Time-interleaved skew reduced pipelined analog to digital converter

This application claims priority to U.S. Provisional Application Ser. No. 61/845,858, filed Jul. 12, 2013, which is incorporated herein by reference in its entirety.

This disclosure relates to analog to digital converters. This disclosure also relates to the time-interleaved skew reduced, pipelined analog to digital converters that may also be SHA-less.

An analog-to-digital converter (ADC) includes an electronic device that receives an analog signal. The analog-to-digital converter samples the received signal and outputs a signal representing a digital value. The digital value may be acquired in operations whose outputs represent a number of bits. The ADC may be used to sample a variety of analog waveforms in the form of radio-frequency wave, sound waves, or voltage signal.

The innovation may be better understood with reference to the falling drawings and description. In the figures, like reference numerals designate corresponding parts throughout the different views.

FIG. 1 shows an example of user equipment that includes an analog to digital converter.

FIG. 2 is a circuit diagram of an exemplary first stage of a pipelined ADC.

FIG. 3 is an exemplary timing diagram for the circuits in FIG. 2 .

FIG. 4 shows a circuit diagram of an example skew reduced and/or SHA-less time-interleaved pipelined ADC.

FIG. 5 shows an exemplary timing diagram of the example skew reduced and/or SHA-less pipelined ADC of FIG. 4 .

FIG. 6 is a circuit diagram of an exemplary lane circuit for the skew reduced and/or SHA-less time-interleaved pipelined ADC of FIG. 4 .

FIG. 7 shows a circuit diagram of an example skew reduced and/or SHA-less time-interleaved pipelined ADC.

FIG. 8 shows an exemplary timing diagram corresponding to an example skew reduced and/or SHA-less pipelined ADC of FIG. 7 .

FIG. 9 shows another exemplary timing diagram corresponding to an example skew reduced and/or SHA-less pipelined ADC of FIG. 7 .

FIG. 10 is a block diagram of an exemplary quantization circuit and exemplary timing diagram for a full-rate sampling with lower speed quantization channels.

FIG. 11 is a block diagram of the exemplary quantization circuit of FIG. 10 and exemplary timing diagram for N quantization channels.

FIG. 12 is an exemplary flowchart 1200 describing a timing for sampling signals.

Battery-mode operation of full-band capture systems can be enabled if the power dissipation is controlled. For high-speed, wide-band analog to digital converters (ADC), a major amount of power can be consumed in the sample and hold amplifier (SHA) stage. This stage may also include noise and signal distortion. Additionally, time-interleaved systems may suffer from clock skew. To mitigate the performance degradation due to clock skew, either a global sampling technique or power consuming digital correction circuitry may be needed. Using a combination of the global sampling technique and the SHA-elimination technique can help control power consumption and mitigate clock skew.

The discussion below makes reference to user equipment that can implement a SHA-less, skew reduced, pipelined ADCs described herein. User equipment may take many different forms and have many different functions. As one example, user equipment may be a 2G, 3G, or 4G/LTE cellular phone capable of making and receiving wireless phone calls, and transmitting and receiving data. The user equipment may also be a smartphone that, in addition to making and receiving phone calls, runs any number or type of applications. User equipment may be virtually any device that transmits and receives information, including as additional examples a driver assistance module in a vehicle, an emergency transponder, a pager, a satellite television receiver, a networked stereo receiver, a computer system, music player, or virtually any other device. The techniques discussed below may also be implemented in a base station or other network controller that communicates with the user equipment.

FIG. 1 shows an example of user equipment (UE) 100 . The user equipment 100 can include among other things system components 112 , system logic 114 and a user interface 118 . System logic 114 can be part of the implementation of desired functionality in the UE 100 . In that regard, the system logic 114 may include logic that facilitates, as examples, running applications; accepting user inputs; saving and retrieving application data; establishing, maintaining, and terminating cellular phone calls or data connections for, as one example, Internet connectivity; establishing, maintaining, and terminating wireless network connections, Bluetooth connections, or other connections; and displaying relevant information on the user interface 118 . The user interface 118 may include a graphical user interface (GUI), touch sensitive display, voice, facial and gesture recognition inputs, buttons, switches, speakers and other user interface elements.

