1. Field of the Invention
This invention is directed to generation of a synchronization clock for a telecommunication system and more particularly to Integrated Timing Systems and Circuits (ITSC) which are used to implement universal transmission synchronizer (UTS).
The ITSC allow the UTS to integrate all digital PLL (DPLL), analog PLL (APLL) and system timing control circuits into a single ASIC solution. The UTS may be used for wireless, optical, or wireline transmission systems and for a wide range of data rates.
2. Background Art
Maintaining accurate timing is critical to the transmission of high speed data via telecommunication networks. Land based, cellular, and satcom networks require precision timing to prevent corruption of the transmitted data.
Timing is derived from external timing devices which are synchronized to a Primary Reference source such as a Cesium Beam Standard. (The element Cesium is extremely stable and can be excited by radio energy to produce a 9.44 GHz reference frequency that is electronically maintained to 1 part in 10E13 stability). The Global Positioning System receives and rebroadcasts a Cesium reference for use by telecommunication systems throughout the world. This same GPS timing signal is used by telecommunication systems or navigation systems.
This primary reference is not always possible. When it is not, alternate sources are used to maintain the performance of the appropriate telecommunication system or navigation system. Some kind of synchronizer is usually used to provide a reference clock to telecommunication equipment. Such synchronizer accepts a primary reference source as one of its inputs. It also accepts a line input such as an optical transmission line.
The synchronizer passes the primary reference source to the network equipment in accordance to a set of performance rules. If the rules are violated, the synchronizer switches to the Line timing source and passes that to the network equipment. The line timing source is generated by a different piece of network equipment which is also synchronized to an external primary reference and therefore should be as accurate as the external input.
The synchronizer has its own clock which is normally synchronized to the External or Line input. The synchronizer clock stores information from the synchronization reference clock.
If the synchronization reference is interrupted, then the synchronizer uses its stored data to maintain the stability of its clock. This is referred as hold-over mode. Once the reference signal is restored the synchronizer will switch back to the reference clock.
If the hold-over clock can not provide the stability required because the stored data is corrupted or some other malfunction, then the synchronizer will switch to free-run mode.
In free-run the accuracy of the timing signal is the basic accuracy of the clock in the synchronizer with no synchronization reference clock.
Current synchronizers use DPLLs for synchronized clock generation. DPLLs allow lowering loop bandwidth in order to comply with the communication standards.
Synchronizer DPLL can be implemented using digital to analog converter (DAC), or direct digital frequency synthesis (DDFS), or direct digital phase synthesis (DDPS).
Current DPLLs typically use microcomputers, EEPROM (electrically erasable programmable read-only memory) and a high resolution DAC (digital phase detector) for controlling the VCXO.
Generally, the use of currently available DACs in DPLL designs necessitates the use of a TCVCXO (temperature compensated voltage controlled crystal oscillator). This special type of oscillator is expensive and must be manufactured with a relatively high frequency of oscillation for providing a telecommunication terminal with a wide range of clock signals derived from the output without having to use additional PLLs. However, this high frequency design makes the oscillator more expensive.
The temperature drift is yet another handicap of DAC-based designs that must be compensated. Also, current DAC phase drift which, as a result, may build up. These limitations demand additional and expensive circuitry for improving the performance of the DPLL.
Other known type of DPLL uses the DDFS method.
The DDFS implies eliminating each n-th pulse in an M-pulses sequence of an incoming digital signal, filtering the resultant signal, eliminating the undesired side bands, and extracting the desired frequency. The circuits based on DDFS are provided with a microcontroller and an EEPROM for determining n, M and effecting the deletion. Also, the DDFS algorithm requires complex logic and long acquisition times. Furthermore, if a low frequency off-shelf oscillator such as for example a temperature compensated crystal oscillator (TCXO) is used in this configuration, an additional analog PLL is necessary for obtaining the desired high frequency by multiplying the frequency of TCXO's fixed reference clock.
Yet another disadvantage of the current DDFS is that the clock has rather high jitters, such that another additional analog PLL is generally used for reducing the jitters.
Still other DPLL implementation can be based on the DDPS method which has been introduced in the U.S. Pat. No. 5,910,753 Bogdan 8 Jun. 1999.
Although DDPS method eliminates the above disadvantages of the DAC and DDFS based solutions and significantly reduces complexity and cost, it still requires external analog amplifiers and VCXO for complete implementation of a transmission synchronizer.
