Mixed-Signal VLSI Design Course Code: EE719 Department: Electrical - - PowerPoint PPT Presentation

▶
mixed signal vlsi design course code ee719 department
SMART_READER_LITE
LIVE PREVIEW

Mixed-Signal VLSI Design Course Code: EE719 Department: Electrical - - PowerPoint PPT Presentation

May 27, 2023 385 likes •608 views

Mixed-Signal VLSI Design Course Code: EE719 Department: Electrical Engineering Lecture 19: March 02, 2020 Instructor Name: M. Shojaei Baghini E-Mail ID: mshojaei@ee.iitb.ac.in 1 2 2 Module 20 Introduction to Performance Parameters of Data

slide-1
SLIDE 1

1

Mixed-Signal VLSI Design Course Code: EE719 Department: Electrical Engineering Lecture 19: March 02, 2020

Instructor Name: M. Shojaei Baghini E-Mail ID: mshojaei@ee.iitb.ac.in

slide-2
SLIDE 2

2 2

IIT-Bombay Lecture 19 M. Shojaei Baghini

Module 20 Introduction to Performance Parameters of Data Converters References

  • Understanding Data Converters, TI, 1995.
  • Chapter 5, the Data Conversion Handbook,

Analog Devices, 2005.

slide-3
SLIDE 3

3 3

IIT-Bombay Lecture 19 M. Shojaei Baghini

Bridge Between Analog and Digital Domains

Analog Domain Digital Domain Conditioning A/D Conversion D/A Conversion Smoothing and Equalization

slide-4
SLIDE 4

4 4

IIT-Bombay Lecture 19 M. Shojaei Baghini

Basic Metrics to Start with

l Full Scale Range l Resolution: Effective Number of Bits (ENOB) l Introduced Noise and Distortion: Signal to

Noise and Distortion Ratio

l Signal Bandwidth l Sampling Frequency

slide-5
SLIDE 5

5 5

IIT-Bombay Lecture 19 M. Shojaei Baghini

Assume N-point FFT of a sine wave test. If the fundamental frequency is in bin m and has amplitude Am then SNDR is given by the following relation.

Signal-to-Noise and Distortion Ratio (SNDR)

slide-6
SLIDE 6

6 6

IIT-Bombay Lecture 19 M. Shojaei Baghini

SFDR - Example

Source: Nicolas Gray, National Semiconductor, 2006 Noise floor

Note: The average value of the noise contained in each frequency bin of M-point FFT is 10 ´ log (M/2) dB below the total noise power.

slide-7
SLIDE 7

7 7

IIT-Bombay Lecture 19 M. Shojaei Baghini

Spurious-Free Dynamic Range (SFDR)

l In general dynamic range is the ratio of

largest absolute value of the signal to the smallest absolute value of the signal.

l Spurious-free dynamic range (SFDR) is

ratio of the input signal to the peak harmonic (or spurious) component. It’s a measure of signal/data fidelity. Example: fractions of fs in !" ADC

slide-8
SLIDE 8

8 8

IIT-Bombay Lecture 19 M. Shojaei Baghini

Intermodulation Distortion (IMD)

l Most of signals contain many frequency components. l Non-linearity causes two or more signal frequencies to

produce intermodulation products.

Source: Analog Devices tutorial MT-053, 2009

Single tone SFDR Multi tone SFDR

Frequency Frequency

slide-9
SLIDE 9

9 9

IIT-Bombay Lecture 19 M. Shojaei Baghini

Ideal Characteristics of DAC/ADC

1 LSB Center point ADC (II) DAC (I)

D: LSB size

  • Analog input range

May be [-VFS/2,VFS/2].

  • Signed output code

may be used. Starting from DAC

Type equation here.

/

012_456 = ∆ 9 :;< =>?

