High-speed receiver architecture

• US 10,097,273 B2
• Filed: 12/12/2017
• Issued: 10/09/2018
• Est. Priority Date: 10/03/2005
• Status: Active Grant

• Current Assignee: Marvell Asia Pte Limited (Marvell Technology Group Limited)
• Original Assignee: Inphi Corporation
• Inventors: Agazzi, Oscar Ernesto, Crivelli, Diego Ernesto, Carrer, Hugo Santiago, Hueda, Mario Rafael, Luna, German Cesar Augusto, Grace, Carl
• Primary Examiner(s): Corrielus, Jean B
• Application Number:
  US15/839,380
• Publication Number:
  US 20180102850A1
• Time in Patent Office: 301 Days
• CPC Class Codes: H03M 13/41 using the Viterbi algorithm or Viterbi processors
• CPC Class Codes: H04B 1/38 Transceivers, i.e. devices in which transmitter and receiver form a structural unit and in which at least one part is used for functions of transmitting and receiving
• CPC Class Codes: H04B 10/25073 using spectral equalisation, e.g. spectral filtering
• CPC Class Codes: H04B 10/2941 using an equalising unit, e.g. a filter H04B10/296 takes precedence
• CPC Class Codes: H04B 10/40 Transceivers
• CPC Class Codes: H04B 10/5059 using a feed-forward signal generated by analysing the optical or electrical input
• CPC Class Codes: H04B 10/6971 using equalisation
• CPC Class Codes: H04B 2001/0441 with linearisation using feed-forward
• CPC Class Codes: H04B 2201/709772 Joint detection using feedforward
• CPC Class Codes: H04B 3/235 combined with adaptive equaliser
• CPC Class Codes: H04B 7/005 Control of transmission; Equalising
• CPC Class Codes: H04B 7/0456 Selection of precoding matrices or codebooks, e.g. using matrices antenna weighting
• CPC Class Codes: H04L 2025/03477 not time-recursive
• CPC Class Codes: H04L 2025/03617 Time recursive algorithms H04L2025/03643 takes precedence
• CPC Class Codes: H04L 25/0202 Channel estimation
• CPC Class Codes: H04L 25/025 using least-mean-square [LMS] method
• CPC Class Codes: H04L 25/03038 with a non-recursive structure H04L25/03031 takes precedence
• CPC Class Codes: H04L 25/03057 with a recursive structure H04L25/03031 takes precedence
• CPC Class Codes: H04L 5/16 Half-duplex systems; Simplex/duplex switching; Transmission of break signals non-automatically inverting the direction of transmission

Abstract

A receiver (e.g., for a 10 G fiber communications link) includes an interleaved ADC coupled to a multi-channel equalizer that can provide different equalization for different ADC channels within the interleaved ADC. That is, the multi-channel equalizer can compensate for channel-dependent impairments. In one approach, the multi-channel equalizer is a feedforward equalizer (FFE) coupled to a Viterbi decorder, for example, a sliding block Viterbi decoder (SBVD); and the FFE and/or the channel estimator for the Viterbi decoder are adapted using the LMS algorithm.

Claims

1. A receiver comprising:

an analog-to-digital converter (ADC) that generates signal samples from a received signal having combined non-Gaussian noise and Gaussian noise;

wherein the ADC is a pipelined ADC including an input track-and-hold stage followed by a plurality of low resolution ADC stages, each ADC stage including

a 1-bit comparator coupled to the input track-and-hold stage,

a 1-bit digital-to-analog converter (DAC) coupled to the 1-bit comparator,

an analog subtractor coupled to the 1-bit DAC and the input track-and-hold stage,

a residue amplifier coupled to the analog subtractor, and

a track-and-hold circuit coupled to the residue amplifier;

a feedforward equalizer coupled to receive the signal samples from the ADC, and to apply equalization to generate equalized samples;

a decoder coupled to an output of the feedforward equalizer, the decoder determining detected symbols from the equalized samples and a channel model by minimizing a cumulative metric that compensates the combined Gaussian and non-Gaussian noise in the received signal; and

a channel estimator to receive the output of the feedforward equalizer and an output of the decoder and to generate the channel model and an error feedback signal to adaptively update coefficients of the feedforward equalizer based on the error feedback signal;

wherein the ADC is configured in a sub-radix architecture having more ADC stages than there are bits in the received signal.

