.5] in that the signal S2 (Figures 1 and 2) is obtained by dividing by two the frequency of the signal clk.
[0151] Furthermore, in this example, the duration of the pulses P2 and P3 is equal to a period Tclk of the signal clk (K2=K3=1) and the amplitudes A2 and A3 are equal. In this example where the PI pulses start with the rising edges of the signal SI and end with the rising edges of the signal S2, the maximum duration Dmax of the PI pulses is then equal to 2*Tclk, from which it follows that the amplitude A1 is equal to the amplitude A2 and to the amplitude A3, the dynamics of the converter 1 also being equal to Dmax=2*Tclk.
[0152] More particularly, the converter 1 of [Fig.7] differs from that of [Fig.5] by the implementation of its circuit C2 and its circuit C1, and by the fact that it further comprises a circuit 700. The circuit 700 is a frequency divider. In other words, the circuit 700 is configured to provide a signal at a frequency divided relative to the frequency of an input signal of the circuit 700.
[0153] The circuit 700 is configured to receive the signal clk, and to provide the signal S2, the latter having, in this example, a frequency twice as low as that of the signal clk. In other embodiments not illustrated, the person skilled in the art may provide a ratio other than two between the frequency of the signal clk and that of the signal S2, by then adapting the amplitudes of the pulses PI, P2 and / or P3 and / or the durations of the pulses P2 and P3 so that the entire dynamic range of the pulses PI can be converted by the converter 1.
[0154] By way of example, as illustrated in [Fig.7], the circuit 700 comprises a D-type flip-flop 702. The flip-flop 702 comprises a clock input C configured to receive the signal clk, the input C being sensitive to the active edges of the signal clk, namely the rising edges in this example, an output Q configured to provide the signal S2, and a data input D configured to receive the inverse of the output Q. By way of example, the inverse of the output Q is provided to the input D of the flip-flop 702 by an inverter 704. The inverter 704 has for example an input connected to the output Q of the flip-flop 702 and an output connected to the input D of the flip-flop.
[0155] In this example, the circuit C1 of [Fig.7], and more particularly its circuit 202, differ from those of [Fig.5] in that the pulses PI that they provide each have an amplitude Al equal to the amplitudes A2 and A3, and not twice the amplitudes A2 and A3 as was the case in the example of [Fig.5]. By way of example, the circuit 202 comprises one or more resistive elements and / or a switch controlled by the signal CMD1.
[0156] Compared to the converter 1 described in relation to figures 2 and 5, in the converter 1 of [Fig.7], the PI pulses, which each begin at an active edge of the signal S1 and each end at the following active edge of the signal S2, have a maximum duration Dmax equal to two periods Tclk of the signal clk. Thus, the circuit C2 is modified so that at each PI pulse, the corresponding P2 pulse does not only begins once the PI pulse has finished.
[0157] In the example of [Fig.7], compared to what was described in relation to [Fig.5], the signal CMD2 is not directly available at the output of the flip-flop 208 of the circuit C2. More particularly, in [Fig.7], the flip-flop 208 always has an input D configured to receive the signal CMD1 and an input C configured to receive the signal clk and to be sensitive to the active edges of the signal clk. However, the circuit C2 comprises a logic circuit configured to provide a pulse CMD2 only when the output Q of the flip-flop 208 has switched following a PI pulse and this PI pulse is completed. For example, this logic circuit is configured to implement the logical AND between the inverse of the signal CMD1 and the output Q of the flip-flop 208, the result of this logic operation being the signal CMD2. As an example illustrated by [Fig.7], this logic circuit comprises an AND gate 706 having an input connected to the output Q of the flip-flop 208 and another input connected to the output of an inverter 708 whose input receives the signal CMD1, the gate 706 having an output providing the signal CMD2. Of course, the person skilled in the art is able to provide other examples of implementation of this logic circuit.
[0158] As in Figures 2 and 5, the signal CMD2 is supplied to the input D of the flip-flop 210 of the circuit Cl.
[0159] [Fig.8] illustrates, by means of timing diagrams, an example of operation of converter 1 of [Fig.7]. In [Fig.8], signal S2 is obtained by dividing by two the frequency of signal clk, and signals SI and clkref are merged.
[0160] In the example of [Fig.8], as in the examples of figures 3 and 6, a positive, respectively negative, pulse PI, P2 or P3 at the input of the integrator 214 causes an increase, respectively a decrease, of the signal RES1, and the bit OUT1 is updated to the high, respectively low state, when at the time of this update the signal RES1 is higher, respectively lower, than the level ref.
[0161] For example, at a time t20, the output RES1 of the circuit INT is at a constant or stored value, in this example negative with respect to the level ref. In addition, the signals PI, P2 and P3 are at their zero levels, the signal clkref is in the low state, and the bit OUT1 is in the high state.
[0162] At a time t21 after time t20, the signal clkref switches to the high state, which which causes a rising edge of the CMD1 signal (not shown in [Fig.8]) and the passage of the PI signal to the +A1 level. In other words, the rising edge of the clkref signal causes the start of a CMD1 pulse and a PI pulse.
[0163] From time t21, because the circuit INT ([Fig.7]) receives the positive PI pulse, the signal RES1 varies, in this example increases, with a slope proportional to the level +A1 of the PI pulse. Each PI pulse here has an amplitude equal to Al.
[0164] At a time t22 after time t21, signal S2 switches to the high state. The rising edge of signal S2 causes signal CMD1 (not shown in [Fig.8]) to switch to the low state and signal PI to switch to its zero level. In other words, the rising edge of signal S2 causes the end of pulse CM1 and pulse PI.
[0165] Furthermore, the end of the CMD1 pulse (not shown in [Fig.8]), causes the start of a CMD2 pulse (not shown in [Fig.8]), therefore the start of a P2 pulse.
[0166] From time t22, because the circuit INT ([Fig.7]) receives the negative pulse P2, the signal RES1 varies, in this example decreases, with a slope proportional to the level -A2 of the pulse P2. Each pulse P2 here has an amplitude equal to A2.
[0167] A time t23 after time t22 corresponds to the next rising edge of the signal clk. This rising edge of the signal clk causes the signal CMD2 (not shown in [Fig.8]) to switch to the low state and the signal P2 to switch to its zero level. In other words, this edge of the signal clk causes the end of the pulses CMD2 and P2.
[0168] Furthermore, the rising edge of the clk signal at time t23 causes the CMD3 signal (not shown in [Fig.8]) to switch to the high state, thus updating the OUT1 bit and starting a P3 pulse.
[0169] In this example, at time t23, bit OUT1 switches to the low state because, at this time t23, the signal RES1 is lower than the ref level. Furthermore, in this example, the low state of bit OUT1 results in a positive pulse P3 of level +A3, and the high state of bit OUT1 results in a negative pulse P3 of level -A3. Thus, in this example, pulse P3 starting at time t23 has a level equal to +A3 because bit OUT1 is in the low state from time t23.
