Delta Sigma Introduction PDF

Delta-Sigma Analog-to-Digital
Converters
Oversampling Data Converters Basics
Vishal Saxena
© Vishal Saxena -1-

Oversampling and Noise-Shaping

Oversampling
fs
v
y OSR Dout
fs
2  OSR
@fs @fs/OSR
Decimation
LPF
8 10
V(f)
6 0
-10  2e=0.29136
4
-20
2
-30
0
-40
-2
-50
-4
-60
-6 -70
-8 -80
0 50 100 150 200 250 300 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
 Oversampling ratio (OSR)

 Conversion bandwidth: fB = fs/2·OSR
 SQNR = 6.02·N + 1.76 + 10·log10(OSR)
 0.5 bits increase in resolution per doubling in OSR

Oversampling
Oversampling: Time-domain waveforms
20
u[n]
10 v[n]
vf [n]
0
-10
-20
0 50 100 150 200 250 300 350 400 450 500
1
e[n]
0.5 ef [n]
-0.5
-1
0 50 100 150 200 250 300 350 400 450 500
File: oversampling1.m
© ©Vishal Saxena
Vishal Saxena -4-
Oversampling
Quantization Noise Spectrum Filtered Quantization Noise Spectrum
10 10
V(f) Vf (f)
0 0
-10  2e=0.30483 -10  2q=0.0082827
-20 -20
-30 -30
-40 -40
-50 -50
-60 -60
-70 -70
-80 -80
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
File: oversampling1.m
© ©Vishal Saxena
Vishal Saxena -5-
Oversampling with Feedback
fs
y v
u A
error
High-gain
 Can we use feedback with high loop-gain (A·kq) to reduce the error
e=|u-v| ?
 Since quantizer output can not be equal to the input
A  kq    e  u  v  
 The loop will be unstable as the error gets unbounded

Oversampling with Feedback
fs
y v
u
error fs
2  OSR
L(z)
 Use large loop-gain in the signal band and small loop-gain at higher
frequencies
 At low frequencies e  u  v  0
 At high frequencies, low loop-gain stabilizes the loop
 L(z) is the loop-filter
 This feedback arrangement is called a (noise) modulator

First-order Modulator
Σ
fs
1 y v
u
1-z-1
Δ
z-1
 Differencing (Δ) followed by an accumulator (Σ)

 ΔΣ modulator
 At low frequencies e  u  v  0

First-order Noise Shaping
e[n]
fs
1 y v
u
1-z-1
z-1
 Linearized model for the modulator
 
V ( z )  z 1U  z   1  z 1 E ( z )
STF NTF
 Noise transfer function (NTF)

 (1-z-1) : first-order differentiator
 High-pass shaping of quantization noise
 Signal transfer function (STF)
 Unit delay

First-order ΔΣ Modulator
e[n]
fs
1 y v
u
1-z-1
z-1
DSM Output Spectrum

DSM time-domain Simulation
0
8
v[n] 6
Sv(f) -20
4 -40
2
-60
0
-80
-2
-100
-4
-120
-6
-140
-8 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
0 100 200 300 400 500 600 700 800 900 1000
File: First_Order_DSM.m
First-order ΔΣ Modulator SQNR

|NTF(ejω)|
OSR
2 IBN  
0
Sv (e j ) d 
2 OSR

12 
0
NTF (e j )  d 
0 π
π 
OSR 2 OSR
  4sin 2  / 2   d 
 In-band quantization noise (IBN) 12 0
 
2 2 
  OSR 3 2 OSR

12 3
SQNR = 6.02·N –3.4 + 30·log10(OSR)

12 0
  2  d

 1.5 bits increase in resolution per doubling in
OSR  
2 3 OSR
 
 Out-of-band noise is filtered out using a 12 3 0
digital filter  2 2

36  OSR 3

Delta-Sigma (ΔΣ) ADC
ΔΣ Modulator
E(z)
vin + vDSM Decimation vout
L(z) ADC Filter
–
DAC
STF
NTF•E
|VDSM(f)| |Vout(f)|
vin vin
E
f f
fs/2•OSR fs/2 fs/2•OSR fs/2
 Use oversampling (fs=2·OSR·BW) to shape the quantization noise

out of the signal band
 Use low-resolution ADC and DAC to higher much higher resolution
 Digitally filter out the out-of band shaped (modulated) noise
 Trades-off SQNR with oversampling ratio (OSR)

