Alignment of sampling phases in a multi-channel time-interleaved analog-to-digital converter
View Patent ↗A multi-channel analog-to-digital (ADC) converter coupled to a clock-and-data-recovery loop that has a plurality of clock-recovery circuits, each configured to set the sampling phase for a respective one of the ADC channels in a manner that causes the different sampling phases to be appropriately time-aligned with one another for time-interleaved operation of the ADC channels. In an example embodiment, an individual clock-recovery circuit comprises a phase detector and a loop filter. Loop filters corresponding to different clock-recovery circuits may be coupled to one another by having shared circuit elements in their frequency-tracking paths and/or by being configured to receive timing gradients from more than one phase detector, including the phase detector of a selected one of the clock-recovery circuits.
1. An apparatus comprising:
M analog-to-digital converters, each configured to generate, at a first rate, digital samples of an electrical input signal modulated with data, where M is an integer greater than one;
an interleaver configured to time-interleave the digital samples generated by the M analog-to-digital converters to generate a sample stream having a second rate that is M times greater than the first rate; and
a clock-and-data recovery loop configured to process the sample stream to generate a data stream carrying the data at the second rate and to set M sampling phases, each for a respective one of the M analog-to-digital converters, wherein the clock-and-data recovery loop comprises:
a first de-interleaver configured to time-de-interleave the data stream to generate M data signals, each having the first rate; and
M clock-recovery circuits, each configured to set a respective one of the M sampling phases based on a respective one of the M data signals.
2. The apparatus of claim 1 , wherein M=4.
3. The apparatus of claim 1 ,
wherein the clock-and-data recovery loop is further configured to process the sample stream to generate an error signal carrying, at the second rate, error values corresponding to the data;
wherein the clock-and-data recovery loop further comprises a second de-interleaver configured to time-de-interleave the error signal to generate M de-interleaved error signals, each having the first rate; and
wherein each of the M clock-recovery circuits is further configured to set the respective one of the M sampling phases based on a respective one of the M de-interleaved error signals.
4. The apparatus of claim 3 , wherein each of the M clock-recovery circuits comprises:
a respective phase detector configured to process the respective one of the M data signals and the respective one of the M de-interleaved error signals to determine a respective timing gradient;
a respective loop filter configured to combine present and past values of the respective timing gradient to generate a respective estimate for a phase change; and
a respective phase interpolator configured to generate a value of the respective one of the M sampling phases based on the respective estimate for the phase change received from the respective loop filter.
5. The apparatus of claim 4 ,
wherein each of the respective loop filters comprises:
a respective first-order path configured to track, based on the respective timing gradient, a phase of a stream of the digital samples generated by the respective one of the M analog-to-digital converters; and
a respective second-order path configured to track, based on the respective timing gradient, a frequency of said stream of the digital samples generated by the respective one of the M analog-to-digital converters; and
wherein each of the respective loop filters is configured to generate the respective estimate for the phase change based on said tracking of the phase and said tracking of the frequency.
6. The apparatus of claim 3 , wherein the M clock-recovery circuits comprise:
M phase detectors, each configured to process the respective one of the M data signals and the respective one of the M de-interleaved error signals to determine a respective timing gradient;
a loop filter configured to combine present and past values of different respective timing gradients to generate M estimates for a phase change; and
M phase interpolators, each configured to generate a value of the respective one of the M sampling phases based on a respective one of the M estimates for the phase change received from the loop filter.
7. The apparatus of claim 6 , wherein M=4.
8. The apparatus of claim 6 ,
wherein the loop filter comprises:
M first-order paths, each connecting a respective one of the M phase detectors and a respective one of the M phase interpolators and configured to track, based on the respective timing gradient, a phase of a stream of the digital samples generated by the respective one of the M analog-to-digital converters; and
M second-order paths, each connecting a common one of the of the M phase detectors and a respective one of the M phase interpolators and configured to track, based on the timing gradient determined by said common one of the of the M phase detectors, an approximated frequency of said stream of the digital samples generated by the respective one of the M analog-to-digital converters; and
wherein the loop filter is configured to generate the M estimates for the phase change based on said tracking of the phases and said tracking of the approximated frequencies.
9. The apparatus of claim 8 , wherein each of the M second-order paths includes a common frequency-update gain element.
10. The apparatus of claim 8 , wherein each of the M second-order paths includes at least one common shift register.
11. The apparatus of claim 8 , wherein each of the M second-order paths includes a common frequency-update gain element and a common shift register.