[0170] From time t23, because the circuit INT receives the positive pulse P3, the signal RES1 varies with a slope proportional to the level +A3 of the pulse P3. More particularly, in this example, the signal RES1 increases with a slope proportional to the level +A3 of the pulse P3.
[0171] A time t24 after time t23 corresponds to the next rising edge of the clk signal. This rising edge of the clk signal causes the CMD3 signal to switch to the low state, thus the end of pulse P3.
[0172] Furthermore, from time t24, as long as the INT circuit no longer receives a PI, P2 or P3 pulse, the RES1 signal stores its value at time t24.
[0173] After time t24, each new rising edge of signal SI (for example at time t25 after time t24) causes a new operating cycle of converter 1, i.e. a new succession of pulses PI, P2 and P3, a modification of signal RES1 from its current value, and an update of bit OUT1 on the basis of signal RES1.
[0174] The person skilled in the art is able to understand that, in the example illustrated by [Fig.8], the frequency of the clk signal is at least four times greater than that of the clkref signal so that the converter 1 has time to generate and integrate three successive PI, P2 and P3 pulses after each active edge of the clkref signal and before the following active edge of the clkref signal.
[0175] When converter 1 of [Fig.7] is implemented in a phase-locked loop of the type of that of [Fig.4], because the maximum duration Dmax that the PI pulses can take is equal to two periods Tclk of the signal clk, the bit OUT1 of converter 1 is then representative of a residual phase error proportional to twice the residual phase error PhiError between the signals clk and clkref. The phase-locked loop 4 described in relation to [Fig.4] is then modified accordingly.
[0176] For example, the circuit accD is modified to be clocked by the signal S2 rather than by the signal clk, i.e. the signal accDOUT is then incremented at the frequency of the signal S2.
[0177] Furthermore, when the circuit accD is modified in this way, so that loop 4 keeps the same frequency transfer function as with converter 1 of [Fig.5], the circuit accD can then be configured to increment the signal accDOUT by the value D / 2 at each active edge of the signal S2, if the value of D is even. Alternatively, so that loop 4 keeps the same frequency transfer function, for example when the value of D is odd, the circuit accN can then be configured to increment the signal accNOUT by twice the value N at each rising edge of the signal clkref.
[0178] Although the case where the signal S2 is obtained by dividing the frequency of the signal clk by two has been described in relation to figures 7 and 8, the person skilled in the art is able to adapt this description to the case where the signal S2 is obtained by a different integer division of two of the signal clk, for example by modifying accordingly the durations and / or the amplitudes of the pulses PI, P2 and P3 so that the entire dynamic range of the pulses PI can be converted by the converter 1.
[0179] Furthermore, an example of an embodiment of the converter 1 has been described in relation to Figures 7 and 8 in the case where it is implemented in a phase-locked loop of the type of that of [Fig.4]. The person skilled in the art is able to adapt the converter 1 of [Fig.7] to an implementation in a circuit other than a phase-locked loop. For example, for this, the circuit C2 of the converter 1 of [Fig.7] is modified to include the flip-flop 206 so as to generate the run signal, this run signal being supplied at the input of the flip-flop 208 and of the logic circuit 708, 706 in place of the signal CMD1.
[0180] The sigma-delta type converter 1 described in relation to figures 1 to 8 is of order 1, that is to say that the quantization error is integrated only once at each operating cycle. In this case, the noise shaping slope is, for example, 20dB per decade.
[0181] The converter 1 may be modified according to a multi-stage noise shaping (MASH) architecture, so that the converter 1 is then of order greater than 1, the order of the converter 1 then corresponding to the number of integrations of the quantization error per operating cycle of the converter 1. In this case, the converter 1 comprises a first stage comprising the circuits INT, C1, C2, C3 and C4 described previously, and a second stage configured to receive the output signal RES 1 from the circuit INT of the first stage, and to generate, at the output of the second stage, a digital signal from the signal RES1, preferably by implementing an integration of a signal determined by the signal RES1. The digital output signal TDCOUT of the converter 1 is then determined at least in part by the bit OUT1 and the digital output signal of the second stage.
[0182] According to one embodiment, the second stage is configured to: generate, after each pulse P3, a pulse CMD4 starting in synchronism with the signal clk; integrate, during each pulse P4, a sum of the output signal RES 1 of the circuit INT and a feedback signal CR, quantify on one bit a result RES2 of said integration and store on a bit OUT2 said binary quantification at the end of the pulse CMD4; generate, during each pulse P3, a pulse P4 from the OUT2 bit, and the signal CR by integrating the pulse P4; and storing the OUT2 bit on an OUT2-1 bit at each active edge of the S2 signal, the second stage output digital signal being generated from the OUT2 and OUT2-1 bits or corresponding to the OUT2 and OUT2-1 bits.
[0183] In other words, the second stage is configured to receive as input the residual quantization error available at the output of the INT circuit after each pulse P3, and to integrate it again. In addition, the second stage implements a negative sigma-delta feedback using the CR signal determined from the OUT2 bit. The determination of the CR signal is implemented after each operating cycle of the converter and before the start of the corresponding CMD4 pulse. During each CMD4 pulse, the CR signal is summed to the RES1 signal and the second stage integrates the sum S of the RES1 and CR signals. The result RES2 of this integration is quantified and stored on the OUT2 bit.
[0184] As an example, the output TDCOUT of the second-order converter 1 is obtained by summing the output OUT1 of the first stage with the mathematical derivative of the signal OUT2 of the second stage.
[0185] For example, the negative feedback, i.e. the determination of the signal CR, comprises the integration of the pulse P4 whose sign is determined by the state of the bit OUT2. The result of this integration then corresponds to the signal CR and is reset before each new integration of a pulse P4, therefore at each operating cycle of the converter 1. For example, because each pulse P4 is generated during a pulse CMD3, the duration of each pulse P4 is equal to K4*Tclk, with K4 a strictly positive integer equal to K3.
[0186] Because the second stage implements a second integration of the quantization error, the noise shaping slope is, for example, at least 40dB per decade and depends on the number of cascaded stages in the converter 1, the slope being, for example, equal to 40 / dB per decade when the 1 sigma-delta converter of the MASH type comprises exactly two stages.
[0187] The greater noise shaping resulting from increasing the order of converter 1 allows for further reduction of low-frequency noise. In addition, increasing the order of converter 1 allows for a purer frequency spectrum with respect to the clk signal.
[0188] Higher than two order MASH type converters 1 are constructed by cascading the stages in the same way as was presented for the first stage and second stage of the second order converter 1. For example, the output of the integrator of the Nth stage N is provided as input to the N+lth stage, and the quantized and derived output of the N+lth stage is summed to the quantized output of the Nth stage. For each additional stage, an additional pulse is added to the operating cycle of the converter 1, to synchronize the operation of this stage with the other stages of the converter 1.