2. The receiver of claim 1 wherein the residue amplifiers of the ADC stages include open-loop amplifiers.

3. A receiver comprising:

an analog-to-digital converter (ADC) that generates signal samples from a received signal having combined non-Gaussian noise and Gaussian noise;

wherein the ADC is a pipelined ADC including an input track-and-hold stage followed by a plurality of low resolution ADC stages, each ADC stage including

a 1-bit comparator coupled to the input track-and-hold stage,

a 1-bit digital-to-analog converter (DAC) coupled to the 1-bit comparator,

an analog subtractor coupled to the 1-bit DAC and the input track-and-hold stage,

a residue amplifier coupled to the analog subtractor, and

a track-and-hold circuit coupled to the residue amplifier;

a feedforward equalizer coupled to receive the signal samples from the ADC, and to apply equalization to generate equalized samples;

a decoder coupled to an output of the feedforward equalizer, the decoder determining detected symbols from the equalized samples and a channel model by minimizing a cumulative metric that compensates the combined Gaussian and non-Gaussian noise in the received signal; and

a channel estimator to receive the output of the feedforward equalizer and an output of the decoder and to generate the channel model and an error feedback signal to adaptively update coefficients of the feedforward equalizer based on the error feedback signal;

wherein the cumulative metric is approximated by

M≈

Σ

_n[(ŷ

_n)^v−

(ŝ

_n)^v]² where M is the cumulative metric, ŷ

_n represents a sample at the output of the feedforward equalizer, ŝ

_n is a noise-free signal component of ŷ

_n, where 0<

v≤

1.

4. The receiver of claim 3, where v is given by

5. The receiver of claim 3 further comprising:

a transformation block to provide an approximate solution to (ŷ

_n)^v, wherein the transformation block implements a function LUT(ŷ

_n)=(ŷ

_n+y_min)^v when the sample ŷ

_n is in an interval between −

y_min and +y_max, where y_min and y_max are the minimum and maximum values of the interval of the sample ŷ

_n, respectively.

6. A receiver comprising:

an analog-to-digital converter (ADC) that generates signal samples from a received signal having combined non-Gaussian noise and Gaussian noise;

wherein the ADC is a pipelined ADC including an input track-and-hold stage followed by a plurality of low resolution ADC stages, each ADC stage including

a 1-bit comparator coupled to the input track-and-hold stage,

a 1-bit digital-to-analog converter (DAC) coupled to the 1-bit comparator,

an analog subtractor coupled to the 1-bit DAC and the input track-and-hold stage,

a residue amplifier coupled to the analog subtractor, and

a track-and-hold circuit coupled to the residue amplifier;

a feedforward equalizer coupled to receive the signal samples from the ADC, and to apply equalization to generate equalized samples;

a decoder coupled to an output of the feedforward equalizer, the decoder determining detected symbols from the equalized samples and a channel model by minimizing a cumulative metric that compensates the combined Gaussian and non-Gaussian noise in the received signal; and

a channel estimator to receive the output of the feedforward equalizer and an output of the decoder and to generate the channel model and an error feedback signal to adaptively update coefficients of the feedforward equalizer based on the error feedback signal;

wherein the ADC further includes a look-up table (LUT) configured to the plurality of ADC stages to store the inverses of non-linear characteristics of the signal samples.

