smith_wangaDAC2

Download Report

Transcript smith_wangaDAC2

Digital to Analog Converter for
High Fidelity Audio Applications
Matt Smith
Alfred Wanga
CSE598A
Project Summary
• High Quality Audio Applications
– Accurate Reproduction [16 bit]
– Low Noise
• Versatile
– Support for Standard Sampling Rates
– Specifications that allow use in Various Audio
Applications
R-2R Ladder Architecture
* Buffer inserted on output for low output impedance
D Flip Flop Schematic
D Flip-Flop Simulation Results
CLK
DIN
Q
QNOT
Pass Switch
• NMOS pass transistor
only
• CMOS Transmission
gate not needed
because we don’t go
near VDD
• Sized to allow proper
operation
Voltage Reference
Voltage Reference Results
• Circuit modified from
Homework #3 provides
2.5V and 1.1V voltage
reference
• Reference output
stable down to ~3.3V
supply voltage
• 2.5V reference varies
by 400mV over -40C to
85C (3.2 mV/˚C)
• 1.1V reference varies
by 150mV over -40C to
85C (1.2 mV/˚C)
Voltage Reference Results
• Power supply
rejection ratio is
48dB
• LSB accuracy
corresponds
with 7mV p-p
supply noise
Resistor
• 200K resistor takes
~6800 um2
• We used values of
400K, 200K, 16K,
and 8K
• Resistors are a
large part of area,
but there is room to
spare in the pad
frame
Output Buffer
• Unity gain opamp
• Open loop gain = 2560
Output Buffer
The high end wasn’t a problem –
we wouldn’t go that high
But what to do about
the low end?
Output Filter
This noise is partly an artifact of
simulation, but a low-pass filter
is a good idea anyway
Complete Design Schematic
Final Layout
Bias
Opamp
400K resistors
0
7
200K resistors
15
8
Transistor Switches
D - Flip Flops
Simulation Results
Complete-circuit simulation with top 7 bits active and the rest tied high
Simulation Results
• Full-circuit simulation
with all 16 bits
operating.
• LSB increments are
28uV
• Noise peaks are
1/4LSB
Design Assessment
•
•
•
•
Maximum output voltage: 3.441V
Minimum output voltage: 1.604V
Voltage swing: 1.836V
LSB voltage change: 28uV
– Calculated and observed agree to +- 0.2uV
• Just a little noisy … simulation problem or
real?
Future Work / Improvements
• Temperature stability (reference)
• Wide swing output
• Decrease noise on output