Maximizing GSPS ADC SFDR Performance: Sources of Spurs and Methods of Mitigation

Similar documents
Analog-to-Digital Converter Survey & Analysis. Bob Walden. (310) Update: July 16,1999

APPLICATION NOTE 3942 Optimize the Buffer Amplifier/ADC Connection

Digital Time-Interleaved ADC Mismatch Error Correction Embedded into High-Performance Digitizers

Noise Power Ratio for the GSPS

A 10 bit, 1.8 GS/s Time Interleaved Pipeline ADC

AD9772A - Functional Block Diagram

ADC07D1520. Low Power, 7-Bit, Dual 1.5 GSPS or Single 3.0 GSPS A/D Converter. General Description. Features. Key Specifications.

CLC Bit, 52 MSPS A/D Converter

Architectures and circuits for timeinterleaved. Sandeep Gupta Teranetics, Santa Clara, CA

ADC and DAC Standards Update

MSP430 Teaching Materials

DATASHEET HI5805. Features. Applications. Ordering Information. Pinout. 12-Bit, 5MSPS A/D Converter. FN3984 Rev 7.00 Page 1 of 12.

Digital Waveform Recorders

12-bit 50/100/125 MSPS 1-channel ADC

ADX216. ADC Interleaving IP-Core

2.5GS/s Pipelined ADC with Background. Linearity Correction

RTH GHz Bandwidth High Linearity Track-and-Hold REV-DATE PA FILE DS_0162PA2-3215

Analog Input Performance of VPX3-530

AD Bit, 20/40/65 MSPS 3 V Low Power A/D Converter. Preliminary Technical Data

Analog-to-Digital i Converters

IP Specification. 12-Bit 125 MSPS Duel ADC in SMIC40L IPS_S40L_ADC12X2_125M FEATURES APPLICATIONS GENERAL DESCRIPTION. Single Supply 1.

Outline. Noise and Distortion. Noise basics Component and system noise Distortion INF4420. Jørgen Andreas Michaelsen Spring / 45 2 / 45

Data Converters. Specifications for Data Converters. Overview. Testing and characterization. Conditions of operation

Lecture 9, ANIK. Data converters 1

SECTION 4 HIGH SPEED SAMPLING AND HIGH SPEED ADCs, Walt Kester

10-Bit, 40 MSPS/60 MSPS A/D Converter AD9050 REV. B. Figure 1. Typical Connections FUNCTIONAL BLOCK DIAGRAM

Low-IMD Two-Tone Signal Generation for ADC Testing

ISSCC 2004 / SESSION 25 / HIGH-RESOLUTION NYQUIST ADCs / 25.4

Discrete Fourier Transform

DATASHEET HI5767. Features. Applications. Pinout. 10-Bit, 20/40/60MSPS A/D Converter with Internal Voltage Reference. FN4319 Rev 6.

Tiny, 2.1mm x 1.6mm, 3Msps, Low-Power, Serial 12-Bit ADC

System on a Chip. Prof. Dr. Michael Kraft

A 130mW 100MS/s Pipelined ADC with 69dB SNDR Enabled by Digital Harmonic Distortion Correction. Andrea Panigada, Ian Galton

LTC2216/LTC Bit, 80Msps/65Msps Low Noise ADC APPLICATIONS TYPICAL APPLICATION

SPT BIT, 30 MSPS, TTL, A/D CONVERTER

A 12 bit 125 MHz ADC USING DIRECT INTERPOLATION

9-Bit, 30 MSPS ADC AD9049 REV. 0. Figure 1. Typical Connections FUNCTIONAL BLOCK DIAGRAM

14-Bit, 80/105/125/150 MSPS, 1.8 V Dual Analog-to-Digital Converter AD9640

DC-Coupled, Fully-Differential Amplifier Reference Design

ADC1006S055/ General description. 2. Features. 3. Applications. Single 10 bits ADC, up to 55 MHz or 70 MHz

High Speed ADC Analog Input Interface Considerations by the Applications Engineering Group Analog Devices, Inc.

