adda analog to digital conversion

1. AD/DA Conversion Techniques
-
An Overview
J. G. Pett
 Introductory tutorial lecture for :-
‘Analogue and digital techniques in
closed-loop regulation applications’
17/09/2002
 for terminology see Analog Devices Inc.

2. AD/DA
 Introduction to the subject
 Understanding conversion methods
 Methods
 Parameters
 The past, the present and the future

3. Introduction
 What are AD/DA Converters
 What are they used for
 Why do you need to know how they work
 Digital coding methods
 Waveform digitising
 CERN examples

4. What are AD/DA
Converters (1)
 An Analog to Digital converter [AD or ADC]
is an electronic circuit which accepts an
analog input signal (usually a voltage) and
produces a corresponding digital number at
the output
 An Digital to Analog converter [DA or DAC]
is an electronic circuit which accepts a
digital number at its input and produces a
corresponding analog signal (usually a
voltage) at the output
 They exist as modules, ICs, or fully
integrated inside other parts, e.g. µCs

5. Photos

6. What are AD/DA Converters (2)
ADC 1 DAC 1
ADC 2
COMPUTER
or µP/µC
12
16
16
Digital
discrete time world
Analog
continuous time world
Analog
continuous time world
The
Real
World
The
Real
World
Typical AD & DA Application
+/-10v
+/-5v
+/-10v

7. What are they used for
 Any time a real world analog signal is
connected to a digital system
 CD players, GSMs, DVMs, Digital Camcorders
etc, etc
 CERN control systems & instruments
 HOWEVER, each application has particular
needs
 Resolution - number of bits
 Speed and Accuracy
 Level of input/output waveforms
 Cost etc

8. Why do you need to know
how they work
 Because the theoretical course you will
shortly undertake assumes perfect
converter products - BUT
 Practical converters have :
 Many conversion methods - why
 Trade-offs between resolution and speeds +
delays
 Different methods of “sampling” the
waveforms
 A large number of basic and method-dependent
error sources
 Manufacturers specifications which ‘differ’ -
AND
 Almost all converters need some analog ‘signal
conditioning’ which is application dependent

9. Digital coding methods (1)
 8,10,12,14,16,18, 20-24bits?
 Most/Least significant bit
MSB/LSB
 Uni-polar, bipolar, straight
binary, 2’s complement -
invert MSB
 Parallel I/O or serial [delay]
 Bytes or words
 Double buffering
 Digital ‘breakthrough’
 Digital correction methods
 Time skewing & jitter
0v
+10v
-10v
0000 FFFF
8000
AD/DA Transfer Characteristic
0000 7FFF
FFFF
8000

10. Digital coding methods (2)
 Resolution = 2n-1 [n = number of bits]
n 2n 1bit ppm [1x10-6]
 8bits 256 3906
 10bits 1024 976
 12bits 4096 244
 14bits 16384 61
 16bits 65536 15
 18bits 262144 3.8
 20bits 1,048576 0.95
 22bits 4,194304 0.24
 24bits 16,777216 0.06

11. Waveform digitising (1)
 A waveform is ‘digitised’ (sampled) at a constant
rate D t
 Each such sample represents the instantaneous
amplitude at the instant of sampling
 Between samples the value remains constant [zero
order hold]
 What errors can occur in this process ?
time
Digital
value

12. Waveform digitising (2)
 A & B show aliasing in the time domain
 C & D show a different case in the frequency
domain
- it is important to understand these effects
A
B
C
D

13. Waveform digitising errors
 For a DAC
 output waveform is a ‘distorted’ version of original
 higher frequencies not reproduced - aliasing ?
 ‘average shape’ displaced in time
 ‘sharp’ edges need filtering
 For an ADC
 converter sampling errors
 with a ‘sample & hold’ circuit ahead of the converter?
 integrating action during part, or all of the sample-time
?
 conversion time
 data ‘available’ delay
 aliasing - [ is multiplication of input spectrum and
fs]
…[must ‘remove’ all spectrum > fs/2 before
sampling]

14. Sampling rate
 Nyquist rate = 2x highest frequency of
interest
 Practically, - always sample at least 5x, or
higher
 Ensure ADCs have input filtering [anti-alias]
where necessary [large hf signals]
 Filter DAC outputs to remove higher
frequencies and switching ‘glitches’
 ‘Over-sampling’ converters sample x4 to
x500 - this may reduce above problems
and/or extend resolution

15. CERN examples
 Many PLCs with analog values, such as
temperature, to measure : 10 - 12bit <10kHz
 PS, SPS, LHC control instrumentation, such
as power converter control, regulation and
monitoring : 16 - 22bit <1kHz
 Beam instrumentation, experiments : high
speed: 10 - 12bit 25ns
 ETC ETC

16. Photos
1969
ISR Beam-Transfer DAC
[5 decimal decades]
Relay switching
Kelvin-Varley divider
1973
ISR Main Bends DAC
[16bit binary
All electronic switching

17. Photos
ADC Sigma-Delta 1998
1989
LEP 16bit Hybrid DAC

18. Understanding Conversion
Methods

19. AD/DA Methods
 Some very simple ideas
 DAC circuits
 Basic ADC circuits
 Successive approximation, flash - S&H
 Integrating - single/dual/multi slope
 Charge balance, D

20. Some very simple ideas
 ADC =
 precise reference voltage
 comparison of divider value with unknown [analog input]
 “digitally adjustable” divider or potentiometer [output
value]
 DAC =
 precise reference voltage ……. {multiplying dac}
 “digitally adjustable” divider or potentiometer [input
value]
 optional output amplifier of pot. value [analog output]
=
‘Digitally set’
potentiometer
dial
Comparator
equal
Vref
Unknown
voltage
DAC ADC
Vdac

21. DAC circuits (1)
 Summation of binary weighted currents
 Modern DACs use the ‘R-2R ladder’
Simplified binary weighted resistor DAC
8.75V
9.375
max.
R - 2R ladder DAC

22. DAC circuits (2)
 Important circuit concepts
 Resistor tracking - temp. & time > ratios
 Switch is part of R [on & off resistance]
 Limits for tracking and adjustment
 Switch transition times - glitches
 Switched current sources are faster
 Other DAC methods
 DC performance not needed for all uses
 Different ladders, Caps. as well as Resistors
 PWM, F>V
 Sigma-Delta
 Performance cannot be better than the Reference
- {multiplying DAC concept}

23. Basic ADC circuits (1)
 Digitising begins with a ‘start’ pulse
 DAC is ramped up from zero
 counter stopped by comparator when Vin = DAC out
 ADC output is counter value
 Tracking ADC
Simple ramp and comparator ADC
start Binary output
Unknown
analog
input

24. Basic ADC circuits (2)
 This ADC circuit is limited and rarely used
WHY -
 slow
 variable time to give result
 input signal can vary during digitising
 Successive Approximation ADC solves these
problems - using
 complex logic to test and retain each DAC bit
 a sample and hold circuit ahead of the
comparator

25. Successive Approximation
ADC
 Fast process - 1 -
100µsecs
 Result always n clocks
after start
 Used extensively for
12-16bit DAQ systems

26. Flash ADC
 The fastest process <50nsecs
 Limited resolution typically 8 -
10bits
 Half-flash technique is cheaper
Flash
Half-Flash
analog
input
analog
input
Vref
Vref

27. Sample & Hold Circuit (1)
 Essential for defining the ‘exact’ moment of
sampling
 Circuit introduces other error sources [ see (2) ]
LF398

28. Sample & Hold Circuit (2)
Storage Capacitor Waveform