Second-Order Noise Shaping
e[n]
u 1 1 y v
1-z-1 1-z-1
z-1
 Linearized model for the modulator


V ( z)  z U  z   1  z 
2
1 1
E( z)
STF NTF
 Second-order noise-shaping
2  4
 In-band quantization noise (IBN):   OSR 5
12 5
 2.5 bits increase in resolution per doubling in OSR
 Can be extended to higher orders

Second-order ΔΣ Modulator
e[n]
u 1 1 y v
1-z-1 1-z-1
z-1
8 20
v[n] 6 Sv(f) 0
-20
4
-40
2
-60
0 -80
-100
-2
-120
-4
-140
-6
-160
-8 -180
0 100 200 300 400 500 600 700 800 900 1000 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
File: Second_Order_DSM.m
Second-order ΔΣ Modulator
Second-Order KD1S Output Spectrum

1 0
0.8 NTF(z) -20 Sv(f)

0.6 -40
0.4
-60
0.2
-80
dB
0
-100
-0.2 NTF X2
-120
-0.4
zeroes SNDR = 73.8 dB
-140 ENOB = 11.97 bits
@OSR = 32
-0.6
-160
-0.8 40 dB/dec
-180
-1 -4 -3 -2 -1
-1 -0.5 0 0.5 1 10 10 10 10
Frequency
 NTF(z) = (1-z-1)2
 Two zeroes at DC
 Out-of-Band Gain (OBG) (i.e gain at ω≈π) = 4

2nd order DSM
1
Magnitude Response
0.8 4.12
3.696
0.6
3.272
0.4 2.848
2.424
0.2
Magnitude
2
0
1.576
-0.2 1.152
0.728
-0.4
0.304
-0.6
-0.12
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Normalized Frequency ( rad/sample)
-0.8
-1
-1 -0.5 0 0.5 1 8
v
6 u
-2
-4
File: Second_Order_DSM_Zero_Opt.m -6
Set variable opt=0. -8

0 50 100 150 200 250 300 350 400 450 500
© ©Vishal Saxena
Vishal Saxena -16-
2nd order DSM: contd.

0
-20
-40
-60
-80
dB
-100
-120
SNDR = 73.8 dB
-140 ENOB = 11.97 bits
@OSR = 32
-160
-180
-4 -3 -2 -1
10 10 10 10
Frequency
© ©Vishal Saxena
Vishal Saxena -17-
Comparison: 1st and 2nd order modulator waveforms
6
6
4
4
2 2
0 0
-2
-2
-4
-4
-6
-6
60 80 100 120 140 160 180 200 220 240

860 880 900 920 940 960 980
 NTF(z) = (1-z-1)  NTF(z) = (1-z-1)2

 OBG = 2  OBG = 4
 Max LSB jump = 1  Max LSB jump = 3
© ©Vishal Saxena
Vishal Saxena -18-
Higher-Order ΔΣ Modulators

Higher-Order NTFs
y v
u L(z)
L( z ) 1
V ( z)  U z  E( z)
1  L( z ) 1  L( z )
STF NTF
 Higher order noise shaping

 Reduced in-band noise, higher SQNR
2  2N
 For NTF=(1-z-1)-N, in-band noise (IBN):   OSR  (2 N 1)
12  2 N  1
 Ideally (N+1/2) bits increase in resolution per doubling in OSR

Higher-Order NTFs
Bode Diagram
50
-50
|NTF(ej )|
-100
-150
-200
0.5 1 1.5 2 2.5 3

Frequency (rad/s)
 NTF gain increases at high frequencies (around ω≈π)

 Can we go on increasing the order?

