At 06:16 PM 9/22/99 -0400, Mario wrote:
>Harold's point is a good one. I will just add that the explicit reason
>why one may choose not to implement an anti-aliasing filter in such a
>case is because one would have no intention of reconstructing the
>orginal signal. After all, the reconstruction of the original signal was
>the context in which Nyquist devised the sampling criterion.

There are basically only two reasons why you want to include an
anti-aliasing filter:

1. You are under-sampling the signal bandwidth, so there's signal power in
frequencies above 1/2 the sampling rate. If you don't anti-alias, you will
fold down the signal frequencies into lower harmonics, converting signal
power above the Nyquist frequency into noise. An example of this would be,
for example, audio input signal frequencies beyond 20 kHz, which can't be
heard by the human ear. This is why CDs sample at 44 ksps (2x 20kHz +
margin) and anti-alias the input to the ADC. (This is what Harold and Mario
and others are discussing.)

2. You are sampling a signal with wide-band noise added. If you don't
restrict the bandwidth, the noise glitches (which may be rail-to-rail in
size) will get sampled as the signal, and you'll get ADC output glitches
with a significant number of bits in error. By limiting the bandpass to the
ADC input, you reduce the noise dramatically. (This is what I've been
discussing.)

The only reason you would not filter the input to an ADC would be that the
input has the following characteristics:

a. Signal bandwidth very low (e.g., DC voltages, or 10x over-sampling).
b. No noise present.

Usually it's (b) that drives you nuts in the end. There's plenty of noise
sources present in a digital system with wires and sensors.

A good design rule is to never allow more bandwidth than you really need.

================================================================
Robert A. LaBudde, PhD, PAS, Dpl. ACAFS  e-mail: ral@lcfltd.com
Least Cost Formulations, Ltd.                   URL: http://lcfltd.com/
824 Timberlake Drive                            Tel: 757-467-0954
Virginia Beach, VA 23464-3239                   Fax: 757-467-2947

"Vere scire est per causae scire"
================================================================