Sampling Time Error Calibration for Time-Interleaved ADCs Nandish Mehta s i s e h T e c n e i c S f o r e t s a M EEMCS Sampling Time Error Calibration for Time-Interleaved ADCs MASTER OF SCIENCE THESIS For the degree of Master of Science in Microelectronics at Delft University of Technology Nandish Mehta August 29,2013 FacultyofElectricalEngineering, MathematicsandComputerScience DelftUniversityof · Technology Copyright ©ElectricalEngineering Allrightsreserved. DELFT UNIVERSITY OF TECHNOLOGY DEPARTMENT OF ELECTRICAL ENGINEERING Theundersignedhereby certify thattheyhaveread and recommendto theFaculty of Electrical Engineering,Mathematicsand ComputerScience foracceptance athesisentitled SAMPLING TIME ERROR CALIBRATION FOR TIME-INTERLEAVED ADCS by NANDISH MEHTA in partialfulfillmentoftherequirementsforthedegreeof MASTER OF SCIENCE MICROELECTRONICS Dated: August29,2013 Supervisor(s): dr. Frank vanderGoes dr. KlaasBult prof. dr. KofiA. A. Makinwa Reader(s): dr.ir.MichielPertijs dr. ing. LeodeVreede Abstract Inthisthesisthedesignofthetimingerrorcalibration loopforatime-interleaved ADCisdescribed. The realization of fully-digital radio transceiver requires a wideband capture ADC that can si- multaneously capture all the commercial wireless bands present in a mobile handset. These ADCs are expected to operate at GHz sampling speed with good energy efficiency. Both of these con- tradicting requirements can be fulfilled by employing time-interleaved architecture. Unfortunately, time-interleaved ADCs suffer from interleaving issues like mismatch in sampling time error. These issuescanbeaddressed bydesigning adedicated calibration loop. In this thesis an attempt is made to design a calibration loop that detects and corrects the sam- pling timeerrors withhigh precision. Thetiming error detection technique relies onintroducing two additional reference ADCs. The correction of timing errors is done using the least mean square it- erative algorithm (LMS). Convergence and stability of such calibration loops are extremely critical. Hence, they are exhaustively investigated in this thesis. Factors that hamper the loop convergence were identified and relevant solutions are applied to overcome them. Furthermore, it was found that the loading effect of the reference ADCs greatly affects the accuracy of the timing error detection. A simple solution using delay lines is shown to remove this effect. Finally, techniques like inserting dummysamplingcircuits,scalingsamplingcapacitance, andmatchingtheclock-paths, areemployed toachievetimingerrorcorrection accuracyintheorderof5fslevel. Some of these techniques are implemented at the architecture level, whereas some are imple- mented at circuit level. The effectiveness of the architecture level techniques is verified through MATLAB modeling while the circuit level techniques are verified through circuit simulations. The sub-blocksforthecalibrationlooparedesignedinindustrial28nmCMOSprocessandrelevantsimu- lationresultsarepresented. Circuitslike11-bit10µWDACwith0.6LSBDNL,atrack-and-hold with HD of 72dB at 1GHz input frequency, and clock-path with a mean delay of 11ps, are designed for 3 thetimingerrorcalibration loop. Keywords: LMS calibration loop, Timing error detection, Time-interleaving, Observer Effect, Referencelanes,Wideband captureADC,Digital-to-Analog converter, Track-and-hold, Low-power. MasterofScienceThesis NandishMehta ii NandishMehta MasterofScienceThesis Table of Contents 1 Introduction 1 1-1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1-2 BasicsofTime-Interleaved ADC . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1-3 Application: Wideband CaptureADC . . . . . . . . . . . . . . . . . . . . . . . . . 5 1-4 TargetTimingErrorCorrection Accuracy . . . . . . . . . . . . . . . . . . . . . . . 7 1-5 ResearchGoalandContributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1-6 ThesisOrganization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2 SamplingTimeErrors in Time-InterleavedA/DConverters 11 2-1 BasicsofTime-Interleaving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2-2 TypesofInterleaving Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2-2-1 OffsetMismatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2-2-2 GainMismatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2-2-3 TimingMismatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2-2-3-1 ImpactofTimingError . . . . . . . . . . . . . . . . . . . . . . . 17 2-2-3-2 SourcesofTimingMismatch . . . . . . . . . . . . . . . . . . . . 18 2-2-4 BandwidthMismatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2-3 TimingErrorDetectionandCalibration . . . . . . . . . . . . . . . . . . . . . . . . 20 2-3-1 Useofcommonsample-and-hold. . . . . . . . . . . . . . . . . . . . . . . . 21 2-3-2 Foreground vsbackground calibration. . . . . . . . . . . . . . . . . . . . . 21 2-3-3 Digitaldetection anddigitalcorrection oftimingerrors. . . . . . . . . . . . 22 2-3-4 Digitaldetection andanalogcorrection oftimingerrors. . . . . . . . . . . . 23 2-4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 MasterofScienceThesis NandishMehta iv TableofContents 3 Timing ErrorCalibrationLoop 25 3-1 PrincipleofOperation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3-2 Description ofCalibration Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3-3 Convergence oftheCalibration Loop . . . . . . . . . . . . . . . . . . . . . . . . . . 29 3-3-1 InputSignalStatistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 3-3-2 Gainmismatchandoffsetofthetworeference lanes . . . . . . . . . . . . . 30 3-3-3 LimitonSpeedofConvergence . . . . . . . . . . . . . . . . . . . . . . . . 32 3-4 StabilityofCalibration Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3-5 ImpactofFiniteQuantization ofthetwoReferenceLanes . . . . . . . . . . . . . . . 34 3-6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 4 Sampling-TimeError Dueto ObserverEffectofReferencelanes 37 4-1 ObserverEffect: SimpleRCModelAnalysis . . . . . . . . . . . . . . . . . . . . . . 38 4-2 ObserverEffect: SamplingInstanceInteractions . . . . . . . . . . . . . . . . . . . . 40 4-3 Isolating theSamplingInteractions usingaWireDelay . . . . . . . . . . . . . . . . 42 4-4 MismatchbetweenDummylaneandREFlane . . . . . . . . . . . . . . . . . . . . 44 4-5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 5 Circuit Implementation 51 5-1 System-LevelDesign . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 5-2 ClockPath . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 5-2-1 Clock-Phase generator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 5-2-2 SamplingEdgeTuningCircuit . . . . . . . . . . . . . . . . . . . . . . . . . 53 5-3 Track-and-hold Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 5-4 HOLDBufferDesign . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 5-5 Digital-to-Analog Converter(DAC) . . . . . . . . . . . . . . . . . . . . . . . . . . 61 5-5-1 EstimatingDynamicRange . . . . . . . . . . . . . . . . . . . . . . . . . . 64 5-5-2 CircuitImplementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 5-5-3 SimulationofDNL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 5-5-4 DesignSummary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 5-6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 6 Conclusion 71 6-1 ProblemDefinition: ARecap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 6-2 Thesiscontribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 6-3 FutureWork . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 A TargetSpecifications 75 B MATLABCodeto SimulateBasicInterleavingIssues 77 C Simulation Setupfor TimingErrors 81 C-1 CoherentFFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 C-2 SettingUpVariables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 NandishMehta MasterofScienceThesis
Description: