Patent Application
20080101510
1. Technical Field of the Invention
The invention relates generally to de-serialization of serial data streams; and, more particularly, it relates to digital signal processing based de-serialization.
2. Description of Related Art
A conventional trend within communications technologies towards trying to achieve higher and higher operating rates has largely been geared towards employing wide parallel bus architectures. These implementations inherently consume a great deal of real estate. In chip-to-chip applications, one of the greatest consumers of real estate includes the wide parallel interconnections themselves that enable the communication and interconnection between the various devices. The constant increases in complexity and speed of digital hardware has turned the problems of interconnecting components increasingly difficult in contexts such as chip-to-chip, connections between multiple circuit boards across back planes, and other connections having a need for high speed communication while also being constrained by real estate and space.
Moreover, aside from the higher data rates desired in the industry, as the complexity of chips continues to increase, there is also commonly an associated requirement to provide a larger number of interconnections. Again, to ensure higher data communication rates, there is often the trend towards providing broad bus width interconnections between the devices.
One conventional approach to arrive at these high communication rates while also trying to address the design considerations of conserving space and real estate is to employ high-speed serial interconnections. A single high-speed serial interconnection may replace a large number of lower speed interconnections. As a result, a high-speed serial interconnection is largely more space and real estate conserving than parallel type interconnections. For this reason, many industries, including the computer and communications industries, have begun the use of high-speed serial transceivers for many applications including chip-to-chip and board-to-board applications. These transceivers, that may be referred to as SERDES (serializer-de-serializer) run at speeds of several hundred Mega-bits per second (Mb/s) to Giga-bits per second (Gb/s). Some products recently introduced run at data rates of 3.125 Gb/s. These SERDES interconnections are commonly implemented using analog based technology. All of the modulation/demodulation in these conventional SERDES is performed in the analog domain. A most common approach to modulation is to perform the modulation/demodulation using baseband signal processing.
However, these conventional developments fail to provide designs that operate at rates sufficiently high for many customer needs and desires. As the data rates increase, the impairments of the transmission medium become more and more important. For example, in the case of a micro-strip transmission line in a printed circuit (PC) board, dispersion (caused by the bandwidth limitations of the transmission line) causes inter-symbol interference (ISI); discontinuities in the transmission line cause reflections which also result in ISI; capacitive coupling between neighboring traces on the PC board causes crosstalk, and other deficiencies as well. Advanced signal processing techniques such as equalization and crosstalk cancellation have been applied for several decades to control similar impairments in communications systems such as voice-band modems, transceivers for the digital subscriber loop, Ethernet transceivers, and so on. However, with the only exception of the most straightforward situations, these techniques are too complex to be implemented using analog circuit design.
In addition, the analog implementation of SERDES may not be easily scaled to integrated circuit (IC) manufacturing technology of smaller dimensions. For example, a recent trend in Complementary Metal Oxide Semiconductor (CMOS) technology has been from a minimum feature size of 0.18 μm to 0.13 μm. Design engineers find the analog implementation of analog based SERDES extremely difficult. As the processing dimensions continue to decrease this situation will only get worse. The inherent non-scalability of analog based SERDES is a major limitation of the existing SERDES art.
The inability of analog based SERDES technology to enable advanced modulation, error correction and signal processing creates a situation where the fundamental limits of data rate that may be supported in backplane interconnections and other SERDES applications may never fully be realized. An analog based SERDES simply does not offer enough capabilities to enable such high data transfer rates.
Further limitations and disadvantages of conventional and traditional systems will become apparent to one of skill in the art through comparison of such systems with the invention as set forth in the remainder of the present application with reference to the drawings.
The present invention is directed to apparatus and methods of operation that are further described in the following Brief Description of the Several Views of the Drawings, the Detailed Description of the Invention, and the claims. Other features and advantages of the present invention will become apparent from the following detailed description of the invention made with reference to the accompanying drawings.
Great effort has been made in the arena of parallel processing implementations of DSP. The present invention particularly employs adaptive DSP compensation techniques to enable very high-speed SERDES operation.
Alternatively, a DSP based SERDES interconnection 199, that may also be implemented through the back plane 190, may communicatively couple two different ICs 120 and 160 on the boards 110 and 150. Each of the ICs 120 and 160 are operable to interface to the DSP based SERDES interconnection 199 using SERDES interface circuitry 121 and 161, respectively. The SERDES interfacing circuitry 121 is placed very closely to the IC 120, and the SERDES interfacing circuitry 161 is resident as part of the IC 160.
Moreover, a DSP based SERDES interconnection 193 may be used to interface directly between two ICs, such as the ICs 120 and 160. In this situation, the ICs 120 and 160 themselves include the proper interfacing functionality. This is often the case is many IC application, where the interfacing is embedded in the IC, requiring only an interconnection and the DSP based functionality is provided within the IC having the receiver to receive data via the DSP based SERDES interconnection 193.
Moreover, the present invention is operable to provide for interconnection between multiple various ICs on a single board as well. For example, the IC 120 is operable to communicatively couple to an IC 130 via a DSP based SERDES interconnection 191 on the board 110. In this example, a SERDES interface circuitry 131 is resident on the IC 130; the DSP based SERDES interconnection 191 communicatively couples to the SERDES interface circuitry 131 (that communicatively couples to the IC 130) and the SERDES interface circuitry 121 (that communicatively couples to the IC 120).
Those having skill in the art will recognize that the DSP based SERDES interconnections, built in accordance with the present invention, are operable to perform high-speed interfacing between a whole host of devices, boards, and circuitries.
The DSP 244 is able to perform adaptive compensation for non-uniformities among various interleaves of the ADC 242 (that may include an ADC array), or possible non-uniformities among various other elements within an analog front-end (AFE) of the receiver 240. As an input signal enters into the various interleaves of the present invention, various offsets mismatches of the different interleaved paths may need compensation. The use of DSP 244 to perform compensation for these various offsets, that may be generated by the non-uniformities of the various interleaves, to enable an extremely fast operating DSP based SERDES. The DSP compensation techniques allow for adaptive compensation of these interleave generated offset mismatches. The ADC 242 and an AFE may have such impairments, in the case of an interleaved array, that may cooperatively generate fixed pattern noise. While this fixed pattern noise will not be existent within the input signal to the DSP based SERDES, the interleaved implementation that makes possible the use of such higher speed operation SERDES, may undesirably create this fixed pattern noise. However, the use of DSP correction techniques is operable to overcome these effects. The DSP compensation techniques are able to compensate for gain, sampling phase and offset errors in the interleaved array that may lead to the fixed pattern noise.
