Transcript Analog to Digital Converters (ADC)

Analog to Digital Converters
(ADC)
1
Technician Series
ADC 1.1
©Paul Godin
Created April 2008
Last Edit Mar 2015
Introduction
◊ Analog to digital conversion is an important aspect
of digital electronics.
◊ ADCs allow the use of real-world values with the
advantages of digital electronics.
◊ There are many examples of ADC converters used
in everyday applications.
Name a few examples of ADC applications
ADC 1.2
Advantages of Digital Values
◊
◊
◊
◊
◊
◊
◊
◊
◊
Relatively less sensitive to distortion (noise and losses)
Can be reproduced much more accurately
Much easier to reconstruct a signal
More storage options
Can be processed mathematically and logically
Easier to standardize
Systems are easier to design (fewer voltage / current issues)
Digital systems can be made small (low current)
Display options
ADC 1.3
Challenges with ADC
◊ Converting an analog value to digital values
comes with disadvantages:
◊ It takes time to convert a signal from Analog to Digital,
and then to process that signal. May be too slow for
some applications.
◊ Never 100% reproduction…always a series of discrete
values.
◊ Requires more complex circuit design
◊ More faithful reproduction requires more bit of
resolution.
◊ Requires other circuit elements such as oscillators and
memory systems.
ADC 1.4
ADC FUNDAMENTALS
ADC 1.5
Sampling
◊ Voltage signals are comprised of amplitude over
time.
◊ The analog signal must be converted to its digital
value at specific periods of time.
◊ Sampling is the process of taking a digital value at
regular time intervals.
◊ The digital values must be stored if they are to be
processed.
ADC 1.6
Sampling
AC Value
Sampling
Pulses
Time
Digital Values at timed intervals
ADC 1.7
Sampling
◊ Increasing the number of binary values
representing a voltage value improves its voltage
resolution. This is called quantization. The
greater the number of bits available, the greater
the quantization level.
◊ Increasing the sampling frequency improves the
time resolution. The more samples taken over
time the more accurate the representation of the
signal.
ADC 1.8
Nyquist
◊ The sampling frequency must be greater than the
highest frequency component of the analog signal.
◊ The Nyquist frequency has a value of twice the
highest analog frequency.
fsample  2fA(MAX)
Where:
◊ fsample is the sampling frequency
◊ fA(MAX) is the maximum analog frequency
ADC 1.9
Sampling Issues
AC Value
Digital Value
Properly Sampled
Under-sampled
ADC 1.10
Sampling Error (animated)
ADC 1.11
Sampling Rates
◊ Sampling rates are selected based on:
◊ application
◊ requirements
◊ standards
◊ As an example, an exterior thermometer needn’t
be sampled at the same rate as an audio
application.
ADC 1.12
Audio Application of ADC
◊ When music is digitized for CDs the sampling
frequency is 44.1 kHz (48 kHz for professional
recording).
◊ According to the Nyquist frequency, 44.1kHz is
acceptable for up to 22 kHz. Since most audio
equipment functions at less than 20 kHz (and is at
the upper limit of human hearing), the 44.1 kHz
sampling rate is acceptable.
◊ The bit depth is 16 for CD audio (up to 24 for
DVD-Audio).
ADC 1.13
Notes on mp3
◊
MP3 audio files refer to their quality as a bit rate. Typical mp3 bit
rates are 128kbps and 192kbps (maximum is 320kbps according to
standards).
◊
MP3 is an encoding format used to compress and reduce the file
size. The file follows protocols and contains various elements such
as headers, file information, the compressed data, bit rate type and
other information.
◊
MP3s do not have a bit depth per say, more of a dynamic
compression of data based on an analysis of the sound. The result
is a high compression rate.
◊
For comparison, bit rate for uncompressed audio (CD) recording is
44.1kHz sampling x 16 bits x 2 channels (1411.2 kbps).
ADC 1.14
Digitizing Voice
◊ Human voice for applications such as telephone
conversations need not be sampled at a similar
rate and bit depth as music.
◊ Typically, 8 bits at 8 kHz sampling rate is used
(64kbps).
ADC 1.15
ASYNCHRONOUS ADC
ADC 1.16
Asynchronous ADC
◊ ADCs can be constructed from comparators.
◊ A comparator is an op amp configuration where
the voltages of two inputs are compared.
◊ If the “+” input is greater than the “-” input, the output
is a logic high.
VDD
ADC 1.17
Comparator-Based ADC
VDD
Analog In
Digital Out
2-bit “weighted” ADC
ADC 1.18
Flash ADC
VDD
Analog In
Priority
Encoder
Digital Out
Enable
3-bit Flash ADC
ADC 1.19
Flash ADC
◊ Flash ADCs are very fast and can convert data at
high frequencies.