12. The apparatus of claim 1 , wherein the clock-and-data recovery loop further comprises:
a feed-forward equalizer configured to process the sample stream to reduce inter-symbol interference therein;
a decision-feedback equalizer configured to further process a processed sample stream generated by the feed-forward equalizer to further reduce inter-symbol interference therein based on previous symbol estimates received from a slicer; and
the slicer configured to generate a data stream based on a further-processed sample stream generated by the decision-feedback equalizer.
13. The apparatus of claim 12 , wherein the clock-and-data recovery loop further comprises an error-estimation circuit configured to generate an error signal corresponding to the data in the data stream based on a comparison of the further-processed sample stream and the data stream.
14. The apparatus of claim 1 , further comprising:
a front-end circuit configured to generate the electrical input signal based on a communication signal received via a communication channel; and
a transmitter coupled to the communication channel and configured to cause the front-end circuit to receive the communication signal.
15. The apparatus of claim 14 , further comprising:
a first electronic chip configured to generate a first plurality of data streams for transmission over the communication channel;
a multiplexer configured to:
multiplex the first plurality of data streams to generate a multiplexed data stream; and
apply the multiplexed data stream to the transmitter to cause the transmitter to generate the communication signal;
a de-multiplexer configured to de-multiplex the data stream to generate a second plurality of data streams;
a second electronic chip configured to receive the second plurality of data streams, each at a different respective pin thereof.
16. A method of sampling an electrical input signal modulated with data, the method comprising:
triggering M analog-to-digital converters, each at a respective one of M sampling phases to cause each of said M analog-to-digital converters to generate, at a first rate, digital samples of the electrical input signal, where M is an integer greater than one;
time-interleaving the digital samples generated by the M analog-to-digital converters in an interleaver to generate a sample stream having a second rate that is M times greater than the first rate; and
setting the M sampling phases using a clock-and-data recovery loop, wherein said setting comprises:
processing the sample stream to generate a data stream carrying the data at the second rate;
time-de-interleaving the data stream in a first de-interleaver to generate M data signals, each having the first rate; and
processing each of the M data signals in a respective one of M clock-recovery circuits to determine a respective one of the M sampling phases.
17. The method of claim 16 , wherein said setting further comprises:
processing the sample stream to generate an error signal carrying, at the second rate, error values corresponding to the data;
time-de-interleaving the error signal in a second de-interleaver to generate M de-interleaved error signals, each having the first rate; and
determining the respective one of the M sampling phases based on a respective one of the M de-interleaved error signals.
18. The method of claim 17 , wherein the M clock-recovery circuits comprise:
M phase detectors, each configured to process the respective one of the M data signals and the respective one of the M de-interleaved error signals to determine a respective timing gradient;
a loop filter configured to combine present and past values of different respective timing gradients to generate M estimates for a phase change; and
M phase interpolators, each configured to generate a value of the respective one of the M sampling phases based on a respective one of the M estimates for the phase change received from the loop filter.
19. The method of claim 18 ,
wherein the loop filter comprises:
M first-order paths, each connecting a respective one of the M phase detectors and a respective one of the M phase interpolators and configured to track, based on the respective timing gradient, a phase of a stream of the digital samples generated by the respective one of the M analog-to-digital converters; and
M second-order paths, each connecting a common one of the of the M phase detectors and a respective one of the M phase interpolators and configured to track, based on the timing gradient determined by said common one of the of the M phase detectors, an approximated frequency of said stream of the digital samples generated by the respective one of the M analog-to-digital converters; and
wherein the loop filter is configured to generate the M estimates for the phase change based on said tracking of the phases and said tracking of the approximated frequencies.
20. A method of sampling an electrical input signal, the method comprising:
triggering M analog-to-digital converters, each at a respective one of M sampling phases, to cause each of said M analog-to-digital converters to generate, at a first rate, digital samples of the electrical input signal, where M is an integer greater than one;
time-interleaving the digital samples generated by the M analog-to-digital converters in an interleaver to generate a sample stream having a second rate that is M times greater than the first rate;
tracking a frequency of a selected one of M streams of the digital samples generated by the M analog-to-digital converters to obtain M−1 frequency estimates, each being an estimate of a frequency of a respective one of the remaining M−1 of the M streams of the digital samples;
determining a sampling phase for an analog-to-digital converter that generates said selected one of the M streams of the digital samples based on said tracking of the frequency and by tracking a phase of said selected one of the M streams of the digital samples using a respective one of M clock recovery circuits; and
determining each of the remaining M−1 of the M sampling phases based on a respective one of the M−1 frequency estimates obtained using said tracking of the frequency and by tracking a phase of a respective one of the remaining M−1 of the M streams of the digital samples using a respective one of the remaining M−1 of the M clock recovery circuits.