Calibrating time-interleaved switched-capacitor track-and-hold circuits and amplifiers
Background calibration techniques can effectively to correct for memory, kick-back, and order-dependent errors in interleaved switched-capacitor track-and-hold (T/H) circuits and amplifiers. The techniques calibrate for errors in both the track/sample phase and the hold-phase, and account for the effects of interleaving, buffer/amplifier sharing, incomplete resetting, incomplete settling, chopping, and randomization on the offset, gain, memory, and kick-back errors. Moreover, the techniques can account for order-dependent and state-dependent hold-phase non-linearities. By correcting for these errors, the proposed techniques improve the noise performance, linearity, gain/offset matching, frequency response (and bandwidth), and order-dependence errors. The techniques also help increase the speed (sample rate and bandwidth) and linearity of T/H circuits and amplifiers while simplifying the analog circuitry and clocking needed. These techniques comprehensively account for various memory, kick-back, and order-dependent effects in a unified framework.
1. A method for addressing errors in time-interleaved sampling networks, comprising:
injecting an additive dither in each time-interleaved sampling network, each time-interleaved sampling network sampling an input signal in a randomized sequence;
quantizing outputs of the time-interleaved sampling networks to generate a digital output;
updating different correction coefficients for a first time-interleaved sampling network based on the additive dither, using samples of the digital output generated according to different orders in which the time-interleaved sampling networks sample the input signal; and
correcting the digital output corresponding to the different orders separately, using the different correction coefficients.
2. The method of claim 1 , wherein updating different correction coefficients comprises:
updating a first correction coefficient for the first time-interleaved sampling network based on the additive dither and samples of the digital output that correspond to a first order of time-interleaved sampling networks sampling the input signal; and
updating a second correction coefficient for the first time-interleaved sampling network based on the additive dither and samples of the digital output that correspond to a second order of time-interleaved sampling networks sampling the input signal.
3. The method of claim 2 , wherein correcting the digital output comprises:
correcting samples of the digital output that correspond to the first order based on the first correction coefficient; and
correcting samples of the digital output that correspond to the second order based on the second correction coefficient.
4. The method of claim 1 , wherein the different orders specify which one of the time-interleaved sampling networks generated one or more of the following: a given sample, a previous sample, and a subsequent sample.
5. The method of claim 1 , further comprising:
injecting a multiplicative dither in each time-interleaved sampling network.
6. The method of claim 5 , wherein updating the different correction coefficients comprises:
updating the different correction coefficients using samples of the digital output generated according to: (1) the different orders, and (2) different values of the multiplicative dither corresponding to one or more of the following: a given sample, a previous sample, and a subsequent sample.
7. The method of claim 5 , wherein correcting the digital output comprises:
correcting the digital output for offset errors before removing the multiplicative dither from the digital output.
8. The method of claim 1 , wherein:
updating the different correction coefficients comprises adaptively adjusting the different correction coefficients, based on corrected samples of the digital output and the additive dither, to reduce the errors.
9. The method of claim 1 , wherein:
one of the errors is defined based on the additive dither and a sample of the digital output with an estimate of the additive dither removed; and
the estimate of the additive dither is based on the additive dither and a corresponding correction coefficient.
10. The method of claim 1 , wherein:
one of the errors is defined based on the additive dither corresponding to a previous sample of the digital output and a sample of the digital output with an estimate of a memory contribution removed; and
the estimate of the memory contribution is based on the previous sample of the digital output and a corresponding correction coefficient.
11. The method of claim 1 , wherein:
one of the errors is defined based on the additive dither corresponding to a previous sample of the digital output generated by the first time-interleaved sampling network and a sample of the digital output with an estimate of a memory contribution removed; and
the estimate of the memory contribution is based on the previous sample of the digital output generated by the first time-interleaved sampling network and a corresponding correction coefficient.
12. The method of claim 1 , wherein:
one of the errors is defined based on a sample of the digital output and an estimate of offset.
13. The method of claim 1 , wherein correcting the digital output comprises:
correcting the digital output for memory errors before correcting for offset and/or gain errors.
14. The method of claim 1 , wherein the different correction coefficients comprise correction coefficients to reduce memory error due to partial resetting of the time-interleaved sampling networks.
15. The method of claim 1 , further comprising:
resetting a summing node of the time-interleaved sampling networks by overlapping a sampling clock signal and a hold clock signal.
16. The method of claim 1 , further comprising:
resetting the first time-interleaved sampling network when the first time-interleaved sampling network remains idle for more than one phase.
17. The method of claim 1 , wherein the errors comprise one or more of: non-idealities in a hold-phase of the first time-interleaved sampling network, hold-phase gain error, hold-phase memory error, hold-phase offset error, and hold-phase non-linearities.
18. The method of claim 1 , wherein the errors comprise one or more of: non-idealities in a track-phase of the first time-interleaved sampling network, track-phase gain error, track-phase memory error, and track-phase offset error.
