Table Of ContentUNIVERSITAT POLITÈCNICA DE CATALUNYA
Design of a 16-bit 50-kHz Low-Power
SC Delta-Sigma Modulator for ADC
in 0.18um CMOS Technology
MASTER THESIS
IMB-CNMSupervisors:
MicheleDei
Author:
FranciscoSerraGraells
JoseCisnerosFernàndez
UPCco-advisor:
XavierAragonèsCervera
Athesissubmittedinfulfillmentoftherequirements
forthedegreeofMasterofElectronicEngineering
inthe
ETSETB
EscolaTècnicaSuperiord’EnginyeriadeTelecomunicacióde
Barcelona
July10,2016
ii
UNIVERSITATPOLITÈCNICADECATALUNYA
ETSETB
MasterofElectronicEngineering
Designofa16-bit50-kHzLow-PowerSCDelta-SigmaModulatorfor
ADCin0.18umCMOSTechnology
JoseCisnerosFernàndez
Abstract
A general purpose 16 Bits Σ-∆ modulator ADC for double precision
audio50kHzbandwidth,targetedforLow-poweroperation,involvingno
additional digital circuit compensation, no bootstrapping techniques and
resistor-less topologies, and relaying on Switched Capacitor Σ-∆ modula-
tortopologiesforrobustoperationandinsensitivitytoprocessandtemper-
aturevariations,ispresentedinthiswork.
Designed in a commercial 180 nm technology, the whole circuit static
current is calculated in 620 µA with a nominal voltage supply of 1.8 V,
performing a Schreier FOM of 174.16 dB. This outstanding state-of-the-art
forseen FOM is achieved by the use of architectural and circuital Low-
power techniques. At the architectural level a single loop Low-distortion
topologywiththeoptimumorderandcoefficientshavebeenchosen,while
atcircuitlevelverynovelOTAbasedonVariableMirrorAmplifiersallows
anefficientClass-ABoperation.
Specially optimized switched variable mirror amplifiers with a novel
design methodology based on Bottom-up approach, allows faster design
stages ensuring feasable circuit performance at architectural level without
the need of large iterative simulations of the complete SC Σ-∆ modula-
tor. Simulationresultsconfirmsthecompleteoptimizationprocessandthe
metionedadvantageswithrespecttothetradicionalapproach.
KEY WORDS- 16 Bits, 50 kHz, Low-Power, Low-Distortion, No boostrapping,
SwitchedCapacitorΣ-∆modulator.
iii
Acknowledgements
Iowemygratitudetoallthosewhocontributedinsomewayinthere-
alizationofthismasterthesis.
Firstly, I would like to thank Dr. Francisco Serra Graells and Dr. Lluis
Teres for the oportunity to develop my master thesis in the ICAS group at
IMB-CNM.
This Thesis would not have been possible without the guidance and the
help of Dr.Michele Dei and Dr. Francisco Serra, from which I’m thankful
fortheirpatience,motivation,andknowhow. Theirguidancehelpedmein
allthetimeofdevelopmentandwritingofthisthesis.
I would like to express my sincere gratitude to my advisor Dr. Xavier
Aragones, which gives me the opportunity and the time I need to learn
fromthisworldofelectronics,withoutexpectinganythinginreturn.
IwouldliketoexpressmydeepestthankstoJoanAymerich,forbeingpart
of this journey through the micro-electronics world and provide support
and advice when needed. With no doubt its a pleasure to find people like
you.
LastbutnottheleastIwouldliketothankmyparentsandbrotherfortheir
unending support and encouragement. Without their support this work
wouldnothavebeenpossible.