Third-order ΔΣ Modulator Example
10
u
5 v
-5
-10
0 100 200 300 400 500 600 700 800 900 1000
7
x 10
4
y
2
-2
-4
0 100 200 300 400 500 600 700 800 900 1000
 NTF(z) = (1-z-1)3
 OBG = 8, Full-scale input.
 Unstable after few samples (look at quantize input (y) blowing up!).
 Signature for ΔΣ instability
 Worst case for a single-bit quantizer. File: Third_Order_DSM.m

Third-order ΔΣ Modulator Example
DSM time-domain Simulation DSM Output Spectrum

10
u
20
5 v
0
0
-20
-5
-40
-10 -60
0 100 200 300 400 500 600 700 800 900 1000
-80
10 -100
y
5 -120
-140
0
-160
-5
-180
-10
0 100 200 300 400 500 600 700 800 900 1000 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
 Stable for 50% of full-Scale amplitude

 Signal dependent stability
 Need to develop intuition for modulator stability
 Reference: Stability theory from the Yellow Bible of delta-sigma
File: Third_Order_DSM.m
Systematic NTF Design
 NTFs of the form (1-z-1)N have stability issues
 The OBG (2N) are too high
 A larger OBG causes more wiggling at the quantizer input
 This saturates the quantizer for even smaller inputs
 Irrecoverable quantizer saturation causes loop instability
 For high-OBG the maximum stable (input) amplitude (MSA)
is small
 The stability is worse for low quantizer resolutions
 Thus we need to reduce OBG while maintaining high in-
band noise shaping

Systematic NTF Design Procedure
Pole-Zero Map
1
0.8
 Introduce poles into the NTF 0.6
0.4
1 N
(1  z ) 0.2
Imaginary Axis
 NTF ( z )  0
D( z ) -0.2
X3
-0.4
 NTF realizability criterion -0.6
-0.8
 No delay-free loops in the modulator -1

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
• First sample of the NTF impulse response (i.e. h[0])=1

Real Axis
 NTF()=1
 D(z=)=1
 Commonly used pole positions: Butterworth, Inverse

Chebyshev and maximally flat poles (maxflat)

NTF Response with Poles
NTF response OBG=8
9 10
u
8 5 v
7 0
6 -5
-10
|NTF(ej )|
5 0 50 100 150 200 250 300 350 400 450 500
4 OBG=3
10
3 u
5 v
2
0
1
-5
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 -10
/ 0 50 100 150 200 250 300 350 400 450 500
 Select appropriate OBG for the NTF to assure stability

 Trade-off between stability and increased in-band noise
 MSA vs SQNR for a given order and quantizer resolution

Systematic NTF Design Example
 Specifications
 SQNR > 120 dB
 A signal bandwidth which results in an OSR = 64
• Study optimal clock rate for the given process and
quantizer design.
 Designer’s Choice
 Order = 3
 Quantizer levels (nLev) = 16
 Butterworth high-pass response for the NTF
 Use MATLAB for finding coefficients of the HPF response.

 [b,a] = butter(order, ω3dB, ‘high’)
 The cutoff frequency ω3dB specifies the transfer function.

Systematic NTF Design contd.
1.5
H(z)
NTF(z)
Magnitude 1
0.5
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/ 
 Start with cutoff frequency ω3dB=π/8, for the butterworth HPF H(z).
 Derive a realizable NTF using NTF(z)=H(z)/b0

 Map the NTF response to a loop-filter architecture (details
later)
 Simulate the modulator for all possible amplitudes and
input tone frequencies.
 Compute the peak SQNR and MSA.
 simulateDSM function in the toolbox.
 Can use Risbo’s method shown later

140
130
120
110
100
90
80
SNR dB
70
60
50
40
30
20 peak SNR = 107.2dB
10
0
-100 -80 -60 -40 -20 0
Input Level, dB
 Peak SNR = 107 dB

 MSA = 0.9

 If SNR is not enough, repeat the entire procedure with a
higher cutoff frequency for the Butterworth HPF
 IBN ↓, SQNR ↑
 OBG ↑ and MSA ↓
 If SNR is too high, repeat the entire procedure with a lower
cutoff frequency for the Butterworth HPF
 IBN ↑, SQNR ↓
 OBG ↓ and MSA ↑

2.5 140
H(z) 130
NTF(z)
120
2 110
100
90
1.5
80
Magnitude
SNR dB
70
60
1
50
40
0.5 30
10
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 -100 -80 -60 -40 -20 0
/  Input Level, dB
 ω3dB=π/4.
 Peak SNR = 119 dB, OBG = 2.25, MSA = 0.8

140
3
H(z) 130
NTF(z) 120
2.5
110
100
2 90
80
SNR dB
Magnitude
1.5 70
60
50
1
40
30
0.5
10
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 -100 -80 -60 -40 -20 0
/  Input Level, dB
 ω3dB=2π/7.
 Peak SNR = 121 dB, OBG = 2.54, MSA = 0.8.
 Design closed !