The system logic 114 can include control instructions 122 , control parameters 124 and operating parameters 130 , e.g., for implementing the SHA-less and/or skew-less pipelined ADC and for controlling the system components 112 . The system components 112 may be implemented, for example, in a system on a chip (SoC), application specific integrated circuit (ASIC), or other circuitry. In the system components 112 , Radio Frequency (RF) transmit (Tx) and receive (Rx) circuitry 130 handles transmission and reception of signals through the antenna(s) 132 .

FIG. 2 is a circuit diagram of an exemplary first stage of a pipelined ADC 200 . In a ping-pong, pipelined ADC, “S” is the input signal sampling, “0” is the sub-ADC quantization, and “A” is the residue amplification by using the residue amplifier (RA) 202 . The circuit 204 details the inside lane A/B. The RA amplifies the difference between the sampled input signal Vin_s 206 and the sub-ADC output Vo_sub 208 . The sampling switches 210 and amplification switches 212 control when sampling and amplification occur, respectively. Therefore, quantization Q can be completed before amplification A occurs.

FIG. 3 is an exemplary timing diagram for the circuits in FIG. 2 . The upper lane LA 214 performs sampling S in phase P 1 , amplification A in phase P 2 and quantization Q in P 3 . The lower lane LB 216 performs sampling S in phase P 2 , amplification A in phase P 1 and quantization Q in P 4 . The RA operates in both phases P 1 and P 2 , and is shared between two lanes LA 214 and LB 216 , e.g., in a time-interleaved way.

In high-speed applications, the time period t 3 −t 2 (non-overlapping time between P 1 and P 2 ) may not be enough for quantization Q. Some approaches may try to increase t 3 −t 2 by either advancing t 2 to t 1 or delaying t 3 to t 4 . In the first case, with a continuously moving input signal without a SHA, Vin_s 206 and Vin_sub 208 may be different since their corresponding sampling instants are different (t 2 and t 1 ). This can result in an over-ranged first stage output. In the second case, the time for amplification A is reduced by amount t 4 −t 3 , which may result in a higher speed requirement of the RA. Additionally or alternatively, an ADC 200 with more than one lane and SHA-less operation may suffer from clock skew.

FIG. 4 shows a circuit diagram of an example skew reduced and/or Sha-less time-interleaved pipelined ADC 400 . The ADC 400 includes a determined number of lanes, e.g., four lanes L 1 402 , L 2 404 , L 3 406 and L 4 408 . More or less lanes may be used depending on an implementation. The lanes L 1 402 , L 2 404 , L 3 406 and L 4 408 can perform sampling S, quantization Q and amplification A on input signal V 1 410 and output a signal V 2 412 . A timing of the sampling S, quantization Q and amplification A is controlled by clock signals F.

FIG. 5 shows an exemplary timing diagram of the skew reduced and/or SHA-less pipelined ADC 400 of FIG. 4 . When F 1 is high, lane L 1 402 performs sampling S, when F 2 is high, lane L 2 performs sampling S, when F 3 is high lane L 3 performs sampling S and when S 4 is high lane 4 performs sampling S. Clock signal F 0 can be used to control skew. The falling edges of F 0 occur slightly before the corresponding falling edges of F 1 /F 2 /F 3 /F 4 , so that the sampling instants of four lanes are controlled by the single clock F 0 , resulting in clock skew reduced operation.

FIG. 6 is a circuit diagram of an exemplary sampling circuit 500 , e.g., lane 1 the skew reduced and/or SHA-less time-interleaved pipelined ADC 400 of FIG. 4 . To reduce power consumption and/or lessen clock skew, serial switches S 1 and S 2 may be used together in the sampling circuit 500 to time the sampling of input signal Vin. One switch, e.g., switch S 2 , may be switched based on a clock signal for sampling Fs, e.g., F 1 of FIG. 5 , for lane 1 of the interleaved system. Each lane can receive an individual Fs clock signal, e.g., F 1 , F 2 , F 3 and F 4 .

