Systems and methods for providing a pipelined analog-to-digital converter

This application claims the benefit of U.S. Provisional Patent Application No. 61/454,217, filed Mar. 18, 2011, which is hereby incorporated by reference herein in its entirety.

This invention was made with government support under grant PHY 0612811 awarded by the National Science Foundation. The government has certain rights in the invention.

Power hungry buffers and/or large decoupling capacitors have long been the solution to providing accurate reference voltages in pipelined Analog-to-Digital Converters (ADCs). Zero-crossing based pipelined ADCs offer a low power alternative to traditional op-amp based solutions, but their reference buffer design is exacerbated since the reference voltage has only a fraction of the hold phase (known as the pre-charge phase) to settle to the desired accuracy.

Accordingly, new systems and methods for providing a pipelined Analog-to-Digital Convener (ADC) are desirable.

Systems and methods for providing a pipelined Analog-to-Digital Convener (ADC) are Provided. In some embodiments, systems for providing a pipelined Analog-to-Digital Converter are provided, the systems comprising: a first multiplying Digital-to-Analog Converter (MDAC) stage comprising: a sub-Analog-to-Digital Converter (ADC) that outputs a value based on an input signal; at least two reference capacitors that are charged to a reference voltage; at least two sampling capacitors that are charged to a sampling voltage; and a plurality of switches that couple the at least two reference capacitors so that they are charged during a sampling phase, that couple the at least two sampling capacitors so that they are charged during the sampling phase, that couple at least one of the reference capacitors so that it is parallel to one of the at least two sampling capacitors during a hold phase, and that couple the other of the at least two sampling capacitors so that it couples the at least one of the reference capacitors and the one of the at least two sampling capacitors to a reference capacitor of a second MDAC stage.

In some embodiments, methods for providing a pipelined Analog-to-Digital Converter are provided, the methods comprising: in a first multiplying Digital-to-Analog Converter (MDAC) stage: outputting from a sub-Analog-to-Digital Converter (ADC) a value based on an input signal; charging at least two reference capacitors to a reference voltage; charging at least two sampling capacitors to a sampling voltage; and using a plurality of switches to couple the at least two reference capacitors so that they are charged during a sampling phase, to couple the at least two sampling capacitors so that they are charged during the sampling phase, to couple at least one of the reference capacitors so that it is parallel to one of the at least two sampling capacitors during a hold phase, and to couple the other of the at least two sampling capacitors so that it couples the at least one of the reference capacitors and the one of the at least two sampling capacitors to a reference capacitor of a second MDAC stage.

FIG. 1 is a block diagram of a pipelined Analog-to-Digital Converter (ADC) in accordance with some embodiments.

FIGS. 2 a - 2 c are block diagrams of a Multiplying Digital-to-Analog Converter (MDAC) stage in accordance with some embodiments.

FIG. 3 is a timing diagram of clock signals in accordance with some embodiments,

FIGS. 4 a - 4 h are circuit diagrams showing a pipelined ADC stage in accordance with some embodiments.

FIG. 5 is a circuit diagram of another pipelined ADC stage in accordance with some embodiments.

FIG. 6 is another timing diagram of clock signals in accordance with some embodiments.

FIG. 7 is a circuit diagram of a final ADC stage in accordance with some embodiments.

Systems and methods for providing a pipelined Analog-to-Digital Converter (ADC) are provided. In accordance with some embodiments, a zero-crossing pipelined ADC that uses a reference pre-charge technique is provided.

Turning to FIG. 1 , an example of a general architecture 100 that can be used for a pipelined ADC in some embodiments is shown. As illustrated, architecture 100 can include a clock generator 102 , non-final ADC stages 104 , 106 , and 108 , and a final ADC stage 110 . Although three non-final ADC stages are illustrated in FIG. 1 , any suitable number of non-final ADC stages can be used in some embodiments.

Clock generator 102 can be any suitable circuit for generating clock signals for controlling the pipelined ADC in some embodiments. Examples of clock signals that can be generated by generator 102 are described below in accordance with some embodiments.

Non-final ADC stages 104 , 106 , and 108 can be any suitable non-final ADC stages, and may include a sub-ADC and a Multiplying Digital-to-Analog Converter (MDAC) such as the MDAC described below in accordance with some embodiments. Final ADC stages 110 can be any suitable final ADC stages, such as those described below in accordance with some embodiments.

FIGS. 2 a , 2 b , and 2 c illustrate example architectures 202 , 216 , and 222 of a 1-bit MDAC with a charge based reference that can be used in a non-final ADC stage in accordance with some embodiments. More particularly, FIG. 2 a illustrates an example 1-bit MDAC architecture during a sample phase, FIG. 2 b illustrates an example 1-bit MDAC architecture during a hold phase, and FIG. 2 c illustrates an alternative 1-bit MDAC architecture to that in FIG. 2 a during a sample phase.

As shown in FIG. 2 a , architecture 202 can include reference capacitors C refp 204 and C refn 206 , sampling capacitors C 1 208 and C 2 210 , a comparator 212 , and a load capacitance 214 .