19. A method for addressing errors in time-interleaved sampling networks, comprising:
injecting an additive dither and a multiplicative dither in each time-interleaved sampling network, each time-interleaved sampling network sampling an input signal in a sequence;
quantizing outputs of the time-interleaved sampling networks to generate a digital output;
updating different correction coefficients for a first time-interleaved sampling network based on the additive dither, using samples of the digital output generated according to different values of the multiplicative dither; and
correcting the digital output corresponding to the different values of the multiplicative dither separately, using the different correction coefficients.
20. The method of claim 19 , wherein updating the different correction coefficients comprises:
updating a first correction coefficient for the first time-interleaved sampling network based on the additive dither and samples of the digital output that correspond to a first value of the multiplicative dither; and
updating a second correction coefficient for the first time-interleaved sampling network based on the additive dither and samples of the digital output that correspond to a second value of the multiplicative dither.
21. The method of claim 19 , wherein correcting the digital output comprises:
correcting samples of the digital output that correspond to the value of the multiplicative dither; and
correcting samples of the digital output that correspond to the value of the multiplicative dither.
22. The method of claim 19 , wherein the different values of the multiplicative dither comprises +1 and −1.
23. The method of claim 19 , wherein injecting a multiplicative dither comprises changing a polarity of a differential signal in the first time-interleaved sampling network when a value of the multiplicative dither is −1.
24. The method of claim 19 , wherein updating the different correction coefficients comprises:
updating the different correction coefficients using samples of the digital output generated according to: (1) different orders in which the time-interleaved sampling networks sample the input signal, and (2) the different values of the multiplicative dither corresponding to one or more of the following: a given sample, a previous sample, and a subsequent sample.
25. The method of claim 19 , wherein:
updating the different correction coefficients comprises adaptively adjusting the different correction coefficients, based on corrected samples of the digital output and the additive dither, to reduce the errors.
26. The method of claim 19 , wherein:
one of the errors is defined based on the additive dither and a sample of the digital output with an estimate of the additive dither removed; and
the estimate of the additive dither is based on the additive dither and a corresponding correction coefficient.
27. The method of claim 19 , wherein:
one of the errors is defined based on the additive dither corresponding to a previous sample of the digital output and a sample of the digital output with an estimate of a memory contribution removed; and
the estimate of the memory contribution is based on the previous sample of the digital output and a corresponding correction coefficient.
28. The method of claim 19 , wherein:
one of the errors is defined based on the additive dither corresponding to a previous sample of the digital output generated by the first time-interleaved sampling network and a sample of the digital output with an estimate of a memory contribution removed; and
the estimate of the memory contribution is based on the previous sample of the digital output generated by the first time-interleaved sampling network and a corresponding correction coefficient.
29. The method of claim 19 , wherein:
one of the errors is defined based on a sample of the digital output and an estimate of offset.
30. The method of claim 19 , wherein correcting the digital output comprises:
correcting the digital output for memory errors before correcting for offset and/or gain errors.
31. The method of claim 19 , wherein correcting the digital output comprises:
correcting the digital output for offset errors before removing the multiplicative dither from the digital output.
32. The method of claim 19 , wherein the different correction coefficients comprise correction coefficients to reduce memory error due to partial resetting of the time-interleaved sampling networks.
33. The method of claim 19 , further comprising:
resetting a summing node of the time-interleaved sampling networks by overlapping a sampling clock signal and a hold clock signal.
34. The method of claim 19 , further comprising:
resetting the first time-interleaved sampling network when the first time-interleaved sampling network remains idle for more than one phase.
35. The method of claim 19 , wherein the errors comprise one or more of:
non-idealities in a hold-phase of the first time-interleaved sampling network, hold-phase gain error, hold-phase memory error, hold-phase offset error, and hold-phase non-linearities.
36. The method of claim 19 , wherein the errors comprise one or more of:
non-idealities in a track-phase of the first time-interleaved sampling network, track-phase gain error, track-phase memory error, and track-phase offset error.
37. A system with calibration, comprising:
track-and-hold circuit to receive an analog input, wherein the track-and-hold circuit comprises randomized time-interleaved sampling networks, and each time-interleaved sampling network has additive dither injection circuitry;
analog-to-digital converter to digitize an output from the track-and-hold circuit; and
digital calibration logic to:
determine different correction coefficients separately using samples of a digital output from the analog-to-digital converter generated according to different orders in which the randomized time-interleaved sampling networks sample the analog input, and
correct a digital output from the analog-to-digital converter using the different correction coefficients.
38. The system of claim 37 , wherein:
each time-interleaved sampling network has a random chopper; and
the digital calibration logic is further to determine the different correction coefficients separately for the different orders and for different states of the random chopper.
39. The system of claim 38 , further comprising:
memory, for each randomized time-interleaved sampling network, to store a last sample of the digital output generated using the given randomized time-interleaved sampling network.