There was a need for a synchronizer and a method of synchronization which will further reduce cost and complexity and allow higher degree of on-chip integration by eliminating the external analog amplifiers and VCXO for a wide variety of telecommunication terminals.
Purpose of the Invention
It is an object of present invention to provide a universal synchronizer for use in variety of telecommunication systems based on digital phase frequency synthesis (DPFS). The synchronizer of the invention may be used for wireless, optical, or wireline transmission systems and works well with a wide ranges of data rates. The synchronizer according to the invention may be used for example for SONET line-timing (frame) clock generation, and may be adapted to provide SONET minimum clock (SMC) hold-over and free-run capabilities, as well as external timing clocks generation with Stratum 3 hold-over and free-run capabilities.
It is other object of the present invention to create systems and circuits which allow a complete on-chip integration by eliminating all the external components like DACs, VCXOs, analog PLLs, microcontrollers and EEPROMs.
Still other object of the invention is to create new digital phase detection techniques,
Accordingly the invention provides DPFS (see
The DPFS method produces similar waveforms as commonly used DDFS method, but DPFS inserts single gates delays into pulses stream instead of eliminating the whole clock cycles from a synthesized clock. Therefore, the phase hits and resulting jitter are reduced by 10 times compared to the DDFS method. The DPFS method allows producing any fS1 clock waveform by using phase steps which are in a range of a gate propagation delay. The gate delays insertions and resulting phase/frequency adjustments can performed by a synthesized clock generator (SCG) which is introduced in
Synthesized Clock Generator (SCG)
The invention also includes the synthesized clock generator (SCG), for carrying out the DPFS method to produce the waveforms which are shown in
The first SCG implementation method is based on moving reference clock entry point; wherein:
The first SCG implementation method is conceptually presented in
The delays density register (DDR) defines a number of fF3 cycles which occur between consecutive increments or decrements of a phase of fF2 clock by a single gate delay time Td.
The delays capture register (DCR) allows capturing a waveform which contains whole fF3 cycle. The delay calibration circuits (DCC) allow an estimation of an average Td, and provide measurements of the captured fF3 positioning along the delay line.
Based on the fF3 positioning measurements, it shall be detected periodically that total delay line propagation time amounts to TTOTAL=Td1+Td2+ . . . +TN=Tperiod of fF3 In such cases amount of active delay elements is scaled down without changing the phase of the fS1 clock, by jumping an entry point of fF3 closer to the end of the delay line by a number of delay elements which corresponds to a period of the fF3 clock.
The second SCG implementation method is based on moving an exit point of the synthesized clock from the reference propagation circuit; in a way which adds gate delays for phase increases, and subtracts gate delays for phase decreases.
The second SCG implementation method is conceptually presented in
Said synthesized clock selection can be implemented in two different ways:
The third SCG implementation method is based on adjusting alignment between an exit point of the synthesized clock from the reference propagation circuit versus an input reference clock; in a way which adds gate delays for phase increases, and subtracts gate delays for phase decreases.
The third method is presented in
The moving exit point from the driven by fF2 phase locked delay line is used as a return clock for the PLLxR multiplier, instead of using fixed output of the Inv((N−1)/2+1) to be the PLL return clock.
The fixed output of the Inv((N−1)/2+1) provides the synthesized clock fS1, instead of the moving reference clock exit point.
The exit point alignments introduce phase jumps which cause synthesized clock jitter. The configuration shown in
While any of the three SCG implementation methods is shown above using a particular type of a reference clock propagation circuit, the SCG invention comprises using all the listed below reference clock propagation circuits by any of the three SCG methods:
The invention also includes a new concept of a digital phase detector DPD 1 which is shown in
The DPD1 uses two symmetrical phase counters buffers A/B (PCBA/PCBB), which perform reverse functions during alternative A/B cycles of the frame clock frS2 which is derived from the synchronized clock fS2. During the A cycle, the PCBA counts the number of incoming fF3 clocks, but during the following B cycle the PCBA remains frozen until its content is read by the MC and subsequently the PCBA is reset before the beginning of the next A cycle. Alternatively, the PCBB performs counting during the B cycle and is read and reset during the following A cycle.
Such symmetrical PCBA/PCBB configuration allows much more time for counters propagation by inhibiting counting long before the actual reading takes place. Therefore, much higher frequencies of counted clocks are allowed for the same IC technology.
