Download - Microchip

Do I Filter Now, Later, or
Never?
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 1

Analog vs Digital Filters
Do I Need an Analog Filter if I am Only
Measuring a DC Signal?
YES - Noise Generators Have no “Knowledge” of Your
Intent
Do I Need an Analog Filter as Opposed to
Digital?
YES - Once the Converted Signal is “Contaminated”
with Aliased Noise, it Takes Time and Memory to
Eliminate (if possible)
YES - Eliminates the Effects of Overdriven Signals that
Usually Occur Beyond the Bandwidth of the Filter
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 2

When to Pick Analog Over
Digital
Analog Filters
Reduces Fold Back Signals (Aliased Signals)
Minimizes Impact of Noise Peaks
Digital Filters
Can Reduce in Band Noise
Removes Noise Injected During Conversion
Easy to Program
Analog
Input
Signal
Analog
Low Pass
Filter
A/D
Conversion
Digital
Filter
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 3

Analog -
Digital -
Butterworth,
Chebyshev,
Bessel, etc.
Applying Filtering Blocks
Infinite Impulse Response (IIR)
Moving Average
Finite Impulse Response (FIR)
Where do you make the cut
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 4

Frequency Response
5th order Butterworth 5th order Chebyshev 5th order Bessel
Step Response
5th order Butterworth 5th order Chebyshev 5th order Bessel
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 5

Nyquist Theorem
Harmonics that Fold Back into the Conversion
Output
Analog Input
N = 0
N = 1
N = 2
(1) (2)
(3)
N = 3
Sampled Output
Representation
N = 0
(2)
(3)
(1)
f S
/2 3f S
/2 5f S
/2
0 f S
2f S
3f S
0 f S
/2 f S
f ALIASED = |f IN -Nf S |
Find N by making f ALIASED < f s / 2
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 6

Analog -
Butterworth,
Chebyshev,
Bessel, etc.
Applying Filtering Blocks
Digital -
Infinite Impulse Response (IIR)
Moving Average
Finite Impulse Response (FIR)
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 7

x(n)
IIR Filter
b0
y(n)
- a 1
z --11
b 1
z --11
- a 2 b2
Direct From II Realization
can implement functions such
Bessel, Butterworth, Chebyshev, etc.
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 8

Analog
Input
Levels
Full-scale
0V
Digital Filters Using
Averaging
Delta sigma
modulator
1-bit data
1-bit data streams
1/2 full scale input 1/4 full scale input 3/4 full scale input
1 1 1
0 Average 0 Average 1 Average
1 = 0.5 0 = 0.25 1 = 0.75
0 0 0
1 1 1
0 0 1
1 0 1
0 0 0
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 9

Oversampling improves SNR
Oversampling and Signal to
Noise
Each oversample by a factor of 4 gives a 6 dB or
1-bit
improvement in SNR
For a 1-bit 1
ADC and oversampling there are:
2-bits for 4 x oversampling (4 1 )
3-bits for 16 x oversampling (4 2 )
4-bits for 64 x oversampling (4 3 )
.. .. .. ..
.. .. .. ..
24-bits for 4 x oversampling !!!!!
Oversampling to 24-bits from a 1-bit 1
ADC is not practical
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 10

Magnitude
1.5
1
0.5
0
-0.5
-1
-1.5
Low Pass Digital Filter (FIR)
Time
Input
w(n)
x(n)
Unit
delay
x(n-1)
x(n-2)
x(n-m)
Unit
delay
Unit
delay
w(n-1)
w(n-2)
w(n-m)
Output
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 11

Sinc 3 Frequency and Step
Response
Frequency Response
Step Response
0
3rd order FIR
3rd order FIR
-40
Attenuation (dB)
-80
-120
-160
0 50 100 150 200 250
FIR Filter Characteristics at 60Hz
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 12

Gain
The Best Design Approach
Mux
LPF
µC
µC
Piecewise
Linearization
Digital
Filter
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 13

Conclusion
Analog Filters
Locate Before the Controller
Reduces Fold Back Signals (Aliased Signals)
Minimizes Impact of Noise Peaks
Digital Filters
Locate in the Controller
Can Reduce InbBand Noise
Removes Noise Injected During Conversion
Easy to Program
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 14

References
Application Notes
ADN003, “Select the Right Operational Amplifier for your Filtering Circuits”,
Bonnie C Baker
Books
AN699, “Anti-Aliasing, Analog Filters for Data Acquisition Systems”, Bonnie
C Baker
AN737, “Using Digital Potentiometers to Design Low Pass Adjustable
Filters”, Bonnie C. Baker
AN852 -“Implementing FIR and IIR Digital Filters Using PIC18
Microcontrollers”, Ananthma Ramu
AN616 - “Digital Signal Processing with the PIC16C74”, Darius Mostowfi
Analog and Digital Filter Design, Second Edition by Steve Winder, Newnes
The Scientist and Engineers Guide to Digital Signal Processing, Steven W
Smith, WWW.DSPguide.com
Feature Articles
EDN, , February 20, 2003, Filtering? Before or after?
FilterLab TM Active Filter Design Software, Microchip
© 2004 Microchip Technology Incorporated. All Rights Reserved. Filter Now, Later, or Never 15