Enhancing Analog Signal Generation by Digital Channel Using Pulse-Width Modulation

AD Bit, 80 MSPS/105 MSPS/125 MSPS, 1.8 V Dual Analog-to-Digital Converter (ADC) FUNCTIONAL BLOCK DIAGRAM FEATURES APPLICATIONS

ADC12D2000RF. 12-Bit, Single 4.0 GSPS RF Sampling ADC. ADC12D2000RF 12-Bit 4.0 GSPS RF Sampling ADC. 1.0 General Description. 3.

ADQ108. Datasheet. Features. Introduction. Applications. Software support. ADQ Development Kit. Ordering information

FUNCTIONAL BLOCK DIAGRAM DIGITAL VIDEO ENGINE

Analog and Telecommunication Electronics

10-Bit, 80Msps, Single 3.0V, Low-Power ADC with Internal Reference

Summary Last Lecture

A 12-bit Interpolated Pipeline ADC using Body Voltage Controlled Amplifier

High Speed System Applications

CMOS Analog to Digital Converters : State-of-the-Art and Perspectives in Digital Communications ADC

Nonlinear Equalization Processor IC for Wideband Receivers and

A 4 GSample/s 8-bit ADC in. Ken Poulton, Robert Neff, Art Muto, Wei Liu, Andrew Burstein*, Mehrdad Heshami* Agilent Laboratories Palo Alto, California

ADC12D1800RF. 12-Bit, Single 3.6 GSPS RF Sampling ADC. ADC12D1800RF 12-Bit 3.6 GSPS RF Sampling ADC. 1.0 General Description. 3.

Reconfigurable 6 GHz Vector Signal Transceiver with I/Q Interface

8-/4-/2-Channel, 14-Bit, Simultaneous-Sampling ADCs with ±10V, ±5V, and 0 to +5V Analog Input Ranges

ADQ214. Datasheet. Features. Introduction. Applications. Software support. ADQ Development Kit. Ordering information

MCP MCP37D10-200

ADC1206S040/055/ General description. 2. Features. 3. Applications. Single 12 bits ADC, up to 40 MHz, 55 MHz or 70 MHz

8-Channel, 10-Bit, 65MSPS Analog-to-Digital Converter

781/ /

Pipeline vs. Sigma Delta ADC for Communications Applications

ADC Bit 65 MSPS 3V A/D Converter

2.4 A/D Converter Survey Linearity

Digital Signal Analysis

Reference Clock Distribution for a 325MHz IF Sampling System with over 30MHz Bandwidth, 64dB SNR and 80dB SFDR

Figure 1. Functional Block Diagram

Workshop ESSCIRC. Low-Power Data Acquisition System For Very Small Signals At Low Frequencies With12-Bit- SAR-ADC. 17. September 2010.

ERROR CORRECTION TECHNIQUES IN HIGH-SPEED A/D AND D/A CONVERTERS

Eliminate Pipeline Headaches with New 12-Bit 3Msps SAR ADC by Dave Thomas and William C. Rempfer

AD9861. Mixed-Signal Front-End (MxFE ) Baseband Transceiver for Broadband Applications. Data Sheet FUNCTIONAL BLOCK DIAGRAM FEATURES

Acquisition Time: Refer to Figure 1 when comparing SAR, Pipeline, and Delta-Sigma converter acquisition time. Signal Noise. Data Out Pipeline ADC

14-Bit, 40/65 MSPS A/D Converter AD9244

9240LP LPTVREF. Memory DESCRIPTION: FEATURES: 14-Bit, 10 MSPS Monolithic A/D Converter with LPT ASIC. 9240LP Block Diagram 9240LP

Low-Power Pipelined ADC Design for Wireless LANs

ADC12D1800. ADC12D Bit, Single 3.6 GSPS Ultra High-Speed ADC. Literature Number: SNAS500L

8 GHz Bandwidth Low Noise 1 GS/s Dual Track-and-Hold

A 42 fj 8-bit 1.0-GS/s folding and interpolating ADC with 1 GHz signal bandwidth

Testing A/D Converters A Practical Approach

Receiver Architecture

Analog and RF circuit techniques in nanometer CMOS

AD Bit, 80 MSPS/105 MSPS/125 MSPS/150 MSPS, 1.8 V Dual Analog-to-Digital Converter FUNCTIONAL BLOCK DIAGRAM FEATURES