7. A receiver comprising:

an analog-to-digital converter (ADC) that generates signal samples from a received signal having combined non-Gaussian noise and Gaussian noise;

wherein the ADC is a pipelined ADC including an input track-and-hold stage followed by a plurality of low resolution ADC stages, each ADC stage including

a 1-bit comparator coupled to the input track-and-hold stage,

a 1-bit digital-to-analog converter (DAC) coupled to the 1-bit comparator,

an analog subtractor coupled to the 1-bit DAC and the input track-and-hold stage,

a residue amplifier coupled to the analog subtractor, and

a track-and-hold circuit coupled to the residue amplifier;

a feedforward equalizer coupled to receive the signal samples from the ADC, and to apply equalization to generate equalized samples;

a decoder coupled to an output of the feedforward equalizer, the decoder determining detected symbols from the equalized samples and a channel model by minimizing a cumulative metric that compensates the combined Gaussian and non-Gaussian noise in the received signal; and

a channel estimator to receive the output of the feedforward equalizer and an output of the decoder and to generate the channel model and an error feedback signal to adaptively update coefficients of the feedforward equalizer based on the error feedback signal;

wherein the ADC further includes a lookahead pipeline configuration in which one or more of the plurality of ADC stages includes a pair of comparators configured to enable parallel interstage amplification and comparator operations.

8. A method for equalizing a received signal, the method comprising:

generating, by an analog-to-digital converter (ADC), signal samples from the received signal having combined non-Gaussian noise and Gaussian noise;

wherein the ADC is a pipelined ADC including an input track-and-hold stage followed by a plurality of low resolution ADC stages, each ADC stage including

a 1-bit comparator coupled to the input track-and-hold stage,

a 1-bit digital-to-analog converter (DAC) coupled to the 1-bit comparator,

an analog subtractor coupled to the 1-bit DAC and the input track-and-hold stage,

a residue amplifier coupled to the analog subtractor, and

a track-and-hold circuit coupled to the residue amplifier;

wherein generating signal samples by the ADC includes generating 1-bit of each sample through each of the ADC stages according to the following;

determining, by the comparator, a contribution voltage from the received signal;

developing, by the analog subtractor, a residue by subtracting the contribution voltage from the received signal;

multiplying, by the residue amplifier, the residue by a gain value; and

holding, by the track-and-hold circuit, the residue;

applying, by a feedforward equalizer, an equalization to the signal samples from the ADC to generate equalized samples;

determining, by a decoder coupled to an output of the feedforward equalizer, detected symbols from the equalized samples and a channel model by minimizing a cumulative metric that compensates the combined Gaussian and non-Gaussian noise in the received signal; and

generating, by a channel estimator, the channel model and an error feedback signal to adaptively update coefficients of the feedforward equalizer based on the error feedback signal.

9. The method of claim 8 wherein the cumulative metric is approximated by M≈

Σ

_n[(ŷ

_n)^v−

(ŝ

_n)^v]² where M is the cumulative metric, ŷ

_n represents a sample at the output of the feedforward equalizer, ŷ

_n is a noise-free signal component of ŷ

_n, where 0<

v≤

1.

10. The method of claim 8, where v is given by

11. The method of claim 10 further comprising:

providing an approximate solution to (ŷ

_n)^v by implementing a function LUT(ŷ

_n)=(ŷ

_n+y_min)^v when the sample ŷ

_n is in an interval between −

y_min and +y_max, where y_min and y_max are the minimum and maximum values of the interval of the sample ŷ

_n, respectively.

12. The method of claim 8 further comprising

storing, by a look-up table (LUT) of the ADC further configured to the plurality of ADC stages, the inverses of non-linear characteristics of the signal samples.

13. The method of claim 8 wherein generating signal samples includes

performing, by the ADC, interstage amplification and comparator operations in parallel.

14. The method of claim 8 wherein generating signal samples includes generating redundancies by the ADC configured in a sub-radix architecture having more ADC stages than there are bits in the received signal.

15. The method of claim 8 wherein multiplying the residue by a gain value includes multiplying the residue by the gain value by open-loop residue amplifiers.