◊ The major disadvantage to flash ADCs is the
complexity of the circuits.
◊ One op amp is required for each output value (minus one
for all zero). This means that:
◊ an 8-bit Flash ADC requires 255 op amps
◊ a 12-bit Flash ADC requires 4095 op amps
◊ a 16-bit flash ADC requires 65,535 op amps
ADC 1.20
Hold/Store
◊
ADCs need to store measured values between the sampling
pulses.
◊
◊
Converting a stored digital signal back to analog requires a
similar clocking frequency (time needs to be reproduced). AD
conversion represents a series of values at specific instances of
time.
The Sample and Hold creates the output “ladder” effect.
ADC 1.21
Flash ADC with Sample/Hold
VDD
Analog In
Priority
Encoder
Latch
Circuit
Clock
Digital Out
Digital Out
Enable
3-bit Flash ADC
ADC 1.22
Basic ADC
◊ A basic ADC contains:
◊ differential analog inputs (VREF) for
scaling
◊ Analog signal input (VIN)
◊ Output Enable for tristate-able outputs
(OE)
◊ Start of Conversion input (SOC) to
trigger the analog signal read cycle.
◊ End of Conversion output (EOC) to
indicate that the conversion is
complete, the data is on the data bus
and a new input may be applied.
◊ Digital output (D0~D7).
ADC 1.23
Dual Slope ADC
◊ Also known as Counter-Ramp or Digital Ramp ADC
◊ A dual slope ADC is commonly used in
measurement instruments (such as DVM’s).
ADC 1.24
Dual Slope
VReference
Capacitor Cycle
Counter Cycle
Charge
Counts from 0
to max
Count Reset
Discharge
Count
Display
Count
Display
Max Count /
Restart Count
ADC 1.25
Dual Slope Function
◊ The Dual Slope ADC functions in this manner:
◊ When an analog value is applied the capacitor begins to
charge in a linear manner and the oscillator passes to the
counter.
◊ The counter continues to count until it reaches a
predetermined value. Once this value is reached the
count stops and the counter is reset. The control logic
switches the input to the first comparator to a reference
voltage, providing a discharge path for the capacitor.
◊ As the capacitor discharges the counter counts.
◊ When the capacitor voltage reaches the reference voltage
the count stops and the value is stored in the register.
ADC 1.26
Dual Slope
◊ The Dual Slope method takes time for the
conversion to occur. Each additional bit improves
resolution but also adds a significant bit to the
counter, costing considerable time. This type of
ADC is therefore unsuitable for rapidly changing
analog input.
◊ Each clocking pulse increments the counter by
one. It takes (2N-1) clock cycles times the clock
period for an output to be produced.
ADC 1.27
Dual Slope
◊ The Dual Slope method is accurate and requires
less circuitry than other methods. Since it uses
the same clock input for both phases of
conversion, a drift in the clocking frequency will
not affect the accuracy of the output.
◊ The Dual Slope is best suited for applications
where the measured value is relatively stable such
as DC voltage measurements.
ADC 1.28
Successive-Approximation ADC
◊ The Successive-Approximation ADC is one of the
most popular types in use today. It has a
relatively simple configuration and an excellent
conversion rate.
◊ If an IC such as a microcontroller has an
integrated ADC, it is typically this type.
ADC 1.29
Successive-Approximation ADC
SOC
Input
Oscillator
Voltage
Comparator
EOC
Control Logic
Digital to
Analog
Converter
Approximation
Register
Output
Register
Digital Output
ADC 1.30
Successive-Approximation ADC
◊
The SAC ADC functions in this manner:
◊
◊
◊
◊
The approximation register is reset to all zero.
When a voltage is applied to the input the approximation
register’s most significant bit is changed from a 1 to a 0. The
digital output of the register is converted back to analog through
the DAC and is compared to the applied analog voltage. If the
value is too low the 1 is left at the MSB. The next MSB is
incremented, the output converted to analog and again compared
to the analog input. Each bit is successively incremented and the
output value compared.
If the voltage from the DAC becomes higher than the applied
analog value the bit is reset to 0 and the next MSB is incremented
and compared.
The process continues in this manner until the LSB value is
reached. At the LSB, if the applied value makes the DAC output
voltage higher the bit is reset to 0. The ADC has completed its
process. It stores the value to the output register and provides an
EOC output to indicate there is a value in the register.
ADC 1.31
SAC ADC Conversion Time
◊
Theoretically each step in the comparison process takes a clock
edge. It therefore takes a SAC ADC approximately the same
amount of clock edges as the number of bits it handles.