FUNCTIONAL BLOCK DIAGRAM REFERENCE AD9255 ADC 14-BIT CORE SERIAL PORT SDIO/ DCS SENSE RBIAS PDWN AGND AVDD (1.8V) LVDS LVDS_RS SVDD SCLK/ DFS

AD Bit, 125 MSPS/105 MSPS/80 MSPS, 1.8 V Analog-to-Digital Converter PRODUCT HIGHLIGHTS FEATURES APPLICATIONS FUNCTIONAL BLOCK DIAGRAM

15-Bit, 65Msps ADC with -78.2dBFS Noise Floor for IF Applications

The need for Data Converters

Lecture #6: Analog-to-Digital Converter

Modern communication and measurement system designs are

Summary Last Lecture

Mixed-Signal-Electronics

Complete 12-Bit 5 MSPS Monolithic A/D Converter AD871

SBAS303C DECEMBER 2003 REVISED MARCH 2004 SPECIFIED TEMPERATURE RANGE

EVALUATION KIT AVAILABLE 65Msps, 12-Bit ADC PART

Sigma-Delta ADC Tutorial and Latest Development in 90 nm CMOS for SoC

CDK bit, 250 MSPS A/D Converter with Demuxed Outputs

Nevis ADC Tes+ng. Tim Andeen. Columbia University. LAr ADC Review. June 4, LAr ADC Review. Tim Andeen

A Low Jitter Analog Circuit for Precisely Correcting Timing Skews in Time-Interleaved Analog-to-Digital Converters

Two- Path Band- Pass Σ- Δ Modulator with 40- MHz IF 72- db DR at 1- MHz Bandwidth Consuming 16 mw

Another way to implement a folding ADC

Transcription:

Maximizing GSPS ADC SFDR Performance: Sources of Spurs and Methods of Mitigation Marjorie Plisch Applications Engineer, Signal Path Solutions November 2012 1

Outline Overview of the issue Sources of spurs Methods of mitigation Summary and recommendations 2

AN OVERVIEW OF THE ISSUE 3

Problem statement How to minimize spurs in the GSPS ADC family in order to maximize SFDR performance? 4

ADC Sampling Rate and Architecture The absolute rate of what low or high is, changes as process technology advances. Higher sampling rate architectures imply lower resolutions. GSPS ADC Certain techniques may be applied to the basic flash architecture in order to improve its performance. 5

Ultra high-speed architecture design (2) Flash ADC Implementation Basic Flash Architecture Can achieve high sampling rates with low conversion latency Basic design requires 2 N comparators and latches Drawbacks are high power consumption, die area What techniques can make a 12-bit 3.6 GSPS ADC practically realizable? Folding and interpolating to improve power consumption, reduce area Folding-Interpolating Architecture Diagrams are from Analog Integrated Circuit Design by Johns and Martin, 1997; Circuit Techniques for Low-voltage and High-speed A/D Converters by Waltari and Halonen, 2002. 6

GSPS ADC Functional Blocks Block Description Pros Cons Track-and-hold Folding Interpolating Interleaving Calibrating Track and hold analog input signal Fold transfer function into sub-ranges Interpolate conversion between series of amplifiers Time-interleave multiple ADC cores Trim bias currents in linear amplifiers Improve performance at low Fin Reduce number of latches to improve power, area Reduce number of amplifiers to improve power, area Achieve higher sampling rates Reduces distortion Can reduce max sampling rate Introduces distortion Introduces distortion Introduce distortion from mismatch factors Time off-line to calibrate For more details on the GSPS ADC architecture, see A 1.8V 1.0Gsps 10b Self-Calibrating Unified-Folding-Interpolating ADC with 9.1 ENOB at Nyquist Frequency by R. Taft, et al. ISSCC 2009 / Session 4 / High-speed Data Converters. 7

