Practical Dynamic Element

Matching Techniques for

3-level Unit Elements
KHIEM NGUYEN
Analog Devices Inc., Wilmington MA
 Outline
• Brief background
• Motivations and Applications
• 3-level vs. 2-level unit element: advantages and challenges
• What problems DEM does and does not solve
• Practical dynamic element matching (DEM) techniques
• 2-level DEM review
• 3-level DEM
• Limitations of each technique
• Solution spaces
• Higher-order DEM
• Issues in 2nd-order DEM
• Solutions
• Application specific concept of DEM

• Q and A
Applications of 3--level unit
elements
In cabin ANC
Safety radar

Active Noise
Cancelling (ANC)
 What is Dynamic Element Matching?

•A technique to achieve perfect matching of

analog elements over time.
 Euclid's 5th definition for plane geometry

“Through a pair of points in space, there exists

one and only one straight line”
 What it means in the converter world ….
Analog out

• Can achieve “perfect

linearity” if the 1

circuit output has -1 Digital In

only 2 states
Year 1995 …..

The “Perfect DAC” !

 Why multi-bit ?
• Drawback of single-bit
• High out-of-band noise Æ Requires stringent post-analog filter Æ costly,
bulky
• Limited usable input range with high-order noise shapers Æ Challenging to
improve dynamic range
• Stringent bandwidth and slew rate requirements for reconstruction amplifier
Æ Challenging to reduce power consumption
• Multi-bit
• Solves all the above but introduces linearity issue due to element mismatch
Æ Mismatch shaping is the remedy
• Each element is 2-level
• Why 3-level element ?
 Current steering DAC cell comparison
I/2 I/2 I/2
+
OUT {I,0,-I} Digital input n z p
- +
Digital input
0 0 OUT {I,0-I}
-
b1 -1 b0 b0 b1 b1 p n p z n
b0 1 z = p NOR n
I/2 I/2 I/2

a) 2-level element architecture b) 3-level element architecture

• 3-level element properties

+ Power efficient
+ Best noise performance: No noise contribution in the “zero” state.
- Needs a new dynamic element matching logic
- Needs a low power solution for ISI problem
- Drain voltage modulation degrades THD
 Example 1:: High performance 6' audio DAC

8 cells 4 cells 4 cells

R2 Process 0.18Pm CMOS

C1 Supply 1.8 V
pi zi ni pi pi zi ni HOLD Full scale differential output 0.9 Vrms
zi ni
S1
+ - Digital power consumption 0.4 mW / channel

-
HOLD S3 OUT Analog power consumption 0.7 mW / channel
S2 + OSR 64
+ -
C2 Clock frequency 3.072 MHz
Vref ni zi pi ni zi pi ni zi pi HOLD
SNR (A-weighted) 108 dB
R3 THD+N -97 dB

16x DAC 4x DAC 1x DAC

 Example 2:: High performance 6' audio ADC
C5
R1
S1
In+ C1 C3
R7 S3
+ - + - Process 0.18Pm CMOS
-
Flash Supply 3.3 V
A3 + cm A1 A2 ADC
- R8 Digital power 0.4 mW / channel
- + - + consumption
S4
Analog power 2 mW / channel
In-
consumption
R2 S2
C2 C6 C4
OSR 128
16 Clock frequency 6.144 MHz
SNR (A-weighted) 111 dB
pos[7:0] Top 8 bits
- 4 bit THD+N -97 dB
3-level
CT-DAC 3-level
+ DEM
neg[7:0] Bottom 8 bits

Therm-to-3-level
encoder
 3-level element in 6' ADC
• Additional advantages:
• Minimum noise contribution from current steering DAC at low level signal
(below -60dB full scale)
• Equivalent to a current noise from a large resistor R,  = /
• I-DAC current noise is not the dominant source anymore
• Dominant noise sources : input R and amplifier
• For the same SNR target : Larger allowable R Æ Smaller integration capacitor size Æ
Silicon area saving
• Less switching activity at low level Æ Less spectral leakage due to ISI
• Cautions
• Signal dependent switching activity Æ A source of harmonic distortions
• Slight noise modulation due to signal amplitude
 DEM: What it does and does not solve
• Solves
• Cell-to-cell static mismatch (born or due to aging)
• Between the +1s’ and the -1s’
• 1/f noise of cells