Again, the DSP 244 is implemented to perform compensation according to the present invention in an adaptive embodiment. There need be no prior knowledge of the specific values of impairments to the interleaves (being ADCs, other portions of the AFE, and/or channel impairments). The DSP techniques are able to identify the appropriate compensation and to perform that compensation adaptively. The adaptive DSP compensation implementation is applicable for both channel impairments and impairments of the ADCs and/or other portions of the AFE.
Analogously, the DSP based SERDES system 200 is operable to perform communication between a transmitter 231 and a receiver 241 via a serial communication link 207. The transmitter 231 is operable to receive data in 211, and if necessary, to perform a parallel to serial conversion of the data in 211. The receiver 241 is operable to receive the serialized data and to perform de-serialization. The receiver 241 generates data out 251. The reverse path (from right to left including the transmitter 231 and the receiver 241), as shown in the
The element 320 contains a SERDES interface circuitry 321, and the element 360 contains a SERDES interface circuitry 361. Both SERDES interface circuitries 321 and 360 are operable to transmit and receive data via the DSP based SERDES interconnection 399. The SERDES interface circuitry 321 includes both transmitter circuitry 322 and receiver circuitry 323. The receiver circuitry 323 includes at least one ADC 324. The ADC 324 could also be multiple ADCs, implemented in an interleaved array. A signal received by the receiver circuitry 323, and after having passed through the ADC 324, is passed to a DSP 327 resident on the element 320. If desired, a DSP 328 may alternatively be employed within the receiver circuitry 323, or a DSP 329 may be implemented on the SERDES interface circuitry 321 itself. The ADC 324 and one, or multiple, of the ADCs 327, 328, and/or 329 operate cooperatively according to the present invention to perform high-speed de-serialization of data received via the DSP based SERDES interconnection 399.
Similarly, the SERDES interface circuitry 361 includes both transmitter circuitry 362 and receiver circuitry 363. The receiver circuitry 363 includes at least one ADC 364. The ADC 364 could also be multiple ADCs, implemented in an interleaved array. A signal received by the receiver circuitry 363, and after having passed through the ADC 364, is passed to a DSP 367 resident on the element 360. If desired, a DSP 368 may alternatively be employed within the receiver circuitry 363, or a DSP 369 may be implemented on the SERDES interface circuitry 361 itself. The ADC 364 and one, or multiple, of the DSPs 367, 368, and/or 369 operate cooperatively according to the present invention to perform high-speed de-serialization of data received via the DSP based SERDES interconnection 399.
The DSP 440 is operable to perform a variety of various digital signal processing operations on the digital samples of the serial data 410 (that are output from the ADC 420). Further, the DSP 440 may also produce compensation controls that are provided to other devices that operate upon the serial data, either in its analog format or after it has been digitally sampled. In performing these operations, the DSP 440 may utilize any of a number of stored adaptation/compensation options 460. These stored adaptation/compensation options 460 may be calculated offline and provided to the DSP 440 for subsequent use in compensating for non-uniformities in the various interleaves of the DSP based de-serializer 400. These stored adaptation/compensation options 460 may include compensation of the channel impairments, such as ISI, and compensation of the impairments of the ADC and possibly other blocks in the analog front-end, such as gain, sampling phase, or offset mismatch among the ADCs of an interleaved ADC array. The stored adaptation/compensation options 460 may include one or more pre-computed compensation operation(s)/compensation control(s) 461, one or more channel responses 462, signal information 463 (that may include any number of types of coding types 464 and modulation schemes 465).
The stored adaptation/compensation options 460 may be calculated and then stored in a memory 450 that is accessible by the DSP 440. The memory 450 may be any number of memory types known to those persons having skill in the art, including random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM), electronically erasable programmable read only memory (EEPROM/E2PROM), Ultraviolet electrically programmable read only memory (UVEPROM), re-programmable PROM (RPPROM), and any other memory types as well. In addition, the stored adaptation/compensation options 460 may alternatively be stored on the receiver 420 of the SERDES 405.
However, the DSP 440 is able to perform real time adaptation to adaptively calculate one or more appropriate feedback signals, shown as the adaptively calculated feedback signal(s) 470. There is typically only one trace through which the serial date 410 (input signal) will propagates at any time. This is inherent in the serial nature of the communication link over which the serial data 410 comes. The DSP is operable to compensate for mismatches among the interleaves of an ADC array (that may be viewed as the ADC 420 in the
In addition and generically speaking, the DSP 440 is operable to compensate for impairment due to the interleaves of any other portions of the AFE, as shown in a functional block 474. The non-uniformities of the various interleaves into which the serial data 410 is placed may require compensation to deal with their inherent variation, that may result from imperfections in processing and fabrication of the devices.
Moreover, the DSP 440 is operable to compensate for NEXT and/or FEXT as shown in a functional block 471. The
In the context of data communications applications, the DSP 440 is operable to compare the digital data received from the ADC 420 against at least the signal information 463. Additionally, the ADC 420 may compare the digital data received from the ADC 420 against the channel responses 462, or a combination of the signal information 463 and the channel responses 462. In one embodiment of making this determination, the DSP 440 first determines the modulation scheme employed by a coupled SERDES transmitter that produced the serial data. Various modulation schemes 465 may be employed including on-off keying (or 2-level pulse amplitude modulation, PAM-2), multilevel PAM (for example, 4-PAM that encodes 2 bits per symbol and therefore double the data rate achievable for a given symbol rate), among other modulation schemes known by persons having skill in the art. In addition, more elaborate modulation schemes, including single carrier quadrature amplitude modulation (QAM: such as 16 QAM, 64 QAM, 256 QAM, and/or 1024 QAM) or multi-carrier modulation are possible, as understood by those persons having skill in the art. Based upon this determination, the DSP 440 adapts its operations to expect a particular modulation scheme, and optionally, a particular channel response. Such adaptation is performed based upon known characteristics of the serial data 410 as provided by the signal information 463 and optionally the channel information.
The DSP 440 is also operable to discern the data coding type of the serial data 410. Similarly, the present invention is operable to adapt to a variety of input coding types 464 known by those persons having skill in the art. For example, the present invention may be used in combination with techniques such as Viterbi decoding, convolutional coding, block coding, trellis coding, and any other form of modulation coding or error correction coding. In addition, turbo coding may be employed as well as various types of pre-coding without departing from the scope and spirit of the invention, including Tomlinson-Harashima pre-coding and Dynamics Limited Pre-coding.
With the DSP 440 adapted to a particular modulation type (and optionally coding type and channel response), the DSP 440 has foreknowledge that it may use to compare the digitized serial data against correct signal information (the signal information 463) stored in the stored adaptation/compensation options 460 (during a training sequence or during normal operations). Based upon this comparison, the DSP 440 determines necessary compensation operations that it will perform on the digitized serial data and/or compensation control operations that it uses to control other components that perform compensation operations on the serial data. With these compensation operations performed, the receiver may then accurately extract digital information from the serial data 410.