SOURCES OF SPURS 8

Harmonic Distortion in an Amplifier What are the harmonic distortion terms for an amplifier? First, let us consider the non-linear system: 2 3 4 5 v ( t) a v ( t) a v ( t) a v ( t) a v ( t) a v ( t)... o 1 in 2 in 3 in 4 in 5 in For differential circuits, the even harmonics are ideally zero and H 3 >>H 5, so we can approximate: v o 3 ( t) a1vin( t) a3vin( t) For a sinusoidal input: v o ( t) a [ a 1 1 Acos( t) a 3 3a A A 4 3 v in cos 3 a A ]cos( t) [ 4 ( t) Acos( t) ( t) ]cos(3 t ) HD HD 1 3 If we define: t) HD cos( t) HD 3 A 3 3 3 v o ( 1 3 t cos(3 ) Example: For a classic non-linear amplifier, if the input level, A, is decreased by 1dB, then HD 3 decreases by 3dB since HD 1 is proportional to A and HD 3 is proportional to A 3. 3 HD a3a 1 a1 A and HD3 9 4

Harmonic Distortion due to Non-Linearity 3. Harmonic distortion contribution from all sources sum at ADC output Distortion location is easy to predict, but amplitude is not Source of Distortion Harmonics Produced Rolls off with Relationship to input power Track-and-hold Lower order harmonics Analog input bandwidth Classic relationship Amplifiers in interpolating, folding architecture Higher order harmonics Folding-interpolating factor Non-linear relationship 10

E.g.: Harmonic Distortion ADC12D800RF (Folding-interpolating) ADS5400 (Pipeline) Fundamental Blue = original data Red = data with harmonics removed Fundamental H 2 H 3 Noise Floor 11

E.g.: ADC12D1600RF Harmonic Levels H 3 is generally the highest level harmonic Lower index harmonics from the track-and-hold roll off with input bandwidth Higher index harmonics from the foldinginterpolating architecture remain present and have highly non-linear level A reduction in the input level is not strongly related to the harmonics level 12

ADC Interleaving Basics Multiple ADC cores sample signal to increase total sampling rate ADC cores sample at same divided frequency but different phase offset Digital outputs are re-aligned in time Input buffer typically drives cores Signal Input Clock Input freq. = Fs Clock Input 0º Clock 90º Clock 180º Clock BUFFER Clock Phase Generator 0º 90º 180º 270º ADC4 ADC3 freq. = Fs/4 ADC2 ADC1 Time Alignment 270º Clock 13

Non-Ideal Interleaving Offset Errors Mismatched ADC core voltage offset Amplitude Errors ADC core gain error ADC reference voltage error Signal Input BUFFER V OFFSET G ERR ADC2 ADC1 Phase Errors Input routing delay Clock Input freq. = Fs Clock Phase Generator 0º 180º freq. = Fs/2 ERR Input BW difference Clock phase error ADC sampling instant 14

Non-Ideal Interleaving Offset Error Different voltage offset at ADC input between different cores Alternating up/down in transient waveform Creates signal independent spurs in spectrum at Fs*n/N for n=1,2,,n-1 where N is # of interleaved cores Example N=2 Power Input Signal Offset spurs F IN F S /4 F S /2 Freq Example N=4 15

Non-Ideal Interleaving Amplitude and Phase errors Gain difference between different cores BW differences or transmission length differences result in phase difference Creates N-1 input signal dependent images from 0 to Fs/2 in a repetitive, mirror-image pattern where N is # of interleaved cores Also creates harmonic distortion images Input Signal Power H2 Input Images H2 Images F IN F S /4 Example N=4 F S /2 Freq 16

E.g.: ADC12D1800RF DES Mode Interleaving Spurs 4x Interleaving Spurs Fs/4-Fin Fs/2-Fin Fs/4+Fin 17

Interleaving by Product Product Number of subconverters in Non-DES Mode Number of subconverters in DES Mode Non-DES Mode Offset Spur Locations DES Mode Offset Spur Locations ADC10D1x00 2 4 DC, Fs/2 DC, Fs/4, Fs/2 ADC12D1x00 2 4 DC, Fs/2 DC, Fs/4, Fs/2 ADC12Dx00RF 1 2 DC DC, Fs/2 ADC12Dxx00RF 2 4 DC, Fs/2 DC, Fs/4, Fs/2 Dual-channel mode is Non-DES Mode ; interleaved mode is DES Mode. The ADC12D800/500RF offer the possibility of one sub-converter per bank in Non-DES Mode. 18