 An advanced version of this iterative process is
implemented as the function synthesizeNTF in the delta-
sigma Toolbox.
 Several ‘opt’ params for NTF zero (and pole) optimization
 Use synthesizeChebyshevNTF for low OSR and low OBG
designs.

NTF-Zero Optimization
 Spread zeros in the signal band to minimize in-band noise
 Complex zeros on the unit circle
 8dB increase in SQNR for 3rd order modulator
 Bandwidth normalized NTF-zero locations obtained by
toolbox function ds_optzeros(order, 1)
 Already implemented in synthesizeNTF function for opt=1
NTF response
20 Pole-Zero Map
1
0.8
0
0.6
-20
0.4
0.2
Imaginary Axis
-40
|NTF(ej )|
-60
-0.2
-0.4
-80
-0.6
-100 -0.8
-1
-120 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-3 -2 -1
10 10 10 Real Axis
/

2nd order DSM: NTF Zero Optimization
1 8
v
0.8 6 u
0.6
4
0.4
2
0.2
0 0
-0.2 -2
-0.4
-4
-0.6
-6
-0.8
-1 -8
-1 -0.5 0 0.5 1 0 50 100 150 200 250 300 350 400 450 500
File: Second_Order_DSM_Zero_Opt.m
Set variable opt=1.
© ©Vishal Saxena
Vishal Saxena -36-
2nd order DSM: NTF Zero Optimization contd.

0
-20
-40
-60
-80
dB
-100
-120
SNDR = 79.3 dB
-140 ENOB = 12.89 bits
@OSR = 32
-160
-4 -3 -2 -1
10 10 10 10
Frequency
•5.5 dB increase in SQNR.

NTF pole (if any) optimization to be discussed later.
© ©Vishal Saxena
Vishal Saxena -37-
Estimating MSA (Maximum Stable Amplitude)
 MSA is found through extensive simulation

 Simulate for input sinusoids of varying amplitudes for all
possible signal frequencies in the signal band.
 For every input amplitude compute in-band SNR.
 Beyond the MSA, the NTF poles move out of the unit circle.
 Noise shaping is disrupted and the in-band SNR drops.
 At this point the quantizer input (y[n]) blows up.
 simulateSNR function in the toolbox does exactly the
same
 Time consuming and often impractical for iterative design

Estimating MSA using Risbo’s Method
FS y v
L
u
0
t
Slow ramp input
 Use a slow ramp input from 0 to FS value.

 Plot log10|y[n]|. Observe where this plot blows up.
 Take 90% of the input amplitude where log10|y[n]| blows up as a
conservative estimate for MSA.
 Estimated MSA is close to that predicted by the sinewave input
method.
 Much quicker than the sinewave technique (simulateSNR
function)

Estimating MSA using Risbo’s Method
MSA Estimation by Simulation

14
12
10
6
MSA = 0.843
log |y|
-2
-4
-6
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Input, u
File: MSA_Risbo_Method.m

Simulation with input with MSA

20
u
10 v
-10
-20
0 100 200 300 400 500 600 700 800
Loop-filter States
20
x1
10 x2
y=x3
0
-10
-20
0 100 200 300 400 500 600 700 800
© ©Vishal Saxena
Vishal Saxena -41-
Simulated SNR with input with MSA
Modulator Output Spectrum

0
-50
-100
dBFS
-150 SNDR = 119.3 dB

ENOB = 19.52 bits
@OSR = 64
-200
-250
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/ 
© ©Vishal Saxena
Vishal Saxena -42-
Simulation with input with 1.2*MSA

20
u
10 v
-10
-20
0 100 200 300 400 500 600 700 800
Loop-filter States
50
x1
x2
y=x3
0
-50
0 100 200 300 400 500 600 700 800
© ©Vishal Saxena
Vishal Saxena -43-
Simulation with input with 1.2*MSA

