HIGH-RESOLUTION SIGNAL RECEPTION

DE502022001949D1Active Publication Date: 2024-11-14ABB (SCHWEIZ) AG
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
DE502022001949
Authority / Receiving Office
DE · DE
Patent Type
Patents
Current Assignee / Owner
Filing Date
2022-06-29
Publication Date
2024-11-14
Estimated Expiration
2042-06-29

AI Technical Summary

Technical Problem

In real-time networks, asynchronous clock frequencies between sending and receiving participants limit signal resolution, and existing synchronization methods face challenges in achieving high resolution and scalability due to oscillator demands and cascading negative effects like jitter.

Method used

A flexible method is introduced where the reception clock of the receiving participant is derived from the counter clock, with its phase position adapted to match the signal clock, using a phase-locked loop to ensure value-synchronous representation of local time, allowing for higher sampling rates and reduced quantization errors.

Benefits of technology

This approach enhances signal resolution and synchronization, reducing quantization errors and allowing for higher bit rates, while maintaining synchronization across multiple participants without significant oscillator demands.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader
Need to check novelty before this filing date? Find Prior Art

Description

[0001] The present invention relates to a method for receiving a signal by a subscriber of a real-time network, wherein the signal has a signal clock with a signal clock frequency and the subscriber comprises a counter which has a counter clock with a counter clock frequency and maps a local time of the subscriber.

[0002] In a real-time network, data is transmitted between participants in the form of digital signals. A sending participant has a first counter with a first counter clock having a first clock frequency, and a receiving participant has a second counter with a second counter clock having a second clock frequency. The second clock frequency often does not match the first clock frequency, i.e. it is neither synchronous nor in phase. A counter increments its counter reading at the counter clock rate with the associated clock frequency, whereby the clock period can also be viewed as the inverse of the clock frequency. In this view, the counter increments its counter reading every clock period.

[0003] The sending station transmits the digital signal to the receiving station. In order to transmit information about the signal, the sending station quantizes the information to be transmitted with the signal at specific times, whereby these times in the signal clock are specified by a signal clock frequency. The quantized values ​​of the signal are thus spaced apart by the signal clock period. Since the sending station has a first counter, it uses the first clock frequency of the first counter for the signal clock frequency. The receiving station samples the signal at a sampling clock frequency to quantize the signal. Since the receiving station has a second counter, it in turn uses the second clock frequency of the second counter for the sampling clock frequency. If the second clock frequency is matched to the signal clock frequency, the signal can be reconstructed without loss.In asynchronous transmissions, it is also common to use a sampling frequency that is a multiple of the bit rate / symbol rate, which is known as oversampling. Sampling frequencies of up to several hundred MHz are possible. However, the signal sampling resolution is limited by the second clock frequency of the receiving station's second counter. For example, if the second counter is operated at a clock frequency of 125 MHz, i.e., a second clock period of 8 ns, this results in a resolution of 8 ns. The resolution can only be improved by increasing the clock frequency, which, however, is not possible or only possible to a very limited extent.

[0004] There are known methods in which a master provides a clock signal, which is then reconstructed by a slave and fed back to the master. The master determines the deviation of the fed-back clock from the original clock using a phase detector and adjusts its clock signal to reduce the deviation. This method is being developed at the CERN Institute as part of the White Rabbit project. To achieve a resolution of less than one nanosecond, an extension of SyncE (Synchronous Ethernet) is used. This enables very high resolution. However, it also places very high demands on the oscillator regarding frequency deviation and jitter behavior. Furthermore, only around 10 devices can be connected in series, since the devices always synchronize their signal with the clock of a connected device.Since the clock of the port to which the other participants are to be synchronized also generates the signal that is sent to these participants, negative effects such as jitter etc. cascade with the clock synchronization.

[0005] US Pat. No. 8,698,530 B2 discloses a global clock frequency that is made available to all participants. To achieve a higher resolution than the global resolution based on the global clock frequency, the local clock frequency is continuously controlled to match the clock frequencies of the participants. For this purpose, the participants' counters are operated in different time domains, meaning that each participant has a different local time.