If desired, and as will be shown in various embodiments, the adaptation/compensation performed by the DSP 440 may be performed using parallel processing techniques on a number of channels into which the now digital form of the serial data 410 has been channeled. The DSP 440 may then determine the proper compensation controls required, if any, from the one or more pre-computed compensation controls 461 that are stored in the stored adaptation/compensation options 460 and to apply such feedback signal(s) in compensation control operations.
Again, the DSP 550A is operable not only to determine which necessary compensation is desirable to achieve a properly sampled signal (compensation determination 551A), but the DSP 550A is also operable to determined those compensation operations 552A, when provided to the ADC 540A, would effectuate a properly sampled signal. These compensation operations 552A are passed to the ADC 540A via a feedback signal shown as compensation control 520A. As described in other of the various embodiments, the particular context of data communications applications, where a vast amount of information is discernible concerning the serial data 510A, and the communication channel over which the serial data 510A has come, one or both of the compensation determination 551A and the compensation operations 552A may be identified using a predetermined number of stored feedback compensation options. This foreknowledge of the incoming serial data 510A allows very effective compensation to curb the effects of fixed pattern noise in the context of data communications applications and other applications where fixed pattern noise arises. The digital correction techniques are very well adapted to performing compensation to the fixed pattern noise problem, again, based on this knowledge of the incoming signal to the ADC 540A and also based on the knowledge of the type of channel over which the incoming signal has come.
From some perspectives, the DSP 550A is able to perform adaptive compensation for non-uniformities among various interleaves of the ADC 540A (that may include an ADC array), or possible non-uniformities among various other elements within an analog front-end (AFE) of the DSP base de-serializer 500A. As an input signal enters into the various interleaves of the present invention, various offsets mismatches of the different interleaved paths may need compensation. The use of DSP 550A to perform compensation for these various offsets, that may be generated by the non-uniformities of the various interleaves, to enable an extremely fast operating DSP based de-serializer 500A. The DSP compensation techniques allow for adaptive compensation of these interleave generated offset mismatches. The ADC 540A and an AFE may have such impairments, in the case of an interleaved array, that may cooperatively generate fixed pattern noise. While this fixed pattern noise will not be existent within the input signal to the DSP based de-serializer 500A, the interleaved implementation may undesirably create this fixed pattern noise. However, the use of DSP correction techniques is operable to overcome these effects. The DSP compensation techniques are able to compensate for gain, sampling phase and offset errors in the interleaved array that may lead to the fixed pattern noise.
Again, the DSP 550A is implemented to perform compensation according to the present invention in an adaptive embodiment. There need be no prior knowledge of the specific values of impairments to the interleaves (being ADCs, other portions of the AFE, and/or channel impairments). The DSP techniques are able to identify the appropriate compensation and to perform that compensation adaptively. The adaptive DSP compensation implementation is applicable for both channel impairments and impairments of the ADCs and/or other portions of the AFE.
Moreover, it is also noted that the adaptive DSP compensation implementation may be performed for NEXT and FEXT crosstalk problems within a communication channel that provides the serial data 510A to the ADC 540A. The embodiment of the
In this embodiment, the DSP 550B is operable not only to determine which necessary compensation is desirable to achieve a properly sampled signal (compensation determination 551B), but the DSP 550B is also operable to determine those compensation operations 552B, and to actually perform the compensation control 553B itself as to effectuate a digital signal having the characteristics of a properly sampled signal. As described in other of the various embodiments, the particular context of data communications applications, where a vast amount of information is discernible concerning the serial data 510B, and the communication channel over which the serial data 510B has come, one or both of the compensation determination 551B and the compensation operations 552B may be identified using a predetermined number of stored feedback compensation options. This foreknowledge of the incoming serial data 510B allows very effective compensation to curb the effects of fixed pattern noise in the context of data communications applications and other applications where fixed pattern noise arises. The digital correction techniques are very well adapted to performing compensation to the fixed pattern noise problem, again, based on this knowledge of the incoming signal to the ADC 540B and also based on the knowledge of the type of channel over which the incoming signal has come.
From some perspectives, the DSP 550B is able to perform adaptive compensation for non-uniformities among various interleaves of the ADC 540B (that may include an ADC array), or possible non-uniformities among various other elements within an analog front-end (AFE) of the DSP base de-serializer 500B. As an input signal enters into the various interleaves of the present invention, various offsets mismatches of the different interleaved paths may need compensation. The use of DSP 550B to perform compensation for these various offsets, that may be generated by the non-uniformities of the various interleaves, to enable an extremely fast operating DSP based de-serializer 500B. The DSP compensation techniques allow for adaptive compensation of these interleave generated offset mismatches. The ADC 540B and an AFE may have such impairments, in the case of an interleaved array, that may cooperatively generate fixed pattern noise. While this fixed pattern noise will not be existent within the input signal to the DSP based de-serializer 500B, the interleaved implementation may undesirably create this fixed pattern noise. However, the use of DSP correction techniques is operable to overcome these effects. The DSP compensation techniques are able to compensate for gain, sampling phase and offset errors in the interleaved array that may lead to the fixed pattern noise.
Again, the DSP 550B is implemented to perform compensation according to the present invention in an adaptive embodiment. There need be no prior knowledge of the specific values of impairments to the interleaves (being ADCs, other portions of the AFE, and/or channel impairments). The DSP techniques are able to identify the appropriate compensation and to perform that compensation adaptively. The adaptive DSP compensation implementation is applicable for both channel impairments and impairments of the ADCs and/or other portions of the AFE.
Moreover, it is also noted that the adaptive DSP compensation implementation may be performed for NEXT and FEXT crosstalk problems within a communication channel that provides the serial data 510B to the ADC 540B. The embodiment of the
From the DSP 650, compensation control 620 is fed back to one or both of the ADC 640 and the compensation circuitry 652. If desired, the compensation circuitry 652 is operable to receive information concerning the compensation determination 651 that is determined by the DSP 650, and then the compensation circuitry 652 may also have intelligence so that it may itself identify compensation operations 653 that should be performed to compensate for deficiencies in the ADC 640. In addition, it is also noted that the ADC 640 is also operable to be adjusted, in similar manner to other ADCs as described in various other embodiments of the present invention to deal with ADC-related non-uniformities. Any gain, phase, offset or other deleterious effects introduced by the analog to digital conversion within the ADC 640 may be addressed by proper adjustment of the ADC 640, as governed by the feedback signal shown as compensation control 620.
