Improving DAC performance

1

• Improving DAC performance
• using redundancy
• and digital pre-correction

Hans Hegt Mixed-signal Microelectronics
group Eindhoven University of Technology
2
Poster contributions
TU/e
Digital pre-correction method for mismatch in
DACs with built-in self-measurement Pieter
Harpe, Jacobus de Meulmeester, Hans Hegt, Arthur
van Roermund 
A parallel current-steering DAC architecture for
flexible and improved performance Richard van den
Hoven, Georgi Radulov, Hans Hegt, Arthur van
Roermund
2
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
3
DAC basics
TU/e
Standard current steering DAC architecture
Control
N-bits input
switches
Analog output
Unit current sources
High accuracy means huge area, or calibration, or
3
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
4
Outline
TU/e

• improve DAC performance using
• redundancy
• self-measurement
• digital pre-correction

• Introduction
• Digital pre-correction
• Method 1 redundancy, self-measurement and
  digital pre-correction
• Method 2 redundancy, self-measurement with
  parallel DACs
• Conclusions

4
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
5
Principle of digital pre-correction
TU/e
Proposed correction technique requires three
function blocks
2. Digital pre-correction circuit determines a
code corresponding to the desired output value
1. DAC core with built-in redundancy prevents
the occurrence of non-correctable errors
3. Self-measurement circuit measures the actual
values of the current sources
5
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
6
Method 1
TU/e

• Digital pre-correction method for mismatch
• in DACs with built-in self-measurement
• Uses
• redundancy
• self-measurement
• digital pre-correction

6
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
7
Method 1 redundancy, self-measurement
and digital pre-correction
TU/e
Analog output
7
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
8
Method 1 redundancy, self-measurement
and digital pre-correction
TU/e
normal (without redundancy)
with redundancy
analog output
analog output
digital input
digital input
With redundancy all steps 1LSB guaranteed
always pre-correction possible ? Price
more current sources needed ? to cover the
same output range
8
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
9
Method 1 redundancy, self-measurement
and digital pre-correction
TU/e
start
select source I
positive
j
initialize wj
(wj
0)
turn o
f
f all smaller source
s
(I
0)
sum
for all smaller sources
select source Ik
negative
(k j-1 down to 0)
( I
I
Ik
)
sum
sum
I
gt I
sum
j
yes
no
turn o
f
f source Ik
add value wk
( I
I
- Ik
)
( wj
wj
wk
)
sum
sum
end
9
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
10
Method 1 redundancy, self-measurement
and digital pre-correction
TU/e
Example of self-measurement technique for current
source I5
10
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
11
Method 1 redundancy, self-measurement
and digital pre-correction
TU/e
Selects suitable combination of current
sourcesfor each possible input code based on
measured digital values wj
successive approximation

• Properties
• Only sign-detection, addition and subtraction
  required
• Fully digital implementation
• Results in good linear relation between input
  code and output current, despite inaccuracies of
  self-measurement

11
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
12
Method 1 redundancy, self-measurement
and digital pre-correction
TU/e
12
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
13
Method 1 redundancy, self-measurement
and digital pre-correction
TU/e
Transistor simulation results fin29MHz fs500MHz
13
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
14
Method 1 redundancy, self-measurement
and digital pre-correction
TU/e
Higher reliability DAC adapts to own
errors less dependant on technology
parameters Better performance improved static
linearity (INL and DNL) improved dynamic
linearity (SFDR) Smaller chip area large
intrinsic mismatch allowed Simple measurement
simple robust algorithm, on-chip, little
overhead
14
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
15
Method 2
TU/e

• A parallel current-steering DAC architecture
  for flexible and improved performance
• Uses
• redundancy
• self-measurement
• digital pre-correction
• parallel DACs

15
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
16
Method 2 redundancy, self-measurement and
pre-correction with parallel DACs
TU/e
Starting point parallel (sub-)DACs for
flexibility

• Separately
• Interleaved
• Parallel
• Combinations

Using in parallel for
-Higher resolution -Improvement (static) linearity
16
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
17
Method 2 redundancy, self-measurement and
pre-correction with parallel DACs
TU/e
Linearity improvement by distributing the DAC
input code optimally among sub-DACs
E.g. Combine two (fixed) sub-DACs
17
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
18
Method 2 redundancy, self-measurement and
pre-correction with parallel DACs
TU/e
Typical example one fixed 8-bit DAC versus
best combinations with four 6-bit sub-DACs
8-bit DAC
INL
best combinationsfour 6-bit sub-DACs
non-linearity reduction better than 5 bit best
combinations found by complete search, result
stored in memory (control part)
18
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
19
Method 2 redundancy, self-measurement and
pre-correction with parallel DACs
TU/e
4 Active DACs
2 Active DACs
19
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
20
Method 2 redundancy, self-measurement and
pre-correction with parallel DACs
TU/e
Characterizing each sub-DAC by its
INL Measurement of INLs of sub-DACs with simple
components
ExampleDAC A measured sub-DACDAC B
(inaccurate) referenceDAC C (downscaled) to
measure DNLj of DAC ADAC D produces 0 or 1 LSB
20
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
21
Method 2 redundancy, self-measurement and
pre-correction with parallel DACs
TU/e
Simulation resultsfour 10-bit sub-DACs
INL improvement
Spectral content
21
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
22
Method 2 redundancy, self-measurement and
pre-correction with parallel DACs
TU/e
Conclusions Flexibility of the parallel DAC
structure totally in the digital
domain Redundancy comes from large number of
possible distributions of the DAC input code
among sub-DACs Linearity improvement example
with 12-bit DAC composed of four 10-bit sub-DACs
shows typical linearity improvement of gt4.5 bit
22
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
23
General conclusions
TU/e

• DAC performance highly improved by applying
  redundancy and digital pre-correction
• Relaxed mismatch requirements ?reduced chip area
• Simple robust algorithms little hardware
  overhead for self-measurements
• Combination of self-measurement and digital
  pre-correction improves robustness DAC adapts to
  its own errors
• Standard CMOS
• Exploits digital processing power to solve analog
  problems
• Both methods can be combined (orthogonal)

23
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction
24
Thank you for your attention
TU/e
24
ProRISC2005 Improving DAC
performance using redundancy and digital
pre-correction