Lec. 3:
Fundamentals of Data Converters
Lecturer: Hooman Farkhani
Department of Electrical Engineering Islamic Azad University of Najafabad Feb. 2016.
Email: [email protected]
In The Name of Almighty
ADC Input-Output Transfer Characteristics
Least-Significant Bit (LSB) in Volts
The Ideal A/D Converter & Quantization Error
1
0
( .2 )
N
m
in ref m Q
m
V V B V
Effect of Quantization error on Signal
Quantization Error
Assuming a uniform distribution
1/ / 2 / 2
( , )
0 P x t x
others
/ 2 / 2 2
2 2 2 2
,
/ 2 / 2
( , ) . 1
Q rms 12
V Ex x P x t dx x dx
VQ rms, 12 VLSB12 2NVFS121LSB
Signal-to-Quantization-Noise Ratio
Assuming uniform distribution of eq and a full-scale sinusoidal
input, we have:
Quantization noise spectrum
Quantization results in an "infinite" number of harmonics
6.02 1.76
SNR N dB
Performance Metrics of ADCs
Static specifications
Offset error
Gain error
Monotonicity
Resolution
Differential nonlinearity (DNL)
Integral nonlinearity (INL)
Dynamic specifications
SNR
THD
SNDR
ENOB
SFDR
ERB
Static specifications
Offset Error
Defined as a constant difference, over the whole range of the ADC, betwe en the actual output value and the ideal output value.
Expressed as number of LSBs (counts) or as percentage of full scale range
Offset error can be removed by measuring a reference point and subtractin g that value from future samples.
Static specifications
Gain Error
Defines as the difference of the slope of the actual output values and the id eal output values.
Expressed as number of LSBs (counts) or as percentage of full scale range
Caused by preamplifiers, attenuators, or signal transducers.
Comments on Offset and Gain Errors
Generally it is non-trivial to build a converter with very good gain/offset specifications
- Nevertheless, since gain and offset affect all codes uniformly, these errors tend to be easy to correct
-E.g. using a digital pre- or post-processing operation - Also many applications are insensitive to a certain level of gain and offset errors
- E.g. audio signals, communication-type signals,…
Static specifications
Monotonicity
Monotonicity is a property of certain types of digital-to-analog converter (DAC) circuits.
In a monotonic DAC, the analog output always increases or remains const ant as the digital input increases.
Static specifications
● Differential nonlinearity (DNL)
LSB
LSB i
in i
in
i
V
V D
V D
D V
DNL
( ) (
) )
(
11. When DNL is large enough, output codes can disappear as the range of inputs able to produce them shrink
2. Any converter with worst-case DNL less than 0.5LSB is guarantee d to have “no missing codes”
3. It doesn’t mean an DNL greater than ½ LSB necessarily produces
Missing code:
k1 i
i
k
DNL
INL
● Integral nonlinearity (INL)
Static specifications
1. INL acts as an input referred high-order polynomial distortion on the in put, usually larger than DNL
2. INL could add distortion terms of high degree so that while 3rd order IM dominates
Dynamic specifications
2 2
( )
( ) 10.log
( )
sig spur
V f SFDR dB
V f
( ) 9
SFDR dB n c
SFDR: Spurious-Free Dynamic Range
Only Quantization Noise:
Dynamic specifications
SNR: Signal-to-Noise Ratio
10 log (10 s ) 6.02 1.76
dB
n
SNR P N dB
P
1 2
2
( . )
( )
NH
j
V j f
sigTHD V f
THD: Total Harmonic Distortion
By convention, total distortion power consists of 2
ndthr ough 7
thharmonic
Only Quantization Noise:
) V
V V
log( V 10 SNDR
dist 2 th
, n 2 Q
2
sig2
SNDR: Signal-to-Noise-and-Distortion-Ratio
ENOB: Effective Number of Bits
02 . 6
76 .
1 dB
ENOB SNDR
SNDR ≤ SNR
Dynamic specifications
Dynamic specifications
ERB: Effective Resolution Bandwidth
Defined as the input frequency at which the SNDR of a converter has dropped by 3dB
Equivalent to a 0.5-bit loss in ENOB
1. SFDR: important for wideband receivers like cellular base-station
due to large interferences; Similarly MTPR for multi-tone receiver s (xDSL)
2. SNDR: important for audio/video applications
3. SNR: important in radar (detecting a target from long distance
means to overcome noise); not too many signal targets for inter- modulations concern
Depending on the application an specific parameter becomes important
Dynamic specifications
Relationship between INL and SFDR
At low input frequencies, finite SFDR is mostly due to INL
Quadratic/cubic bow gives rise to second/third order harmonic
Rule of Thumb: SFDR≈20 log (2
N/INL)
E.g. 1 LSB INL, 10 bits SFDR ≈ 60dB
References
Professor Boris Murmann Course slides 2012, Stanford University- EE315B course
Dr. Reza Lotfi, ADC course slides 2008.