[0006] Furthermore, WO 2020 / 261066 A1 discloses a method for determining a time in a digital processing system for high-precision time stamps. For this purpose, a receiver contains a reference time counter, which is incremented by a fixed amount with each cycle of a digital reference clock, and a network time counter, which can be incremented by a variable amount with each cycle of a digital network clock. The variable amount by which the network time counter can be incremented can be influenced by the reference clock, which runs at a lower frequency but with a higher accuracy than the network clock, in order to synchronize the network time counter and the reference time counter.

[0007] It is an object of the present invention to provide a flexible method for receiving signals.

[0008] This object is achieved according to the invention in that a receive clock of a receive counter of the subscriber is provided for sampling the signal, wherein the receive clock is derived from the counter clock, whereby the receive counter also maps the local time of the subscriber, wherein a phase position of the receive clock is adapted to the phase position of the signal clock when derived from the counter clock, and wherein the signal is sampled with a receive clock frequency of the receive counter.

[0009] The signal can be further processed after sampling. Adjusting the phase position of the receive clock to the phase position of the signal clock can be achieved, for example, by shifting the phase position of the receive clock toward the phase position of the signal clock.

[0010] Since it is not possible to transfer a signal from one clock domain to another asynchronous clock domain without special mechanisms (without generating inaccuracies), according to the invention, not only the counter in a counter clock domain is provided at the receiver to represent the local time, but also an additional receive counter, whose receive clock is derived from the counter clock, in a receive clock domain. This receive counter is clock-asynchronous but value-synchronous and, taking into account a phase shift compared to the counter clock, also represents the local time. Thus, not only is a counter clock with a counter clock frequency provided at the subscriber to represent the local time, but also a receive counter with a receive clock frequency. After the time stamp has been formed in the receive clock domain, the information can be transferred to the counter clock domain via an interface.It is therefore desirable to ensure a sufficiently small phase shift between the counter clock and the receive clock so that the time information of the two domains is synchronized.

[0011] Preferably, the receive clock is derived from the counter clock using a phase-locked loop.

[0012] The receive clock frequency can be the same as the counter clock frequency. However, the receive clock frequency is preferably a multiple of the counter clock frequency, which allows the receive counter to have a multiple of the counter's resolution, since the signal can be sampled at a higher bit rate.

[0013] It can be assumed that the receive clock frequency differs at least slightly from the signal clock frequency, causing the receive clock and signal clock to drift apart, resulting in a phase shift. This can be remedied, at least temporarily, by a one-time adjustment of the phase position of the receive clock to the signal clock. It is advantageous to also adjust the counter reading of the receive counter to the adjusted phase position, which changes the quantization time of the receive counter and allows the signal to be sampled with less uncertainty.

[0014] Preferably, a sign of a phase shift between the receive clock and the signal clock is determined, and based on this, the phase position of the receive clock is changed in order to adapt the phase position of the receive clock to the phase position of the signal clock. By determining the sign of the phase shift, it is determined in which direction the phase position of the receive clock must be shifted to reduce the phase shift. However, if the phase position is shifted by an amount greater than the phase shift, an overshoot occurs, which in turn changes the sign of the phase shift.

[0015] The phase position of the receive clock can be changed over several receive clock periods. This allows a successive adjustment of the phase position of the receive clock to the phase position of the signal clock. If the phase position is changed in one direction based on the previously determined sign of the phase shift and an overshoot occurs (i.e. the phase difference changes its sign), the phase position of the receive clock is shifted in the other direction again in the next receive clock period, which can cause an overshoot in the other direction. It is therefore possible that the phase position of the receive clock is changed alternately in one direction and thus overshoots alternately in both directions, i.e. oscillates around the phase position of the signal clock frequency.

[0016] If a phase-locked loop is used to generate the receive clock from the counter clock, it is advantageous if it supports so-called dynamic phase shifts. The size of these phase shifts is preferably constant. This allows the phase position of the receive clock relative to the counter clock (which serves as the reference clock) to be shifted in small steps (<< 1 ns, depending on the FPGA or ASIC used), which in turn allows the phase position of the receive clock to be continuously adjusted to the phase position of the signal clock.

