Converter Fundamentals Leicester U March 2003

1
Converter Fundamentals

James Bryant
University of Leicester
March 2003
Converter Fundamentals Leicester U March 2003


2
Converters
Converter Fundamentals Leicester U March 2003


3
The Size of an LSB
THE SIZE OF AN LSB
RESOLUTION
(N) 2
N Voltage
(10 V FS)
ppm FS % FS dB FS
2-bit 4 2.5 V 250,000 25 -12
4-bit 16 625 mV 62,500 6.25 -24
6-bit 64 156 mV 15,625 1.56 -36
8-bit 256 39.1 mV 3,906 0.39 -48
10-bit 1,024 9.77 mV (10 mV) 977 .098 -60
12-bit 4,096 2.44 mV 244 .024 -72
14-bit 16,384 610 V 61 .0061 -84
16-bit 65,536 153 V 15 .0015 -96
18-bit 262,144 38 V 4 .0004 -108
20-bit 1,048,576 9.54 V (10 V) 1 .0001 -120
22-bit 4,194,304 2.38 V .24 .000024 -132
24-bit 16,777,216 596 nV (.6 V)* .06 .000006 -144
*600 nV is the Johnson noise in a 10 KHz bandwidth of a 2.2 K

resistor at room temperature (300 K)

(A simple technique to memorise this table is to remember that at
10-bits and 10 V FS an lsb is approximately 10 mV, 1,000 ppm or 0.1%.
All other values may be calculated by powers of 2.)
Converter Fundamentals Leicester U March 2003


4
Ideal Transfer Characteristics
Converter Fundamentals Leicester U March 2003


5
Quantization Uncertainty
Converter Fundamentals Leicester U March 2003


6
Unipolar & Bipolar Converters
Converter Fundamentals Leicester U March 2003


7
Offset & Gain Error
Converter Fundamentals Leicester U March 2003


8
Linearity Error Measurement
Converter Fundamentals Leicester U March 2003


9
Differential Non-Linearity (DNL)
Converter Fundamentals Leicester U March 2003


10
Combined Effects of Transition Noise
& DNL
Converter Fundamentals Leicester U March 2003


11
Sampled Data Systems
Converter Fundamentals Leicester U March 2003


12
DAC Settling Time
Converter Fundamentals Leicester U March 2003


13
DAC Transitions
Converter Fundamentals Leicester U March 2003


14
Harmonic Distortion
Converter Fundamentals Leicester U March 2003


15
Intermodulation Distortion
Converter Fundamentals Leicester U March 2003


16
Third Order Intercept Point
Converter Fundamentals Leicester U March 2003


17
Quantization Noise
Converter Fundamentals Leicester U March 2003


18
Large Signal Bandwidth
With small signals, the bandwidth of a circuit is limited by its
overall frequency response.
At high levels of signal, the slew rate of some stage (generally the
output stage) may control the upper frequency limit.
In amplifiers, there are so many variables that Large Signal
Bandwidth needs to be redefined in every individual case and
slew rate is a more useful parameter for a data sheet.
In ADCs, the maximum signal swing is the ADCs full-scale span,
and is therefore defined so Full Power Bandwidth may appear on
the datasheet.
HOWEVER, the Full Power Bandwidth specification says
nothing about distortion levels. ENOB is much more useful in
practical applications
(If Full Power Bandwidth is specified and ENOB is not,
somebody is probably trying to hide something!)
Converter Fundamentals Leicester U March 2003


19
ENOB
Converter Fundamentals Leicester U March 2003


20
SNR Due to Sampling Clock Jitter
Converter Fundamentals Leicester U March 2003


21
Components for Data Converters
Data Converters require:
Good logic
Good switches
Good analog circuitry (amplifiers, comparators and
references)
Good resistors
Converter Fundamentals Leicester U March 2003


22
Hybrid Converters
Early Data Converters used hybrid technology to
achieve performance unavailable from any single
monolithic technology.
Even today, some of the best converters cannot use
any available monolithic technology and are hybrid
Compound Monolithic is a marketers term for a
simpler (and cheaper) hybrid technology where two
monolithic chips from different technologies are
mounted together in a single package, but without a
ceramic substrate or other components.
Converter Fundamentals Leicester U March 2003


23
Monolithic Converter Processes
Bipolar processes have good analog performance but
less good logic and switches.
CMOS processes make excellent logic and switches
but relatively poor amplifiers and lousy references.
Processes combining the two (BIMOS , LCCMOS, etc.)
tend to be more complex and expensive and have
slightly less performance than the sum of the two but
are very convenient.
Good designers choose the best process for the
circuit to be designed.
Converter Fundamentals Leicester U March 2003