• Does not solve

• Magnitude error between “+1” and “-1” of a single cell
• Circuit design technique exists as remedy for this issue
• Cell-to-cell mismatch due to switching dynamics
 Recap
• Brief background
• Motivations and Applications
• 3-level vs. 2-level unit element: advantages and challenges
• What problems DEM does and does not solve
• Practical dynamic element matching (DEM) techniques
• 2-level DEM review
• 3-level DEM
• Limitations of each technique
• Solution spaces
• Higher-order DEM
• Issues in 2nd-order DEM
• Solutions
• Application specific concepts of DEM

• Q and A
 Review: Noise shaping concept
• High precision multi-level analog quantity is “difficult” to reconstruct

• Noise shaping: Trade off frequency band usage for word-width

• 16b @ 44.1kHz Æ 1b at 5.6488MHz

Mag Mag

Shaped quantization
noise (out-of-band)

48k freq 2.8224MHz

freq
 1-bit versus multi-bit DAC

„ Multi-bit advantages
‹ 1-bit Advantages
† Tone-free quantization noise
z Linear – no matching issue (can be dithered)
‹ 1-bit problems † Lower-order loops can often be
z Large step size Æ jitter sensitivity used (easier stability)
z Tonal quantization noise can cause † Small steps (low jitter
䇾idle tones䇿 (quantizer can䇻t be sensitivity, less filtering
dithered properly) required)
z High-order loops become unstable „ Multi-bit problems
with large inputs † Matching; DAC element errors
cause distortion + noise
 Multi-bit with element mismatch
Out
Mag

In

Harmonics

freq
 Example of multi-bit audio DAC

3-level 3-level
data shuffler 1x DAC
24 bit @ 6.144MHz 3

Extended- ISI-free
IN Modulator
8 3 3-level 3-level I-to-V Analog
noise-shaped data shuffler
M1 4x DAC output Output
24b splitter
stage
128Fs
4
3-level 3-level
data shuffler 16x DAC

8b @ 6.144MHz Analog
Digital
 The essence of DEM
• Linearizing the transfer function
• Making each element appears as average value Æ Reconstructed values are
“perfectly” linear in the band of interest

• Requirements
• Use thermometer coded data (uniformly weighted elements)

• Must have multiple ways (redundancy) to reconstruct any value

• Have an “out-of-band” region to push shaped mismatch errors to. i.e., oversampling
ratio larger than 1 (practical OSR >= 16)
 Multi-bit transfer function
All elements are used
Æ no mismatch error
Out

Average value

Actual elements

-1 1 In

All elements are used

Æ no mismatch error
 Spectral contents with and without DEM

Amplitude
No DEM

Frequency

Random Selection DEM

OUR GOAL
Noise shaped DEM
 Approaches to 2-level unit element DEM
1. Data directed
2. Rotational
3. Tree structure
4. Vector quantization
5. Real-time
1. Data directed shuffling Actual output
2 + eA + e B
• Consider a 2-bit thermometer code DAC
• Cell A = 1 + e1 (10)
• Cell B = 1 + e2 e A- e B
Input code
00 11
eB - eA
AB Out Instantaneous (01)
error (ei)
00 -(2+e1+e2) 0
-2 - eA - eB
01 -e1 + e2 -e1+e2
Data A OUT0
10 e1 – e2 e1-e2
Data B OUT1
11 2+e1+e2 0
Swap/No-Swap

REG
• Goal: Make the cumulative error approach
zero. Data A
• Done by alternating use of OUT1 and OUT2 DataB

Swapper cell
How to make a multi-bit shuffler

From Thermom Decode

To Unit-element DAC
• Butterfly network works as “usage” correlator
• Every input can reach every output
• Decision to swap is made by swapper cell
 Spectral view of mismatch shaping
Spectrum of a single
thermometer-encoded
bit: Input + harmonics
+ shaped mismatch
+

......
++

If addition is perfect, the

original modulator spectrum is
recovered!
 2. Rotational DEM
• Example: Digital in = 1, element selection in 4 clock cycles

1 + e1 Pointer(2) 1 + e1 1 + e1 1 + e1

1 + e2 1 + e2 1 + e2 1 + e2
Pointer(3)