[0017] Preferably, the phase position of the receive clock is changed for each receive clock period by shifting it by a positive or negative phase increment. The current phase position can be determined based on a signal edge. For example, with 8B10B coding, signal edges always occur. If, based on the sign of the phase shift, the phase position of the receive clock frequency is changed by one phase increment each over several receive clock periods, a phase deviation of less than one (positive or negative) phase increment remains after a sufficient number of shift steps.

[0018] Preferably, the counter reading of the receive counter is changed analogously to the phase position of the receive clock. Thus, the counter reading adjusts the quantization time of the receive counter to the local time, which ensures that the clock is correctly quantized and displayed. By adjusting the phase position of the receive clock to the phase position of the signal clock and the resulting adjustment of the receive counter, it can be ensured that the signal is sampled with minimal quantization errors. The resolution with which the received signal can be measured is thus no longer directly dependent on the counter clock, but rather on the currently occurring phase shift.

[0019] The signal clock frequency is known in advance and can preferably be reconstructed from the signal by the subscriber. If the signal is line-coded, for example, with 8B10B line coding, reconstructing the signal clock frequency from the signal is particularly easy.

[0020] The present invention is described below with reference to the Figuren 1 and 2 which show exemplary, schematic and non-limiting advantageous embodiments of the invention. Fig.1 an exemplary real-time network, Fig.2 an adaptation of the counter clock to the phase shift.

[0021] In Fig. 1 1 shows a real-time network 1, with two participants 11, 12 being shown here only as examples. A transmitting participant 11 comprises a first counter Z1 with an associated first counter clock ZT1, wherein the first counter clock ZT1 has a first clock frequency f1. The counter reading of the first counter Z1 increases with the first counter clock ZT1 and thus represents a first local time t1 of the first participant 11. The first local time t1 of the first participant 11 is thus quantized with the first counter clock ZT1 of the first participant 11.

[0022] The first participant 11 further comprises a transmission counter Z11, which has a transmission clock ZT11. The transmission clock ZT11 is derived from the first counter clock ZT1 and can, for example, be a multiple thereof. The transmission clock ZT11 has a transmission clock frequency f11.

[0023] The second participant 12 comprises a second counter Z2 with an associated second counter clock ZT2, which has a second clock frequency f2. The counter reading of the second counter Z2 increases with the second counter clock ZT2 and thus represents a second local time t2 of the second participant 12, wherein the second local time t2 of the second participant 12 is quantized with the second counter clock ZT2.

[0024] A first clock period T1 can be regarded as the inverse of the first clock frequency f1 and a second clock period T2 as the inverse of the second clock frequency f2.

[0025] The transmitting participant 11 transmits a signal S to the receiving participant 12, wherein the signal S is quantized with the first transmission clock ZT11 of the first participant 11. The first transmission clock ZT11 is thus referred to as the signal clock ZT11, and the first transmission clock frequency f11 is referred to as the signal clock frequency f11.

[0026] Since the reception of the signal S by the second participant 12 is considered in the following, the second participant 12 is referred to simply as participant 12, the second counter Z2 as counter Z2, the second counter clock ZT2 as counter clock ZT2, the associated second counter clock frequency f2 as counter clock frequency f2 and the second local time t2 as local time t2.

[0027] According to the invention, subscriber 12 has a receive counter Z20 in addition to counter Z2. Receive counter Z20 has a receive clock ZT20 with a receive clock frequency f20, with the counter reading of receive counter Z20 increasing with receive clock ZT20. Receive clock ZT20 is provided for sampling signal S and is derived from counter clock Z2. Thus, receive counter Z20 maps the local time t2 of subscriber 12, whereby local time t2 is also quantized with receive clock frequency f20.

[0028] Preferably, the receive clock ZT20 is formed from the counter clock ZT2 using a phase-locked loop.

[0029] However, since the receive clock ZT20 is intended for sampling the signal S, the phase position of the receive clock ZT20 is adjusted to the phase position of the signal clock ZT11 when derived from the counter clock ZT2. Thus, the signal S is sampled by the receive clock ZT20 with a low quantization error.