20
u
10 v
-10
-20
0 100 200 300 400 500 600 700 800
6
x 10 Loop-filter States
2
x1
1 x2
y=x3
0
-1
-2
0 100 200 300 400 500 600 700 800
© ©Vishal Saxena
Vishal Saxena -44-
Bode Sensitivity Integral
C1=0.083472, C2=0.083569
0.3
0.2
0.1
0
log|H|
-0.1
-0.2
-0.3
-0.4
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/ 
Single pole/zero transfer function with pole/zero inside the unit circle.
Area above and below the 0-dB axis are equal.
File: BodeSensitivity1.m
© ©Vishal Saxena
Vishal Saxena -45-
C1=3.27, C2=3.367
20
-20
-40
10 log|NTF|
-60
-80
-100
-120
-140
-160
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/ 
Butterworth NTF.
© ©Vishal Saxena
Vishal Saxena -46-
C1=3.949, C2=4.0776
20
10
-10
10 log|H|
-20
-30
-40
-50
-60
-70
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/ 
Inverse Chebyshev NTF.

© ©Vishal Saxena
Vishal Saxena -47-
20
10
-10
10*log|H|
-20
-30
-40
-50
-60
-70
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/ 
Better in-band performance results in worse out-of-band performance.
© ©Vishal Saxena
Vishal Saxena -48-
20
10
-10
10*log|H|
-20
-30
-40
-50
-60
-70
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/ 
Complex NTF zeros result in better in-band performance for the same OBG.
© ©Vishal Saxena
Vishal Saxena -49-
20
10
-10
10*log|H|
-20
-30
-40
-50
-60
-70
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/ 
Higher-order NTF results in better in-band performance for the same OBG.
© ©Vishal Saxena
Vishal Saxena -50-
Loop Filter Architectures

Loop-Filter Architectures
 Several loop-filter discrete-time architectures possible

 Toolbox function realizeNTF maps the synthesized NTF to
loop-filter co-efficients
[a,g,b,c] = realizeNTF(H, form);
 Dynamic Range Scaling (DRS) performed to scale loop-

filter states to a bounded value
 Scaling performed using ABCD matrix representation of the loop-
filter
 See any introductory text on Linear Systems
ABCD = stuffABCD(a,g,b,c,form);
[ABCDs umax] = scaleABCD(ABCD, nLev, f0, xLim);
[a,g,b,c] = mapABCD(ABCDs, form);

CIFB (Cascade of Integrators with Distributed Feedback)
u(n)
b1 b2 b3 b4
-g1
z-1 x1(n) z-1 x2(n) z-1 x3(n) y(n) v(n)

c1 c2 c2
1-z-1 1-z-1 1-z-1
-a1 -a2 -a3 -a4
DAC
 Cascade of delaying integrators:

 Feedback coefficients a’s realize the zeros of L1 and thus the
NTF and STF poles.
 Feed-in coefficients b’s determine zeros of L0 and thus the STF
zeros.
 State scaling coefficients c’s are used for dynamic range
scaling.

CRFB (Cascade of Resonators with Distributed Feedback)
u(n)
b1 b2 b3 b4
-g1
z-1 x1(n) 1 x2(n+1) z-1 x3(n) y(n) v(n)

c1 c2 c2
1-z-1 1-z -1
1-z-1
-a1 -a2 -a3 -a4
DAC
 Combine a non-delaying and a delaying integrator with

local feedback around them, to form a stable resonator
 Local feedback coefficients g’s realize the complex zeros in the
NTF.
 Implements NTF with complex zeros.
 For odd-order, use an integrator in the front to avoid
noise coupling due to g

CIFF (Cascade of Integrators with Feed-Forward Summation)
u(n)
b1 b2 b3 b4
-g1
z-1 x1(n) z-1 x2(n) z-1 y(n) v(n)

c2 c3 a3
1-z-1 1-z-1 1-z-1 x3(n)
a2
-c1
a1
DAC
 Feedforward summation of states

 For b1=bN+1=1 and b2 to bN=0, STF=1
 Loop-filter only processes quantization noise, low power and
distortion
 Feedforward loop-filters typically result in lower-power
implementation

CRFF (Cascade of Resonators with Feed-Forward Summation)
u(n)
-b1 -b2 -b3 -b4

-g1
z-1 x1(n) 1 x2(n) z-1 x3(n) y(n) v(n)

c1 c2 c2
1-z-1 1-z-1 1-z-1
-a1 -a2 -a3 -a4
DAC
 Use resonators with feedforward summation

 Implements NTF with complex zeros
 For odd-order, use an integrator in the front to avoid
noise coupling due to g