1 + e3 1 + e3 1 + e3 1 + e3
Pointer(4)
Pointer(1) 1 + e4 1 + e4 1 + e4 1 + e4

E(1) = -e1 + e2 + e3 +e4 E(2) = e1 - e2 + e3 +e4 E(3) = e1 + e2 - e3 +e4 E(4) = e1 + e2 + e3 - e4

‹ Cumulativeerror after 4 cycles = 2(e1 + e2 + e3 + e4)

‹ (e1+e2+e3+e4) : all elements are used equally Æ cumulative error = zero!
 3. Tree structure DEM
‹ DEM algorithm
1b z Start with N-bit binary word
3b 2b
z Divide by 2 ( shift left by 1)
BINARY
4b z If LSB = 1, use 6' modulator to decide which
sub-word to receive the LSB
MSB z Update modulator

‹ Observation
z Difference
between 2 words is a noise shaped
sequence Æ Gain mismatch is high-passed
LSB ‹ Note:
z Needs headroom to not overflow
z Works fine for first–order shaping, what
happens if input is EVEN for a while!
 4. Vector quantization
• Generalized version of all DEM algorithms Æ Excellent model to study DEMs
• Hardware intensive

uN-1

uN-1
DAC input D  {0,..,N} V=
.
.
min() .
u2
u0
sort () D > 0, V = top D elements ON
u0 ui  {0,1}
element selector
-1

delay

Example of a 1st-order DEM (using the error feedback 6' structure)

 Element selection logic
• Rank N input vectors with least used first
• Turn on (assert “1”) D output elements in output vector V
• Feedback the elements of V to the corresponding integrators (or
loops)
• Can apply the use of a dithering vector to the input of the sorting
routine to cure idle tone behavior
 5. Real-time DEM

• Use all elements in every modulator clock cycle

• Advantage:
• Very simple logic
• Drawbacks:
• Need a fast clock (2N times) for an N-bit DAC
• Cannot go beyond 1st-order shaping
 Recap
• Brief background
• Motivations and Applications
• 3-level vs. 2-level unit element: advantages and challenges
• What problems DEM does and does not solve
• Practical dynamic element matching (DEM) techniques
• 2-level DEM review
• 3-level DEM
• Limitations of each technique
• Solution spaces
• Higher-order DEM
• Issues in 2nd-order DEM
• Solutions
• Application specific concepts of DEM

• Q and A
 Approaches to 3-level unit-element DEM
1. Data directed shuffling
2. Rotational
3. Tree structure
4. Vector quantization
 Encoding style for 3-level data
• Best to treat data as positive and negative thermometer codes
separately
• Results in simplest logic manipulation and decoding logic in the
current steering cells
• Truth table
Positive Negative Analog value
thermometer thermometer
code code
1 0 +1
0 1 -1
0 0 0
1 1 Unused
 Approach #1: Data directed shuffling
IN_P[7] INA_P OUTA_P INA_P OUTA_P INA_P OUTA_P OUT_P[7]
IN_N[7] INA_N OUTA_N INA_N OUTA_N INA_N OUTA_N OUT_N[7]
Binary to thermometer encoder

IN_P[6] INB_P OUTB_P INB_P OUTB_P INB_P OUTB_P OUT_P[6]

IN_N[6] INB_N OUTB_N INB_N OUTB_N INB_N OUTB_N OUT_N[6]
IN_P[5] INA_P OUTA_P INA_P OUTA_P INA_P OUTA_P OUT_P[5]
IN_N[5] INA_N OUTA_N INA_N OUTA_N INA_N OUTA_N OUT_N[5]
IN_P[4] INB_P OUTB_P INB_P OUTB_P INB_P OUTB_P OUT_P[4]
IN_N[4] INB_N OUTB_N INB_N OUTB_N INB_N OUTB_N OUT_N[4]
4
IN A 4-bit example
IN IN_P[3] INA_P OUTA_P INA_P OUTA_P INA_P OUTA_P OUT_P[3]
IN_N[3] INA_N OUTA_N INA_N OUTA_N INA_N OUTA_N OUT_N[3]
IN_P[2] INB_P OUTB_P INB_P OUTB_P INB_P OUTB_P OUT_P[2]
IN_N[2] INB_N OUTB_N INB_N OUTB_N INB_N OUTB_N OUT_N[2]
IN_P[1] INA_P OUTA_P INA_P OUTA_P INA_P OUTA_P OUT_P[1]
IN_N[1] INA_N OUTA_N INA_N OUTA_N INA_N OUTA_N OUT_N[1]
IN_P[0] INB_P OUTB_P INB_P OUTB_P INB_P OUTB_P OUT_P[0]
IN_N[0] INB_N OUTB_N INB_N OUTB_N INB_N OUTB_N
OUT_N[0]