iv
Contents
Abstract ii
Acknowledgements iii
1 Introduction 1
1.1 ADCFundamentals . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 ADCarchitectures . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 State-of-the-artADCs . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Σ-∆ADC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4.1 Σ-∆ModulationBasics . . . . . . . . . . . . . . . . . 7
1.4.2 CTandDTΣ-∆Modulation . . . . . . . . . . . . . . 9
1.4.3 ClassificationofΣ-∆Modulators . . . . . . . . . . . 10
1.5 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.6 Designmethodologies . . . . . . . . . . . . . . . . . . . . . . 12
1.7 StructureoftheWork . . . . . . . . . . . . . . . . . . . . . . . 16
2 Measurements 17
2.1 Σ-∆performancemetrics . . . . . . . . . . . . . . . . . . . . 17
2.2 Σ-∆ModulatorexpectedPerformance . . . . . . . . . . . . . 19
2.3 Σ-∆TestVehicle. . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.1 TestVehicleEquipment . . . . . . . . . . . . . . . . . 20
2.3.2 TestVehicleBoard . . . . . . . . . . . . . . . . . . . . 21
2.3.3 TestVehicleSoftware . . . . . . . . . . . . . . . . . . . 22
2.4 Σ-∆ModulatorMeasurements . . . . . . . . . . . . . . . . . 23
2.4.1 MeasurementProcedure . . . . . . . . . . . . . . . . . 23
2.5 MeasurementResults . . . . . . . . . . . . . . . . . . . . . . . 23
2.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3 HighlevelΣ-∆modelling 26
3.1 Σ-∆modulatortopologies . . . . . . . . . . . . . . . . . . . . 27
3.1.1 DistributedfeedbackΣ-∆modulator . . . . . . . . . 27
3.1.2 FeedforwardΣ-∆modulator . . . . . . . . . . . . . . 30
3.2 Σ-∆ModulatorTopologyElection . . . . . . . . . . . . . . . 32
3.3 CoefficientOptimization . . . . . . . . . . . . . . . . . . . . . 34
3.4 Mismatchrobustnesstest. . . . . . . . . . . . . . . . . . . . . 37
3.5 Settlingrobustnesstest . . . . . . . . . . . . . . . . . . . . . . 39
3.6 Clockjitterrobustnesstest . . . . . . . . . . . . . . . . . . . . 40
3.7 Referencenoiserobustnesstest . . . . . . . . . . . . . . . . . 41
3.8 Thermalnoise . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4 Circuitleveldesign 44
4.1 SwitchedCapacitorΣ-∆Modulator . . . . . . . . . . . . . . 44
4.2 Modulatorbuildingblocks . . . . . . . . . . . . . . . . . . . . 50
4.2.1 PhaseSplitter . . . . . . . . . . . . . . . . . . . . . . . 50
v
4.2.2 Single-bitQuantizer . . . . . . . . . . . . . . . . . . . 51
4.2.3 SwitchedVariable-MirrorAmplifiers . . . . . . . . . 52
4.2.3.1 SwitchedVariable-MirrorAmplifiersprinci-
pleofoperation . . . . . . . . . . . . . . . . 53
4.2.3.2 SwitchedVariable-MirrorTypes . . . . . . . 54
4.2.3.3 SwitchedVariable-MirrorSCimplementation 56
5 Circuitoptimizationandsimulationresults 58
5.1 SimulationTestbench . . . . . . . . . . . . . . . . . . . . . . . 58
5.2 Switches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.2.1 Switchesoptimizationprocess . . . . . . . . . . . . . 62
5.3 SwitchedVariableMirrorAmplifier . . . . . . . . . . . . . . 63
5.3.1 Designspaceandalternatives . . . . . . . . . . . . . . 63
5.3.2 Slewrateandlinearrelaxation . . . . . . . . . . . . . 65
5.3.3 SVMAType1Optimization . . . . . . . . . . . . . . . 66
5.3.4 SVMAType2Optimization . . . . . . . . . . . . . . . 68
5.3.5 2nd and3rd SVMAstages . . . . . . . . . . . . . . . . 71
5.4 SCΣ-∆Melectricalsimulations . . . . . . . . . . . . . . . . . 71
6 Conclusions 74
6.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6.2 Futurework . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
References 76
vi
List of Figures
1.1 ADC basic block diagram and Analog-to-Digital conversion
process. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 (A) 6-level quantizer characteristic and its quantization er-
ror, (B) equivalence of the quantizer block diagram with its