If a SAC ADC has an output of 8 bits and an input clocking
frequency of 500 kHz, it takes approximately:
8•(1/500kHz) = 8•2µs=16µs
◊
In actual practice it may take more than one clock edge per
step, but this is still faster than some other methods.
ADC 1.32
The ADC08
◊ The ADC08 family is a relatively popular SAC ADC.
VDD
+VIN
D0 to D7
-VIN
Digital Output
Vref/2
CS
RD
ClkOUT
WR
ClkIN
INTR
GNDAnalog
A
GNDDigital
D
ADC 1.33
ADC08
◊
+VIN and –VIN : Differential analog voltage.
◊
Vref/2 : Used to change the input voltage range. Normally at
2.5V when VDD = 5V, if 1.5V is applied the input range is 3.0
Volts and the resolution is changed accordingly.
◊
ClkIN: Input clock. External clocking edges can be provided
to the ADC.
◊
ClkOUT: Output Clock. This ADC has an internal clocking
circuit that requires external connection to an RC.
T = 1.1RC
Typical values: 10kΩ & 150ρF
ADC 1.34
ADC08
◊
CS’ : Chip Select (input), tri-states the digital output for bus
applications
◊
RD’: Read enable (input), enables the output from the
Approximation register to the output register.
◊
WR’: Write enable (input), used to request the start of a new
conversion.
◊
INTR: Interrupt, output high when the ADC is in the process
of converting an input. Used to signal microprocessors or
microcontrollers. Conversion time is approximately 100µs.
ADC 1.35
ADC08
◊ Questions:
◊ What is the purpose of two grounds?
◊ How would the device be configured for an input of:
◊ 0 to 5 Volts
◊ 0 to 3 Volts
◊ -2.5 to +2.5 Volts
◊ What is the purpose of the Vref/2 input?
ADC 1.36
SAC ADC
◊ The SAC ADC is a fast, accurate device.
◊ It has few disadvantages over other methods.
◊ It requires a high clock frequency as each bit takes up an
edge.
◊ It is not overly complex, and can be scaled more easily.
ADC 1.37
Other Methods of ADC
◊ Up/Down Ramp
◊ Similar to the Ramp ADC but instead of starting the
count at zero for each measurement it increments the
counter up or down based on the direction of the analog
input. The time it takes for conversion is variable.
◊ Voltage-Frequency ADC
◊ This device utilizes a Voltage Controlled Oscillator (VCO).
The higher the input voltage the higher the output
frequency of the VCO. The ADC enables the output of
the VCO for a specific period of time and the output of
the oscillator is connected to a counter. The result of the
count is supplied as the digital value.
ADC 1.38
Other Methods of ADC
◊ Sigma/Delta
◊ Creates an internal bitstream based upon the difference
(Delta) between the analog input and a feedback
comparator. The running total (Sigma) is reported as
the digitized value. Operates at a high internal
frequency.
See http://www.analog.com/Analog_Root/static/techSupport/designTools/interactiveTools/sdtutorial/sdtutorial.html
for more information.
ADC 1.39
Errors
◊ Analog to Digital systems have several possible
sources of error.
◊ The errors can be the result of:
◊
◊
◊
◊
Quantization
Noise
Slew
Under-Sampling
ADC 1.40
Quantization Error
◊ Quantization errors are a normal occurrence for
ADCs. The input voltage will commonly have a
value that is between the LSB voltage step and
must be rounded up or down.
◊ For instance, if the resolution is 20mV per bit and
a value of 10 mV is applied, the ADC will interpret
this as either 0v or 20 mV, depending on the ADC
design.
ADC 1.41
Quantization
◊ There are several additional quantization issues:
◊ If the total number of bits representing the amplitude of
the signal is too small the quantization error increases.
◊ If the voltage applied to the ADC is not properly
compensated then the full range of output values is not
used.
ADC 1.42
Quantization and Sampling
◊ Sampling-Related Quantization Errors
◊ Relatively low sampling frequencies can lead to
quantization errors
◊ If the sampling frequency is slightly different from the
analog signal’s frequency (or a harmonic..a factor of the
frequency), then the values may not indicate all the
peaks and valleys, leading to other harmonics (also see
aliasing).
ADC 1.43
Not quite perfect...lower sample rate
Above: original signal with samples
Below: imperfections in the digital output quantization
ADC 1.44
Not quite perfect...higher sample rate
ADC 1.45
Noise
◊ Noise is a factor in analog communications and it
causes problems in ADC.
◊ The sources for noise are varied and can be
somewhat complex.