ƒ Butterfly network serves as the usage correlator

N §N·
# of cells log¨ ¸ , for N  1 output levels
4 ©2¹
 3-level mismatch error analysis
• eA and eB: individual error of
element A and B
Actual output
2 + eA + e B • Error delivered: ­( eA  eB ) / 2
1 + eA Ideal - Actual  °®( eB  eA ) / 2
ideal = 1 + (eA + eB)/2 ° 0
1 + eB ¯
Input
-2 -1 1 2
• Goal: Select elements so that
-1 - eB the long term average of each
-1 - (eA + eB)/2 = ideal output level approaches its
-1 - eA ideal value
-2 - eA - eB

Transfer function of a DAC formed by a • A 2b state machine to keep

pair of analog elements A and B track of cumulative error
 Cell implementation
INA, INB, OUTA, OUTB
are 2b quantities
Value Code (PN)
INA_P OUTA_P
INA OUTA
INA_N OUTA_N
+1 10
-1 01
INB_P OUTB_P
INB OUTB_N OUTB
INB_N
0 00
unused 11

Ƈ Output value = OUTA + OUTB, ^ 2,..0..,2`

Ƈ Simple modification of thermometer code
ƒ A and B always have the same sign
Ƈ Input can be rerouted to either output
Ƈ Needs a state machine
 State diagram
• Cumulative error:
INA = 0, INB = 0 or “00” : zero
INA = 1, INB = 1 or
INA = -1, INB = -1 “01” : ( eB  e A ) / 2
00
“10” : ( e A  eB ) / 2
INB = 1 or INA = 1 or
INA = -1 INB = -1
• Apply the DEM rules to
INB = -1 or reroute data and keep
INA = 1 INA = -1 or
INB = 1 01
cumulative error at zero
10

INA = 0, INB = 0 or INA = 0, INB = 0 or • The shuffler cell operates

INA = 1, INB = 1 or INA = 1, INB = 1 or similarly to the
INA = -1, INB = -1 INA = -1, INB = -1
generalized DEM model
 Idle tones in first-order data-direct shuffling
• Reasons:
• Each shuffling cell is a 1st-order noise shaper (can create large inband tones)
• With gain mismatch between cells, these result in idle tones
• At low level (below -60dBFS), input has a tendency to travel through only a
certain path due to the construct of the butterfly network
• At low level input, some cells in the network become “starve” in activity
• Can be detected by sweeping the input amplitude with small DC range
• Remedy
• Precede the shuffler network with a randomizer
• Ensure that each cell will have equal amount of data activity over time
• Mismatch shaping effectiveness is slightly degraded but still can perform very
well with high OSR (>=64)
Measured result ADAU1361 at -0.5dbFS
+0

-10

-20

-30

-40

-50

-60

d
B -70
dBFS

r
-80
A

-90

-100 THD + N = -97dB

-110

-120

-130

-140

-150
2k 4k 6k 8k 10k 12k 14k 16k 18k 20k
Hz
Freq in Hz
Measured result ADAU1361 at -60dbFS
+0

-10

-20

-30

-40

-50

-60

d
B -70
dBFS

r
-80
A

-90

-100
Dynamic range (A-weighted) = 108dB
-110

-120

-130

-140

-150
2k 4k 6k 8k 10k 12k 14k 16k 18k 20k
Hz
Freq in Hz
 Approach #2a: Dual-pointer Rotational DEM
• Use 2 separate pointers: one for positive and one for negative
thermometer data
• Use the same encoding scheme previously shown