The above new concept of a digital phase detector, represents one of several possible DPD solutions; which are based on counting a first signal clock during every second signal frame, wherein the second signal frame contains a fixed number of the second signal clocks.
For all said DPD solutions, the invention further includes improving a DPD resolution by introducing a phase capture register. The phase capture register captures a state of outputs of multiple serially connected gates which the first signal clock is continuously propagated through, at the leading edge of the second signal frame.
Such resolution improvement is implemented in the DPD1, by using the phase capture register (PCR) to measure positioning of a last frS2 edge versus fF3 waveform. The PCR and its frame edge decoder (FED), significantly improve phase detection resolution.
Said improvement of a DPD resolution further comprises two different solutions for obtaining the first clock propagation functionality:
The first mentioned solution is shown in the
The second mentioned solution can be implemented as it is explained below. Instead of using the added propagation circuits (APC) from the
The second solution allows using shown in
The second solution eliminates any need for delay calibration of the added propagation circuits (APC), because the replacing inverters Inv(1) to Inv(N) have their delays controlled very accurately by the VCO Control Voltage.
Integrated Synchronizer
The invention further includes a synchronizer which is completely integrated into a single chip (see also
A first/second set of reference signals is named FR1/FR2 and their single representatives are named fR1/fR2 accordingly, throughout this document.
The synchronizer invention further comprises three different configurations which are explained below.
The first synchronizer configuration is based on the SCG which does not have an internal frequency multiplier (see
The above listed status control circuits and phase transfer control circuits can be implemented as separate on-chip microcontrollers or with a single on-chip microcontroller (MC).
The first synchronizer configuration is carried out by an UTS configuration which is based on the DPFS, the SCG, the DPD1 and the DPD2.
As it is shown in
The on-chip implementation of a DPLL mode is explained below.
The DPD1 measures a phase error between TCXO's frequency multiplication fF3 and synthesized clock derivative frS2, and the DPD2 measures a phase error between the fF3 and the DPLL reference derivative fR1.
The MC reads the above phase errors and uses them to calculate a new contents of SCG's delay density register (DDR), which shall fulfill a phase transfer function (PTF) which is preprogrammed on the MC input.
When UTS is working in the DPLL mode, the synthesized output clock fS2 is further applied as a reference for the on-chip APLL which is implemented with the programmable reference selector (RFS) and reference divider (RFD), output PLL (OUTPLL), output clock generator (OCG), programmable return selector (RTS) and return divider (RTD).
The on-chip implementation of an APLL mode uses an alternative reference clock fR2 as a reference for otherwise unchanged the above explained APLL; by selecting the fR2 on the RFS input, instead of the fS2 derivative of the SCG's output which would be selected for the DPLL mode.
It shall be noticed that the first synchronizer configuration uses lower frequency TCXO in order to reduce cost, and uses on-chip PLL cells to multiply TCXOs fF1 clock to a highest frequency which can be still feasible for a particular technology (see
Since the time period of the fF3 clock is reduced to a few nS by TCXO frequency multiplications; fewer delay elements are used for fS2 generation and power supply jitter introduced by the delay elements is proportionally decreased.
The invention further includes a simplified version of the first synchronizer configuration; which can be implemented by eliminating the first digital phase detector (DPD1), and by replacing it with calculations of the first phase error based on analysis of SCG control signals.
The invention of the first synchronizer configuration further includes a DPLL integrated synchronizer, which provides DPLL functions only. The DPLL integrated synchronizer can be obtained from the universal integrated synchronizer by eliminating the reference selector (RFS) and the programmable frequency dividers for reference and return signals of the APLL (RFD and RTD), by applying the fS2 signal directly to the OUTPLL reference input REF.
As it is shown in
The second synchronizer configuration comprises the same circuits and functions as the listed above for the first configuration, with the exceptions which are specified below.
Said second configuration uses an SCG which comprises a frequency multiplier PLLxR for producing a base frequency for the fS1 clock.
The internal SCG PLLxR multiplier provides a frequency increase which is sufficient for achieving a reasonable reduction of a physical size of the SCG. Consequently the single PLLxK frequency multiplier is sufficient to provide the SCG driving clock fF2.