◊ Sources include:
◊ External Sources
◊ Internal circuit sources
◊ Aliasing
ADC 1.46
Aliasing
◊
Higher quality ADC (and DAC) will contain anti-aliasing filters
to remove frequencies that are aliasing (“posing”) as
signal.
◊
Aliasing occurs when additional frequencies are inadvertently
produced in the AD process.
◊
For instance, if there are some frequencies present in the
original signal that are higher than the Nyquist sample rate,
a different lower frequency may appear on the conversion.
◊
Aliasing also occurs with adequate sampling frequencies.
ADC 1.47
Aliasing due to Inadequate Sampling Rate
ADC 1.48
Aliasing with an Adequate Sampling Rate
ADC 1.49
Noise Sources
◊ Analog values are susceptible to noise (an
advantage of digital electronics!). This noise
causes problems for the AD conversion, as the
noise will be included in the conversion process.
◊ There are external and internal sources of noise.
◊ External: use filters and proper techniques to minimize
noise (such as decoupling, isolation, etc)
◊ Internal: switching noise occurs with AD converters.
Use filters, decoupling and ground isolation. Thermal
noise (aka white noise) is caused by a variety of reasons.
◊ Other solutions include processing the signal (DSP),
injecting additional white noise to help average out
the noise and ignoring the LSB.
ADC 1.50
Filter
◊ The analog input to an ADC should contain a lowpass filter to ensure high frequencies cannot pass
and create aliasing problems.
◊ Advanced filtering techniques are also employed.
These include DSP (Digital Signal Processing) to
mathematically analyze and adjust the digital
values.
ADC 1.51
Gain Error
◊
A Gain Error occurs when the ADC produces an output at a
different quantization level than desired. The output is
linear but the steps are either larger or smaller than they
should be. The resolution is poor and there is the possibility
of clipping (the binary number stops changing with a change
in analog input). The voltage values will be inaccurately
represented.
Red: Ideal
Blue: Error
Green: Error
Binary Code
Input Volts
ADC 1.52
Gain Error
Typical Causes:
•wrong VDD/VEE/Vreference
voltage to the ADC
•Improper Analog input
voltage
Red: Ideal
Blue: Error
Green: Error
ADC 1.53
Offset Error
◊ An Offset Error occurs when the output has the
same voltage per step but the starting voltage is
different.
Red: Ideal
Blue: Error
Green: Error
Binary Code
Input Volts
ADC 1.54
Offset Error
Typical Causes:
•Improper ground reference
•Improper analog voltage
range
Red: Ideal
Blue: Error
Green: Error
ADC 1.55
Non-Linearity: Non-Monotonic
◊ A Monotonic Error occurs when the individual
voltage steps are non-linear.
Binary Code
Red: Ideal
Blue: Error
Input Volts
ADC 1.56
Non-Linearity: Non-Monotonic
Typical Causes:
•defective ADC
•wrong outputs measured
•significant noise on ac input
Red: Ideal
Blue: Error
ADC 1.57
Slew
◊ Slew represents the maximum rate of change of a
signal. Slew rate is measured in V/t.
Black: ideal binary step from ADC
Grey: binary step from ADC with slew
ADC 1.58
Analog Sample & Hold
◊ The AD conversion process may be adversely
affected when an analog signal changes voltage
during conversion.
◊ Analog sample & hold circuits are able to capture
the analog value and retain it, allowing the ADC to
perform the conversion without input changes.
Digitally
controlled
switch
Analog In
Storage
Capacitor
Analog
to ADC
ADC 1.59
Specification Sheet
◊ View the specification sheet for the ADC08 from
National Instruments.
◊ Definitions:
◊ Common-Mode noise is noise that occurs on both
conductors at the same time.
◊ Ratiometric means measures the ratio, or takes ratio into
account when measuring.
ADC 1.60
DSP
◊ Digital Signal Processors perform operations on
the digitized signals including:
◊ removing interference, noise and other frequency and
transmission effects
◊ encoding the values for transmission
◊ reconstruct signals by enhancing or decreasing specific
frequencies
◊ DSPs are complex devices that rely on
programming code (algorithms) to perform
operations.
ADC 1.61
ADC-DAC with DSP
Filter
ADC
DSP
DSP
DAC
Filter
ADC 1.62
DSP
◊ Analog signals contain noise. Noise can be filtered
but sometimes the filtering process can eliminate
wanted frequencies.
◊ The better means of removing noise is to take an
average, as illustrated below as the red line.
Analog Signal with Noise
ADC 1.63
END ADC1
©Paul R. Godin
prgodin°@ gmail.com
ADC 1.64