@:×2: bi: bit # i

slide-10
SLIDE 10

10 10

IIT-Bombay Lecture 19 M. Shojaei Baghini

Classification Based on Sampling Rate of Low-pass Signals

  • fs > 2 × BW (Nyquist rate sampling but not much higher

than 2) – "Nyquist Converters“ – It requires high-performance anti-aliasing filtering.

  • fs >> 2 × BW Oversampling (Oversampled converters with

noise shaping) – Simple anti-aliasing filter and smoothing filter is enough. – Oversampling reduces quantization noise.

  • fs < 2 × BW (Undersampling or sub-sampling)

Useful for signals of which power is concentrated in a narrow band around a frequency.

slide-11
SLIDE 11

11 11

IIT-Bombay Lecture 19 M. Shojaei Baghini

Common Performance Measures

Static Measures

  • Monotonicity
  • Offset
  • Full-scale gain error and gain error
  • Diff. nonlinearity (DNL) (mainly for Nyquist converters)
  • Integral nonlinearity (INL) (mainly for Nyquist converters)

Dynamic Measures

  • Delay, settling time
  • Aperture uncertainty
  • Distortion- harmonic content
  • Signal-to-(noise and distortion) ratio (SNDR)
  • Idle channel noise
  • Spurious-free dynamic range (SFDR)
slide-12
SLIDE 12

12 12

IIT-Bombay Lecture 19 M. Shojaei Baghini

Static Specifications Offset Error (Zero Scale Error)

1 LSB Center point

Offset error is the shift of the first transition point.

  • Similar notion for full scale error
slide-13
SLIDE 13

13 13

IIT-Bombay Lecture 19 M. Shojaei Baghini

Static Specifications – Simple Definition

  • f Gain Error for DAC

Gain error (in units of LSB) = V111 / (VFS – 1 LSB) - 1

Source: The data conversion handbook, Analog Devices, 2005

Is this the best definition?

slide-14
SLIDE 14

14 14

IIT-Bombay Lecture 19 M. Shojaei Baghini

Static Specifications - Concept of Gain

Connecting center of segments will not results in a line.

  • Which definition does present the gain better?
slide-15
SLIDE 15

15 15

IIT-Bombay Lecture 19 M. Shojaei Baghini

Static Specifications Differential Nonlinearity (DNL) of DAC

Source: The data conversion handbook, Analog Devices, 2005 Variation of Difference with the code results in nonlinearity: SQNR degradation

slide-16
SLIDE 16

16 16

IIT-Bombay Lecture 19 M. Shojaei Baghini

Static Specifications Differential Nonlinearity (DNL) of ADC

Source: The data conversion handbook, Analog Devices, 2005

DNLmin > -1 LSB (no missing code).

slide-17
SLIDE 17

17 17

IIT-Bombay Lecture 19 M. Shojaei Baghini

Static Specifications: Integrated Nonlinearity (INL) of ADC and DAC

Understanding Data Converters, TI, 1995

Variation of transition points

slide-18
SLIDE 18

18 18

IIT-Bombay Lecture 19 M. Shojaei Baghini

Static Specifications Integrated Nonlinearity (INL) of DAC

Source: The data conversion handbook, Analog Devices, 2005

7 5 1 3

After offset correction

slide-19
SLIDE 19

19 19

IIT-Bombay Lecture 19 M. Shojaei Baghini

End of Lecture 19

slide-20
SLIDE 20

4 4

IIT-Bombay Lecture 14 M. Shojaei Baghini

Quantization Error in Ideal ADC

!

"# #

$%& $, ( )$ = !

"∆/% ∆/%

$% 1 ∆ )$ = ∆% 12

( )

dB N q x SQNR V x V q

e rms FS rms N FS e

76 . 1 02 . 6 log 10 2 2 12 2 12

2 2 2 2 2 2

+ = ÷ ÷ ø ö ç ç è æ = = ´ = D =

For a sinusoidal signal x(t) and approximate uniform distribution of qe.

Example: N=6 bits Þ SQNR=37.9dB N=10 bits Þ SQNR=62.0dB

… …

t qe(t)