Still another PLLxL frequency multiplier is used with the multiplication factor L which is significantly different than the above mentioned factor R, in order to produce the fF3 clock. The fF3 drives digital phase detectors like the DPD1 and the DPD2, which represent extensive heavy loads which can introduce significant on-chip noise.
The above explained spacing between the fF3 versus the fS1 frequency reduces impact of inter-modulation products.
The third synchronizer configuration is based on the return clock synthesizer (RCS) (see the
The third synchronizer configuration is carried out by an UTS configuration which is based on the DPFS, the RCS, and the DPD1 and DPD2.
As it is shown in
As it is further shown in
Therefore the SSP and the PTP together represent the whole functionality of the MC as it has been defined above for the first and the second synchronizer configurations.
While this part of specification refers to the third synchronizer configuration, the invention includes using the above MC to SSP and PTP splitting for the first and for the second synchronizer configurations as well.
The on-chip implementation of a DPLL mode is explained below.
The SSP controls the input reference selector (INPREFSEL) and the reference divider (REF_DIV) which select and divide the TCXO's fOUT1 clock, in order to provide selected reference clock fREFSEL which references the analog phase detector (APD) which drives the JF VCXO.
The JF VCXO provides low jitter clock fFILX, which is applied as the reference clock for the output PLL (OUTPLL) via the output reference selector (OUTREFSEL).
The OUTPLL output fOUTPLL is applied as the return clock for the OUTPLL via the output return selector (OUTRETSEL).
The OUTPLL supplies the fOUTPLL clock for the OUTCLKGEN and the RCS.
The OUTCLKGEN provides the required set of output clocks FOUT.
The RCS allows implementation of the phase synthesis process as it is explained below.
The RCS's output clock fRCS is applied to 1/R divider which converts the fRCS into a return clock for the APD.
The DPD1 measures a phase error between TCXO's frequency derivative frF1 and the output clock multiplication fOUT\T. The DPD2 measures a phase error between the DPLL reference derivative frR1 and the output clock multiplication fOUT\T.
The phase transfer processor (PTP) reads the above phase errors and uses them to calculate a new contents of RCS's delay density register (DDR), which shall fulfill a phase transfer function (PTF) which is preprogrammed on the PTP input.
The on-chip implementation of an APLL mode (see the
The invention includes providing slave mode implementation which replaces the external fR2 clock with the mate UTS output clock fMATE, in order to drive the above described APLL configuration. The slave mode allows maintaining phase alignment between active and reserve UTS units, for the purpose of avoiding phase hits when protection switching reverts to using clocks from the reserve UTS unit.
While this part of specification refers to the third synchronizer configuration, the invention includes using the above mentioned method of slave UTS phase alignment for the first and for the second synchronizer configurations as well.
The invention further includes a simplified version of the third synchronizer configuration, which can eliminate the JF VCXO as it is described below.
The frequency of the fREFSEL clock is multiplied by S by the reference PLL (REFPLL), and is selected with the output reference selector (OUTREFSEL) to provide the reference clock for the OUTPLL.
The RCS output fRCS is selected by the output return selector (OUTRETSEL) to provide the return clock for the OUTPLL.
Synchronizer Configuration Based on SCG.
DPLL configurations are explained below.
TCXOs fF1 fixed output is multiplied by PLLxK cell and by PLLxL cell up to fF3 frequency which is used as a frequency reference by the digital phase detectors DPD1 and the DPD2.
Programmable 1/M divider (1/M DIV) allows the same input pin of the reference clock fR1 to be used for a variety of applications having different frequencies of DPLL reference clocks. The 1/M division ratio is programmed by MC_OUT contents being written into reference programming register (RPR).
The DPD1 measures a phase error between the synthesized clock frS2 and the fF3 clock, as Δφ1=φ_frS2−φ_fF3.
The DPD2 measures a phase error between a DPLL reference clock frR1 and the fF3 clock, asΔφ2=φ_frR1−φ_fF3.
Based on the measurements of Δφ1 and Δφ2, microcontroller (MC) calculates control codes for the delay density register (DDR) of the synthesized clock generator (SCG), which shall implement its preprogrammed transfer function between the synthesized clock and the DPLL reference clock.
While the synthesized clock fS2 is selected by the reference selector (RFS) and having the same frequency output clock fOUTY is selected by the return selector (RTS), corresponding to them reference divider (RFD) and return divider (RTD) are set to the same division ratio (usually these dividers are set to 1) in order to drive output PLL (OUTPLL) and output clock generator (OCG).