CIFB Example 1
 CIFB, order = 4
 All NTF zeros at z=1, i.e. opt =0.
 OBG = 3, OSR = 16, nLev = 15. NTF(z) STF(z)
 Only single input coupling is used 1 1
 b(2:end) = 0 0.5 0.5

 Maxflat poles in STF
0 0
 a = [0.16 0.86 1.9 2.1]
-0.5 -0.5
 b = [0.16 0 0 0]
 c = [1 1 1 1] -1 -1
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
 g = [0 0]
L0(z) L1(z)
1 1
0.5 0.5
0 0
-0.5 -0.5
-1 -1
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
File: CIFB_4th_Order_1.m
© ©Vishal Saxena
Vishal Saxena -57-
CIFB Example 1 contd. : NTF and STF
3
50
NTF
NTF
STF
STF
2.5
0
-50
Mag, dB
Mag
1.5
-100
1
0.5 -150
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 -200
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/
/
© ©Vishal Saxena
Vishal Saxena -58-
States
Transient Simulation Integrator States

10 20
u x1
8 v x2
15
x3
6
x4
10
4
2 5
0 0
-2
-5
-4
-10
-6
-8 -15
-10 -20
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000
samples, n samples, n
© ©Vishal Saxena
Vishal Saxena -59-
Spectrum

0
-20
-40
-60
-80
dBFS
-100
-120
-140
SNDR = 80.7 dB
-160 ENOB = 13.11 bits
@OSR = 16
-180
-200
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/ 
© ©Vishal Saxena
Vishal Saxena -60-
Topologies
 CRFB with single feed-in
 CRFB_4th_Order_1.m
 Low-distortion CRFB topology
 CRFB_4th_Order_2.m
 CIFB with single feed-in and optimized NTF zeros
 CIFB_Opt_4th_Order_1.m
 Low-distortion CIFB topology with optimized NTF zeros
 CIFB_Opt_4th_Order_2.m
© ©Vishal Saxena
Vishal Saxena -61-
CIFF Example 1
 CIFF, order = 4
NTF(z) STF(z)
 OBG = 3, OSR = 16, nLev = 15. 1 1
 Low-distortion topology 0.5 0.5
 b(1) =b(5)= 1
 b(2:4)=0 0 0
 a = [2.1 1.9 0.86 0.16] -0.5 -0.5
 b = [1 0 0 0 1] -1 -1
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
 c = [1 1 1 1]
 g = [0 0] L0(z) L1(z)
1 1
0.5 0.5
0 0
-0.5 -0.5
-1 -1
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
© ©Vishal Saxena
Vishal Saxena -62-
CIFF Example 1 contd. : NTF and STF
3 50
NTF NTF
STF STF
2.5
0
2
-50
Mag, dB
Mag
1.5
-100
1
-150
0.5
0 -200
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/ /
File: CIFF_4th_Order_1.m
© ©Vishal Saxena
Vishal Saxena -63-
States

10 6
u x1
8 v x2
4 x3
6
x4
4
2
2
0 0
-2
-4 -2
-6
-4
-8
-10
0 100 200 300 400 500 600 700 800 900 1000 -6
0 100 200 300 400 500 600 700 800 900 1000
samples, n
samples, n
© ©Vishal Saxena
Vishal Saxena -64-
Spectrum

0
-20
-40
-60
-80
dBFS
-100
-120
-140
SNDR = 80.4 dB
-160 ENOB = 13.06 bits
@OSR = 16
-180
-200
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/ 
© ©Vishal Saxena
Vishal Saxena -65-
CIFF Example 2
 CIFF, order = 4
 OBG = 3, OSR = 16, nLev = 15. NTF(z) STF(z)
1 1
 Only single input feed-in used
 b(2:end)=0 0.5 0.5
 a = [2.1 1.9 0.86 0.16] 0 0
 b = [1 0 0 0 0]
-0.5 -0.5
 c = [1 1 1 1]
 g = [0 0] -1 -1
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
L0(z) L1(z)
1 1
0.5 0.5
0 0
-0.5 -0.5
-1 -1
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
© ©Vishal Saxena
Vishal Saxena -66-
CIFF Example 2 contd. : NTF and STF
3.5 50
NTF NTF
STF STF
3
0
2.5
-50
2
Mag, dB
Mag
1.5
-100
-150
0.5
0 -200
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/ /
 Notice the significant STF peaking !
© ©Vishal Saxena
Vishal Saxena -67-
States