[0189] [Fig.9] illustrates an alternative embodiment of the converter 1 of [Fig.1]. More particularly, [Fig.9] illustrates an example of implementation of the converter 1 when it is of the second-order MASH type. In this [Fig.9], the first stage of the converter 1 is referenced STAGE1 and delimited by dotted lines, the second stage of the converter 1 being referenced STAGE2 and delimited by dotted lines.
[0190] Stage STAGE1 comprises circuits C1, C2, C3, C4 and INT, shown and connected in [Fig.9] in the same way as in [Fig.l]. The output signal of stage STAGE 1 is bit OUT1. As an example, circuits C1, C2, INT, C3 and C4 are implemented in the manner described in relation to [Fig.2].
[0191] Stage STAGE2 comprises, in this example, circuits C5, INT2, C6, C7 and DER, examples of implementation of which are described in more detail below. The person skilled in the art is however able to provide other examples of implementation of at least one of these circuits C5, INT2, C6, C7 and DER, from the functional description which will be given.
[0192] The circuit C5 is configured to generate, after each pulse P3, a corresponding pulse CMD4. Each pulse CMD4 has a duration equal to K4*Tclk, with K4 a strictly positive integer and equal to K3. For example, the circuit C5 receives the signal CMD3 indicating the start and end of each pulse P3, which allows the pulses received by the circuit C5 to all have the same sign. In the example illustrated by [Fig.9], the circuit C5 comprises a flip-flop 900, for example of type D. For example, the flip-flop 900 comprises an input D receiving the signal CMD3, a clock input C receiving the signal clk, and an output Q providing the pulses CMD4 in this example where K4=K3=1.
[0193] The circuit INT2 is an integrator circuit configured to integrate, during each pulse CMD4, the sum S of the signal RES 1 and the signal CR. The circuit INT2 provides a signal RES2 corresponding to the result of this integration.
[0194] For example, the circuit INT2 receives the signals RES1, CR and CMD4. At each pulse CMD4, for the entire duration of the pulse P4, the circuit INT2 is for example configured to sum the signals RES1 and CR and to integrate this sum S.
[0195] As an example of implementation illustrated by [Fig.9], the circuit INT2 comprises two switches 902 and 904 controlled by the signal CMD4, for example configured to be on during each pulse CMD4, a circuit 906 and an integrator circuit 908, for example implemented from an operational amplifier. More particularly, the switch 902 is configured to receive the signal RES1 and to transmit this signal to the circuit 906 only during each pulse CMD4, the switch 904 is configured to receive the signal RES1 and to provide this signal to the circuit 906 only during each pulse CMD4, the circuit 906 is configured to provide the signal S equal to the sum of the signals that it receives from the switches 902 and 904 and the circuit 908 is configured to integrate the signal S. The circuit 908 provides the signal RES2.
[0196] According to another example not illustrated, in the circuit INT2, the signals RES1 and CR are first summed, and the signal corresponding to this sum is supplied to the integrator circuit 908 during each pulse CMD4, for example by means of a switch controlled by the pulses CMD4.
[0197] The circuit C6 is configured to quantize the signal RES2 on one bit and to store the result of this binary quantization on the bit OUT2 at the end of each pulse P4. The quantization of the signal RES2 is carried out by comparing the signal RES1 to the ref level. For example, the circuit C6 generates a bit comp2 corresponding to the result of the binary quantization of the signal RES2, and updates the bit OUT2 from the bit comp2.
[0198] As an example of implementation illustrated by [Fig.9], the circuit C6 comprises a comparator 910 configured to receive the signal RES2 and the level, or potential, ref, and to provide the bit comp2. For example, the bit comp2 is in a first binary state when the signal RES2 is greater than the level ref, and in a second binary state when the signal RES2 is less than the level ref. The circuit C6 also comprises a storage element 912, for example a flip-flop, for example of type D, configured to provide the bit OUT2 from the signal comp2, and update the state of the bit OUT2 at the end of each pulse P4. In other words, the element 912 is configured to sample the bit comp2, for example at each falling edge of the signal CMD4, and store on the bit OUT2 the sampled level of the bit comp2.For example, the D-type flip-flop 912 comprises an input D configured to receive the signal comp2, a clock input C sensitive to rising edges and configured to receive a signal CMD4b complementary to the signal CMD4, and an output Q configured to provide the signal OUT2, the signal CMD4b being for example available at the output of an inverter 914 receiving the signal CMD4 as an input signal.
[0199] The circuit C7 is configured to provide the signal CR from the bit OUT2. More particularly, the circuit C7 is configured, at each operating cycle of the converter 1 and before the start of the pulse CMD4 of this operating cycle, for example during the pulse CMD3 of this operating cycle, to generate a pulse P4 of the same duration as the pulses P3 and to integrate this pulse P4. Furthermore, each pulse P4 is signed and has its sign determined by the binary state of the signal OUT2. In other words, the circuit C7 is configured, at each pulse CMD3, to provide the signal CR by integrating, during this pulse CMD3, the pulse P4 determined from the bit OUT2. The pulses P4 all have an amplitude A4, for example equal to the amplitude A3 of the pulses P3.
[0200] As an exemplary implementation illustrated by [Fig.9], the circuit C7 comprises an INTP circuit configured to integrate the P4 pulses, and a circuit 915 (block +A4 / -A4 in [Fig.9]) configured to provide the P4 pulses from the OUT2 bit and the CMD3 signal. The circuit 915 provides the P4 pulses to the INTP circuit. Unlike the integrator circuits INT (not shown in [Fig.9]) and INT2, the output of the INTP integrator circuit is reset after each CMD4 pulse, for example during the next P2 pulse. For example, the INTP circuit receives the CMD2 signal and its output is reset at each CMD2 pulse.
[0201] As an example, we consider the case where a positive, respectively negative, pulse S at the input of the integrator 908 causes a decrease, respectively an increase, of the signal RES2, and where a positive, respectively negative, pulse P4 at the input of the integrator INTP causes a decrease, respectively an increase, of the signal CR. In this example, the pulses P4 are negative and have a level equal to -A4 when the bit OUT2 is in a binary state indicating that the signal RES2 is greater than the ref level, and are positive and have a level equal to +A4 when the OUT2 bit is in a binary state indicating that the RES2 signal is less than the ref level. The amplitude A4 of the P4 pulses is preferably determined so that K3*A2=K4*A4, and is therefore equal to the amplitude A3 of the P3 pulses in this example where K3=K4.
[0202] The person skilled in the art will be able to deduce from the example above, the polarity of the pulses P4 as a function of the state of the bit OUT2 in the cases where a positive, respectively negative, pulse S at the input of the integrator 918 causes an increase, respectively a decrease, of the signal RES2 and / or a positive, respectively negative, pulse P4 at the input of the integrator INTP causes an increase, respectively a decrease, of the signal CR.