For most configurations the output clocks set (FOUT) is sufficient to drive all the system timing without any additional jitter filtering.
Only for some jitter sensitive applications, the output clock fOUTY can be used as a reference for an external narrowband Jitter Filter PLL which is implemented with a bandwidth adjusting programmable filter divider (FLD), an Analog Phase Detector (APD) and an external jitter filter crystal oscillator JFVCXO. The FLD allows MC to reprogram the bandwidth of the Jitter Filter PLL for different type of applications and for different synchronization modes. Output of the JFVCXO is named fFILX, and is available to be applied to a jitter sensitive circuit of a synchronized network element.
APLL implementations use analog portions of the above DPLL configurations, but the above described synthesized clock fS2 is not used as a reference for the output PLL (OUTPLL).
In the APLL mode, the reference selector RFS uses an alternative reference clock fR2 instead of the synthesized clock fS2, as its reference clock.
The above mentioned reference and return selectors and dividers (RFS, RTS, RFD, RFD), allow diversified APLL configuring for a wide variety of applications and synchronization modes.
The DPD3 measures a phase error between an output clock fOUTZ and the fF3 clock, asΔφ3=φ_fOUTZ−φ_fF3.
The Δφ3 measurements allow the synchronizer; to detect any “fOUTZ out of range” condition, and to switch from the APLL mode to a “free-run mode” Additionally the Δφ3 and the Δφ1 measurements, allow the MC to work out SCG/DDR control codes which provide coherence of the fS2 signal versus the fOUTZ signal. Therefore the invention allows switching from the APLL mode to a “hold-over mode”, by freezing the DDR content when activity monitor detects a failure of a presently used reference clock.
While this part of specification refers to the second synchronizer configuration: the invention includes using the above mentioned circuits and methods, of switching from the APLL mode to the free-run or the hold-over, for the first and for the third synchronizer configurations as well.
Similarly as for the DPLL, APLL may be configured with or without the jitter filter dependent of jitter levels requirements.
SCG Block Diagram and Circuits Description
The above mentioned third SCG implementation is selected for the preferred embodiment, and it is shown in the
Details of the time critical Delay Shifting Register and the Delay Number Register are shown in
SCG selects outputs of the ring oscillator, based on the inverters Inv(1) to Inv(N), to be applied as PLL return clock fSRC.
Moving the selected output forward by 2 inverters provides delayed fSRC return clock; which causes the PLL to speed up the synthesized clock by the delay of the two inverters, in order to maintain phase locking between the fF2 and the fSRC Using the return clock fSRC instead of the synthesized clock fS1, provides additional filtering of high frequency jitters in the fS1 by the PLL Filter.
Said oscillator output selection is made by a single active high output of the delay number register DNR(1:N).
The DNR bits are controlled by the delay flip-flops DFF(1:N) which are loaded from the delay shifting register DSR(1:N) by their corresponding outputs of the ring oscillator Inv(1) to Inv(N).
In the selector shown in the
In order to eliminate any kind of glitches during the selection switching of the fSRC clock; all the switching of a presently active DNR bit must be completed while selected oscillator output clocks remain in a low half-cycle condition.
During UTS power-up sequence, the DSR(1) bit is preset to 1 and all the other DSR(2:N) bits are reset.
Consequently, the delay shifting register DSR(1:N) always contains a single bit active high, while all the other bits are reset to 0.
DSR content is usually shifted right/left for INC=1/0, by a falling edge of the fSRC; when zero content of the delay density counter DDC(1:N) is detected by the zero decoder (ZERDEC).
However said DSR shifting will not occur and DSR content remains frozen, if the STOP signal is set active high in the DDC.
The DDC(1:N) content is decreased by 1, by a falling edge of the fSRC; when a non zero content of the delay density counter DDC(1:N) is detected by the zero decoder (ZERDEC).
The DDC(1:N+2) content is loaded with a content of the delay density register (DDR(1:N+2), by a falling edge of the fSRC; when a zero content of the delay density counter DDC(1:N) is detected by the zero decoder (ZERDEC).
Additionally the ZERDEC=1 condition is signaled to the MC as the MC_INT, in order to allow more accurate phase control by MC phase transfer algorithms.