16. A transceiver chip for communication over an optical fiber, the transceiver chip comprising:

a host interface for providing encoded data in a host format for transmitting over the optical fiber and for receiving decoded data in the host format originating from the optical fiber;

transmit path circuitry to receive the encoded data from the host interface and generate an electrical signal suitable for driving a laser coupled to the optical fiber;

an input port for receiving a received signal originating from the optical fiber having combined non-Gaussian noise and Gaussian noise; and

receive path circuitry coupled between the input port and the host interface, the receive path circuitry comprising;

an analog-to-digital converter (ADC) that generates signal samples from the received signal having the combined non-Gaussian noise and Gaussian noise;

wherein the ADC is a pipelined ADC including an input track-and-hold stage followed by a plurality of low resolution ADC stages, each ADC stage including

a 1-bit comparator coupled to the input track-and-hold stage,

a 1-bit digital-to-analog converter (DAC) coupled to the 1-bit comparator,

an analog subtractor coupled to the 1-bit DAC and the input track-and-hold stage,

a residue amplifier coupled to the analog subtractor, and

a track-and-hold circuit coupled to the residue amplifier;

a feedforward equalizer coupled to receive the signal samples from the ADC, and to apply equalization to generate equalized samples;

a decoder coupled to an output of the feedforward equalizer, the decoder determining detected symbols from the equalized samples and a channel model by minimizing a cumulative metric that compensates the combined Gaussian and non-Gaussian noise in the received signal, the decoder to generate the decoded data; and

a channel estimator to receive the output of the feedforward equalizer and an output of the decoder and to generate the channel model and an error feedback signal to adaptively update coefficients of the feedforward equalizer based on the error feedback signal;

wherein the ADC is configured in a sub-radix architecture having more ADC stages than there are bits in the received signal.

17. The transceiver chip of claim 16 wherein the residue amplifiers of the ADC stages include open-loop amplifiers.

18. A transceiver chip for communication over an optical fiber, the transceiver chip comprising:

a host interface for providing encoded data in a host format for transmitting over the optical fiber and for receiving decoded data in the host format originating from the optical fiber;

transmit path circuitry to receive the encoded data from the host interface and generate an electrical signal suitable for driving a laser coupled to the optical fiber;

an input port for receiving a received signal originating from the optical fiber having combined non-Gaussian noise and Gaussian noise; and

receive path circuitry coupled between the input port and the host interface, the receive path circuitry comprising;

an analog-to-digital converter (ADC) that generates signal samples from the received signal having the combined non-Gaussian noise and Gaussian noise;

wherein the ADC is a pipelined ADC including an input track-and-hold stage followed by a plurality of low resolution ADC stages, each ADC stage including

a 1-bit comparator coupled to the input track-and-hold stage,

a 1-bit digital-to-analog converter (DAC) coupled to the 1-bit comparator,

an analog subtractor coupled to the 1-bit DAC and the input track-and-hold stage,

a residue amplifier coupled to the analog subtractor, and

a track-and-hold circuit coupled to the residue amplifier;

a feedforward equalizer coupled to receive the signal samples from the ADC, and to apply equalization to generate equalized samples;

a decoder coupled to an output of the feedforward equalizer, the decoder determining detected symbols from the equalized samples and a channel model by minimizing a cumulative metric that compensates the combined Gaussian and non-Gaussian noise in the received signal, the decoder to generate the decoded data; and

a channel estimator to receive the output of the feedforward equalizer and an output of the decoder and to generate the channel model and an error feedback signal to adaptively update coefficients of the feedforward equalizer based on the error feedback signal;

wherein the cumulative metric is approximated by

M≈

Σ

_n[(ŷ

_n)^v−

(ŝ

_n)^v]² where M is the cumulative metric, ŷ

_n represents a sample at the output of the feedforward equalizer, ŝ

_n is a noise-free signal component of ŷ

_n, where 0<

v≤

1.