Positive data Negative data

positive pointer negative pointer

 Rules for rotational DEM
• Rules of DEM
• For positive data, only positive
pointer moves forward. Vice
versa for negative data
• Both pointers move in the
same direction
• Pointers wrap around in a
Pointer advancement Pointer wraps around circular manner
direction • Observations
• If a pointer wraps arounds,
the respective cumulative
positive pointer negative pointer error is equal zero
• If both pointers are at the
same location, the cumulative
error is equal to zero
 Example
negative out positive out

positive pointer
positive pointer

negative pointer
positive pointer

negative pointer positive pointer negative pointer negative pointer

data = 0 data = 1 data = 2
data = -2
Cycle 0 Cycle 1 Cycle 2
Cycle 4

positive pointer
positive pointer
negative pointer
negative pointer

positive pointer

negative pointer

data =-3 data = -4

data =3

Cycle 5 Cycle 7
Cycle 6
 Implementation of dual pointer DEM
Barrel
. shifter .
2N-1 positive
. .
thermometer inputs . .

shift

Signed
binary data positive pointer
DEM decision logic Signed shuffled
(dual counters) thermometer outputs

CLK
negative pointer

shift
. .
. .
2N-1 negative
thermometer inputs
. Barrel .
shifter
 Implementation of dual pointer DEM with
sign-magnitude data
• Area efficient single barrel
implementation
. Barrel
shifter
. • Only 1 pointer is active and
2N-1
thermometer inputs
.
.
.
.
Signed shuffled
thermometer outputs used at a time
shift
Signed
• Barrel shift control is
positive pointer
binary data DEM decision
logic
dependent on sign
CLK
negative pointer
• Barrel shifter only takes in
magnitude
• Sign bit is used as a decoding
 Idle tones in dual pointer DEM
• Reasons:
• Similar to those found in 2-level single pointer DEM
• Static born-mismatch pattern shows up at low input level
• Remedy
• Observe that when both pointers are at the same location, the cumulative
error is zero
• Both pointers can be repositioned to a randomly chosen location
• Static mismatch error pattern will be broken up
• A slight degradation to the SNR due to the introduction of the randomness
into the first-order single-bit noise-shaped sequences
• Mismatch shaping effectiveness is still very good with high OSR (>=64)
 Example: With idle tone remedy
negative out positive out

positive pointer
positive pointer

negative pointer
positive pointer

negative pointer positive pointer negative pointer negative pointer

data = 0 data = 1 data = 2
data = -2
Cycle 0 Cycle 1 Cycle 2
Cycle 4

positive pointer
positive pointer
negative pointer
negative pointer When positive and
new negative pointer new positive pointer negative pointers
positive pointer are equal,
randomize the
old negative pointer old positive pointer position of both
negative pointer
pointers.
data =-3 data = -4 data =3
data =3
Cycle 7 Cycle 8
Cycle 5 Cycle 6
 Measured result ADAU1966 at -0.5dbFS
dBFS

THD+N = -100dB

Freq in Hz
 Measured result ADAU1966 at -0.5dbFS
dBFS

Dynamic range (A-weighted) = 120dB

THD
THD + N = -97dB
-9
97dB

Freq in Hz
Approach #2b: Single-pointer rotational DEM
• One pointer that moves forward when data is
positive and backward when data is negative

Positive data
• Possible remedy for idle tone
• Set up a reference point
Bi-directional pointer • Keep track of wrapping around the reference point
Negative data • After wrapping around (single or multiple times)
and ending at the reference point, reposition
pointer to a new randomly chosen reference point.
• Repeat the entire procedure
 Approach #3: Tree structure DEM
3b 2b
BINARY
• Use the tree structure to split the word until
4b
{1,0,-1} the last 2MSBs
MSB • Decode the leaf cells to get the {+1,0,-1}
• Remedy for idle tone
• If data is even for a while, consider add 1 to one
branch and subtract 1 in the other branch. This
LSB
increases the data activity.
{1,0,-1} • Watch out for headroom issue in the digital
words
Example of a 4b implementation • Note: that each bit is still a 1st-order noise
shaped sequence which has the
fundamental component and shaped noise.
Other possible solutions to tree-structure DEM
• Applicable when input word width is large for any straight
implementations of DEM
• Consider noise shaped segmentation principle
• Only one component in the sub-level has the signal fundamental. Easier to
derive solution for idle tones.
• May be done recursively
• A and B can then be followed by a butterfly, rotational or tree-structure
shuffler