It is also noted that in this embodiment, as well as in other embodiments, the DSP block (shown as DSP 650 in the
From certain perspectives, this embodiment shows the ability of the present invention to perform accommodation of fixed pattern noise related problems by dealing with these problems before any analog to digital conversion of the analog signal. Other embodiments have shown that the present invention is operable to deal with the fixed pattern noise problem by adjusting the operational parameters of the ADC itself, by performing mathematical processing of the digital data coming out of an ADC, and this embodiment shows the ability to deal with the fixed pattern noise problem by performing compensation of the analog signal using a compensation circuitry 652 that is situated before the ADC 640. In addition, this embodiment shows the ability to perform hybrid compensation as well, by performing some compensation in the compensation circuitry 652, and some compensation by adjusting the operational parameters of the ADC 640. Similarly, the present invention is also operable to perform hybrid compensation in other embodiments as well. For example, the present invention allows implementation of combinations of compensation to be performed in various sections of a system that is built according to the present invention. Compensation may be performed before the analog to digital conversion, within the ADC performing the analog to digital conversion, and also in a DSP situated after the ADC as well. The deleterious effects of fixed pattern noise and other recurring noise problems are aptly dealt with using these compensation aspects of the present invention.
In addition, each of the ADCs within the ADC array 720 may individually be controlled to effectuate a proper sampling of the serial data 710. For example, the gain, phase, offset, and other operational parameters of each of the ADCs within the ADC array 720 may be adjusted to ensure that the sampling of the various ADCs is performed uniformly and with proper phase and time delays between the samples, thereby ensuring that the digital data is a true rendition of the serial data 710. Each of the ADCs within the ADC array 720 may be viewed as generating a channel, properly spaced in time from the other channels, that is generated by the other ADCs within the ADC array 720. Each of these channels is fed to a DSP 750.
Again, it is also noted that in this embodiment, as well as in other embodiments, the DSP block (shown as DSP 750 in the
The DSP 750 is operable to determine any parallel based compensation that may be required to transform the now digital, parallel form of the serial data 710, so as to generate digital data having the proper characteristics including proper gain, phase, offset, and other characteristics (as shown in the functional block 751). The parallel based compensation determination 751 creates compensation controls individually for each of the ADCs within the ADC array 750. That is to say, a parallel based feedback signal, shown as compensation control 715, may be fed back to the ADC array 720, such that different components of a digital word may be used to adjust the operational characteristics of the individual ADCs of the ADC array 720. The use of a digital word, having various bits or bit segments within the digital word, to adjust different ADCs within the ADC array 720 will be understood by those persons having skill in the art. In addition, if desired, the DSP 750 is operable not only to perform the parallel based compensation determination 751, but also to identify precisely those parallel based compensation operations 752 that will effectuate the desired operation of the ADC array 750.
Moreover, in even other embodiments, the DSP 750 is operable to identify and also operable to perform parallel based compensation control 753 within the DSP 750 itself. The DSP 750 is operable to perform the compensation control 753 alone, without feeding back and compensation control 715, yet it is also operable to perform compensation in cooperation with the ADC array 720 to which compensation control 715 is passed. Such an embodiment shows the ability of the present invention to perform hybrid type of compensation for the many deleterious effects present in the prior art that have, until now, prevented an effective implementation of a DSP-based de-serializer.
It is also noted that the parallel based compensation determination 751, the parallel based compensation operations 752, and the parallel based compensation control 753 may all be performed independently for various interleaves within the DSP based de-serializer 700. That is to say, the various interleaves may undergo independent determination, identification of operations, and control according to the present invention.
It is also noted that the combined functional block compensation/operations 851 may be performed independently for various interleaves within the DSP base de-serializer 800. That is to say, the various interleaves may undergo independent determination, identification of operations, and control according to the present invention.
From the DSP 850, compensation control 820 is fed back to one or both of the ADC array 820 and the compensation circuitry 852. The compensation control, when fed to the ADC array 820, is operable to perform adaptation of the entire ADC array 820, or alternatively, to each of the individual ADCs within the ADC array 820. If desired, the compensation circuitry 852 is operable to receive information concerning the compensation determination that is determined by the DSP 850, and then the compensation circuitry 852 may also have intelligence so that it may itself identify compensation operations 854 that should be performed to compensation for deficiencies in the incoming serial data 810. The compensation circuitry 852, when operable to identify the compensation operations 854, is able to identify the compensation operations 854 independently or in conjunction with the DSP 850. In addition, it is also noted that the ADC array 820 is also operable to be adjusted, in similar manner to other ADCs as described in various other embodiments of the present invention to deal with ADC-related non-uniformities. Any gain, phase, offset or other deleterious effects introduced by the analog to digital conversion performed within the ADC array 840 (or any of the individual ADCs within the ADC array 840), may be addressed by proper adjustment of the ADC 840, as governed by the feedback signal shown as compensation control 820.
It is also noted that in this embodiment, as well as in other embodiments, the DSP block (shown as DSP 850 in the
As also described in other of the various embodiments of the present invention, the present invention is operable to ascertain various characteristics of an incoming analog signal, once it has been sampled and transformed into the digital realm. A DSP is operable to perform analysis of the digital signal against a number of stored parameters. In the context of data communications applications, a large degree of knowledge may be discerned concerning the channel response of a communication link over which an analog data signal has come before arriving at an ADC that performs analog to digital conversion of an analog signal. Given the context of data communications applications, a predetermined number of pre-computed feedback signal, shown as functional block 961, may be calculated based on the finite number of feedback options that would be required to compensate for problems of gain, phase, offset, and other deleterious effects that may arise from analog to digital conversion in an ADC, and also to other deficiencies that may be introduced within the system before the signal has made its way to the ADC.
In some situations, when a signal is made up of various signal components that are combined into a single signal, some will arrive at an ADC having some fixed pattern noise related deficiencies. The present invention is operable to curb these effects. This particular ability of the present invention shows the ability of the present invention to compensate for deficiencies of fixed pattern noise that may be introduced in analog to digital conversion within an ADC, and also to compensate for fixed pattern noise related problems in the overall system itself.
In addition, the present invention is operable to perform compensation for deficiencies and non-uniformities within an ADC array. This is particularly applicable in the context of systems employing an ADC array, in which multiple ADCs perform sampling at different times, and the operation of all of the ADCs within the ADC array may be non-uniform. In addition, the various ADCs may introduce non-uniform differences of phase between the various channels serviced by the ADC array. The present invention is operable to compensate for this by adjusting the phase of the various channels appropriately, as shown in a functional block 1022. The phase of some channels may be advanced, and the phase of other channels in an ADC array may be delayed, to ensure that the channels each sample the incoming signal appropriately, with a proper spacing of time and phase between the various channels. One additional problem that may arise is when the various ADCs each introduce a non-uniform amount of offset into the signal during the analog to digital conversion; the various ADCs may have slightly different characteristics that generate this offset. The present invention is also operable to compensate for deficiencies in non-uniform offset (as shown in the functional block 1025), as well as eliminate any amount of offset at all, that may be introduced as an offset into the now-digital data that represents the incoming analog signal that is provided to an ADC. In addition, the present invention is operable to perform compensation of crosstalk in the form of NEXT 1021 and/or FEXT 1024. The present invention is also operable to perform compensation of ADC gain mismatch 1023, as well as to accommodate any problems or deficiencies of channel response 1027 as well. Some problems may arise when the incoming signal, or at least portions of the incoming signal, arrives at the ADC across different signal paths. However, more common that that will be problems arising due to non-uniformities between the various ADCs within an ADC array that is employed according to the present invention.