The DDR is loaded by the MC_OUT content, which is determined by MC phase transfer algorithms based on measurements provided by the digital phase detectors.
6.2 SCG Timing Analysis
The timing analysis is based on the timing diagrams which are shown in
For INC=1 and ZERDEC=0:
The fSRC keeps subtracting 1 from the content of the delay density counter (DDC), and the DNR(1) continues selecting the output of the Inv(1) to be the source of the fSRC. For this stage the listed below timing requirements shall be fulfilled:
The propagation delay from fSRC falling edge to eventual ZERDEC rising edge, must be lesser than fSRC cycle minus DDC set-up time; where the ZERDEC propagation delay includes ZERDEC propagation through the COUNTER/DDR SELECTOR.
For INC=1 and ZERDEC=1:
When ZERDEC=1 is reached and signals that DDC content has been reduced to zero; the fSRC falling edge shall load a content of the delay density register (DDR(1:N+2)) into the DDC(1:N+2), and the reloading of the DDC with a non zero content shall reset the ZERDEC signal.
Additionally, the fSRC falling edge shall shift right the delay shifting register DSR, in order to deactivate the DSR(1) bit and to activate the DSR(2) bit. Consequently the next falling edge of the Inv(1) will reset the DNR(1) bit and the next falling edge of the Inv(2) will set the DNR(2) bit.
For this stage the listed below timing requirements shall be fulfilled.
The propagation delay from the FSRC falling edge to eventual ZERDEC falling edge, must be lesser than the delay between the fSRC falling edge cycle minus DDC set-up time; where the ZERDEC propagation delay includes ZERDEC propagation through the COUNTER/DDR SELECTOR.
The total propagation delay from the Inv(1) falling edge to the fSRC falling edge plus from the fSRC falling edge to the DSR(1)/DSR(2) falling/rising edge; must be lesser than the Inv(1) cycle minus DNR(1)/DNR(2) set up time.
It shall be noticed that for INC=1; every DFF output is inhibited from activating a corresponding DNR output, for as long as the previous DFF output is still active. Said inhibition prevents a premature activation of the next DNR bit, before the presently active DNR bit is reset. However even without the inhibition, the premature activation might happen only for extremely fast selector and DSR combined with extremely slow oscillator inverters.
For INC=0 and ZERDEC=0:
The fSRC keeps subtracting 1 from the content of the delay density counter (DDC), and the DNR(2) continues selecting the output of the Inv(2) to be the source of the fSRC. For this stage the listed below timing requirements shall be fulfilled.
The propagation delay from fSRC falling edge to eventual ZERDEC rising edge; must be lesser than fSRC cycle minus DDC set-up time; where the ZERDEC propagation delay includes ZERDEC propagation through the COUNTER/DDR SELECTOR.
For INC=0 and ZERDEC=1:
When ZERDEC=1 is reached and signals that DDC content has been reduced to zero; the fSRC falling edge shall load a content of the delay density register (DDR(1:N+2)) into the DDC(1:N+2), and the reloading of the DDC with a non zero content shall reset the ZERDEC signal.
Additionally, the fSRC falling edge shall shift left the delay shifting register DSR, in order to activate the DSR(1) bit and to deactivate the DSR(2) bit. Consequently the next falling edge of the Inv(1) will set the DNR(1) bit and the next falling edge of the Inv(2) will reset the DNR(2) bit. For this stage the listed below timing requirements shall be fulfilled:
The prop. delay from the fSRC falling edge to eventual ZERDEC falling edge; must be lesser than the delay between the fSRC falling edge cycle minus DDC set-up time; where the ZERDEC propagation delay includes ZERDEC propagation through the COUNTER/DDR SELECTOR.
The total propagation delay from the Inv(2) falling edge to the fSRC falling edge plus from the fSRC falling edge to the DSR(2)/DSR(1) falling/rising edge, must be lesser than the Inv(2) falling edge to the Inv(1) falling edge minus DNR(2)/DNR(1) set up time.
Digital Phase Detectors (DPD1/DPD2)
Since both digital phase detectors are identical, only DPD1 is described below, based on its presentation in
Two major digital phase detector circuits are explained below.
A symmetrical twin pair PCBA/PCBB configuration allows higher counting speeds by eliminating all problems related to counters propagation delays.
The PCBA/PCBB configuration allows measurements of frS2 versus fF3 phase errors, with a resolution of a single fF3 period.