As shown, in the sample phase, the reference voltages V refp and V refn can be sampled onto the reference capacitors and input voltage V in can be sampled onto the sampling capacitors.

During a hold phase, as shown in FIG. 2 b , one of two switches 218 and 220 can be closed based on whether a sub-ADC output bit associated with the stage is a zero or a one. That is, if the bit is a zero, then switch 218 will be closed, and if the bit is a one, then switch 220 will be closed. Charge stored on the corresponding capacitor 204 (which corresponds to switch 218 ) or capacitor 206 (which corresponds to switch 220 ) can be added to parallel sampling capacitor C 2 210 and transferred along with charge from sampling capacitor C 1 208 to load capacitance 214 .

As shown FIG. 2 c , in some embodiments, the charge that is supplied to the reference capacitors need not come from a voltage source. Instead, currents I refp and I refn from current sources 224 and 226 can be used to charge capacitors C refp and C refn for a period of time T ref and thereby provide the same amount of charge. The value of the reference voltage can be given by V ref =(I ref /C ref )T ref . When using a current source to charge a reference capacitor, the exact value of the reference voltage sampled may not be known beforehand, as the parameters I ref , C ref and T ref can be process dependent. But because this value is input signal independent, the error can be combined with stage gain error and corrected with Digital Gain Error Correction.

In some embodiments, each MDAC can use a 1 bit/stage architecture with a nominal gain of 1.75.

Turning to FIG. 3 , a timing diagram of clock signals that can be generated by a clock generator for controlling the operation of a pipelined ADC in accordance with some embodiments is illustrated. As shown, each stage of the pipelined ADC has a sample phase and a hold phase represented by Φ s and Φ h , respectively. During the sample phase, the stage input V in is sampled onto the sampling capacitors. During the hold phase, charge from the stage is transferred to the next stage.

During a pre-charge phase of the reference phase, represented by Φ rpe for even numbered stages and Φ rpo for odd numbered stages, the reference capacitors can be discharged.

During a reference charge phase, represented by Φ re for even numbered stages and Φ ro for odd numbered stages, reference current sources I refp and I refn can charge the reference capacitors C refp1 , C refp2 and C refin1 , C refin2 to a nominal reference voltage V refp and V refin , respectively

During a pre-charge phase for the sampling phase, represented by Φ po for odd numbered stages and Φ pe for even numbered stages, the sampling capacitors can be discharged.

FIGS. 4 a - 4 h illustrate an example of a design for a pipelined ADC stage of a circuit 400 that uses a 1-bit MDAC architecture as described above, in FIGS. 2 a - 2 C in accordance with some embodiments. As shown, the ADC stage is a differential stage and receives input voltages V inp and V inn . Charge from these input voltages is stored in sampling capacitors C 1p , C 2p and C 1n , C 2n , respectively. Input voltages V inp and V inn are also compared by comparator 402 (which acts as a 1-bit sub-ADC) to provide an output bit b N . Reference capacitors C refp1 , C refp2 , C refn1 , and C refn2 are charged by current sources I refp and I refn . At the end of the sample phase, based on the value of b N output by comparator 402 , the charge from one pair of reference capacitors C refp1 , C refn1 or C refp2 , C refn2 is transferred to the output capacitors C 1p , C 1n during the hold phase, by shorting one of their plates to V cm .

In circuit 400 , the sampling of the input voltages, the charging of the reference capacitors, and the transfer of charge from these capacitors to the next stage is controlled by the illustrated switches that open or close based on the value of Φ 1i , Φ rpe , Φ rpo , Φ re , Φ ro , Φ pe , Φ po , Φ N−1 , Φ dN−1 , Φ N , Φ dN , Φ N+1 , Φ dN+1 , b N , and b N+1 .

As described above, Φ h , Φ rpe , Φ rpo , Φ re , Φ ro , Φ pe , Φ po can be generated by a clock generator in some embodiments. The eight FIGS. 4 a - 4 h show the states of switches for Φ h , Φ rpe , Φ rpo , Φ re , Φ ro , Φ pe , Φ po at the eight sub-stage periods 1-8, respectively, shown in FIG. 3 , in some embodiments.

Φ N−1 , Φ dN−1 , Φ N , Φ dN , Φ N+1 , Φ dN+1 can be generated by the ZCDs and delay circuits shown in some embodiments. b N and b N+1 can be generated by the comparators shown in sonic embodiments. Although specific states for the switches for Φ N−1 , Φ dN−1 , Φ N , Φ dN , Φ N+1 , Φ dN+1 , b N , and b N+1 are shown, these switches can be opened or closed in some of the sub-stage periods based on the operation of the circuit and the stage input voltage in some embodiments. b N and b N+1 can correspond to switches that close when these signals are high and b N * and b N+1 * can correspond to switches that close when the signals b N and b N+1 are low.

Any suitable zero-crossing detector (ZCD) can be used in circuit 400 in some embodiments. For example, a ZCD similar to the ZCD described in L. Brooks et al., “A 12b 50 MS/s, Fully Differential Zero-Crossing Based Pipelined ADC,” IEEE Journal of Solid-State Circuits, December 2009, which is hereby incorporated by reference herein in its entirety, can be used in some embodiments. In some embodiments, this ZCD can use static inverters at its output to generate the necessary logic levels. Overshoot can be corrected by adjusting the relative strengths of the PMOS load pair of the ZCD in some embodiments.