8 5
IN B Only shaped noise

1st-order 4
modulator A Fundamental and shaped noise
M2
 Approach #4: Vector quantization
uN-1

uN-1
DAC input D  {0,..,N}
D V= .
.
min() .
u2
element selector u0
sort () u0 ui  {-1,0,1}
-1

delay

• A slight change in the ESL to accommodate 3-level data

• Rank N input vectors with least used first
• Assert {-1,0,1} in D output elements in output vector V
• Feedback the elements of V to the corresponding integrators (or loops)
• Can apply the use of a dithering vector to the input of the sorting routine to
cure idle tone behavior
 Approach #5: Real-time DEM
• Direct concept can be re-used
• Requires fast clock
• Switching loss due to logics may become significant to not be considered for
low power implementation
 Recap
• Brief background
• Motivations and Applications
• 3-level vs. 2-level unit element: advantages and challenges
• What problems DEM does and does not solve
• Practical dynamic element matching (DEM) techniques
• 2-level DEM review
• 3-level DEM
• Limitations of each technique
• Solution spaces
• Higher-order DEM
• Issues in 2nd-order DEM
• Solutions
• Application specific concepts of DEM

• Q and A
 Why need nd
2 -order DEM
• 1st-order DEM
• Effective when OSR >= 64x
• At lower OSR
• Shaped mismatch noise degrades SNR.
• DEM tones appear inband.
• Over-clocking DEM (2x) may offer small improvement (not enough). Switching
dynamics become a source of noise.

• Solution: 2nd-order shaping

• Issues with 2nd-order:

• Ineffective for low input amplitude.
• Area intensive especially with number levels >8.
 Approaches to nd
2 -order DEM
• Data directed : Cannot produce second-order mismatch shaping
effectively
• When inputs are both 11 or 00, mismatch shaper cells get “overloaded”

• Rotational: Can produce a second-order, but extremely hardware

intensive, tends to revert to first-order shaping

• Tree-structure: Can produce second-order shaping but tends to revert

back to first-order in real life use case (signal conditions)

• Vector quantization: Very hardware intensive due to realization of

sorter and element selector
• Hardware implementation at size = 2-bit (4 elements) is affordable
 Design challenges and solutions

• Full implementation of vector quantization is extremely hardware

intensive
• Solution: Segmentation (or tree structure)
• Break it down to smaller chunks where realization is affordable
 2nd-order DEM for 3-level unit elements
­00 for 0
°
• Each leaf cell needs 2 bits: ®01 for  1
°11 for  1
¯

• At one level higher in the tree, minimum size = 3 bits

• Will need 4 cells to represent {+3,…, -4}

Operations:
signed 1) OUT[1:0] = IN >> 2
2b out
1, 0 -1 2) RESIDUE = 2 LSBs of IN
in 3) Use a 2nd-order NS loop to
signed decide how to distribute the
3b 1, 0 -1
residue to the outputs
Block diagram of a 3-bit, 2nd-order DEM
IN
Signed
>> 2
Signed
3b 2 LSBs 2b
x[3]
z 1 z 1 OUT[3]
1  z 1 1  z 1 2
x[2]
feedback[3] OUT[2]

x[1]
Vector OUT[1]
quantizer
z 1 z 1 x[0]
OUT[0]
1  z 1 1  z 1 {+1,0,-1}
2

feedback[0]
{+1,0,-1}

Feedback is ONLY produced when there is an instantaneous error made

 Block diagram of a 5-bit, nd
2 -order DEM
1, 0 -1
1. A + B + C + D = IN.
A 2. The spectral difference
3b between every pair in layer
1 must show a 2nd-order
Signed binary 1, 0 -1 highpass function
IN {+/-16} B • DAC-to-DAC gain error
will be 2nd-order shaped.
3b OUT 3. The spectral difference
{16,….,-16} between any pair of unit
1, 0 -1 elements in layer 2 must
5b also show a 2nd-order
C highpass function
3b • Element mismatch will
LAYER 1 be 2nd-order shaped.
1, 0 -1
D
3b
LAYER 2
 Simulation result