[0203] In this example, the second stage STAGE2 provides a digital output signal NUM from the bits OUT2 and OUT2-1. In this example, the circuit DER is configured to provide the signal NUM from the bit OUT2. More particularly, the circuit DER is configured to generate the bit OUT2-1 from the bit OUT2, and to provide the signal NUM from the bits OUT2 and OUT2-1. The circuit DER implements a derivative function on the signal OUT2, the result of which is the digital signal NUM. The circuit DER is a digital circuit clocked by the signal SL. In other words, the circuit DER is a derivative circuit clocked by the signal SI and configured to provide the signal NUM from the signal OUT2.
[0204] Since the OUT2 bit represents the result of a double integration while the OUT1 bit represents the result of a single integration, implementing a derivative function on the OUT2 signal to obtain the NUM signal allows the NUM signal to be consistent with the OUT1 signal.
[0205] As an exemplary implementation, the DER circuit comprises a D-type flip-flop 920 comprising an input D configured to receive the signal OUT2, a clock input C configured to receive the signal SI and an output Q configured to provide the signal OUT2-1. The DER circuit further comprises a circuit 922 configured to subtract the signal OUT2-1 from the signal OUT2, the result of this subtraction corresponding to the output bits NUM of the stage STAGE2. As an example, the signal NUM corresponds to two bits and is coded in 2's complement.
[0206] In this example where the second stage provides the signal NUM which corresponds to the result of the derivative of the bit OUT2, the converter 1 further comprises a circuit 924 configured to provide the output signal TDCOUT of the converter 1 from the signals, or bits, OUT1 and NUM. For example, the circuit 924 is configured to sum the signals NUM and OUT1 and to provide the signal TDCOUT which then corresponds to the result of this sum. For example, the signal TDCOUT is a three-bit binary word, coded in 2's complement.
[0207] In another example, not illustrated in [Fig.9], the second stage STAGE2 does not include the circuit 922, and then provides a digital output signal corresponding to the two bits OUT2 and OUT2-1. Furthermore, the converter 1 then does not include the circuit 924, but instead includes an arithmetic circuit configured to provide the output signal TDCOUT from the bits OUT1, OUT2 and OUT2-1. This arithmetic circuit is configured to simultaneously implement the mathematical derivative of the signal OUT2 from the bits OUT2 and OUT2-1, and the sum of this derivative with the bit OUT1.
[0208] In relation to [Fig.9], the case where the signal SI does not correspond to the signal clkref received by a phase-locked loop of the type of that of [Fig.4] has been considered. Based on the descriptions made in relation to figures 4 to 9, the person skilled in the art will be able to adapt the description made in relation to [Fig.9] to the cases where: - the duration Dmax of the PI pulses is different from one period Tclk of the clk signal and is, for example, equal to two periods Tclk of the clk signal, and / or - the SI signal corresponds to the clkref signal of a phase-locked loop of the type described in relation to [Fig.4], and / or - the signal S2 corresponds to the signal clk of a phase-locked loop of the type described in relation to [Fig.4] or is obtained by integer division, for example equal to two, of the frequency of this signal clk, and / or - converter 1 is of order greater than 2.
[0209] For example, the person skilled in the art will be able to modify, on the basis of the description given in relation to Figures 5 to 8, the converter 1 of [Fig.9] to implement it in a phase-locked loop of the type described in relation to [Fig.4].
[0210] For example, when the 2nd order converter 1 described in relation to [Fig.9] is implemented in a phase-locked loop of the type of that of [Fig.4], and the first stage STAGE1 of the 2nd order converter 1 corresponds to the 1st order converter 1 described in relation to [Fig.5], the frequency of the clk signal then being at least 4 times greater than that of the SI signal. Alternatively, when the 2nd order converter 1 described in relation to [Fig.9] is implemented in a phase-locked loop of the type of that of [Fig.4], and the first stage STAGE1 of the 2nd order converter 1 corresponds to the 1st order converter 1 described in relation to [Fig.7], the frequency of the clk signal then being at least 5 times greater than that of the SI signal.
[0211] [Fig. 10] illustrates an example of implementation of a part of the converter 1 of the previous figures. More particularly, [Fig. 10] illustrates an exemplary embodiment of an RS flip-flop, for example the RS flip-flop of circuit C1, configured to take into account possible metastability issues.
[0212] In this example, the RS flip-flop is active on rising edges, or, in other words, the S and R inputs of the flip-flop are both sensitive to the rising edges of the signals they receive.
[0213] The flip-flop comprises a NAND gate 1000 having an input connected to input S and an input configured to receive a signal sigl. The output of gate 1000 provides a signal sig2.
[0214] The flip-flop comprises three identical sets RS1, RS2 and RS3 of two NAND gates 1002 and 1004. Each set RS1, RS2 and RS3 comprises an input s connected to a first input of its gate 1002 and an input r connected to a first input of its gate 1004, the output of gate 1002 being connected to a second input of gate 1004, and the output of gate 1004 being connected to a second input of gate 1002.
[0215] The input s of the RS1 set receives the signal sig2, and the output of the gate 1002 of the RS1 set constitutes the output Q of the RS flip-flop of [Fig. 10]. In addition, the input r of the RS1 set receives a signal sig3.
[0216] The input s of the RS2 set is connected to the output of the gate 1004 of the RS1 set, the input r of the RS2 set is connected to the input S of the RS flip-flop of [Fig. 10], and the output of the gate 1004 of the RS2 set provides the signal sigl.
[0217] The input s of the set RS3 is connected to the output of the gate 1002 of the set RS1, therefore to the output Q of the flip-flop, the input r of the set RS3 is connected to the input R of the flip-flop, and the output of the gate 1002 of the set RS3 provides a signal sig4.
[0218] The flip-flop further comprises a circuit 1006, for example a series inverter chain, having an input connected to the R input of the flip-flop and an output providing a signal sig5. The circuit 1006 is configured to delay the signal received by the R input by a duration DT, the sig5 signal corresponding to this delayed signal.
[0219] The RS flip-flop of [Fig. 10] further comprises a NAND gate 1008 having one input connected to the R input of the flip-flop, another input connected to the output of the circuit 1006 so as to receive the signal sig5, and an output configured to provide a signal sig6.
[0220] The flip-flop further comprises an OR gate having an input receiving the signal sig6, an input receiving the signal sig4 and an output providing the signal sig3.