24
Thin Film Resistors
One of the key technologies for making many types of
monolithic data converters is the ability to deposit
accurate, stable SiCr resistors on monolithic chips.
Some converters use these resistors as fabricated;
others require the additional accuracy and economy
of laser trimming.
Parameters include matching to 0.005%, TC<20 ppm,
Diff TC<0.2 ppm, and long term stability of the order
of
1 ppm/1000 hours (drunkards walk).
Converter Fundamentals Leicester U March 2003


25
Changeover Switches
Converter Fundamentals Leicester U March 2003


26
Kelvin Dividers
Converter Fundamentals Leicester U March 2003


27
Simplest Current OP DAC
Converter Fundamentals Leicester U March 2003


28
Segmented Voltage DACs
Converter Fundamentals Leicester U March 2003


29
Current Segment 4-Bit DAC
Converter Fundamentals Leicester U March 2003


30
Binary Weighted DAC
Converter Fundamentals Leicester U March 2003


31
DAC Using Cascaded Binary Quads
Converter Fundamentals Leicester U March 2003


32
4-Bit R-2R Ladder Network
Converter Fundamentals Leicester U March 2003


33
Voltage-Mode Ladder Network DAC
Converter Fundamentals Leicester U March 2003


34
Current-Mode Ladder Network DAC
Converter Fundamentals Leicester U March 2003


35
Multiplying DACs (MDACs)
In all DACs, the output is the product of the reference
voltage and the digital code.
Most DACs work only over a limited range of
reference voltages
DACs which work with reference voltages which
include zero volts are known as multiplying DACs
Many MDACs work with bipolar and AC references
DACs which work with a large range of reference
voltages, but not down to zero, are not true MDACs
but are sometimes called MDACs. It is better to use
the term semi-multiplying DACs.
Converter Fundamentals Leicester U March 2003


36
Segmented Ladder DAC
Converter Fundamentals Leicester U March 2003


37
Audio DAC with Offset MSB Transition
Converter Fundamentals Leicester U March 2003


38
Sigma-Delta DAC
Converter Fundamentals Leicester U March 2003


39
Double-Buffered DAC
Converter Fundamentals Leicester U March 2003


40
Serial DACs
If data is loaded serially into a DAC, it requires fewer
data pins.
This saves space and also reduces capacitive noise
coupling from data lines to the analog output .
If the shift register of a serial DAC has an output pin, a
number of DACs may be connected in series (daisy-
chained) to a single serial data port
Converter Fundamentals Leicester U March 2003


41
Types of Analog-to-Digital Converters
Comparator: 1-bit ADC
Flash: Fast, low-resolution, power-hungry
Magamp: A new architecture with lower power and
complexity but speed approaching that of a
flash ADC
Subranging: Quite fast, high-resolution, complex
Integrating: Slow, accurate, low-power
VFC: High-resolution, low-power, ideal for
telemetry
Tracking: Fast and slow, high-resolution
Successive Approximation: Versatile, general purpose
Sigma Delta: Complex, low-power, very accurate
Converter Fundamentals Leicester U March 2003


42
Beware of ADC Logic Pitfalls!
After power-up, one or two conversions may be necessary
before the ADC runs right. EOC cannot always be trusted
at this time. An ADC may not behave the same way every
time it starts.
EOC says conversion is finished. DRDY says that data is
valid. There may be tens of nS difference between the two.
CS may not just enable the data--it may reset things for the
next conversion. In some converters, you cant not read
the data. In some converters you cant read the data twice.
In some converters, you cant strap CS and forget it.
FIND OUT WHAT SORT YOURE USING.
ALWAYS READ THE DATASHEET, OR ELSE...
Converter Fundamentals Leicester U March 2003


43
Comparators
Converter Fundamentals Leicester U March 2003


44
Flash or Parallel ADCs
Converter Fundamentals Leicester U March 2003


45
Flash ADC Input Model and Its Effect on
ENOB
Converter Fundamentals Leicester U March 2003


46
OUTPUT
-FS +FS
0
INPUT
+FS
0
-FS
Fig. 1. TRANSFER CHARACTERISTIC OF X1 AMPLIFIER
Mag Amps 1
Converter Fundamentals Leicester U March 2003