The ADC gain mismatch 1023 may be generated in all, or in part, by the analog to digital conversion operation. The ADC gain mismatch 1023 may alternatively be generated in all, or in part, by the other parts of the system before the analog to digital conversion in an ADC. Similarly, the timing mismatch 1026 may be generated in all, or in part, by the analog to digital conversion operation. The present invention is operable to compensate for these deleterious effects using DSP-based compensation techniques, by performing mathematical processing of the digital data in the DSP domain, or alternatively, to perform analog-based signal processing by modifying certain analog circuitries before the ADC, as identified and determined, at least in part, by a DSP that is placed after the ADC (which may be an ADC array) that performs the analog to digital conversion. The DSP is the entity that is operable, at least, to identify the compensation (if any) to be performed. In certain embodiments, the DSP performs the compensation in the digital domain, performing mathematical processing on the digital data.
Again, the compensation of the present invention may be performed using parallel based compensation 1030 by performing array compensation/adjustment 1040 that is operable to be implemented on an individual device basis 1042. Alternatively, the array compensation/adjustment 1040 may also be performed on an entire array basis 1044. The array may very well be an ADC array, as described in various embodiments, but the array may be an alternative device. For example, an array of filters or an array of voltage scaling circuitry (to adjust gain) may be placed before the ADC that performs analog to digital conversion. The array may be selectively switched in and operated to perform modification of certain channels of an analog signals that are fed to various ADCs in an ADC array. These devices may be PGAs that may be adjusted and controlled using AGC. In addition, as described in various embodiments, the parallel based compensation 1030 may be performed after an analog to digital conversion 1054 where the parallel based compensation 1030 may be performed using digital signal processing 1055. Alternatively, the parallel based compensation 1030 may be performed before an analog to digital conversion 1052 where the parallel based compensation 1030 may be performed using analog circuitry 1053. Again, the direction and control of precisely how to configure the analog circuitry 1053 may also be determined and controlled using digital signal processing.
The outputs of the channels, after they have passed through the precursor filter 1140, are fed to a DFE 1150 that is implemented in a parallel manner. The output from the DFE 1150 is provided as parallel digitized data 1130. In addition, the parallel digitized data 1130 is also sampled and provided to two feedback functional blocks, namely, a timing recovery functional block 1160 and an AGC functional block 1170. The timing recovery functional block 1160 is operable to provide clock signals that are adapted as necessary, to each of the ADCs 1131-1139. In addition, the AGC functional block 1170 is operable to provide gain control signals to the PGAs 1121-1129.
The timing recovery functional block 1160 is operable to perform adjustment of the sampling times at which the various ADCs 1131-1139 may have sampled the incoming serial input data 1110. For example, the ADCs 1131-1139 may not operate ideally and may sample the serial input data 1110 at non-uniformly spaced times. To compensate and correct for this deficiency in a non-ideal array of ADCs, the timing recovery functional block 1160 is operable to adjust the time at which the ADCs effectively sample the serial input data 1110; more precise clock signals may be provided to the ADCs via the timing recovery functional block 1160. The AGC functional block 1170 is operable to provide input to the PGAs 1121-1129 to ensure that the gain of the serial input data 1110 is appropriate for the particular data type, channel response, and application context. The gain of the various channels into which the serial input data 1110 is partitioned, after having passed through the PGAs 1121-1129, may then be appropriately adjusted.
The following
Sanjay Kasturia and Jack H. Winters, “Techniques for High-Speed Implementation of Nonlinear Cancellation,” IEEE Journal on Selected Areas in Communications, vol. 9, no. 5, June 1991.
As shown in the 1-tap DFE embodiment of the
The number of pre-computed feedback signals are stored in the MUX 1230, and the appropriate feedback signal is selected, as provided by the output signal an−1 that is fed back to select “Sel” from the MUX 1230. Rather that need to calculate the feedback signal, the present invention is operable to switch in the appropriate feedback signal based on information it may easily acquire including the input signal type and the channel response. In contradistinction to other implementations of a DFE, the present invention is operable to move the comparators 1211 and 1212 out of the feedback loop. The time delays 1221 and 1222 are added in to compensate and give time for the slowness of the comparators 1211 and 1212. For example, in some embodiments, the operational speed of the comparators 1211 and 1212 and the operational speed of the MUX 1230 may be of longer duration than a clock cycle. The present invention benefits greatly by the introduction of the time delays 1221 and 1222 to compensate for this possibility.
From certain perspectives, the
A second step is to perform the look-ahead transformation in doing the serial to parallel implementation of the DFE. The look-ahead transformation involves employing the following equations:
an=Anan−1+B
an−1=An−1an−2+Bn−1
Then, the second expression is replaced into the first expression as follows:
an=(AnAn−1+BnAn−1)an−2+(AnBn−1+Bn
Then, we have the following result:
an−1=(An−1An−2+Bn−1An−2)an−3+(An−1Bn−2+Bn−1
The look-ahead transformation is understood in the context of parallel processing by those persons having skill in the art. By getting a in terms of an−2, the clock speed may effectively be slowed by a factor of 2. If desired, the look-ahead transformation may also be performed multiple times, getting a in terms of an−m, the clock speed may effectively be slowed by a factor of m. Those persons having skill in the art will understand the extendibility of such operations to effectively allow operation at slower clock frequencies while still maintaining performance.
Here, the present invention shows its adaptability to operate with a clock signal that is reduced by a factor of 2. Two parallel paths of the DFE are used in the
Analogously, an input signal yn−1 is fed simultaneously along two channels, one of which is a comparator 1351 and one of which is a comparator 1352. The comparator thresholds may be set to any desired values (−h1 and +h1) as required or desired for various applications. The output of the comparator 1351 is fed to a time delay z−2 1361, from which an output signal shown an An−1, is provided to combinatorial logic 1370; the output of the comparator 1352 is fed to a time delay z−2 1362, from which an output signal shown as Bn−1, is provided to the combinatorial logic 1370. The output from the combinatorial logic 1370, shown as an−1, is fed to a time delay z−2 1380, from which an output signal an−3 is ultimately output as the second output from the 2-parallel implementation of a 1-tap DFE 1300.