[0221] If we assume that initially the output Q is at the low level, the signal sigl is at the high level. Thanks to this, the arrival of a high level on the input S can propagate through the gate 1000, which then provides a low level of the signal sig2. The RS1 assembly then switches the output Q to the high level. The signal sig3 is initially at the high level and switching it to low level will switch output Q back to low level 0. At the same time, since signal sig3 is still high, input s of set RS2 receives a low level. Since input r of set RS2 is connected to input S which has gone high, signal sigl switches low, which causes signal sig2 to switch high. This mechanism makes the flip-flop sensitive only to rising edges of S, output Q remaining stored at high level until signal sig3 has gone low. To activate output Q again, input S will have to switch low and signal sigl will have to switch high. In parallel, set RS3 initially provides signal sig4 at high level, since output Q is initially low. When signal Q switches high, there are two possibilities.Either the R input was already high at that moment and the sig4 signal remains high, or the R input was low and in this case the sig4 signal switches to low. This mechanism is important because the rising edge of the R input can occur at any time relative to the rising edge of the Q output. In particular, the Q output and the R input can have a rising edge at the same time. In this case, the RS3 assembly enters a metastable state, in which the sig4 signal takes longer to "choose" between a low level and a high level. In the worst case, the sig4 signal will only switch due to the electronic noise present in the circuit and therefore at a random time. Since the standard deviation of this noise can be known, it is possible to know the maximum time Tmax that can be necessary for the sig4 signal to switch.The purpose of circuit 1006 and signal sig6 is to allow time for RS3 to come out of metastability and switch. Gate 1008 allows signal sig6 to be set low only when input R has been high for a duration at least equal to Tmax. Since signal sig3 is the result of the logical OR between signals sig4 and sig6, if sig6 has finally switched low, it will be necessary to wait for duration DT for signal sig3 and then output Q to go low. If it can be guaranteed that duration DT is greater than duration Tmax, signal sig3 will not undergo the random delay due to metastability but only a constant delay equal to DT. In this way, the falling edge of the pulse on output Q can avoid being affected by jitter, which could be detrimental to the result.
[0222] For example, when the flip-flop 200 of the circuit C1 is implemented in the manner described in relation to [Fig. 10], the duration Dmax of the PI pulses is increased by DT, and this duration DT must be compensated by the P2 pulses, by increasing the duration of the P2 pulses by a duration DT2. This results in DT*A1=DT2*A2, therefore DT2=(DT*A1) / A2. For example, in the case where A1=2*A2, then DT2=2*DT.
[0223] Of course, although the flip-flop described in relation to [Fig. 10] is particularly suitable for implementation in the circuit C1 of the converter 1, a simpler flip-flop, for example an RS flip-flop similar to one of the sets RS1, RS2 and RS3, but not addressing the metastability issues can be used to implement circuit Cl.
[0224] [Fig. 11] illustrates an example of implementation of another part of the converter 1 of the preceding figures. More particularly, [Fig. 11] illustrates an example of implementation of the circuit 214, it being understood that the circuit 912 can be implemented in the same way.
[0225] In this example, the circuit 214 comprises an operational amplifier 1100 having a first input, in this example the inverting input -, configured to receive the pulses to be integrated, a second input, in this example the non-inverting input +, receiving any potential, preferably the potential ref as illustrated in [Fig. 11], and an output providing the result of the integration, namely the signal RES1 for the circuit 214. A capacitor C is connected between the output of the amplifier 1100 and the non-inverting input - of the amplifier 1100.
[0226] In this example, the circuit 214 receives the pulses to be integrated on the inverting input - of the amplifier 1100. As a result, a positive or negative pulse causes a decrease or an increase in the signal RES 1 with a slope proportional to the amplitude of the pulse.
[0227] In practice, the circuit 214 of [Fig.l 1] is a low-pass filter with cut-off frequency fO having the behavior of a pure integrator for frequencies greater than the frequency fO. The frequency fO depends in particular on the value of the capacitance C. Thus, it will always be possible to choose the frequency fO sufficiently low compared to the frequency of the signal clk so that the circuit 214 behaves, in the converter 1, like a pure integrator.
[0228] It will be noted that if the inverting input - of the amplifier 1100 is left in the high impedance state, no current flows in the capacitor C and, as the leakage currents are low, the output of the amplifier memorizes its state, that is to say it remains constant and equal to the value it had at the moment when the inverting input - went into the high impedance state.
[0229] [Fig. 12] illustrates an example of implementation of yet another part of the converter 1 of the preceding figures. More particularly, [Fig.l 1] illustrates an example of implementation of the INTP circuit.
[0230] The circuit INTP differs from the circuit 214 in that the output of the amplifier 1100 provides the signal CR and in that a switch 1200 is connected in parallel with the capacitor C. The switch 1200 is configured, when in the on state, to reset to the level of the non-inverting input +, preferably the level ref as illustrated in [Fig. 12], the output of the amplifier 1100. The switch 1200 is, for example, controlled by the signal CMD2 and is then configured to be on at each CMD2 pulse.
[0231] In this example, as in [Fig. 10], the INTP circuit receives the pulses to be integrated on the inverting input of the amplifier 1100. The result is that a positive, respectively negative, pulse causes a decrease, respectively an increase, of the signal CR with a slope proportional to the amplitude of the pulse.
[0232] The implementation of the circuits 214, 912 and INTP is not limited to the examples described in relation to FIGS. 11 and 12, and the person skilled in the art will be able to use standard integrators to implement the circuits INT, INT2 and INTP. In particular, the person skilled in the art is able to provide integrators whose output signal increases, respectively decreases, when they receive a positive, respectively negative pulse to be integrated, as has been described by way of example in relation to FIGS. 3, 6 and 8.
[0233] [Fig. 13] illustrates an example of implementation of yet another part of the converter 1 of the preceding figures. More particularly, [Fig. 13] illustrates an example of implementation of the circuit 218, it being understood that the circuit 910 can be implemented in the same way. In this example, the low supply level of the converter 1, therefore of the comparator 218 is the ground GND, and the ref level is equal to Vdd / 2.
[0234] In this example, the circuit 218 comprises a conventional differential amplifier 1300 (dotted in [Fig. 13]) having an inverting inl input, an inverting in2 input, and an N output.
[0235] In this example, the circuit 218 further comprises, connected to the output N of the amplifier 1300, a stage 1302 comprising an inverter 1304 and two transistors 1306 and 1308, respectively P-channel and N-channel. The inverter 1304 has an input connected to the output N of the stage 1300 and an output providing the complete output signal of the circuit 218. The transistor 1306 is connected between the node N and the potential Vdd, the transistor 1308 being connected between the node N and the low supply potential, here the ground GND. The transistors 1306 and 1308 are controlled by the output of the inverter 1304, that is to say by the output of the circuit 218.
[0236] Transistors 1306 and 1308 implement a positive feedback from the output of inverter 1304 to output N of stage 1300, which makes it possible to accelerate the switching of the signal compl. This makes it possible to obtain a signal compl having very distinct high and low levels, even when the value of input inl is close to that of input in2.
[0237] Of course, the implementation of circuits 218 and 910 is not limited to the example illustrated in [Fig. 13], and the person skilled in the art may use other usual threshold comparators to implement circuits C3 and C6. In particular, the implementation of threshold comparators is not limited to the case where ref is equal to half of the supply voltage of the comparator.
[0238] [Fig. 14] illustrates a detailed example of implementation of converter 1 described in relation to [Fig.9]. In this example, stage STAGE1 corresponds to converter 1 described in relation to [Fig.5], i.e. signal SI corresponds to signal clkref and signal S2 corresponds to signal clk. In this example, K2=K3=1.
