1/16/2014

17.2 – Sample Aliasing

Analog to Digital Conversion
MABE 345 • Analog signals are continuous in time and
output.
Lecture #2 • Digital Signals are broken into discrete “steps”
Chapter 7 Conversion Issues – How many of these “steps”
are required to accurately represent an analog
Sampling, Digital devices signal? This depends on:

and Data Acquisition 1) Frequency content of measure analog system

2) Increment size between each discrete sample
3) The total sample period
4) Behavior of signal to be measured (periodic?)

Sampling Rate and Aliasing

Sample Rate
1
fs =
dt
Aliasing
If an insufficient sample rate is used the signal
measured will differ from the actual signal.

Nyquist Sampling Theorem

A sample rate of at least two times the
maximum signal frequency is require to
prevent aliasing.
f smin > 2 f m
fs
Figure 7.1 from “Theory and Design for Mechanical Engineers” R.
Figliola and D. Beasley
Nyquist _ Frequency = f N =
2

Determining Alias Frequencies

Figure 7.3 Folding Diagram from “Theory and Design for Mechanical
Engineers” R. Figliola and D. Beasley

1
 1/16/2014

Using the Folding Diagram Using the Folding Diagram

Example

Actual Signal to be Measured

fsignal = 168 Hz

Sample Frequency
fs = 200 Hz
Figure 7.3 Folding Diagram from “Theory Figure 7.3 Folding Diagram from “Theory
and Design for Mechanical Engineers” R. and Design for Mechanical Engineers” R.
Figliola and D. Beasley Nyquist Frequency Figliola and D. Beasley

• fN (fs/2) is the maximum fN = fs/2 = 100 Hz

frequency that can be
represented (measured). Result
*Divide fsignal by fN to get fsignal/fn = 1.68
• If the signal is a higher * find this point on the folding diagram and drop straight down to get 0.32
frequency, this will result in an * So, our output (what we read) will be 0.32*fN = 32 Hz, so fmeasured = 32 Hz
aliased signal with an output <fN. Due to low sampling rate, we measure a wave with f = 168 Hz as f = 32 Hz!!!

7.3 Digital devices • We use different combinations of bits to form words

representing decimal (base 10) numbers.
Systems use:
Word Combinations Decimal Integers
2 bit 22 = 4 0, 1, 2, 3
• discrete steps in time and amplitude.
4 bit 24 = 16 0 to 15
8 bit 28 = 256 0 to 256
• binary system 16 bit 216 = 65,536 0 to 65,535
- Base 2 system. It is on or off.
- A digit or bit is the smallest unit of measure (0, 1) • Conversion from Binary to Decimal
- A word is a collection of bits used to express a number. a) Straight Binary Code (5 bit example)
- An 8 bit word is a byte.
1 0 1 0 1
• a physical location used to store a word is a register. bit 4 bit 3 bit 2 bit 1 bit 0
1 x 24 = 16 0 x 23 = 0 1 x 22 = 4 0 x 21 = 0 1 x 20 = 1
16 + 0 + 4 + 1 = 21

b) Binary Coded Decimal

- In this system, each digit of a decimal

1 0 number is individually coded into binary.
bit 1 bit 0
1 x 21 = 2 0 x 20 = 0 - EX. 532 is represented by
2+0=2 0101 0011 0010
(5) (3) (2)

Image from www.ThinkGeek.com

2
 1/16/2014

7.4 Transmitting Digital Numbers 7.5 Voltage Measurements

Analog to Digital Converter
• Binary code is signaled using voltage
- Analog Side : EFSR (Full Scale Range)
“switches”
Ex. 10 V
• High voltage represents 1 (on) and low 0 (off)
- Digital Side: 2M binary numbers
Ex. 8 bit (256 steps)
- Resolution

E FSR 10V
Q= = 8 = 0.03906 "count
V
"
2M 2
Saturation
If a signal that is above or below the EFSR is
measured it will be converted at the limit
Figure 7.5(b) 4-bit register from “Theory and Design
for Mechanical Engineers” R. Figliola and D. value.
Beasley
Ex. 11 V will give the same reading as 10 V

A/D Errors A/D Errors

1) Quantization Quantization (cont.)
Error between the actual voltage and the indicated
voltage. Ex. 3 V at 22 give outputs of 0, 1, 2, 3 Signal-to-Noise Ratio (SNR) due to Quantization
Consider the power of the signal given by Ohm’s law
Ex. 0-4 Volt board with a 0-10 “count” output. E2
R
and the power that can be resolved by the converter
E2
R2 M
(where M is the number of bits)
then the ratio of these in terms of decibels is
SNR[dB ] = 20 log(2 M )

A/D Errors
2) Saturation Error
Error associated with exceeding the limits of the A/D
converter.
Ex. Using the above, 5 V will be converted as 3 V
for an error of 5 - 3 = 2 V

3) Conversion Error
- A/D Problems such as - settling time, signal noise,
temperature effects, power fluctuations, etc.

- Yield errors like hysteresis, linearity, zero drift,

repeatability