Although not shown, additional calibration logic for gain calibration can be implemented in some embodiments.

In some embodiments, the output current sources I p , I n that perform the charge transfer can employ local negative feedback to provide a constant ramp rate for good linearity.

In some embodiments, the bias currents for a first set of stages (e.g., stages 1-3) can be analog and digitally programmable, while those of the rest of the stages can share a common digital control.

In some embodiments, pipelined ADC 100 of FIG. 1 can be implemented so as to provide a multi-bit output at each ADC stage. Multi-bit non-final ADC stages 104 , 106 , and 108 in such a pipelined ADC can be any suitable multi-bit ADC stages, and can include a sub-ADC path and a multi-bit MDAC stage in some embodiments.

In some embodiments, the sub-ADC path can include a set of signal and reference capacitors and a sequential search a successive approximation register (SAR) ADC that can obtain its thresholds by capacitive division between its input and its reference, and hence may not require a reference buffer.

In some embodiments, the multi-bit MDAC stages can be implemented using a 2.5 bit zero-crossing MDAC stage, each with a gain of four, such as stage 500 of FIG. 5 in some embodiments.

Similarly to the circuits described above, in stage 500 , during the sample phase Φ s , the differential input V inp , V inn is sampled on the signal (or sampling) capacitors C sigp , C sign . Concurrently, during the reference pre-charge phase Φ ref , six reference current sources, I refp <5:0> and I refn <5:0>, pre-charge six reference capacitors, C refp <5:0> and C refn <5:0>, to the nominal reference voltages. At the end of the pre-charge phase Φ p , based on outputs S<5:0> of a sub-ADC corresponding to the stage (not shown), the reference capacitors are connected appropriately to V p and V n by a switch 502 . By the end of the hold phase Φ h , the charge is transferred to the next stage capacitors C′ sigp , C′ sign . The value of the reference voltage in the case of reference pre-charging is given by V ref =(I refp /C refp −I refn /C refn )T ref , where T ref is the duration for which the reference current sources I repf <5:0> and <5:0> charge the reference capacitors, C refp <5:0> and C refn <5:0>. The exact value of V ref may not be known beforehand in some embodiments, but, because V ref is input-signal independent, the variation can be combined with stage gain error and corrected with digital gain error correction.

The loading can be reduced in some embodiments by connecting the reference capacitors to V p and V n only when required.

In some embodiments, each C ref <5:0> can be chosen to be half the size of C sig as a tradeoff between the extra loading and power supply noise rejection. In some embodiments, the reference current sources I refp <5:0> and I refn <5:0> can be implemented as digitally programmable cascoded current sources. Switches (not shown) can be used to short the reference capacitors to the common-mode voltage V cm during Φ pr , and pre-charge the outputs V outp and V outn to GND and V DD respectively, during Φ p , in a similar manner to that shown in FIGS. 4 a - 4 h in some embodiments.

The output current sources, I p and I n , perform the actual charge transfer and can be implemented as regulated cascodes for good linearity performance in some embodiments. The zero-crossing detector (ZCD) can include a two-stage differential-to-single-ended amplifier for improved common-mode rejection. The overshoot at the output of the MDAC due to the finite delay of the ZCD can be compensated for by introducing an offset through a digitally programmable capacitor array connected to V p and V n as described in J. Chu et al., “A zero-crossing based 12b 100 MS/s pipelines ADC with decision boundary gap estimation calibration,” VLSI Circuits (VLSIC), 2010 IEEE Symposium on, June 2010, which is hereby incorporated by reference herein in its entirety.

A timing diagram for clock signals that can be used control stage 500 of FIG. 5 in accordance with some embodiments is shown in FIG. 6 . The illustrated clock signals can be generated by any suitable clock generator in some embodiments.

An example of a circuit 700 that can be used to implement a final ADC stage 110 in accordance with some embodiments is shown in FIG. 7 . As illustrated, reference pre-charging can be implemented in circuit 700 similarly to that shown and described above in connection with FIG. 5 . Circuit 700 can also use a binary search SAR 702 to resolve the final eight bits of the pipelined ADC in some embodiments. Depending on the outputs D<7:1> of SAR 702 , the reference capacitors are connected appropriately by switch 704 to implement a binary search in some embodiments. The SAR timing can be controlled using a delay locked loop in some embodiments.

In some embodiments, a pipelined ADC as described here can be implemented using only two external voltages: a V DD and a common-mode voltage V cm , which can be nominally V DD /2 formed using a resistive divider from V DD .

Although the invention has been described and illustrated in the foregoing illustrative embodiments, it is understood that the present disclosure has been made only by way of example, and that numerous changes in the details of implementation of the invention can be made without departing from the spirit and scope of the invention, which is only limited by the claims which follow. Features of the disclosed embodiments can be combined and rearranged in various ways.