10 50
u x1
8 v 40 x2
x3
6 30
x4
4 20
2 10
0 0
-2 -10
-4 -20
-6 -30
-8 -40
-10 -50
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000
samples, n samples, n
 Last integrator output has significant signal content

 Use dynamic range scaling.
 Last integrator will burn more power in this case.
© ©Vishal Saxena
Vishal Saxena -68-
Spectrum

0
-20
-40
-60
-80
dBFS
-100
-120
-140
SNDR = 82.1 dB
-160 ENOB = 13.34 bits
@OSR = 16
-180
-200
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
/ 
© ©Vishal Saxena
Vishal Saxena -69-
Topologies
 Low-distortion CRFF topology
 CRFF_4th_Order_1.m
 CRFF with single feed-in
 CRFF_4th_Order_2.m
 Low-distortion CIFF topology with optimized NTF zeros
 CIFF_Opt_4th_Order_1.m
 CIFF with single feed-in and optimized NTF zeros
 CIFF_Opt_4th_Order_2.m
 STF peaking in FF topologies with single feed-in is an issue
 CT FF DSM will have STF peaking as full-feedforward branch can’t be
used.
 The feed-in coefficients b’s can be strategically used to realize CIFF/CRFB
topology with better out-of-band STF attenuation.
© ©Vishal Saxena
Vishal Saxena -70-
ΔΣ Modulator Architectures
 Cascade/ MASH architecture:
•Eg. Two first order modulators are used to implement second order
modulator.
•Stability concerns are relaxed but mismatch in the two forward paths
should be properly monitored.

ΔΣ Modulator Architectures
 Feedforward modulators
•Most popular architecture.

•Input signal is summed at Nth stage integrator output.
•Summation block may be required at higher order modulators.
•Multibit quantizer is necessary.

Key Terminologies :
 SQNR – Signal to quantization noise ratio
 Thermal/electrical noise are not included.
 SNR - Signal to noise ratio
 Distortion is not included.
 SDNR - Signal to noise and distortion ratio
 All noise sources are included.
 ENOB – Effective number of bits (resolution)
 This is very important than actual number of output bits
 Dynamic Range (DR)
 Measured with input of the modulator shorted.
 Harmonic Distortion
 THD is usually total harmonic distortion. Or Third??
 Spur Free Dynamic Range (SFDR)
 Very key parameter in communication systems

Frequency Domain Measurements
Peak SNR
Peak SNDR
Dynamic range

Spurious (tone) Free Dynamic Range (SFDR)
SFDR

References
Data Conversion Fundamentals
A.1 M. Gustavsson, J. Wikner, N. Tan, CMOS Data Converters for
Communications, Kluwer Academic Publishers, 2000.
A.2 B. Razavi, Principles of Data Conversion System Design, Wiley-IEEE Press, 1994.
A.3 ADC and DAC Glossary by Maxim.
A.4 B. Murmann, "ADC Performance Survey 1997-2009," [Online].
A.5 S. Pavan, N. Krishnapura, EE658: Data Conversion Circuits Course at
IIT Madras [Online].
A.6 The Fundamentals of FFT-Based Signal Analysis and Measurement, T.I. App Note here.

References
Delta-Sigma Data Converters
B.1 R. Schreier, G. C. Temes, Understanding Delta-Sigma Data Converters,
Wiley-IEEE Press, 2005 (the Green Bible of Delta-Sigma Converters).
B.2 S. R. Norsworthy, R. Schreier, G. C. Temes, Delta-Sigma Data Converters:
Theory, Design, and Simulation, Wiley-IEEE Press, 1996 (the Yellow Bible of
Delta-Sigma Converters).
B.3 S. Pavan, N. Krishnapura, “Oversampling Analog to Digital Converters
Tutorial,” 21st International Conference on VLSI Design, Hyderabad, Jan,
2008.
B.4 S. H. Ardalan, J. J. Paulos, “An Analysis of Nonlinear behavior in Delta-Sigma
Modulators,” IEEE TCAS, vol. 34, no. 6, June 1987.
B.5 R. Schreier, “An Empirical Study of Higher-Order Single-Bit Delta-Sigma
Modulators,” IEEE TCAS-II, vol. 40, no. 8, pp. 461-466, Aug. 1993.
B.6 J. G. Kenney and L. R. Carley, “Design of multibit noise-shaping data
converters,” Analog Integrated Circuits Signal Processing Journal, vol. 3, pp.
259-272, 1993.
B.7 L. Risbo, “Delta-Sigma Modulators: Stability Analysis and Optimization,”
Doctoral Dissertation, Technical University of Denmark, 1994 [Online].
B.8 R. Schreier, J. Silva, J. Steensgaard, G. C. Temes,“Design-Oriented Estimation
of Thermal Noise in Switched-Capacitor Circuits,” IEEE TCAS-I, vol. 52, no.
11, pp. 2358-2368, Nov. 2005.