E.g.: ADC12D1000RF Harmonic Distortion Interleaved Fundamental System Clock Blue = original data Red = data with harmonics removed Fs/2 - Fin Note the symmetrical appearance of the spurs around Fs/4 due to the 2x interleaving FFT Details: ADC12D1800RF Non-DES Mode 2x interleaving NDM DDR clock Ain = 124.47MHz at -1dBFS 19

IMD 3 Concept amplitude IMD 2f 1 -f 2 f 1 f 2 2f 2 -f 1 frequency IMD3 3 f min( f1, f2) max( f2 f1, f1 2) Although there are 4 3 rd order inter-modulation distortion products, only the 2 near-in ones are typically considered for an ADC: 2f 2 -f 1 and 2f 1 -f 2. The amplitude of the fundamentals, f 1 and f 2, should be set: At, but not above -7dBFS As close to each other as possible, preferably <0.1dB The difference between f 1 and f 2 in the diagram is exaggerated to show which is chosen to measure IMD 3. 20

E.g. ADC12D1800RF IMD 3-40 -45 0 500 1000 1500 2000 2500 3000 3500 4000-50 IMD3 [dbc] -55-60 -65-70 IMD3-7dBFS IMD3-10dBFS IMD3-13dBFS IMD3-16dBFS IMD3-19dBFS -75-80 -85 Fin [MHz] 21

DCLK Spur Product NDM SDR DCLK (MHz) NDM DDR DCLK (MHz) Demux SDR DCLK (MHz) Demux DDR DCLK (MHz) DES Mode Spur Locations Non-DES Mode Spur Locations ADC10D1500 N/A 500 N/A 250 N/A Fs/4 N/A Fs/8 N/A Fs/2 N/A Fs/4 ADC12D1800 N/A 900 N/A 450 N/A Fs/4 N/A Fs/8 N/A Fs/2 N/A Fs/4 ADC12D800RF 800 400 400 N/A Fs/2 Fs/4 Fs/4 N/A Fs Fs/2 Fs/2 N/A ADC12D1800RF N/A 900 N/A 450 N/A Fs/4 N/A Fs/8 N/A Fs/2 N/A Fs/4 Depending upon the mode, the Data Clock (DCLK) can couple back into the analog circuitry to appear in the output. 22

Sub-converter Clock Spur Product Fclk (MHz) Subconverters / Channel Subconverter Clock (MHz) DES Mode Spur Location Non-DES Mode Spur Location ADC1xDxxxx Fclk N Fclk / N Fs / (2*N) Fs / N ADC12D1800RF 1800 2 900 Fs / 4 Fs / 2 ADC12D500RF 500 1 500 Fs / 2 Fs = DC ADC12D1800RF. DES Mode. No Fin Fs/4 = -75dBFS 23

Spur Source Summary and Conclusions Spur sources Non-linearities from the track-and-hold Non-linearities from the folding-interpolating architecture Interleaving images from gain mismatch and timing skew Fixed frequency spurs: DCLK, offset mismatch in interleaving, subconverter clock Conclusions Testing with single tone inputs is the traditional method for assessing the non-linear performance of an ADC, but It may not be adequate or relevant for most real world applications The next section illustrates how some non-linearities discussed in this section are either mitigated or reduced based on input signal type 24

METHODS OF MITIGATION 25

Calibration (1) Calibration is the primary way to address spurs due to nonlinearities in the conversion process (2) A foreground calibration addresses the following: (1) trim the analog input resistance (2) trim amplifier bias currents Minimize full-scale error, offset error, DNL and INL, which results in the maximum dynamic performance 26

Before-and-After Calibration Without Calibration ENOB = 6.62 With Calibration ENOB = 8.85 ADC12D1800RF Fin = 997.47MHz Non-DESI Mode 27

Dithering to Improve Harmonics Fs/2-Fin Without Dither Fin Adding dither or bandlimited noise improves harmonic performance E.g. ADC12D1800RF Ain=-13dBFS Fs/2-Fin With Dither Fin Harmonic No Dither (dbc) With Dither (dbc) Improvement (db) H2-65 -70 5 H3-64 -74 10 H4-77 -77 0 H5-70 -75 5 H6-67 -77 10 H7-66 -78 12 28