[0030] However, the counter clock frequency f2 and the receive clock frequency f20 do not have to match; rather, the receive clock frequency f20 can be a multiple of the counter clock frequency f2. Preferably, the counter clock frequency f2 corresponds to the symbol clock frequency of the signal S, whereby the symbol clock frequency is one x-th of the signal clock frequency f11, where x represents the number of bits per symbol. For 8B10B coding, for example, x=10 bits per symbol is provided, which, with a signal clock frequency f11 of 512 MHz, results in a symbol clock frequency of 51.2 MHz. If a single data rate is provided (SDR - single data rate), the receive clock frequency f20 corresponds to the signal clock frequency f11. With a double data rate (DDR - double data rate), the receive clock frequency f20 corresponds to half the signal clock frequency f11. This results in a reception clock frequency f20 of 256 MHz for a signal clock frequency f11 of 512 MHz.

[0031] The resolution of a counter usually corresponds to the clock period of the associated counter, so for a counter with a clock period of 20ns this means a resolution of 20ns.

[0032] In Fig. 2 The real time t is shown on the abscissa. The counter reading of counter Z2 and the receive counter Z20, as well as the signal clock ZT11, are shown on the ordinate as steps. Each step of counter Z2 represents an increment of the corresponding counter reading, and each step of the receive counter Z20 represents an increment of the corresponding counter reading.

[0033] The unquantized local time t2 is shown as a straight line. The counter Z2 increments its counter reading at the counter clock ZT2, i.e., at the counter clock frequency f2 (or, in other words, at each counter clock period T2), by a plurality of counter increments I2. The number of counter increments I2 depends on the duration of a counter increment I2 and the desired counter clock period T2. The second local time t2 is only accurately reproduced at the times of this counter reading increment. Thus, the local time t2 is only accurately reproduced by the counter Z2 when the counter Z2 intersects the local time t2. At all other times, the counter Z2 exhibits quantization errors.

[0034] The second participant 12 receives a signal S, which is quantized with the signal clock ZT11, i.e., with the signal clock frequency f11. According to the invention, a receive clock ZT20 of a receive counter Z20 is derived from the counter clock ZT2, wherein the receive clock frequency f20 of the receive clock ZT20 corresponds to the counter clock frequency f2 of the counter clock ZT2 or is a multiple thereof. The receive counter Z20 increments its counter reading with the receive clock ZT20, i.e., with the receive clock frequency f20, or in other words, each receive clock period T2 by a plurality of receive counter increments I20, wherein the number of receive counter increments I20 depends on the duration of a receive counter increment I20 and the desired receive counter clock period T2.

[0035] The reception counter Z20 is in Fig. 2 shown in dashed lines. Furthermore, Fig. 2 The signal clock ZT11 is shown dotted. The signal clock ZT11 and thus also the signal clock frequency f11 may be unknown or known to the second participant 12. However, the (nominal) signal clock frequency f11 is preferably known to the participant 12 in advance.

[0036] In the example shown, the receive clock frequency f20 corresponds to twice the counter clock frequency f2, which means that the receive clock period T20 corresponds to half the counter clock period T2. It is advantageous if the receive clock frequency f20 is a multiple, e.g., five times, of the counter clock frequency f2, which also means that the resolution of the receive counter Z20 is a multiple of the counter Z2, and the signal S can be sampled with this increased resolution.

[0037] The reception counter Z20 increases its counter reading with the reception clock frequency f20, i.e. per reception clock period T20 by a plurality of reception counter increments I20, whereby the local time t2 is only reproduced exactly at these moments, i.e. only when Fig. 2 the reception counter Z20 intersects the local time t2.

[0038] Under real-world conditions, a phase shift dP can be assumed, which is why the change in the counter reading of the receive counter Z20 occurs at slightly different times than the change in the counter reading of counter Z2. It can also be assumed that the receive clock frequency f20 differs from the signal clock frequency f11 under real-world conditions.