The operations of the combinatorial logics 1330 and 1370 are operable to perform the combination of the two parallel paths in this embodiment. As the number of parallel paths increases, as may be performed in alternative embodiments, the complexity of the combinatorial logics associated will also be increased. However, the increase in complexity of the combinatorial logic is slight when compared to advantages of enabling high-speed operation at significantly reduced clock rates. The combinatorial logics are operable to ensure proper implementation of the equations that enable the representation of a present sample in terms of samples 1 or 2 samples away. For example, the look-ahead transformation described above shows the implementation where a present sample may be made using samples that are at least two samples away. Those persons having skill in the art will appreciate that the present invention may be extended so as to permit the implementation of representing a present sample by a sample that is any number n samples away.
In addition, the two parallel paths of the 2-parallel implementation of a 1-tap DFE 1300 are coupled to one another. For example, the intermediate output signal An from the time delay z−2 1321 is also fed through a time delay z−2 1391, from which an output signal An−2 is fed to the combinatorial logic 1370; the intermediate output signal Bn from the time delay z−2 1322 is also fed through a time delay z−2 1392, from which an output signal Bn−2 is also fed to the combinatorial logic 1370.
In somewhat analogous manner, the intermediate output signal An−, from the time delay z−2 1361 is fed as one of the inputs to the combinatorial logic 1330; the intermediate output signal Bn−1, from the time delay z−2 1362 is fed as one of the inputs to the combinatorial logic 1330.
It is also noted that the architecture of the embodiment shown in the
However, in one embodiment, if the thresholds are sorted in increased order, the operation done by the comparators may be viewed as being a “non-linear thermometer code” encoder. The non-linearity stems from the fact that, in general, the thresholds are not evenly spaced. To accomplish the non-linear mapping, a look up table storing the non-linear mapping, stored in memory that may be Random Access Memory (RAM), could be used. The output of the look up table would then need to be “reverse sorted” to compensate for the sorting operation performed on the thresholds. The combinatorial logics 1330 and 1370 may be employed that is operable to undo this sorting of the thresholds. The combined operation of this non-linear mapping, in conjunction with the operation of the combinatorial logic, by re-ordering the thresholds back to their original ordering, may be used in this and in other embodiments as well.
The outputs of the channels, after they have passed through the precursor filter 1440, are each fed to a RAM mapper, shown as a RAM mapper 1461, a RAM mapper 1462, a RAM mapper 1463, . . . , and a RAM mapper 1469. The RAM mappers 1461-1469 may include the non-linear mapping of a “non-linear thermometer code”. Other non-linear mappings may also be implemented using the RAM mappers 1461-1469 as well without departing from the scope and spirit of the invention. The channeled output from the RAM mappers 1461-1469 are each fed to a switch matrix, shown as a switch matrix 1471, a switch matrix 1472, a switch matrix 1473, . . . , and a switch matrix 1479. The switch matrices 1471-1479 are each operable to perform the undoing of the mapping performed in the RAM mappers 1461-1469; that is to say, the switch matrices 1471-1479 are operable to reorder the sorting of the thresholds back to their original ordering for subsequent processing purposes.
The outputs of the switch matrices 1471-1479 are each provided to a decision feedback equalizer 1450 that is implemented in a parallel manner. The output from the decision feedback equalizer 1450 is provided as parallel digitized data 1430. In addition, the parallel digitized data 1430 is also sampled and provided to an adaptation functional block 1470. The adaptation functional block 1470, in this embodiment, is shown to provide correction to each of the RAM mappers 1461-1469 and to each of the switch matrices 1471 based on error information provided by the decision feedback equalizer 1450.
In addition, the embodiment shown in the
Input data 1510 is fed to a single PGA 1510, where any necessary gain adjustment of the input data 1510 may be performed. In alternative embodiments, multiple PGAs may be employed in place of the PGA 1510, as described in many of the other various embodiments of the present invention. The output of the PGA 1510 is simultaneously fed to a number of ADCs, shown as an ADC01531, an ADC11532, an ADC21533, . . . , and an ADCn 1539, where it is digitally sampled. The outputs of each of the ADCs form channels for the serial input data 1510. The sampling of the ADCs 1531-1539 may be adapted to be performed at delayed times. If desired, each of the ADC 1531-1539 covers a different portion of a cycle of the input data 1510. Together, the ADCs 1531-1539 operate to achieve an effective sampling rate of the sum total of each of the sampling rates of the ADCs 1531-1539. Each of the ADCs 1531-1539 operates at substantially the same sampling rate. Additional intelligence may be employed in various embodiments to accommodate the situation where some of the ADCs operate at different sampling rates as the other of the ADCs.
The outputs of each of the ADCs are fed to a re-synchronization functional block 1540. The re-synchronization functional block 1540 is operable to synchronize the outputs of all the ADCs in the interleaved array so that the DSP can operate with a single clock. As a result of the interleaved operation, each ADC outputs its samples at a different time instant. The re-synchronization functional block 1540 outputs all samples at the transitions of a single clock, which is the same clock used to operate the DSP.
The channels coming out of the re-synchronization functional block 1540 are fed into a feed forward equalizer (FFE) 1545, from which the channeled outputs are fed to a decision feedback equalizer (DFE) 1550. The DFE 1550 provides information concerning slicer error and decision to an adaptation functional block 1585. The adaptation functional block 1585 is operable to perform PGA adaptation 1586 and ADC adaptation 1587. The PGA adaptation 1586 is operable to provide control to the PGA 1520. In addition, the ADC adaptation 1587 is operable to provide timing recovery 1589 to the ADCs 1531-1539. The timing recovery 1589 may be performed individually for each of the ADCs 1531-1539, as shown by the individual timing recovery functional blocks, including a timing recovery 0 functional block 1590, a timing recovery 1 functional block 1591, a timing recovery 2 functional block 1592, . . . , and a timing recovery n functional block 1599. The timing recovery 1589 is fed to a clock generator functional block 1588 whose output may be fed individually to the ADCs 1531-1539.
Input data 1610 is fed simultaneously to a number of programmable gain amplifiers (PGAs), shown as a PGA01621, a PGA11622, a PGA21623, . . . , and a PGAn 1629. The outputs of each of the PGAs form channels for the serial input data 1610, after undergoing any gain adjustment using the PGAs 1621-1629. The output of each of these PGAs is fed to an ADC where the serial data is sampled using a number of ADCs, shown as an ADC01631, an ADC11632, an ADC21633, . . . , and an ADCn 1639. The sampling of the ADCs 1631-1639 may be adapted to be performed at delayed times. If desired, each of the ADC 1631-1639 covers a different portion of a cycle of the input data 1610. Together, the ADCs 1631-1639 operate to achieve an effective sampling rate of the sum total of each of the sampling rates of the ADCs 1631-1639. If desired, each of the ADCs 1631-1639 operates at substantially the same sampling rate. Additional intelligence may be employed in various embodiments to accommodate the situation where some of the ADCs operate at different sampling rates as the other of the ADCs. The outputs of each of the ADCs are fed to a re-synchronization functional block 1640.