The lanes L 1 /L 2 /L 3 /L 4 of the time-interleaved pipelined, e.g., ADC 400 of FIG. 4 , can be skew-less or reduced skew by including another switch, e.g., switch S 1 of FIG. 6 . Switch S 1 switches based on a global clock Fse signal, e.g., F 0 in FIG. 5 . The sampling circuit 500 can represent one lane, e.g., lane 1 in the interleaved sampling. The sampling in all the lanes L 1 /L 2 /L 3 /L 4 is synchronized to the global clock signal Fse as one clock signal is assigned for all lanes of the time interleaved system. The falling edge of Fse (F 0 in FIG. 5 ) comes slightly earlier than the corresponding edge of Fs, e.g. F 1 /F 2 /F 3 /F 4 . Therefore, the input signal is sampled onto both Cs 1 and Cs 2 at the falling edge of Fse, instead of Fs. Since the same clock (F 0 in FIG. 5 ) is assigned to Fse in all four lanes L 1 /L 2 /L 3 /L 4 , the clock skew in a time-interleaved system can be resolved. When switches S 1 and S 2 are closed, the sampling occurs at a proper time for the lane (switch S 2 ) and without skew (switch S 1 ). Serial switches S 1 and S 2 together can also be implemented by other components, e.g., multi-gate finfet/mosfet devices.

Additionally or alternatively, the ADC 400 can also be SHA-less. The Vin signal is held on capacitor Cs 1 and capacitor Cs 2 upon signal F 1 ( FIG. 5 ) for lane 1 sampling. During the phase when clock sampling signal Fs is high, the bottom plate of Cs 1 (node n 1 ) and the bottom plate of Cs 2 (node n 2 ) are connected with Vin. The top plates of Cs 1 and Cs 2 (node n 4 ) are connected to AC ground via the two serial switches S 1 and S 2 (closed switches in the dashed circles) which are controlled by Fs and Fse, respectively. The capacitor Csub can sample the same way. Since the global clock signal Fse is used to control the time of sampling for Cs 1 and Csub, there is no need to place a SHA preceding the circuit, resulting in SHA-less operation.

A residue amplification stage 502 can connect with node 4 . A switch Fa can control timing for amplification of the sampled signal. The residue amplification stage can output signal Vo. Output signal Vo is the amplified difference between the input signal Vin and the quantized signal Vdac.

A dedicated full clock phase Fq may be used for the quantization Q stage 504 , thus the speed requirement on the sub-ADC may be greatly relaxed compared to systems in FIG. 2 . At the falling edge of Fq, the sub-ADC can start quantization. The sub-ADC quantization can also begin after the falling edge of Fs, if top-plate sampling is used in the sub-ADC sampling. The SHA-less design may allow for power saving and the lane interleaving provides for high-speeds of the wide-band ADC's, e.g., in the multiplying digital to analog converter (MDAC) or successive approximation (SAR) ADC stage.

In one example, total power dissipation can be reduced, e.g., about halved, by using this SHA-less architecture and additional power reduction can occur due to the clock skew reduced operation. Power utilizing digital clock skew error correction circuitry can be eliminated and better noise and distortion performance may be achieved with the same power dissipation. For the same power dissipation, the speed of the ADC can be faster, e.g., two times faster, than other ADC's that are not using this architecture. During exemplary chip measurements of the time-interleaved skew reduced and/or SHA-less technique, the total harmonic distortion (THD) may not degrade abruptly with increasing input signal frequency. For chip de-cap, de-layer, a clock tree layout may use multiple clock phases.

Other advantages may include that the chip area of the ADC may be reduced due to SHA-less operation. Background calibrations may not be needed to compensate the voltage difference between Vin_s and Vin_sub if SHA-less operation is enabled, and residue amplification time or the input sampling time to perform sub-ADC quantization need not be sacrificed. A sampling difference can be eliminated, and both the input sampling time and residue amplification time need not be sacrificed. Clock skew reduced operation is integrated, while maintaining SHA-less operation.

FIG. 7 shows a circuit diagram of an example skew reduced and/or Sha-less time-interleaved pipelined ADC 400 . FIG. 8 shows an exemplary timing diagram corresponding to an example skew reduced and/or SHA-less pipelined ADC of FIG. 7 . There are four lanes (L 1 through L 4 ). When F 1 is high, lane L 1 performs sampling S. When F 5 is high, lane L 1 performs quantization Q. When F 9 is high, lane L 1 performs amplification A. Other lanes can operate in a similar way. The falling edges of F 0 are slightly before the corresponding falling edges of F 1 /F 2 /F 3 /F 4 , so that the sampling instants of four lanes are controlled by the single clock F 0 , resulting in clock skew reduced operation.