Ideal
With 0.3% rms mismatch
Input = -60dBFS at 100kHz
Fclk = 64MHz
40dB/dec SNR is severely degraded !
Closer look at the integrator outputs at low input level
 The cause: low data activity
z 1 z 1
1  z 1 1  z 1 2 Vector
quantizer

• Very large dynamic range integrator required

• Clipping makes the matter worse
 Design challenges and solutions

• Full implementation of vector quantization is extremely hardware

intensive
• Solution: Segmentation (or tree structure)
• Break it down to smaller chunks where realization is affordable

• Inadequate data activity: cause large growth in integrator word

widths
• Solution: Trade off a small thermal noise performance: Dynamics
enhancement
 Dynamics enhancement (DE)
• Increases the data activities while maintaining the digital value
unchanged.
• Add +1 and -1 to digital input when possible

• Trades off thermal noise for mismatch

• Goal: keep # of ON cells as low as possible

• Reduces the integrator output swing tremendously. Very hardware

efficient.
 Dynamics enhancement (DE)
• If 2LSB = “00”, start counting
• If count exceeds a designated number “wait_time”
• Check if signal(s) are in safe range
• Add +1 and -1 to the branches. Reset wait_time.

• Parameters to adjust
• Wait_time: affects integrators word width
• Amount of +1/-1 added, restricted by thermal noise performance
• Choice of where to direct +1 and -1 to.
Combining nd
2 -order DEM and dynamics
enhancement
3b signed binary

{4,..0,..-4}

Signed binary
3 MSBs

5b

{1,0,-1}

2nd-order
Dynamics
DEM
enhancement
6' mods (4)
Simulation: With dynamics enhancement

Input: 106.5dB

Input = -60dBFS at 100kHz

Fclk = 64MHz
Mismatch: 0.3% rms
Output: 106.4dB
Integrator outputs with DE

LAYER 1

Much smaller dynamic

range needed !

LAYER 2
Number of DAC cells used at -60-dB input

Average 1 cells on Average ~ 5 cells on

(31% of total)

WITHOUT dynamics enhancement WITH dynamics enhancement

Layout of a Multibit Continuous Time 6' ADC”

DEM

A CT sigma delta 3rd-order modulator with 2nd-order DEM

 Recap
• Brief background
• Motivations and Applications
• 3-level vs. 2-level unit element: advantages and challenges
• What problems DEM does and does not solve
• Practical dynamic element matching (DEM) techniques
• 2-level DEM review
• 3-level DEM
• Limitations of each technique
• Solution spaces
• Higher-order DEM
• Issues in 2nd-order DEM
• Solutions
• Application specific concept of DEM

• Q and A
 Problem definition

• Reduce power consumption in the 3-level unit element DAC

• In particular, when the input is small (-60dB) , only 1 or 2 cells are

used, the rest of the cells are not used but still burn power.

• Goal: to turn off those unused cells without degrading performance.

• Note: Dynamically turning off cells will cause degradation due to

power consumption constraints and other dynamic effects.
 Variable-length DEM
• Group size changes according to the envelop of the input signal
• Depends on the input data amplitude, the full group, half-group or quarter-
group of the elements is used to create the signal
• The switch over to smaller group sizes is done at the time when the
CUMMULATIVE ERROR is zero Æ DOES NOT interfere with the DEM activities
• Once the smaller group is selected, the unused elements (outside of the
smaller group) will be turn off to save power
• With small group of elements, there is LESS mismatch error Æ Smaller shaped
mismatch error in output.
• With less elements, the dynamics of the DEM is more Æ the DEM becomes
more effective!
 Variable-length DEM
• Example of a 16 units 3-level thermometer-code DAC.