The channeled outputs of the re-synchronization functional block 1640 are fed along a path in which a summer is placed, followed by a time delay z−1. In addition, each of the channeled paths is coupled to another of the channeled paths. A scaled version of each of the channeled paths is added in the summers to the channel below it, except for the bottom-most channel, whose scaled version is added to the top-most channel after having passed through a time delay z−1 1679. The scaling factor employed in this embodiment is shown as y, and those persons having skill in the art will appreciate that any other factor may be used to scale the channels before they are added to another channel.
For example, the top-most channel is fed to a summer that receives a y-scaled and delayed (via the time delay z−1 1679) version of the bottom-most channel, and it is then passed to a time delay z−1 1671, before being fed to a decision feedback equalizer 1650. The 2nd channel from the top is fed to a summer that receives a y-scaled version of the top channel, and it is then passed to a time delay z−1 1672, before it is fed to the decision feedback equalizer 1650; the 3rd channel from the top is fed to a summer that receives a γ-scaled version of the 2nd channel from the top, and it is then passed to a time delay z−1 1673, before it is fed to the decision feedback equalizer 1650, whose output is provided to a de-scrambler and interface. This continues for each of the channels within the embodiment shown in the
The decision feedback equalizer 1650 provides information concerning slicer error and decision to an adaptation functional block 1685. The adaptation functional block 1685 is operable to perform PGA adaptation 1686 and ADC adaptation 1687. The PGA adaptation 1686 is operable to provide individual control to each of the PGAs 1621-1629, as shown by the individual PGA adaptation functional blocks, including a PGA0 functional block 1680, a PGA1 functional block 1681, a PGA2 functional block 1682, . . . , and a PGAn functional block 1689. Analogously, the ADC adaptation 1687 is operable to provide individual control to each of the ADCs 1631-1639, as shown by the individual timing recovery functional blocks, including a timing recovery 0 functional block 1690, a timing recovery 1 functional block 1691, a timing recovery 2 functional block 1692, . . . , and a timing recovery n functional block 1699.
The control signal, generated by the AGC 1700, may be employed in any of the various embodiments to perform adaptation/compensation of a single AGC, or an array of AGCs, according to the present invention. The
The proportional path signal is also provided to a summer in an integral path as well (top right hand side of
The phase control signal, generated by the timing recovery 1800, may be employed in any of the various embodiments to perform adaptation/compensation of a single ADC, or an array of ADCs, according to the present invention. The
The scrambler 1900 is employed to ensure a random looking data stream, as will be understood by those persons having skill in the art. If desired, the same polynomial as that used in 1000BaseT Ethernet may be employed, as used on the slave side as follows:
gs(x)=1+x20+x33
For simplicity, this may be implemented as a self-synchronizing scrambler. The self-synchronizing scrambler may be viewed as being essentially a recursive filter (in modulo 2 arithmetic), and it can be parallelized using a number of transformations including a look-ahead transformation.
In addition, the present invention is operable to employ a start-up controller in various embodiments. A start-up controller is a state machine that sequences the control signals during start-up to facilitate convergence of various components including DFE, AGC, and timing recovery. The start-up controller is operable to run at relatively low clock speeds, so as not to contribute significantly to power dissipation in the overall system. For example, a nominal clock rate of less than 1 MHz would be sufficient in most instances.
Then, based upon this knowledge or foreknowledge of the input signal, and the communication channel over which the input signal has come, all possible values of feedback signal (that provides adaptation/compensation) are pre-computed in a block 2220. Then, in a block 2230, the adaptation/compensation is applied to the appropriate blocks via a feedback signal. The feedback signal may be applied to one or more ADCs (say to an ADC array) as shown in a functional block 2231, to one or more PGAs as shown in a functional block 2232. The feedback signal may be fed to a device residing before an ADC that performs analog to digital conversion 2233, or to a device residing after the ADC that performs analog to digital conversion 2234. Moreover, parallel based techniques may be employed according to the present invention to deliver this feedback signal, as shown in a functional block 2235.
Then, based upon this knowledge or foreknowledge of the input signal, and the communication channel over which the input signal has come, all possible values of adaptation/compensation may be pre-computed in a block 2240. Then, the adaptation/compensation is performed directly using a DSP, as shown in a functional block 2250.
In addition, it is also noted that some of the adaptation/compensation may be performed, in part, using a DSP and also performed, in part, using a device to which a feedback signal is provided. The present invention is adaptable to perform hybrid adaptation/compensation.
Then, in a block 2330, any necessary adaptation/compensation is performed. The necessary adaptation/compensation may be performed after one or more ADCs, as shown in a functional block 2334, and it may be performed individually to all of the devices, as shown in a functional block 2350. In the ADC array context, the necessary adaptation/compensation is performed on an ADC basis as shown in a functional block 2351. Alternatively, the necessary adaptation/compensation may also be performed before any ADCs, as shown in a functional block 2332.
The present invention is operable to perform the DSP based SERDES de-serializer method 2300 in a manner that any necessary adaptation/compensation may be performed to any number of devices, such as one a to one or more PGA basis (as shown in a functional block 2352), and it may also be performed individually to any other type of device 2359. The PGA basis 2352 may be performed using AGCs in certain embodiments. Alternatively, parallel based techniques 2360 may also be performed in doing the necessary adaptation/compensation. In addition, the necessary adaptation/compensation may also be performed jointly to all devices as if they are they were a single entity or device. This may be performed to all ADCs 2341, to all PGAs 2342 (perhaps using a number of AGCs as shown in a functional block 243), or alternatively to any other number of devices 2349 that are configured to operate as a single entity.
In a block 2440, it is determined whether there is any necessary adaptation/compensation control needs to be performed using a DSP, after analyzing the now digital data in each of the various channels. In certain embodiments, the necessary adaptation/compensation control may be determined whether it should or can be performed before or after the analog to digital conversion of the serial/analog, channeled data. Then, in a block 2460, the compensation/determination is performed using analog circuitry that is situated before the ADCs of the various channels. The control signals, to each of the various ADCs, are passed from the DSP to the analog circuitry. If desired, the control to the analog circuitry is delivered using parallel based techniques.
Alternatively, the compensation/determination is performed using a DSP, as shown in a block 2450; the compensation/determination is performed by performing mathematical manipulation of the now-digital data. In addition, any number of multiple DSPs may also be employed, as shown in a functional block 2451.