Referring to FIGS. 6 through 8 , since the same signal is sampled onto Cs 1 and Cs 2 , there is no need to put a SHA before the sampling circuit. During the phase when Fq is high, node n 3 is connected to AC ground, so −Vin is presented at node n 5 for the sub-ADC at the falling edge of Fq. Assuming the frequency and period of F 0 is f 0 and T 0 (T 0 =1/f 0 ), the sub-ADC quantization time starting from the falling edge of Fq is T 0 . At the end of sub-ADC quantization, Vdac is available for the residue amplification coming afterwards.

Take lane L 1 as an example. At the falling edge of F 0 (slightly before F 1 ) the input is sampled onto Cs 1 , Cs 2 , and Csub. During the falling phase F 5 (L 1 Q), the voltage present at sub-ADC input is −Vin. The falling edge of F 5 , triggers the quantization operation of the sub-ADC which is shared between lanes L 1 and L 3 . At the rising edge of F 9 , the residue amplification begins at L 1 A. Other lanes can use shifted versions of clocks.

The residue amplifier 502 need not be shared between lanes and can be powered down when not in amplification phase to maintain 100% power efficiency. The sub-ADC also need not be shared between lanes. Compared to other systems where only a small fraction of clock period T 0 is used for sub-ADC quantization, this system has T 0 for sub-ADC quantization, which may relax the speed and power requirements of the sub-ADC. The quantization time can be further increased if top-plate sampling is used in sub-ADC to start the quantization after the falling edge of Fs. For the residue amplification time (2*T 0 ), the residue amplifier (RA) can be shared between two lanes. The sampling time of about (0.5*T 0 ) may not suffer from clock skew.

The idea can be generally extended to multiple-lane time-interleaved ADCs, where the sampling instants of each lane are synchronized by the same global clock. A determined time period may be allocated for sub-ADC quantization and the rest for residue amplification afterwards. Residue amplifiers and sub-ADCs may be shared between lanes.

FIG. 9 shows another exemplary timing diagram corresponding to an example skew reduced and/or SHA-less pipelined ADC of FIG. 7 . In FIG. 9 , the residue amplifier 502 can be powered down when not in amplification phase to maintain high power efficiency. Two residue amplifiers can be used with one being shared between lanes L 1 and L 3 and the other being shared between lanes L 2 and L 4 . The residue amplification clocks (F 9 -F 12 ) can be different from those in FIG. 8 as an alternative to keep residue amplifiers fully occupied. The sub-ADC can also be shared between lanes. The idea can be extended to multiple-lane time-interleaved ADCs, where the sampling instants of each lane are synchronized by the same global clock, with some time allocated for sub-ADC quantization, and the rest for residue amplification afterwards.

FIG. 10 is a block diagram of an exemplary circuit 1100 and exemplary timing diagram for a full-rate sampling with lower speed quantization channels. The circuit 1100 receives a radio frequency input signal (RF IN) which is buffered by input buffer 1102 . In this time interleaved system, RF IN can be sent to line 1 including capacitor C 1 , line one buffer 1106 and MDAC 1110 . RF IN is sent to line 1 when switch φ 1 is closed. The circuit 1100 can also include line 2 including capacitor C 2 , line two buffer 1104 and MDAC 1108 . RF IN is sent to line 2 when switch φ 2 is closed. Switch φ 0 connects between the input buffer 1102 and switches φ 1 and φ 2 to avoid or mitigate timing errors that may otherwise require complicated, digital-intensive and power-hungry lane-imbalance corrections (LICs). Such timing errors between the clocks sampling the signal in the N lanes can otherwise also create images of the desired signal, located at fs/N±fin (where fs is the interleaved sampling frequency and fin is the input signal frequency).

The switch φ 0 controlled by signal φ 0 operates at full-speed, e.g., 5 GHz. The switches φ 1 and φ 2 controlled by signals φ 1 and φ 2 interleave the samples onto the capacitors C 1 and C 2 . Because φ 0 is opened earlier than φ 1 and φ 2 , the sampling instant is determined by φ 0 . Therefore, the timing mismatch between φ 1 and φ 2 do not cause skew or timing errors. By removing timing errors, the only remaining factor that can produce unwanted images is the bandwidth (BW) mismatch between the two interleaved sampling networks. Because the BW is typically made much wider than the max signal frequency, the BW mismatch has a very minor effect, e.g., the image performance can be greatly improved compared to traditional interleaved designs. No digital or analog correction of timing mismatch is necessary, and a reduction in power, area and complexity can be obtained.

