Abstract:
Described embodiments provide for, in a SerDes device, an adaptation process that adjusts data path gain through programmable-bias based on process, voltage, temperature (PVT) and data rate changes. Such adaptation process extends bias current dynamic range, and low frequency gain can be programmed to a desired target range of values for a given variable gain amplifier (VGA) setting at any PVT and data rate corner. A receive (RX) data path structure auto-adapts data path gain through programmable bias based on sensed PVT and data rate changes. The low frequency attenuation/gain range is extended, and can be programmed to a desirable targeted range by a SerDes device RX adaptive process for a given VGA and linear equalizer (LEQ) setting at any given PVT and data rate condition.
Abstract:
Described embodiments provide for de-coupling between adaptation of decision feedback equalizer (DFE) filter taps and transmitter (TX) post cursor filtering in group delay (GD)-based adaptation. Consequently, an excessive build-up of transmitter post cursor effects and its excessive equalization cancellation by the DFE may be substantially reduced or eliminated. By breaking this coupling, a transmitter does not over equalize a signal, the DFE does not attempt to “undo” the over equalization, and a variable gain amplifier (VGA) in the receiver front end data path generally does not apply gain to amplify the signal back again due to the reduced DC level. GD-based TX post cursor adaptation may reduce over equalization effect and hence save power and increase performance by not over equalizing the signal.
Abstract:
A method of adjusting a post-cursor tap weight in a transmitter FIR filter in a high-speed digital data transmission system. A receiver, over a forward channel, receives a signal from the transmitter and equalizes the received signal using an adaptive analog equalizer coupled to the forward channel and a decision feedback equalizer (DFE) coupled to the analog equalizer. A gain coefficient used to adjust the peaking by the analog equalizer is adapted using an error signal generated by the DFE. The post-cursor tap weight of the transmitter filter is adjusted up or down based on a comparison of the gain coefficient to a set. of limits. The post-cursor tap weight is transmitted to the transmitter over a reverse channel and then equalizers in the receiver readapt. Alternatively, eye opening data and a DFE tap coefficient are used to determine whether the post-cursor tap weight is adjusted up or down.
Abstract:
A SerDes data sampling controller that includes a gear shifting data sampling clock that zeroes the data sampling skew at the center of the unit interval during the CDR phase lock stage, and then skews the data sample timing away from the center of the unit interval as the DFE coefficients adapt during the data transfer stage. This allows the controller to implement the best (unskewed) data sample timing during the CDR phase locking stage, and then skew the data sample timing after the DFE coefficients have adapted to provide the best (skewed) data sample timing for data bit sampling during the data transfer stage. The data sampling gear shifter may apply a variable skew value to the transition sampling or quadrature (Q) data sampling clock differentially varying the quadrature (Q) transition sampling clock from the inphase (I) data sampling clock.
Abstract:
A SerDes data sampling controller that includes a gear shifting data sampling clock that zeroes the data sampling skew at the center of the unit interval during the CDR phase lock stage, and then skews the data sample timing away from the center of the unit interval as the DFE coefficients adapt during the data transfer stage. This allows the controller to implement the best (unskewed) data sample timing during the CDR phase locking stage, and then skew the data sample timing after the DFE coefficients have adapted to provide the best (skewed) data sample timing for data bit sampling during the data transfer stage. The data sampling gear shifter may apply a variable skew value to the transition sampling or quadrature (Q) data sampling clock differentially varying the quadrature (Q) transition sampling clock from the inphase (I) data sampling clock.
Abstract:
In described embodiments, process, voltage, temperature (PVT) compensation in a serializer/deserializer (SerDes) device employs a closed loop adaptation compensation that is incorporated into the SerDes receiver adaptation process. A detection method, where the adapted decision feedback equalizer (DFE) target level (e.g., tap H0) is monitored, employs this DFE target level when implementing a closed loop variable gain amplifier adaptation. The DFE target level in conjunction with the VGA level is used to control the PVT setting to maintain target SerDes data path gain by detecting aPVT corner condition. The detected PVT corner condition is employed to generate a control signal to further adjust the LEQ and DFE data path differential pair gain as required by the PVT condition.
Abstract:
Described embodiments provide for, in a clock and data recovery (CDR) circuit, detection of loss of acquisition and CDR restarting with corrective integral accumulator register seeding and gearshift restarting. In described embodiments, a mechanism is employed to cause faster loss of lock condition if the CDR circuit directed on an incorrect acquisition trajectory, actual loss of CDR lock is then detected, and CDR acquisition is recovered with corrective integral accumulator seeding.
Abstract:
Described embodiments include a process and apparatus that takes into account the operating voltage and temperature (VT) variations of a SERDES receiver implemented in an integrated circuit (IC) or system-on-chip (SoC). An analog equalizer (AEQ) adaptation loop and a decision feedback equalizer (DFE) adaptation loop are disabled after the loops have converged or stabilized the parameters of the AEQ and DFE. While the AFE and DFE adaptation loops are disabled, certain monitor coefficients related to signals corrected by the AFE and DFE are adapted and metrics derived therefrom are generated. The metrics are compared to threshold values to check if they have sufficiently changed over time to warrant re-enabling of the AFE and DFE adaptation loops.
Abstract:
A tap coefficient control circuit and a method for controlling a tap coefficient for a decision feedback equalizer are disclosed. The method includes adjusting a correction voltage applied to the tap coefficient based on a first tap quantization and detecting a decision feedback equalizer tap convergence. After the decision feedback equalizer tap convergence is detected, the method adjusts the correction voltage applied to the tap coefficient based on a second tap quantization, wherein the second tap quantization is different from the first tap quantization.
Abstract:
An apparatus comprises a clock and data recovery system, and a loss of lock detector at least partially incorporated within or otherwise associated with the clock and data recovery system. The loss of lock detector is configured to generate a loss of lock signal responsive to phase adjustment requests generated for a clock signal in the clock and data recovery system. By way of example, the loss of lock signal may have a first logic level indicative of the phase adjustment requests occurring at a first rate associated with a lock condition and a second logic level indicative of the phase adjustment requests occurring at a second rate lower than the first rate. Absolute values of respective phase increments each associated with multiple up and down phase requests may be accumulated, and the loss of lock signal generated as a function of the accumulated phase increment absolute values.