Then, once the DSP based SERDES is in an operational mode, any number of events and/or conditions may occur that would initiate a re-training of the DSP based SERDES. If no such event and/or condition occurs, then the DSP based SERDES will simply run indefinitely, as mentioned above. However, some situations necessitate the re-training of the DSP based SERDES to achieve desirable operation once again.
One such example that may be used to initiate the retraining of the DSP based SERDES includes a power loss in the system, as shown in a functional block 2551. This power loss may be determined as having been for a predetermined duration, as indicated in a functional block 2552. If desired, brief or nearly instantaneous power losses may be ignored, as desired in certain embodiments. Alternatively, after power losses of certain duration, an abbreviated re-training may be performed. In such situations, those system operational parameters that have a low likelihood of changing may be ignored, yet those parameters that are more volatile may be re-calculated, and those parts of the system may then be retrained, as shown in a functional block 2550. Then, once the DSP based SERDES is re-trained, the method returns to the operating mode of the DSP based SERDES, as shown in the functional block 2530. Alternatively, the method 2500 may end. The ending may be user controlled, or as controlled by the DSP based SERDES.
Another such example that may be used to initiate the retaining of the DSP based SERDES includes a loss of the coefficients of an equalizer that is employed in the system, as shown in a functional block 2561. Any number of techniques may be employed to detect a loss of the coefficients of any equalizer. Then, once the DSP based SERDES is re-trained, the method returns to the operating mode of the DSP based SERDES, as shown in the functional block 2530. Alternatively, the method 2500 may end. The ending may be user controlled, or as controlled by the DSP based SERDES.
Another such example that may be used to initiate the retraining of the DSP based SERDES includes a detection of an error in the system, as shown in a functional block 2571. Any number of error detection techniques, as understood by those persons having skill in the art, may be employed to detect an error in the system. The error detection may include degraded system performance (functional block 2572), where there is evidence of some change in the system that has contributed to compromised performance. Another situation is where the equalizer ceases to provide proper operation because the equalizer has diverged (functional block 2573), and it no longer converges or properly operates. Then, once the DSP based SERDES is re-trained, the method returns to the operating mode of the DSP based SERDES, as shown in the functional block 2530. Alternatively, the method 2500 may end. The ending may be user controlled, or as controlled by the DSP based SERDES.
Moreover, any other event and/or condition (functional block 2581) may be used to initiate the re-training of the DSP based SERDES. Examples of some other events and/or conditions include seemingly changed operating conditions, including environmental changes (including temperature, humidity) that have substantially altered the performance of the overall system.
Moreover, it is noted that the SERDES based interfacing between devices may be performed using alternatively means as well, as shown by the connectivity options between devices 2670. For example, there may be other possibilities in which at least two devices may be communicatively coupled that may benefit from the present invention. Two devices may be communicatively coupled via twisted pair cabling 2671, coaxial cabling 2672, and/or twin-ax cabling 2673, among others.
Generically speaking, the compensation that may be performed according to the present invention includes a number of compensation types, many of which have been described herein. The present invention is able to compensate for various types of errors 2610, including ISI 2611, and attenuation 2650. The attenuation 2650 may be generated due to the variations among various channels (CHs) 2651, and it may be compensated using one or more PGAs and/or one or more ADCs 2652. Those persons having skill in the art will appreciate how the use of such devices may be used to compensate for such degradation of attenuation 2650.
Moreover, the present invention is operable to compensate for various types of crosstalk using crosstalk cancellation 2660. The crosstalk cancellation 2660 may be employed using a crosstalk canceller within any of the various transmitters, receivers, and/or transceivers that may be arranged and employed according to the present invention. The crosstalk cancellation 2660 may be performed to overcome the effects of NEXT 2662 and/or FEXT 2661.
In addition, various offsets mismatches 2620 of the different interleaved paths may need compensation, and the present invention is operable to overcome these effects as well. These offset mismatches 2620 may result due to non-uniformities of various interleaves 2621 in the various embodiments of the present invention. The non-uniformities of various interleaves 2621 may result in fixed pattern noise that must be addressed and may be substantially eliminated using the DSP compensation and correction techniques performed according to the present invention.
Various embodiments and aspects of the present invention have been described above. Some of the aspects are geared towards data communications applications. Those persons having skill in the art will recognize the extendibility of the present invention to any application where operational parameters and knowledge of the type of an incoming signal, the channel type, the channel response and other information may be discerned may also benefit from the various aspects of the present invention.
In view of the above detailed description of the invention and associated drawings, other modifications and variations will now become apparent to those skilled in the art. It should also be apparent that such other modifications and variations may be effected without departing from the spirit and scope of the invention.
The present U.S. Utility patent application claims priority pursuant to 35 U.S.C. § 120, as a continuation, to the following U.S. Utility patent application which is hereby incorporated herein by reference in its entirety and made part of the present U.S. Utility patent application for all purposes: 1. U.S. Utility application Ser. No. 10/086,267, entitled “DIGITAL SIGNAL PROCESSING BASED DE-SERIALIZER,” (Attorney Docket No. BP1904), filed 03-01-2002, pending, which claims priority pursuant to 35 U.S.C. § 119(e) to the following U.S. Provisional Patent Applications which are hereby incorporated herein by reference in their entirety and made part of the present U.S. Utility patent application for all purposes: a. U.S. Provisional Application Ser. No. 60/273,215, entitled “High-speed analog to digital conversion system for communications applications,” (Attorney Docket No. BP1860), filed 03-01-2001, now expired. b. U.S. Provisional Application Ser. No. 60/290,263, entitled “Digital signal processing based transceiver for high-speed interconnections,” (Attorney Docket No. BP1904), filed 05-11-2001, now expired. The following U.S. Utility patent applications are hereby incorporated herein by reference in their entirety and made part of the present U.S. Utility patent application for all purposes: 1. U.S. Utility patent application Ser. No. 09/844,441, entitled, “HIGH-SPEED SERIAL DATA TRANSCEIVER AND RELATED METHODS,” (Attorney Docket No. BP1423.4), filed 04-30-2001, now U.S. Pat. No. 7,058,150, issued on 06-06-2006. 2. U.S. Utility patent application Ser. No. 10/085,071, entitled “METHODS AND SYSTEMS FOR DSP-BASED RECEIVERS,” (Attorney Docket No. BP1860), filed 03-01-2002, now U.S. Pat. No. 7,245,638, issued on 07-17-2007.
| Number | Date | Country | |
|---|---|---|---|
| 60273215 | Mar 2001 | US | |
| 60290263 | May 2001 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 10086267 | Mar 2002 | US |
| Child | 11968450 | Jan 2008 | US |