The interleaving clocks φ 1 and φ 2 can include a 25% duty-cycle, for example. When φ 1 or φ 2 is high, the sample is stored onto C 1 or C 2 respectively. When φ 0 goes low, and then φ 1 or φ 2 goes low, and the sample is ready for the quantization that follows (e.g. for an MDAC in a pipeline ADC). The figure shows the timing for an exemplary interleaved MDAC (clocks P 0 , P 2 ). While the signal is tracked (e.g. φ 1 is high), the amplifier differential outputs can be reset to the common-mode. Benefits of doing this include memory effects are removed and the amplifier power can be reduced, because the worst-case slewing requirement is reduced to half. By resetting, the amplifier holding starts from the common-mode rather than from the previous sample.

FIG. 11 is a block diagram of the exemplary quantization circuit 1100 of FIG. 10 and exemplary timing diagram for N quantization channels. The exemplary quantization circuit 1100 can be generalized to an arbitrarily large set of N quantizers. The sampler operates at full speed φ 0 , and lower-rate interleaved holding circuits timed φ 1 , φ 2 and φN interface the sampled data to each quantizer timed P 1 , P 2 and PN. The falling edge of the full-rate clock φ 0 happens earlier with respect to the falling edge of the interleaving holding blocks, e.g., φ 0 determines the sampling instant. The example is for MDAC-based quantizers, but it can be further generalized to any quantizer implementation. The holding blocks are optional and the full-rate sampling operation can be extended to the case of SHA-less interleaved quantizers.

FIG. 12 is an exemplary flowchart 1200 describing a timing for sampling signals. In an interleaved sampling system, for example the ones described above, a first switch is activated to determine a lane to sample the input signal, e.g., an RF signal ( 1202 ). A second switch is activated, e.g., based on a global clock that is same for all lanes of the interleaved system, to determine an instance that the signal is sampled ( 1204 ). When both the first switch and the second switch are activated ( 1206 ), the signal is sampled for the determined lane ( 1208 ). If either the first switch or the second switch are not activated, the signal is not sampled ( 1210 ). The sampled signal can be quantized and output as Vdac ( 1212 ). The difference between the signal and the quantized signal can be amplified and output as Vo ( 1214 ).

The methods, devices, and logic described above may be implemented in many different ways in many different combinations of hardware, software or both hardware and software. For example, all or parts of the system may include circuitry in a controller, a microprocessor, or an application specific integrated circuit (ASIC), or may be implemented with discrete logic or components, or a combination of other types of analog or digital circuitry, combined on a single integrated circuit or distributed among multiple integrated circuits. All or part of the logic described above may be implemented as instructions for execution by a processor, controller, or other processing device and may be stored in a tangible or non-transitory machine-readable or computer-readable medium such as flash memory, random access memory (RAM) or read only memory (ROM), erasable programmable read only memory (EPROM) or other machine-readable medium such as a compact disc read only memory (CDROM), or magnetic or optical disk. Thus, a product, such as a computer program product, may include a storage medium and computer readable instructions stored on the medium, which when executed in an endpoint, computer system, or other device, cause the device to perform operations according to any of the description above.

The processing capability of the system may be distributed among multiple system components, such as among multiple processors and memories, optionally including multiple distributed processing systems. Parameters, databases, and other data structures may be separately stored and managed, may be incorporated into a single memory or database, may be logically and physically organized in many different ways, and may implemented in many ways, including data structures such as linked lists, hash tables, or implicit storage mechanisms. Programs may be parts (e.g., subroutines) of a single program, separate programs, distributed across several memories and processors, or implemented in many different ways, such as in a library, such as a shared library (e.g., a dynamic link library (DLL)). The DLL, for example, may store code that performs any of the system processing described above.

While various embodiments have been described, it can be apparent that many more embodiments and implementations are possible. Accordingly, the embodiments are not to be restricted.