References
CAD for Mixed-Signal Design
D.1 K. Kundert, “Principles of Top-Down Mixed-Signal Design,” Designer’s Guide
Community [Online].
D.2 R. Schreier, Matlab Delta-Sigma Toolbox, 2009 [Online], [Manual], [One page
summary].
D.3 Jose de la Rosa, “SIMSIDES Toolbox: An Interactive Tool for the Behavioral
Simulation of Discrete-and Continuous-time SD Modulators in the MATLAB,”
University of Sevilla, Spain, [Contact the authors for the software].
D.4 P. Malcovati, Simulink Delta-Sigma Toolbox 2, 2009. Available [Online].
Example Datasheets
E.1 A 16-bit, 2.5MHz/5 MHz/10 MHz, 30 MSPS to 160 MSPS Dual Continuous Time
Sigma-Delta ADC – AD9262, Analog Devices, 2008.
E.2 A 24-bit, 192 kHz Multi-bit Audio ADC – CS5340, Cirrus Logic, 2008.

Delta Sigma Introduction PDF

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Delta Sigma Introduction PDF

Загружено:

Авторское право:

Доступные форматы

Delta-Sigma Analog-to-Digital

© Vishal Saxena -1-

© Vishal Saxena -2-

 Oversampling ratio (OSR)

© Vishal Saxena -3-

-10  2e=0.30483 -10  2q=0.0082827

© Vishal Saxena -6-

© Vishal Saxena -7-

 Differencing (Δ) followed by an accumulator (Σ)

© Vishal Saxena -8-

 Linearized model for the modulator

 Noise transfer function (NTF)

© Vishal Saxena -9-

DSM Output Spectrum

© Vishal Saxena -11-

 Use oversampling (fs=2·OSR·BW) to shape the quantization noise

© Vishal Saxena -12-

 Linearized model for the modulator

© Vishal Saxena -13-

Second-Order KD1S Output Spectrum

0.8 NTF(z) -20 Sv(f)

© Vishal Saxena -15-

Set variable opt=0. -8

Second-Order KD1S Output Spectrum

DSM time-domain Simulation

60 80 100 120 140 160 180 200 220 240

 NTF(z) = (1-z-1)  NTF(z) = (1-z-1)2

© Vishal Saxena -19-

 Higher order noise shaping

© Vishal Saxena -20-

 NTF gain increases at high frequencies (around ω≈π)

© Vishal Saxena -21-

© Vishal Saxena -22-

DSM time-domain Simulation DSM Output Spectrum

 Stable for 50% of full-Scale amplitude

© Vishal Saxena -24-

 Introduce poles into the NTF 0.6

 NTF realizability criterion -0.6

 No delay-free loops in the modulator -1

• First sample of the NTF impulse response (i.e. h[0])=1

 Commonly used pole positions: Butterworth, Inverse

© Vishal Saxena -25-

5 0 50 100 150 200 250 300 350 400 450 500

 Select appropriate OBG for the NTF to assure stability

© Vishal Saxena -26-

 Use MATLAB for finding coefficients of the HPF response.

© Vishal Saxena -27-

© Vishal Saxena -28-

© Vishal Saxena -29-

 Peak SNR = 107 dB

© Vishal Saxena -30-

© Vishal Saxena -31-

© Vishal Saxena -32-

© Vishal Saxena -33-

© Vishal Saxena -34-

© Vishal Saxena -35-

Second-Order KD1S Output Spectrum

•5.5 dB increase in SQNR.

 MSA is found through extensive simulation

© Vishal Saxena -38-

 Use a slow ramp input from 0 to FS value.

© Vishal Saxena -39-

MSA Estimation by Simulation