47
Mag Amps 1b
Converter Fundamentals Leicester U March 2003


48
Fig. 2. TRANSFER CHARACTERISTIC OF FULL WAVE RECTIFIER
FULL-WAVE
RECTIFIER

X2
-FS
OUTPUT
-FS +FS
0
INPUT
+FS
0
-FS
PLUS X2 AMPLIFIER
PLUS HALF-SCALE OFFSET
PLUS COMPARATOR
THIS ARRANGEMENT IS KNOWN AS A MAGNITUDE AMPLIFIER
OR MAGAMP
Mag Amps 2
Converter Fundamentals Leicester U March 2003


49
OUTPUT
-FS +FS
0
INPUT
+FS
0
-FS
Fig. 3A. TRANSFER CHARACTERISTICS OF CASCADED MAGAMPS
If we cascade several magamps, connecting the analog OP of each to the IP of the next, the transfer
characteristic between the first input and the various analog outputs will be as shown.
Mag Amps 3
Converter Fundamentals Leicester U March 2003


50
OUTPUT
-FS +FS
0
INPUT
+FS
0
-FS
Fig. 3B. TRANSFER CHARACTERISTICS OF CASCADED MAGAMPS
If we cascade several magamps, connecting the analog OP of each to the IP of the next, the transfer
characteristic between the first input and the various analog outputs will be as shown.
Mag Amps 4
Converter Fundamentals Leicester U March 2003


51
OUTPUT
-FS +FS
0
INPUT
+FS
0
-FS
Fig. 3C. TRANSFER CHARACTERISTICS OF CASCADED MAGAMPS
If we cascade several magamps, connecting the analog OP of each to the IP of the next, the transfer
characteristic between the first input and the various analog outputs will be as shown.
Mag Amps 5
Converter Fundamentals Leicester U March 2003


52
OUTPUT
-FS +FS
0
INPUT
+FS
0
-FS
Fig. 3D. TRANSFER CHARACTERISTICS OF CASCADED MAGAMPS
If we cascade several magamps, connecting the analog OP of each to the IP of the next, the transfer
characteristic between the first input and the various analog outputs will be as shown.
Mag Amps 6
Converter Fundamentals Leicester U March 2003


53
OUTPUT
-FS +FS
0
INPUT
+FS
0
-FS
Fig. 4. AN A.D.C. USING CASCADED MAGAMPS
If we look at the digital (comparator) outputs of cascaded magamps
(and the output of a comparator on the original input line) we find
that we have an ADC with a Gray Code output representing the
value of the voltage on this original input line.
Mag Amps 7
Converter Fundamentals Leicester U March 2003


54
Fig. 5A. AN A.D.C. USING CASCADED MAGAMPS
WITH DIGITAL DELAYS TO SYNCHRONISE O/P DATA.
Fig. 4 did not consider timing. There is a delay through each magamp.
The timing problems arising from these delays may be addressed in
several different ways. In this diagram digital delays in the data lines
give a parallel digital output with minimal data skew.
DLA x1
DLA x 2
DLA x 3
DLA x 4
Mag Amps 8
Converter Fundamentals Leicester U March 2003


55
Fig. 5B. AN A.D.C. USING CASCADED MAGAMPS
WITH PIPELINED SHA's AND SHIFT REGISTERSTO
SYNCHRONISE O/P DATA.
Fig. 4 did not consider timing. There is a delay through each magamp.
The timing problems arising from these delays may be addressed in
several different ways. In this diagram clocked digital delays in the data
lines (shift registers) and SHAs between the MAGAMPS give a parallel
digital output with no data skew, but a pipeline delay of N-1 clock cycles
for an N-bit converter.
SHA SHA SHA
S/R
S/R
S/R S/R
S/R
S/R
MAG-
AMP
MAG-
AMP
MAG-
AMP
CONVERSION
CLOCK
Mag Amps 9
Converter Fundamentals Leicester U March 2003


56
Subranging (Half-Flash) ADC
Converter Fundamentals Leicester U March 2003


57
Subranging ADC with Digital Error Correction
Converter Fundamentals Leicester U March 2003


58
Integrating ADC
Converter Fundamentals Leicester U March 2003


59
Integrating ADC
Converter Fundamentals Leicester U March 2003


60
VFCs
Converter Fundamentals Leicester U March 2003


61
Current-Steering VFC
Converter Fundamentals Leicester U March 2003


62
Charge-Balance VFC
Converter Fundamentals Leicester U March 2003


63
Synchronous VFC
Converter Fundamentals Leicester U March 2003


64
VFC & SVFC Waveforms
Converter Fundamentals Leicester U March 2003


65
SVFC Non-Linearity
Converter Fundamentals Leicester U March 2003


66
VFCs
It is possible to use the PERIOD of a VFC, rather than
its frequency, to measure its input
VFCs have other applications than as ADC elements:
these include isolation and use as FVCs
Converter Fundamentals Leicester U March 2003


67
Tracking ADCs
Converter Fundamentals Leicester U March 2003


68
Successive Approximation ADCs
Converter Fundamentals Leicester U March 2003


69
Successive Approximation ADCs
In modern successive approximation, ADCs the DAC
is frequently constructed from capacitors (this is
called a charge redistribution DAC).