[0239] In this example, the low supply potential of converter 1 is ground GND, and the ref level is equal to Vdd / 2. However, the person skilled in the art is able to adapt the following description to the case where the low supply potential is a potential Vss equal to -Vdd and where the ref level is ground GND, and, more generally, to the case where the ref level is different from half the supply voltage of converter 1.
[0240] In this example, the circuit C1, as in [Fig.5] comprises the RS type flip-flop 200 configured to provide the signal CMD1. Furthermore, in this example, the circuit 202 of the circuit C1 comprises a switch IT1 in series with a resistor RI, between the high supply potential Vdd and the node 216, the switch IT1 being preferably connected to the potential Vdd. The switch IT1 is controlled by the signal CMD1 so as to be conducting only during the CMD1 pulses. Thus, each CMD1 pulse results in a positive current pulse PI on the node 216, of level +A1. The amplitude Al of the PI pulses is the difference between the potentials Vdd and ref divided by the value of the resistor RL
[0241] In this example, circuit C2 is implemented in the manner described in relation to [Fig.5]. Furthermore, circuit 212 of circuit C2 comprises, in this example, a switch IT2 in series with a resistor R2, between the low supply potential, here ground GND, and node 216, switch IT2 preferably being connected to the low supply potential GND. Switch IT2 is controlled by signal CMD2 so as to be conducting only during pulses CMD2. Thus, each pulse CMD2 results in a negative current pulse P2 on node 216, of level -A2. The amplitude of pulses P2 is determined by the difference between potentials ref and GND divided by the value of resistor R2. In this example where the amplitude Al of the PI pulses is equal to twice the amplitude A2 of the P2 pulses, the resistor R2 has a value twice that of the resistor RL
[0242] In this example, the circuit INT is implemented in the manner described in relation to [Fig. 11], and therefore comprises an amplifier 1100 and a capacitor C. The inverting input - of the amplifier 1100 of the circuit INT is connected to the node 216, the non-inverting input + of the amplifier 1100 of the circuit INT being connected to the potential ref. Thus, in this example, the signal RES1 increases, respectively decreases, when the amplifier 1100 of the circuit INT receives a negative pulse, respec- tively positive, on its entry
[0243] In this example, the circuit C3 is implemented in the manner described in relation to [Fig. 5]. For example, the circuit 218 is implemented in the manner described in relation to [Fig. 13] and receives the ref level on its inverting input (inl in [Fig. 13] and - in [Fig. 14]) and the signal RES1 on its non-inverting input (in2 in [Fig. 13] and + in [Fig. 14]). Thus, in this example, the signal compl is at a high, respectively low, level when the signal RES1 is higher, respectively lower, than the ref level, and the bit OUT1 is updated with the level of the signal compl at each start of pulse CMD3. As a result, compared to the operation described in relation to the timing diagrams of figures 3, 6 and 8, the direction of variation of the signal RES1 as a function of the polarity of the pulses received by the circuit INT is reversed, which causes an inversion of the bit OUT1.
[0244] In this example, circuit C4 (or 222 or block -A3 / +A3) comprises a switch IT3 and a resistor R3 in series between the output of circuit C3 and node 216. Switch IT3 is controlled by signal CMD3 so as to be conductive only during CMD3 pulses. Thus, each CMD3 pulse results in a pulse P3 on node 216, which has a negative level -A3 or positive +A3 depending on whether the output of circuit C3 (bit OUT1) is respectively at the low level equal to the low supply potential or at the high level equal to the high supply potential. In this example, because the variations of the signal RES1 relative to the polarity of the pulses are inverted, and therefore the state of the bit OUT1 is inverted, compared to the operation described in relation to figures 3, 6 and 8, the polarity of the pulses P3 relative to the high or low state of the signal OUT1 is also inverted compared to what was described in relation to figures 3, 6 and 8.Furthermore, the amplitude A3 of the P3 pulses is determined by half the difference between the Vdd and GND potentials divided by the value of the resistor R3. In this example where the amplitude Al of the PI pulses is equal to twice the amplitude A3 of the P3 pulses, the resistor R3 has a value twice as large as that of the resistor RI.
[0245] In this example, the circuit C5 is implemented in the manner described in relation to [Fig.9], i.e. by the flip-flop 900.
[0246] In this example, the circuit C6 is implemented in the manner described in relation to [Fig.9]. The circuit 910 is for example implemented by a comparator as described in relation to [Fig. 13] having its inverting input (- in [Fig. 14], inl in [Fig.13]) receiving the ref level, its non-inverting input (+ in [Fig.14], in2 in [Fig. 13]) receiving the output signal RES2 of the circuit INT2, and its output providing the signal comp2.
[0247] In this example, circuit C7 is implemented in the manner described in relation to [Fig.9],
[0248] In this example, the INTP circuit is implemented in the manner described in relation to [Fig. 12], the inverting input - of the amplifier 1100 of the INTP circuit receiving the potential ref and the non-inverting input + of this amplifier 1100 receiving the pulses P4. The switch 1200 of the INTP circuit is, for example, controlled by the signal CMD2, so as to be conducting during each pulse CMD2.
[0249] In this example, the circuit INT2 is implemented in the manner described in relation to [Fig. 9]. More particularly, the circuit 908 is implemented in the manner described in relation to [Fig. 1 1], by an operational amplifier 1100 and a capacitor C, the inverting input - of the amplifier 1100 receiving the signal ref, the non-inverting input + of the amplifier 1100 receiving the signal S, and the output of the amplifier 1100 providing the signal RES2. Furthermore, circuit 906 corresponds, in this example, to two resistors R5 and R6, of the same value, each connected between a respective input of circuit 906 and the inverting input - of amplifier 1100 of circuit 908, signal S being available on the inverting input - of amplifier 1100. Circuit 902 corresponds, in this example, to a switch IT5 connected between the output of circuit INT and an input of circuit 906.The switch IT5 is controlled by the CMD4 pulses so that it is only conductive during the CMD4 pulses. Thus, the signal RES 1 is only supplied to the input of the circuit 906 during each CMD4 pulse. Similarly, in this example, the circuit 904 corresponds to a switch IT6 connected between the output of the circuit C7 and another input of the circuit 906. The switch IT6 is controlled by the CMD4 pulses so that it is only conductive during the CMD4 pulses. Thus, the signal CR is only supplied to the input of the circuit 906 during each CMD4 pulse. As a result, the signal S corresponding to the sum of the signals CR and RES 1 is only supplied to the input of the circuit 908, that is to say to the inverting input - of the amplifier 1100 of this circuit 908, during each CMD4 pulse.