simplifiedlinearmodelaccountingfortheinjectionofauni-
formlydistributedwhitenoise. . . . . . . . . . . . . . . . . . 3
1.3 Murmann survey of ADC performance: representation on
theENOBversussignalbandwidthplane. . . . . . . . . . . . 5
1.4 Murmann survey of ADC performance: representation on
theenergy-per-conversionversustheENOB. . . . . . . . . . 6
1.5 In-bandnoisepowerrepresentationsinthreedifferentcases:
Nyquist-rateADC,oversampledADCandoversampledwith
noiseshapingADC. . . . . . . . . . . . . . . . . . . . . . . . . 8
1.6 BasicsingleloopΣ-∆Modulatorblockdiagram. . . . . . . . 8
1.7 Discretetime(A)andContinuoustime(B)Σ-∆modulators
blockdiagram. . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.8 Design flow diagram adopted in this work. Design tasks
havebeenmappedtotheirrespectivedesignenvironment. . 13
2.1 Typical Σ-∆ Modulator power spectral density with perfor-
mancemetrics. . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2 Graphicalrepresentationofthemostimportantperformance
metricsofΣ-∆ADC. . . . . . . . . . . . . . . . . . . . . . . . 19
2.3 TestvehicleboardoftheΣ-∆ADCof[32]. . . . . . . . . . . 21
2.4 Σ-∆ModulatorPSDmeasurementoftheΣ-∆ADCof[32]. 22
3.1 DesignflowofHighlevelmodulatorimplementation. . . . . 27
3.2 L-thorderdistributedfeedbackΣ-∆Modulator. . . . . . . . 28
3.3 2nd orderdistributedfeedbackΣ-∆Modulator. . . . . . . . . 28
3.4 Transient simulation and signal distortion at integrator out-
putsfora2nd orderdistributedfeedbackΣ-∆modulator. . . 30
3.5 L-thorderFeedforwardΣ-∆Modulator. . . . . . . . . . . . 30
3.6 2nd orderFeedforwardΣ-∆Modulator. . . . . . . . . . . . . 31
3.7 Transient simulation and signal distortion at integrator out-
putsfora2nd orderfeedforwardΣ-∆modulator.. . . . . . . 32
3.8 Block diagram of the 3rd order feedforward Σ-∆ modulator
developedinthiswork. . . . . . . . . . . . . . . . . . . . . . 34
3.9 Transientsimulationsandstatevariablesdistributionfortwo
differentcandidatesofthegridsearchalgorithm. . . . . . . . 36
3.10 High level simulation results of 3rd order Σ-∆ modulator
modelofthiswork. . . . . . . . . . . . . . . . . . . . . . . . . 37
3.11 Mismatchcoefficienttestresults. . . . . . . . . . . . . . . . . 38
3.12 SNDRdegradationasfunctionoftheintegratorsettlingerror. 39
vii
3.13 Graphical representation of a non-uniform sampling of a si-
nusoidalsignalduetoclockjitter. . . . . . . . . . . . . . . . . 40
3.14 SNDRdegradationasfunctionofJitter. . . . . . . . . . . . . 41
3.15 Referencenoisetestresults. . . . . . . . . . . . . . . . . . . . 42
3.16 High level PSD-spectrum with (Grey) and without (Black)
thermalnoise. . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.1 InterleavedoperationofswitchedOpAmpsina3rd orderΣ-
∆ Modulator. This implementation modifies the loop trans-
ferfunctioninawaythatmakethemodulatordysfunctional.
Itisshownhereforillustrationpurposes. . . . . . . . . . . . 45
4.2 3rd order Σ-∆M SC based on switched OpAmps. Here the
switched OpAmps blocks are labelled as SVMA accounting
fortheswitchedvariablemirroramplifiers. . . . . . . . . . . 46
4.3 Rearrangement of half delays in a cascade of two half delay
integrators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.4 Σ-∆MSC-switchedOpAmpschematicduringφ . . . . . . 48
1−2
4.5 Σ-∆MSC-switchedOpAmpschematicduringφ . . . . . . 49
3−4
4.6 Σ-∆Mphasegenerator. . . . . . . . . . . . . . . . . . . . . . 50
4.7 Σ-∆Moperationchronogram. . . . . . . . . . . . . . . . . . . 51
4.8 Σ-∆modulatorsinglebitquantizer. . . . . . . . . . . . . . . 52
4.9 Σ-∆modulatorsinglebitquantizerwaveformrepresentation. 52
4.10 GeneralarchitectureoftheproposedVMA[33]. . . . . . . . . 53
4.11 Qualitativelarge-signalClass-ABoperation[33]. . . . . . . . 54
4.12 Type1Class-ABcurrentamplifier[33]. . . . . . . . . . . . . . 55