Wideband Input Signals behave like dither ADC12D1800RF WCDMA at Fc = 847MHz In an application with wideband input signals, each signal will act as dither on the others and improve its harmonics Noise Power Ratio (NPR) can be used to measure how quiet one unused channel in a wideband system remains in the presence occupied channels RMS Noise Level [db] NPR NPR 10*log10 P P Ni No Frequency f s /2 29

Frequency Planning Frequency planning can be used to make sure lower order harmonics do not interfere with the desired signal. The ADC Harmonic Calculator can be found at the link below. Please note that this tool does not yet include the effects of interleaved harmonics. http://www.ti.com/lsds/ti/analog/dataconverters/tools.page 30

Analog Correction of Interleaving Spurs Timing Spur Magnitude (dbfs) 0-10 -20-30 -40-50 -60-70 -80-90 Timing Spur H3 0 20 40 60 80 100 120 140 DES Timing Adjust Code Example: Using the DES Timing Adjust on the ADC12D1800RF to adjust the level of the interleaving spur at Fs/2-Fin. At its relative minimum, the power in the spur is due to gain mismatch between the I/Q-channel GSPS ADC Feature DES Mode Spur Spur Source Addressed I/Q-ch Offset Adjust Fs/2 I/Q-ch offset mismatch I/Q-ch FSR Adjust Fs/2 - Fin I/Q-ch gain mismatch Duty Cycle Correct Fs/2 - Fin I/Q-ch timing skew DES Timing Adjust Fs/2 - Fin I/Q-ch timing skew 31

Digital Correction of Interleaving Spurs Interleave correction reduces spectrum offset spurs and images Dithering does not affect the level of interleaving spurs Correction in analog /digital domain For resolution > 8 bits, achieving the level of matching required in the analog domain is extremely difficult Digital correction: estimate the errors and correct the data with coefficients Estimation Detection in time-domain or frequency domain Convergence Calibration time Foreground: Calibration interrupts normal operation Background: Calibration runs continuously 32

Reducing Fixed Frequency Spurs Fixed frequency spurs occur from the DCLK, offset mismatch, and subconverter clock Example solutions include DCLK Mode choice, system architecture choices, and ADC selection Example Spur Source Solution A spectrum analyzer uses the ADC12D800RF interleaved and cannot tolerate a strong spur in the middle of the spectrum A wideband communications application uses the ADC12D1800RF interleaved to achieve high sampling bandwidth A long-range tactical radar with Fs = 750Msps cannot tolerate interleave images DCLK running in Demux SDR Mode causes a spur at Fs/4 DCLK produces spur at Fs/8 or Fs/4 Offset mismatch spur at Fs/4 Sub-converter clock spur at Fs/4 Most members of the GSPS ADC family have interleaved channels, which produce image spurs Choose instead the Non- Demux SDR DCLK, which moves the spur to Fs/2 Adjust the sampling clock so that the fixedfrequency spurs land on channel boundaries Use the ADC12D800RF, which has only 1 converter per channel 33

SUMMARY AND RECOMMENDATIONS 34

Summary and conclusions In order to achieve high-resolution, high-sampling rate ADCs, certain techniques were chosen which also generate spurious content. Spurs in the GSPS ADC family come from non-linearities, interleaving, and system clocks. Techniques to address these spurs include ADC features such as calibration, dithering, and frequency planning. Input signals that consist of multiple wideband signals or single tones act like dithering and reduce the impact of nonlinearities. 35

Solutions Recommendation Spur Dominant Source Solution Lower order harmonics Higher order harmonics IMD 3 DC, Fs/4, Fs/2 Fs/2 Fin, Fs/4 ± Fin Non-linearity in track-andhold Folding-interpolating architecture Non-linearity in track-andhold, folding-interpolating architecture Sub-converter offset mismatch in interleaving architecture Sub-converter gain mismatch and timing skew in interleaving architecture Calibration Frequency planning Calibration Dithering ADC selection ADC features Digital correction Fs/8, Fs/4, Fs/2 Coupling from DCLK DCLK selection DC, Fs/2 Coupling from subconverter clock ADC selection 36

Questions? Thank you for attending! Any questions? 37