[0039] Therefore, according to the invention, the phase position of the receive clock frequency f20 (and the receive clock period T20) is adapted to the phase of the phase position of the signal clock period T11 (and thus to that of the signal clock frequency f11). It may happen that the absolute phase shift dP between the receive clock period T20 and the signal clock period T11 is unknown, although the sign of the phase shift dP is known. For this purpose, it can be determined, for example, whether the phase shift dP increases or decreases when the receive clock period T20 changes. This allows conclusions to be drawn as to whether the phase shift dP between the receive clock period T20 and the signal clock period T11 is positive or negative.

[0040] This results in a synchronization of the receive clock period T20 to the signal clock period T11. It can be assumed that, despite synchronization, an absolute phase shift dP remains, resulting in a small quantization error after synchronization. To ensure that the signal clock period T11 and the receive clock period T20 do not diverge, it is advantageous for the synchronization to be permanent.

[0041] The phase position of the receive clock pulse is preferably shifted by a phase increment PI equal to a counter increment I10 of the receive counter Z20. The sign of the phase shift dP is determined, for example, using a phase detector. The sign of the phase shift dP can be determined, for example, by first shifting the phase of the receive clock pulse in one direction and then determining whether the phase shift dP has increased or decreased. Thus, after each introduction of a phase increment PI, it can be determined whether a positive or negative phase increment PI will be added in the next receive clock period T20. If, for example, the phase shift dP is negative but less than a phase increment PI, then adding a positive phase increment PI will result in a positive phase shift dP (which is also less than a phase increment PI).Therefore, in the next receive clock period T20, a negative phase increment -PI is added, etc. This alternating addition of positive and negative phase increments would, for example, occur permanently if the receive clock period T20 corresponds exactly to the signal clock period T11, but a phase shift dP smaller than a phase increment PI remains.

[0042] In Fig. 2 Initially, a negative phase shift dP occurs, which, for example, corresponds in magnitude to three and a half phase increments Pl and is thus greater in magnitude than a phase increment PI, which is why a phase increment Pl is added in the second receive clock period T20 shown. In the next two receive clock periods T20, it is determined that the phase shift dP is still negative in each case, which is why the phase of the receive clock period T20 is shifted by a further positive phase increment Pl. Thus, the phase shift dP is successively reduced to minus two and a half phase increments PI, minus one and a half phase increments PI, and finally minus half a phase increment PI. After a further phase increment PI has been added, a phase shift dP of half a phase increment Pl occurs.

[0043] The transient response is now complete, as the phase shift dP is positive and has thus changed sign. In the next receive clock period T20, the phase of the receive clock period T20 is shifted by a negative phase increment -PI. Subsequently, the phase shift dP is negative, which is why the phase of the receive clock period T20 is shifted by a positive phase increment PI, and so on.

[0044] Shifting the phase of the receive clock period T20 by a phase increment PI can be done, for example, by a phase-locked loop.

[0045] The receive clock frequency f20 becomes asynchronous to the counter clock frequency f2 by shifting the phase position of the receive clock frequency T20. However, since the phase shifts dP are known, they can be taken into account in further processing.

[0046] It is particularly advantageous if, in addition to changing the phase of the receive clock period T20, the counter reading of the receive counter Z20 is also changed by a counter difference dZ corresponding to the phase shift dP, ie, with a positive phase shift dP, it is lengthened by a counter difference dZ=dP and with a negative phase shift dP, it is shortened by a counter difference dZ=dP. In particular, if the phase shift dP is changed by a phase increment PI (corresponding to a counter increment I20), the counter difference dZ is then changed by the counter increment I20. Fig. 2 There is always a change in the phase shift dP by one phase increment PI and an equivalent change in the counter reading of the receive counter Z20 by one counter increment I20 in the amount of one phase increment PI. If the counter reading of the receive counter Z20 in Fig. 2 are not adjusted, then the receive counter would no longer be quantized in the local time t2.