4.13 Type2Class-ABcurrentamplifier[33]. . . . . . . . . . . . . . 55
4.14 SVMAschematicduringoffphase(A)andonphase(B). . . 57
5.1 Reduced Testbench schematic used in the optimization pro-
cesses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.2 Differential output voltage waveforms during the integra-
tion phase for input differential voltage ranging from 0 to
-1V. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.3 Differential output voltage (Red) and its derivative (Yellow)
representationobtainedfromFig.5.2. . . . . . . . . . . . . . . 60
5.4 HighlevelsimulationresultsfortwodesigncasesofFig.5.1. 61
5.5 SCΣ-∆modulatorfirststageschematicshowingrelevantis-
suesregardingthesamplerswitches. . . . . . . . . . . . . . . 62
5.6 SC Σ-∆ modulator second and third stage schematic illus-
tratingtheswitchedOpAmpactionontheswitchingscheme. 63
5.7 Qualitative transient response of the VMA differential out-
putvoltage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.8 Type 1 differential output voltage waveform with K = 4
AB
andI = 150µA. . . . . . . . . . . . . . . . . . . . . . . . . . 67
tail
5.9 Type 1 differential output voltage waveform with K = 8
AB
andI = 150µA. . . . . . . . . . . . . . . . . . . . . . . . . . 67
tail
5.10 Type1differentialoutputvoltagewaveformsforafixedVin
DIFF
stepof1VandI equalto150,100and50µA. . . . . . . . 68
tail
5.11 Type 2 differential output voltage waveform with K = 4
AB
andI = 150µA. . . . . . . . . . . . . . . . . . . . . . . . . . 69
tail
viii
5.12 Type2differentialoutputvoltagewaveformsforafixedVin
DIFF
stepof1VwithI of150µAandK equalto3,4and6.. 69
tail AB
5.13 Type 2 differential output voltage waveform with K = 4
AB
andI = 100µA. . . . . . . . . . . . . . . . . . . . . . . . . . 70
tail
5.14 SCΣ-∆Melectricalsimulations(Cadence),whichrepresents
7 days of computing for 16 cycles with the ideal versions of
the quantizer and the switching network, versus the maxi-
mum of 20 minutes spent in the bottom-up approach (sim-
plifiedCadencetestbenchandPythoncodesimulation). . . . 73
ix
List of Tables
1.1 Initial specifications and characteristics of the Σ-∆ Modula-
torofthisthesis. . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1 Σ-∆ModulatorSNDRsimulatedunderprocessandtemper-
aturevariations. . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2 Σ-∆ModulatorSNDRmeasurementsacrosstemperature. . 24
3.1 Σ-∆ Modulator simulation time comparison on 12 GHz/24
CPUsx86_64machine. . . . . . . . . . . . . . . . . . . . . . . 26
3.2 SNDRpeakvalueforvariousΣ-∆Modulatorconfigurations,
consideredinthiswork. . . . . . . . . . . . . . . . . . . . . . 33
3.3 ExtractedoutputsamplesoftheGridsearchalgorithm. . . . 35
3.4 Selected set of coefficients for the 3rd order Σ-∆ Modulator
ofthiswork. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.5 Gaincoefficientsgeneratedforthemismatchsensitivitytest. 38
3.6 CapacitorsizingfortheSCΣ-∆M. . . . . . . . . . . . . . . . 43
5.1 CapacitorvaluesusedinFig.5.1. . . . . . . . . . . . . . . . . 59
5.2 SVMAcandidatesperformanceforK = 6andI = 200µA. 65
AB tail
5.3 Type1exploreddesignspace. . . . . . . . . . . . . . . . . . . 68
5.4 Type2exploreddesignspace. . . . . . . . . . . . . . . . . . . 70
5.5 2nd and3rd SVMAstagesscaledown. . . . . . . . . . . . . . 71
5.6 Simulated dynamic current consumption of each SC Σ-∆M
implementedforbothSVMAtypes. . . . . . . . . . . . . . . 71
5.7 SCΣ-∆MType2simulationacrosscorners. . . . . . . . . . . 72
Description:A general purpose 16 Bits Σ-∆ modulator ADC for double precision audio 50 kHz additional digital circuit compensation, no bootstrapping techniques and .. In the case of ADCs there are two main FOMs which are the Walden .. block is described using Verilog-A to form the complete system. When.