19. The transceiver chip of claim 18, where v is given by

20. The transceiver of claim 18 further comprising:

a transformation block to provide an approximate solution to (ŷ

_n)^v, wherein the transformation block implements a function LUT(ŷ

_n)=(ŷ

_n+y_min)^v when the sample ŷ

_n is in an interval between −

y_min and +y_max, where y_min and y_max are the minimum and maximum values of the interval of the sample ŷ

_n, respectively.

21. A transceiver chip for communication over an optical fiber, the transceiver chip comprising:

a host interface for providing encoded data in a host format for transmitting over the optical fiber and for receiving decoded data in the host format originating from the optical fiber;

transmit path circuitry to receive the encoded data from the host interface and generate an electrical signal suitable for driving a laser coupled to the optical fiber;

an input port for receiving a received signal originating from the optical fiber having combined non-Gaussian noise and Gaussian noise; and

receive path circuitry coupled between the input port and the host interface, the receive path circuitry comprising;

an analog-to-digital converter (ADC) that generates signal samples from the received signal having the combined non-Gaussian noise and Gaussian noise;

wherein the ADC is a pipelined ADC including an input track-and-hold stage followed by a plurality of low resolution ADC stages, each ADC stage including

a 1-bit comparator coupled to the input track-and-hold stage,

a 1-bit digital-to-analog converter (DAC) coupled to the 1-bit comparator,

an analog subtractor coupled to the 1-bit DAC and the input track-and-hold stage,

a residue amplifier coupled to the analog subtractor, and

a track-and-hold circuit coupled to the residue amplifier;

a feedforward equalizer coupled to receive the signal samples from the ADC, and to apply equalization to generate equalized samples;

a decoder coupled to an output of the feedforward equalizer, the decoder determining detected symbols from the equalized samples and a channel model by minimizing a cumulative metric that compensates the combined Gaussian and non-Gaussian noise in the received signal, the decoder to generate the decoded data; and

a channel estimator to receive the output of the feedforward equalizer and an output of the decoder and to generate the channel model and an error feedback signal to adaptively update coefficients of the feedforward equalizer based on the error feedback signal;

wherein the ADC further includes a look-up table (LUT) configured to the plurality of ADC stages to store the inverses of non-linear characteristics of the signal samples.

22. A transceiver chip for communication over an optical fiber, the transceiver chip comprising:

a host interface for providing encoded data in a host format for transmitting over the optical fiber and for receiving decoded data in the host format originating from the optical fiber;

transmit path circuitry to receive the encoded data from the host interface and generate an electrical signal suitable for driving a laser coupled to the optical fiber;

an input port for receiving a received signal originating from the optical fiber having combined non-Gaussian noise and Gaussian noise; and

receive path circuitry coupled between the input port and the host interface, the receive path circuitry comprising;

an analog-to-digital converter (ADC) that generates signal samples from the received signal having the combined non-Gaussian noise and Gaussian noise;

wherein the ADC is a pipelined ADC including an input track-and-hold stage followed by a plurality of low resolution ADC stages, each ADC stage including

a 1-bit comparator coupled to the input track-and-hold stage,

a 1-bit digital-to-analog converter (DAC) coupled to the 1-bit comparator,

an analog subtractor coupled to the 1-bit DAC and the input track-and-hold stage,

a residue amplifier coupled to the analog subtractor, and

a track-and-hold circuit coupled to the residue amplifier;

a feedforward equalizer coupled to receive the signal samples from the ADC, and to apply equalization to generate equalized samples;

a decoder coupled to an output of the feedforward equalizer, the decoder determining detected symbols from the equalized samples and a channel model by minimizing a cumulative metric that compensates the combined Gaussian and non-Gaussian noise in the received signal, the decoder to generate the decoded data; and

a channel estimator to receive the output of the feedforward equalizer and an output of the decoder and to generate the channel model and an error feedback signal to adaptively update coefficients of the feedforward equalizer based on the error feedback signal;

wherein the ADC further includes a lookahead pipeline configuration in which one or more of the plurality of ADC stages includes a pair of comparators configured to enable parallel interstage amplification and comparator operations.