When an frS2 rise signals the end of the current phase measurement in a currently active phase counter (PCBA or PCBB), counting of fF3 clock is inhibited and the phase counter content remains frozen, until the next rise of the frS2 signal when the counted clock will be enabled again. The whole frS2 cycle is a very long freeze period, which is more than sufficient to accommodate; any kind of counter propagation, and the counter transfer to phase processing MC, and the counter reset. During the freeze period a mate phase counter is kept enabled and provides measurement of frS2 phase.
Phase Capture Register (PCR) and its control and detection enhance phase detection resolution to a single inverter delay (i.e. by 10 times compared with conventional methods based on clock counting). This enhanced phase resolution is achieved by capturing fF3 propagation over inverters chain with a rising edge of frS2 in the PCR, which is later decoded and transferred to the microcontroller (MC).
More detailed operations of the PCBA/PCBB configuration for both alternatives STOPA=1 and STOPB=1, are explained below.
When STOPA signal is active, DPD circuits perform listed below functions.
PCBB counts all rising edges of fF3 clocks.
PCBB generates SEL9 signal (when PCBB(9) goes high), which activates RD_REQ which initiates MC to read PCBA via CNTR(15:0).
MC calculates previous frS2 versus ff3 phase error, by subtracting from the newly read PCB, the number T of fF3 clocks which nominally should correspond to the frame frS2 (as it is shown in the
PCBB generates SEL14 signal (when CTRB(14) goes high), which activates RST_PCBA which initiates PCBA reset circuits after its content has been read by MC.
When frS2 rise occurs, STOP signal is activated and inverts STOPA/STOPB signals.
When STOPB signal is active all the above functionality is fulfilled with reversed roles of STOPB&PCBA versus STOPA&PCBB.
Detailed timing analysis of the enhanced phase capture circuits is shown in
High Clock Region (HCR) signal shall be interpreted as it is defined below.
The HCR is set to 1: if fF3—rise at frS2=high is detected by the STOP FF, after fF3—fall at frS2=high was detected by the STDI FF (see
The HCR is reset to 0: if fF3—rise at frS2=high is detected by the STOP FF, before fF3—fall at frS2=high is detected by the STDI FF (see
PCR decoders are used for enhancing a phase detection resolution, and they are defined below.
Last Rise Decoder (LRD) provides a binary encoded position of fF3 rising edge, which has been captured at the most right location of the PCR.
Last Fall Decoder (LFD) provides a binary encoded position of fF3 falling edge, which has been captured at the most right location of the PCR.
Cycle Length Decoder (CLD) provides a binary encoded lengths of the fF3 wave, which has been captured between these 2 falling or 2 rising edges of the fF3 wave which occurred at the most right locations of the PCR.
MC algorithms for HCR, LRD, LFD and CLD interpretation are shown in
Calculated by MC measured_phase (MEA_PHA) represents an actual phase error between frS2 versus the equivalent fF3 frame; and consists of the listed below components.
CNTR-1/CNTR/CNTR-2 is an invalidated contents of a counter value CNTR which has been read by MC (all the invalidation algorithms are detailed in
Remaining_phase (REM_PHA) is calculated based on present measurement results, but MC stores and uses it to the correct next measurement result (all the REM_PHA calculation algorithms are shown in
−T=−N×P (see
It shall be noted that in most cases a first fF3 rise which occurs after frS2 rise, will set STOP FF=1 and freeze the previously active counter by inverting STOPA/STOPB signals. Since the first fF3 rise will still add 1 to the previously active counter; MC shall subtract 1 from the counter it reads, while a newly activated mate counter will begin with a correct 0 value. Therefore the first component of a calculated by MC MEA_PHA shall be CNTR-1.
When frS2 rise occurs during tSU of the STOP FF and HCR=1 (see the region “CNTR-2” in
When frS2 rise occurs during t1, of the STOP FF and HCR=0 (see the region “CNTR” in
While the LRD/CLD represents normalized PCR captured extension of the CNTR(15:0) captured phase, and is added to MEA_PHA; the remaining phase error between the frS2 and the next fF3 rise, amounts to (CLD-LRD)/CLD and it is added to the REM_PHA in order to modify next measurement's MEA_PHA.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2,364,506 | Dec 2001 | CA | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/CA02/01873 | 12/2/2002 | WO | 6/7/2004 |