Pulse or digital communications – Equalizers – Automatic
Reexamination Certificate
2000-07-24
2004-07-20
Tse, Young T. (Department: 2634)
Pulse or digital communications
Equalizers
Automatic
C375S229000, C375S230000
Reexamination Certificate
active
06765958
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates generally to VLSI circuits. More particularly, the invention relates to low cost receiver structures and methods to provide high-speed inter-chip or inter-module communication links.
2. Description of the Related Art
Communication between chips on a circuit board traditionally use very simple binary zero-one logic. A high voltage is sent to represent a binary one, and a zero voltage is sent to represent a binary zero. The receiver maintains clock synchronization with the transmitter and at the appropriate time decides a binary one if the voltage on the communication wire is above a threshold and decides a binary zero if the voltage is below another threshold. More recently it has been proposed to use multilevel signaling such as pulse-amplitude modulation in order to increase the data rate between chips. U.S. Pat. No. 6,005,895 discusses such a scheme. Another multilevel signaling approach for inter-chip interconnects is described in J. Zerbe et al., “1.6 Gb/s/pin 4-PAM signaling and circuits for multi-drop bus,” 2000 Symposium on VLSI Circuits, pp. 128-131, IEEE Press. This reference is referred to as the “Zerbe reference” henceforth.
While these multilevel signaling approaches are advantageous, inter-chip communication speeds are eventually limited by a phenomenon known in the art as “inter-symbol interference” or, “eye-closing.” Eye closing occurs when distortions introduced by the communication channel make it impossible to discern the transmitted signal levels by sampling the received waveform. The so-called “eye” refers to a pattern observed on an oscilloscope. When the eye is open, distinct signal levels can be viewed. When the eye is closed, the signal levels have run together and therefore distinct signal values cannot be observed. The problem of eye closing becomes more severe on a given connection as the data rate is increased. While for a fixed data rate it may be possible to assure the eye will stay open for short and well engineered point-to-point connections, this is not the case for multi-drop busses and/or longer runs as may be needed to support various system topologies. In future wafer scale designs, the same problems may occur for longer runs between intra-wafer modules.
In the field of wireline and wireless communications, various approaches to recovering data streams from received waveforms having closed eyes are known. Typically equalizers are used to open the closed eye so that the data may be properly recovered. Equalization approaches are multiply-accumulate intensive and rely on DSP (digital signal processing). Hence prior art solutions are too expensive for inter-chip applications where the symbol rates are presently in the 800 MHz region. To cross beyond the 800 MHz barrier, improved equalizing receiver structures are needed, but these would need to be able to operate at symbol rates in excess of 800 MHz. Such high-speed equalizers might also need to be able to differentiate more than two signal levels in a multilevel PAM (pulse-amplitude modulation) scheme. Prior art DSP-based equalizers are not suited to solve such inter-chip equalization problems in a cost efficient way.
It would be desirable to have a receiver structure for inter-chip communications that could perform equalization to open a closed eye pattern in a received signal. It would be desirable if such a receiver could be low cost in terms of silicon area and power consumption. It would be desirable for the receiver to not require, multiplications as are usually needed in equalizers, because multiplcations are very expensive. It would be desirable to have a receiver structure that could loosen design constraints on the physical channel between the chips by allowing reliable communications over channels involving longer runs and multiple drops. This would allow a given data transfer rate to be supported over a wider variety of wire-routing topologies, thereby increasing design flexibility. It would also be desirable to have a receiver that could increase the sustainable data transfer rate on a well-engineered circuit path. Accompanying system level application architectures that make use of the high speed interconnect are also taught. Methods of training and operating the receivers and systems of the present invention are also developed.
SUMMARY OF THE INVENTION
The present invention overcomes difficulties with prior art inter-chip and inter-module interconnects by introducing a low cost and low power equalizing receiver structure for inter-chip communications. The novel receiver structure allows binary and multilevel signaling to be received at greater speeds and over more diverse paths by processing the received signal prior to signal symbol detection. The equalizer structure can inherently operate at high speeds due to its multiply-free architecture. Both serial and parallel circuit structures are taught. Either of these structures or a hybrid of the two can be selected in light of design constraints. Systems based on symbol-spaced and fractionally-spaced sampling are taught. The equalizer is adaptive, but only needs to be adapted at power-up and can be optionally retuned periodically.
REFERENCES:
patent: 5459432 (1995-10-01), White et al.
patent: 5818655 (1998-10-01), Satoh et al.
patent: 6057792 (2000-05-01), Eastty et al.
patent: 6421381 (2002-07-01), Raghavan
patent: 6480534 (2002-11-01), Gatherer et al.
patent: 6556637 (2003-04-01), Moriuchi
Ahn Sam K.
Dowling Eric M.
Tse Young T.
LandOfFree
High-speed adaptive interconnect architecture does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with High-speed adaptive interconnect architecture, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and High-speed adaptive interconnect architecture will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3249706