16 elements
8 elements
4 elements
If envelop is < 1/4 FS If envelop is < ½ full scale (FS)
16 element
& cumulative error = 0 & cumulative error = 0

Rotation rings of DEM

 Example: Application in audio DAC

DEM length info

Fs

Interpolation Variable 3-level

Sigma-delta
Length Unit element I-to-V
Filter modulator
128xFs DEM Multi-bit DAC
Fs
128xFs

• Peak detector is used at the interpolation filter input

• Hysteresis is needed to avoid chatting (in and out of variable length mode)
 Simulation case
• 2nd order sigma delta , 128x OSR, 6b output
• Input frequency = 1kHz
• Analog elements have 0.2% rms mismatch error
• Amplitude changes from full scale to -60dB FS
• Demonstrate the switching from full-size group to quarter-
size group
• Demonstrate the improvements in SNR
• Demonstrate the ability to turn off the unused elements
to save power
 Simulation result: No Variable-length
Full scale -60dB Full scale

Input signal

Time (in sample)

Peak detector output

Time (in sample)

Number of elements used

= 32
Time (in sample)
No Variable-length: FFT of the -60dB input

Tones due to large number of

Elements -> had to wait for
long time to apply remedy.

Amplitude (dB)

Frequency (Hz)
 Simulation: With Variable-length DEM
Full scale -60dB Full scale

Input signal

Time (in sample)

Peak detector output

Time (in sample)

32 elements used
Number of elements
used
Time (in sample)

16 elements
ONLY 8 elements used (can turn OFF 24 elements completely)
With Variable-length: FFT of the -60dB input

Amplitude (dB) Very clean inband noise floor

Frequency (Hz)
 Performance compared to 2nd order DEM @ -60dBFS, 0.25% rms
mismatch

Enhanced 2nd-
1st-order VLDEM
order DEM

• Performs comparably to a properly done 2nd-order DEM (Both @ 103dB SNR)

• Much lower hardware requirement
 Notes
• Optimized power consumption for DAC at low input level
• Very effective first order DEM for the sub segment(s)
• Very effective for DEM idle tone remedy
• Trade off the gain accuracy across signal amplitude range (there is NO
distortion because there is no switching back and forth between
smaller DACs and the large DAC , on the fly).
• Suitable for applications that require high AC linearity and not DC
accuracy
 Recap
• Brief background
• Motivations and Applications
• 3-level vs. 2-level unit element: advantages and challenges
• What problems DEM does and does not solve
• Practical dynamic element matching (DEM) techniques
• 2-level DEM review
• 3-level DEM
• Limitations of each technique
• Solution spaces
• Higher-order DEM
• Issues in 2nd-order DEM
• Solutions
• Application specific concepts of DEM

• Q and A
 References
• Richard Schreier et al., “Delta-Sigma Data Converter”, IEEE press 1977
• Tom Kwan and Bob Adams, “ A stereo multibit DAC with asynchronous master clock interface”, JSSC Dec 1996.
• Khiem Nguyen et al., “A 108dB SNR, 1.1mW Oversampling DAC with Three-level DEM technique”, ISSCC 2008
• A. Bandyopadhyad , K Nguyen, “A 120dB SNR 21.5mW CT SD DAC”, ISSCC 2011
• I. Fujimori et al., “A Multibit Delta–Sigma Audio DAC with 120-dB Dynamic Range”, JSSC vol. 38, no. 8, Aug 2000
• I. Galton et al., “Why Dynamic-Element-Matching DACs Work”, Trans on Circuits and Systems II, vol. 57, no. 2, pg. 69-
74, Feb 20
• I. Galton et al., “Simplified Logic for First-Order and Second-Order Mismatch-Shaping Digital-to-Analog Converters”,
Trans on Circuits and Systems II, vol. 48, no. 11, pg. 1014-1027, Nov 2001
• P. Rombouts et al., “A Study of Dynamic Element-Matching Techniques for 3-Level Unit Elements”, Trans on Circuits and
Systems II, vol. 47, no. 11, pg. 1077-1087, Nov 2000
• A. Bandyopadhyad , K Nguyen, “A 96dB SNR 600kHz CT SD ADC”, Symposium on VLSI Circuits Digest of Technical Papers
2014
• E. Folgman et al., “A Dynamic Element Matching Technique for Reduced-Distortion Multibit Quantization in Delta–
Sigma ADCs”, Trans on Circuits and Systems II, vol. 48, no. 2, pg. 158-170, Feb 2001
• R. Wang et al., “Split-set data weighted averaging”, Electronic Letters, 16th February 2006 Vol. 42 No. 4.
 References
• Anas Hamoui and K. Martin, “Linearity enhancement of multi-bit SD modulators using pseudo data weighted
averaging”, ISCAS 2002, pg 285-288.