[0250] In this example, circuit 915 or block +A4 / -A4 includes a switch IT4, an inverter 1402 and a resistor R4 in series between the output of circuit C6 and circuit INTP. Switch IT4 is controlled by signal CMD3 so that it is only on during pulses CMD3. Thus, each pulse CMD3 results in a pulse P4 supplied to circuit INTP, which has a negative level -A4 when bit OUT2 is high and a positive level +A4 when bit OUT2 is low. The amplitude A4 of the pulses P4 is determined by half the difference between potentials Vdd and GND divided by the value of resistor R4. In this example where the amplitude Al of the pulses PI is equal to twice the amplitude A4 of the pulses P4, resistor R4 has a value twice as large as that of resistor RL
[0251] In this example, the signal RES2 increases, respectively decreases, when the circuit 908 receives a negative pulse S, respectively positive, the signal comp2 is in the high binary state, respectively low, when the signal RES2 is higher, respectively lower, than the level ref, the signal OUT2 is updated with the state of the signal comp2 at the end of each pulse CMD4, and the signal CR increases, respectively decreases, when the pulse P4 received by the circuit INTP is negative, respectively positive. Thus, the person skilled in the art understands that, to implement the negative feedback in the stage STAGE2 (not referenced in [Fig. 14]) according to the example of [Fig. 14], the pulses P4 must be negative when the bit OUT2 is in the high binary state, and positive when the bit OUT2 is in the low binary state. The person skilled in the art is able to modify the polarity of the pulses P4 relative to the binary state of the bit OUT2 in cases where, compared to the example of [Fig.14], the signal RES2 increases, respectively decreases, when the pulses S are positive, respectively negative, and / or the signal comp2 is in the binary state high, respectively low, when the signal RES2 is lower, respectively higher, than the level ref, and / or the bit OUT2 is updated with the complementary binary state of the signal comp2 at each end of pulse CMD4, and / or the signal CR increases, respectively decreases, when the pulses P4 are positive, respectively negative.
[0252] In this example, the DER circuit and the 924 circuit described in relation to [Fig.9] are partly merged, or, in other words, the output of the STAGE2 stage of the converter 1 of [Fig. 14] corresponds to the two bits OUT2 and OUT2-1. More precisely, the 922 circuit of the DER circuit and the 924 circuit are implemented by a 9224 circuit configured to receive the signals OUT, OUT2 and OUT2-1 and to provide the TDCOUT signal, in this example coded in 2's complement on three bits TDCOUT0, TDCOUT1 and TDCOUT2. The 9224 circuit is a digital logic circuit, implemented from combinational logic gates, for example from EXCLUSIVE OR gates ("XOR" in English) and / or inverting gates and / or AND gates ("AND" in English) and / or OR gates, etc. In [Fig. 14], the circuit 9224 is represented in the form of a block, its implementation being within the reach of the person skilled in the art.
[0253] In the detailed implementation example illustrated by [Fig. 14], the circuit 9224 is for example configured to provide the bits TDCOUT0, TDCOUT 1 and TDCOUT2 from the bits OUT1, OUT2 and OU2-1 according to Table 1 below. In this example, the binary state T', respectively '0', of a bit corresponds to a high, respectively low, level of this bit, and the bit TDCOUT2 represents the sign of the signal TDCOUT which corresponds to the concatenation of the bits TDCOUT2, TDCOUT 1 and TDCOUT0.
[0254] [Tables 1] OUT1 OUT2 OUT2-1 TDCOUT 2 TDCOUT 1 TDCOUT0 Value of TDCOUT T' '0' '0' '0' '0' '0' 0 '0' '0' '0' '0' '0' T' 1 T' T' '0' '0' '0' T' 1 '0' T' '0' '0' T' '0' 2 T' '0' T' T' T' T' -1 '0' '0' T' '0' '0' '0' 0 T' T' T' '0' '0' '0' 0 '0' T' T' '0' '0' T' 1
[0255] As for the second order converter 1 described in relation to [Fig.9], based on the descriptions made in relation to figures 1 to 9, the person skilled in the art will be able to adapt the description made in relation to [Fig. 14] to cases where: - the duration Dmax of the PI pulses is different from one period Tclk of the clk signal and is, for example, equal to two periods Tclk of the clk signal, and / or - converter 1 is not implemented in an all-digital phase-locked loop, and / or - converter 1 is of order greater than 2.
[0256] Various embodiments and variations have been described. Those skilled in the art will understand that certain features of these various embodiments and variations could be combined, and other variations will occur to those skilled in the art.
[0257] In particular, the person skilled in the art is able to adapt the description given above to the case where all polarities of the pulses are inverted. More generally, the person skilled in the art is able to invert the high and low states of any of the binary signals and / or any of the pulses CMD1, CMD2, CMD3 and CMD4 and / or the signs of the pulses P1, P2, P3 and P4 by adapting the circuits which receive or supply these signals so that the functionalities implemented by these circuits remain those described in the present application.
[0258] Furthermore, the person skilled in the art will be able to provide other examples of values for the numbers K1, K2, K3 and K4 by adapting the amplitudes A1, A2, A3 and A4 according to the dynamics of the converter 1.
[0259] The person skilled in the art will also be able to foresee other examples of implementation of the circuits described above. For example, circuits C3 and C6 could each be implemented solely by a D flip-flop whose output Q is set to a first binary state, respectively a second binary state, if its input D is lower, respectively higher, than the level ref when its input C receives an edge to which this input C is sensitive, the flip-flop then simultaneously implementing a binary quantization and the memorization of this quantization.
[0260] Finally, the practical implementation of the embodiments and variants described is within the reach of those skilled in the art from the functional indications given above.
Claims
Claims
1. Digital time converter comprising: a first circuit (Cl) configured to provide, at each active edge of a first signal (SI), a first pulse (PI) of duration determined by a difference between said active edge of the first signal (SI) and an immediately successive active edge of a second signal (S2); an integrator circuit (INT) configured, at each first pulse (PI), to integrate: - the first impulse (PI), - a second pulse (P2) starting after the first pulse (PI) and in synchronism with a clock signal (clk), and - a third pulse (P3) starting after the second pulse (P2) and in synchronism with the clock signal (clk), a second circuit (C2) configured to generate, for each first pulse (PI), the corresponding second pulse (P2); a third circuit (C3) configured to quantize on one bit (compl) an output signal (RES1) of the integrator circuit (INT) and, at the start of each third pulse (P3), store on a first bit (0UT1) said binary quantization; and a fourth circuit (C4) configured to generate, for each first pulse (PI), the corresponding third pulse (P3) from the first bit (0UT1), in which the second circuit (C2) is configured to receive the clock signal (clk) and to generate at least one third signal (CMD3) configured to control the memorizations by the third circuit (C3) and the generations of the third pulses (P3) by the fourth circuit (C4), and wherein the first circuit (Cl) is configured to provide a fourth signal (CMD1) indicating a start and an end of each first pulse (PI), the second circuit (C2) being configured to receive the fourth signal (CMD1) and to generate the third signal (CMD3) from the clock signal (clk) and the fourth signal (CMD1).