The architecture is smaller, cheaper, faster and easier
to manufacture than traditional resistive DACs but
capacitor leakage may (not always) necessitate a
minimum clock rate
Converter Fundamentals Leicester U March 2003


70
-D
Sigma-Delta ADCs have a very high resolution, and
theyre very cheap.
But the theory of the operation is hard.
Their bandwidth is not marvellous either.
Converter Fundamentals Leicester U March 2003


71
Sampling ADC Quantization Noise
Converter Fundamentals Leicester U March 2003


72
Oversampling and Filtering Improves
ENOB
Converter Fundamentals Leicester U March 2003


73
First-Order D ADC
Converter Fundamentals Leicester U March 2003


74
D Modulators Shape Quantization
Noise
Converter Fundamentals Leicester U March 2003


75
Second-Order D ADC
Converter Fundamentals Leicester U March 2003


76
Bandpass D ADCs
Converter Fundamentals Leicester U March 2003


77
Sample-Hold Amplifiers (SHAs)
Converter Fundamentals Leicester U March 2003


78
MicroConverter
TM
Definition
High Performance Analog I/O
+
FLASH Memory
+
Microcontroller
=
MicroConverter
TM
1
2
3
Converter Fundamentals Leicester U March 2003


79
Introducing the ADuC812
12bit,8ch ADC & dual 12bit DAC
+
8Kbyte Program & 640byte Data FLASH
+
Industry Standard 8052
=
ADuC812
1
2
3
Converter Fundamentals Leicester U March 2003


80
ADuC812 - Analog I/O
8channel, 12bit, 5s, Autocalibrating ADC
DMA Controller for High Speed Capture
True 12bit Performance (INL, SNR, etc.)
Two 12bit, 4s, Voltage Output DACs
Guaranteed 12bit Monotonicity
On-Chip 2.5V Precision Bandgap Reference
On-Chip Temperature Sensor
Simple ADC & DAC Control Through Software or
Hardware
1
Converter Fundamentals Leicester U March 2003


81
ADuC812 - Flash Memory
RETAIN DATA WITHOUT POWER!
8Kbytes Nonvolatile Program Memory
Stores Program and Fixed Lookup Tables
In-Circuit Serial Programmable or External Parallel
Programmable
640bytes Nonvolatile Data Memory
User Scratch Pad for Storing Data During Program Execution
Simple Read/Write Access Through SFR Space
Built-In Security Features for Both Program & Data FLASH
Programming Voltage (V
PP
) Generated On-Chip
2
Converter Fundamentals Leicester U March 2003


82
ADuC812 - Microcontroller
Industry Standard 8052 Core
12 Clock Machine Cycle w/ up to 16MHz Clock
32 Digital I/O Pins
Three 16bit Counter/Timers
UART Serial Port
...Plus Some Useful Extras
SPI or I2C Compatible Serial Interface
WatchDog Timer
Power Supply Monitor
3
Converter Fundamentals Leicester U March 2003


83
References
[1] "HIGH SPEED SEMINAR" ANALOG DEVICES INC. 1990 $20
[2] "MIXED SIGNAL SEMINAR" ANALOG DEVICES INC. 1991 $20
[3] "1992 AMPLIFIER APPLICATIONS GUIDE" ANALOG DEVICES INC.
1992 $20
[4] "DATA CONVERTER REFERENCE MANUAL (VOL II)" ANALOG
DEVICES INC. (FREE)
[5] APPLICATION NOTE: "FREQUENCY-VOLTAGE CONVERTERS" BY
JAMES M. BRYANT (IN PREPARATION)
ANALOG DEVICES INC. (FREE WHEN AVAILABLE - TYPESCRIPT
ALREADY AVAILABLE FROM JAMES BRYANT)
[6] "A 4TH-ORDER BANDPASS SIGMA-DELTA MODULATOR"
S.A.JANTZI, M.SNELGROVE & P.F.FERGUSON JR.
PROCEEDINGS OF THE IEEE 1992 CUSTOM INTEGRATED CIRCUITS
CONFERENCE. PP 16.5.1-4
[7] "ANALOG-DIGITAL CONVERSION HANDBOOK" DANIEL H.
SHEINGOLD (ED.) PRENTICE-HALL, 3RD EDITION. 1986