[0047] By adjusting the receive counter Z20, it is ensured that the local time t2 continues to be correctly represented even after the phase shift dP by the receive counter Z20. Fig. 2The change in the phase position and the counter reading of the receive counter Z20 occurs in the same receive clock period T20 of the receive counter Z20. However, the counter reading can also be changed in a later receive clock period T20 by the counter difference dZ, or even piecewise over several counter increments I20. Of course, the application of the added counter difference dZ is only possible if the receive counter Z20 has a resolution that is greater than a receive frequency f20. For example, a resolution of 1 µs / 2^16, i.e., approximately 15.26 ps, can be provided, which corresponds to a frequency of 65.536 GHz. This resolution is 256 times higher than the receive clock frequency f20. If the phase shift dP corresponds, for example, to one 24th of the reception clock period T20, the counter difference dZ corresponds to 65.536 GHz / (256 MHz * 24) = 10.6 (6 is periodic).To avoid accumulation of rounding errors, all decimal places are preferably taken into account and the counter difference dZ is exactly adjusted to the phase shift dP.

[0048] The receive counter Z20 requires a resolution that is at least equal to the counter difference dZ. An even higher resolution is desirable, especially for the use of digital filters. If the receive counter Z20 has a receive clock frequency f20 of 51.2 MHz, it will be incremented 1280 times in one receive clock period T20 at a resolution of 15.26 ps.

[0049] The receive counter Z20 and the counter Z2 thus represent the same local time t2 and are therefore "value-synchronous," even though the receive clock frequency f20 and the counter clock frequency f2 are asynchronous. However, since the phase shifts are known and arranged before the clock data recovery, they can be taken into account in further processing.

[0050] In reality, counter oscillators exhibit a maximum deviation of + / - 50 ppm, meaning that two signals can deviate by a maximum of 100 ppm. Therefore, it is sufficient if the (accumulated) phase shift dP is not corrected after each counter increment I20 of the receive counter Z20. It is therefore advantageous if the phase shift dP is only corrected after a maximum phase deviation, preferably equal to half a phase increment PI, has been exceeded.

Claims

1. A method for receiving a signal (S) by a subscriber (12) of a real-time network (1), the signal (S) having a signal clock (ZT11) having a signal clock frequency (f11) and the subscriber (12) comprising a counter (Z2), which comprises a counter clock (ZT2) having a counter clock frequency (f2) and which maps a local time (t2) of the subscriber (12), wherein a reception clock (ZT20) of a reception counter (Z20) of the subscriber (12) is provided for sampling the signal (S) and the reception clock (ZT20) is derived from the counter clock (ZT2), whereby the reception counter (Z20) maps the local time (t2) of the subscriber (12), wherein a phase position of the reception clock (ZT20) is adapted to the phase position of the signal clock (ZT11), when said reception clock is derived from the counter clock (ZT2), and wherein the signal (S) is sampled at a reception clock frequency (f20) of the reception counter (Z20).

2. The method according to claim 1, characterized in that the reception clock (ZT20) is derived from the counter clock (ZT11) using a phase-locked loop.

3. The method according to any of claims 1 to 2, characterized in that the reception clock frequency (f20) of the reception clock (ZT20) corresponds to a counter clock frequency (f2) of the counter clock (ZT2).

4. The method according to any of claims 1 to 2, characterized in that a reception clock frequency (f20) of the reception clock (ZT20) is a multiple of a counter clock frequency (f2) of the counter clock (ZT2).

5. The method according to any of claims 1 to 4, characterized in that a sign of a phase shift (dP) between the reception clock (ZT20) and the signal clock (ZT11) is determined, and based on this the phase position of the reception clock (ZT20) is changed to adapt the phase position of the reception clock (ZT20) to the phase position of the signal clock (ZT11).

6. The method according to claim 5, characterized in that the phase position of the reception clock (ZT20) is changed for each reception clock period (T20).

7. The method according to any of claims 1 to 6, characterized in that the phase position of the reception clock per reception clock period (T20) is changed by shifting said phase position by a positive or negative phase increment (PI).

8. The method according to any of claims 1 to 7, characterized in that the counter reading of the reception counter (Z20) is changed analogously to the phase position of the reception clock (ZT20).

9. The method according to any of claims 1 to 8, characterized in that the subscriber (12) reconstructs the signal clock (ZT11) from the signal (S).

10. The method according to claim 9, characterized in that the signal (S) is line-coded.

11. The method according to claim 10, characterized in that the signal (S) is coded with the 8B10B line coding.