2. A converter according to claim 1, wherein: the first (Cl) and second (C2) circuits are configured so that the second pulses (P2) place a conversion zero of the converter (1) in the middle of the dynamics of the converter (1); and the fourth circuit (C4) is configured so that the integration of each third pulse (P3) implements a subtraction of the stored binary quantization (0UT1) at the start of said third pulse (P3).
3. Converter according to claim 1 or 2, in which: the converter (1) is of the sigma-delta type; the first (Cl) and second (C2) circuits are configured so that each second pulse (P2) causes, on the output signal (RES1) of the integrator circuit (INT), a variation of amplitude identical and of opposite sign to an amplitude variation that would cause a first pulse (PI) of duration determined by half the dynamic range of the converter (1); and the fourth circuit (P4) is configured so that each third pulse (P3) implements a negative sigma-delta feedback.
4. A converter according to any one of claims 1 to 3, wherein: the first, second and third pulses (PI, P2, P3) are signed; the first and second circuits (Cl, C2) are configured so that a sign of the second pulses (P2) is opposite to a sign of the first pulses (PI) and a maximum duration of the first pulses that the converter (1) can convert multiplied by an amplitude of the first pulses (PI) is equal to twice a duration of the second pulses (P2) multiplied by an amplitude of the second pulses (P2); and the fourth circuit is configured so that a sign of each third pulse (P3) is determined by a state of the first bit (OUT) and that an amplitude of the third pulses (P3) multiplied by a duration of the third pulses is equal to the duration of the second pulses (P2) multiplied by the amplitude of the second pulses (P2).
5. Converter according to any one of claims 1 to 4, in which the second and third pulses (P2, P3) each have a duration which is a multiple of a period of the clock signal (clk), preferably equal to a period of the clock signal (clk).
6. Converter according to any one of claims 1 to 5, wherein an output (TDCOUT) of said converter (1) is at least partly determined by the first bit (0UT1).
7. Converter according to any one of claims 1 to 6, wherein the converter (1) is of the multi-stage noise-shaping sigma-delta type, the converter (1) comprising a first stage (STAGE 1) comprising the integrator circuit (INT) and the first, second, third and fourth circuits (C1, C2, C3, C4), and a second stage (STAGE2) configured to receive an output signal (RES1) of the integrator circuit (INT) of the first stage (STAGE 1) and to provide a digital output signal (NUM; 0UT2, 0UT2-1) of the second stage from the output signal (RES1) of the integrator circuit (INT) of the first stage, an output (TDCOUT) of said converter (1) being determined at least in part by the first bit (0UT1) and the digital output signal (NUM; 0UT2, 0UT2-1) of the second stage (STAGE2).
8. Converter according to claim 7, wherein the second stage (STAGE2) is configured to: generate, after each third pulse (P3), a fourth pulse (CMD4) starting in synchronism with the clock signal (clk); integrate, during each fourth pulse (CMD4), a sum (S) of the output signal (RES1) of the integrator circuit (INT) of the first stage (STAGE1) and a feedback signal (CR), quantize on a bit (comp2) a result of said integration and store on a second bit (0UT2) said binary quantization (comp2) at the end of the fourth pulse (CMD4); generate, during each third pulse (P3), a fifth pulse (P4) from the second bit (0UT2), and the feedback signal (CR) by integrating said fifth pulse (P4); and store the second bit (0UT2) on a third bit (0UT2-1) at each active edge of the first signal (SI), the digital output signal (NUM;0UT2, 0UT2-1) of the second stage (STAGE2) being generated from the second and third bits or corresponding to the second and third bits.;
9. Converter according to claim 7, in which the second stage (STAGE2) comprises: a fifth circuit (C5) configured to generate, after each third pulse, a fourth pulse (CMD4) starting in synchronism with the clock signal (clk); an integrator circuit (INT2) configured to integrate, during each fourth pulse (CMD4), a sum (S) of the output signal (RES1) of the integrator circuit (INT) of the first stage (STAGE 1) and a feedback signal (CR); a sixth circuit (C6) configured to quantize on a bit (comp2) an output (RES2) of the integrator circuit (INT2) of the second stage (STAGE2) and store said binary quantization (comp2) on a second bit (OUT2), at the end of each fourth pulse (CMD4); a seventh circuit (C7) configured, at each third pulse (P3), to provide said feedback signal (CR) by integrating, during said third pulse (P3), a fifth pulse (P4) determined from the second bit (OUT2); and a circuit configured to store the second bit (OUT2) on a third bit (OUT2-1) at each active edge of the first signal (SI), the digital output signal (NUM;OUT2, OUT2-1) of the second stage (STAGE2) being generated from the second and third bits or corresponding to the second and third bits.;
10. Converter according to any one of claims 1 to 9, wherein the converter (1) comprises a first input configured to receive the first signal (SI) and a second input configured to receive the clock signal (clk), and wherein: the second signal (S2) and the clock signal (clk) are merged, or the converter (1) comprises a frequency divider circuit (700), preferably by two, configured to receive the clock signal (clk) and to provide the second signal (S2).
11. An all-digital phase-locked loop (4) configured to receive a signal (clkref) at a first frequency and to provide a signal (clk) at a second frequency equal to N / D times the first frequency, with N and D two positive values, the phase-locked loop (4) comprising a converter (1) according to claim 10, the first input of the converter being configured to receive the signal (clkref) at the first frequency and the second input of the converter being configured to receive the signal (clk) at the second frequency.
12. Phase-locked loop (4) according to claim 11 comprising a circuit (accN) configured to provide a first digital signal (accNOUT) incremented by the value N at the first frequency; a circuit (accD, reg) configured to provide a second digital signal (accDOUT) incremented by the value D at the second frequency and a third digital signal (accDOUT) corresponding to a storage of the second digital signal (accDOUT) clocked at the first frequency; a digital filter (402) configured to receive a subtraction of an output (TDCOUT) of the converter (1) and the third digital signal (accDOUT1) from the first digital signal (accNOUT) and to control an oscillator (404) configured to provide the signal (clk) at the second frequency, in which the second signal (S2) of the converter (1) is merged with the signal (clk) at the second frequency.
13. Phase-locked loop according to claim 11 comprising either a circuit (accN) configured to provide a first digital signal (accNOUT) incremented by the value N at the first frequency, and a circuit configured for a second digital signal incremented by the value D / 2 at the second frequency, with D even, or a circuit configured to provide a first digital signal incremented by the value 2N at the first frequency and a circuit (accD) configured to provide a second digital signal (accDOUT) incremented by the value D at the second frequency; a circuit (reg) configured to provide a third digital signal (accDOUT1) corresponding to a storage of the second digital signal (accDOUT) clocked by the second signal (S2) of the converter; a digital filter (402) configured to receive a subtraction of an output (TDCOUT) of the converter (1) and the third digital signal (accDOUT1) from the first digital signal (accNOUT) and to control an oscillator (404) configured to provide the signal (clk) at the second frequency, wherein the second signal (S2) of the converter (1) is obtained by dividing by two the frequency of the signal (clk) at the second frequency.
