Know Them by
Their True Names
The transducer signals you measure are what you
know about what you test. When these signals
are digitized, however, the sampling process creates
aliases—out-of-band signals disguised as in-band
signals of interest. At once inherent and insidious,
you won’t even know they are there in your digitized
data, rendering it invalid. Low-pass filtering prior to
A/D conversion will remove aliases, but not all filters
are created equal. Precision Filters offers the sharpest
filters on the market, with superior roll-off to maximally
preserve the signals you need and reject the aliases
you don’t. Available with all of Precision’s signal
conditioning products.
Accept no aliases.
Know your signals by their true names.
The Precision 28000
Signal Conditioning
System
Transducer Conditioning Systems Filter/Amplier Systems Signal Switching Systems
Learn more at www.pfinc.com, call 607-277-3550 or email sales@pfinc.com
AEROSPACETESTINGINTERNATIONAL.COM // SHOWCASE 2020 171
4
1002644_V3_ATI Supp_Precision Filters Inc.indd 1 20/09/2019 11:37
The cutoff and sampling frequencies may be set as
follows: 1) The filter cutoff is set so that the highest
frequency in the signal, fx, is attenuated by not more than
X dB. 2) Referring to Figure 4, the sampling frequency is
set high enough to permit the filter to attenuate the first
alias of fx, fy = FS - fx, by at least Y dB. The sampling
frequency FS may be expressed in terms of fx and its first
alias as: FS = (fy/fx +1)fx.
As an example, we assume fx = F3dB, the filter cutoff
frequency, and that 80 dB minimum attenuation of its first
alias, fy, is required. We define F80dB as the frequency of the
filter where 80 dB of attenuation is reached. With these
assumptions, the equation for setting the sampling
frequency becomes: FS = (F80dB/F3dB +1)F3dB.
Values of F80dB for various filters are provided in Table 5.
The resulting sampling frequency when using an 8-pole
Butterworth (BU8) filter would be FS = 4.16*F3dB. Similarly,
for the sharper Precision Filters LP8F under the same
constraints, FS = 2.75*F3dB, which is a 34% reduction in
sampling rate compared to the BU8. The Precision Filters
LP6F 6-pole filter results in FS = 3.61*F3dB, a 13% reduction
in sampling rate when compared to the BU8.
CONCLUSION
Aliasing is an inescapable attribute of A/D conversion and
of digital data acquisition generally. Without proper
intervention, digital data – without any indication – can
easily be rendered meaningless. Barring advance
knowledge of the full spectrum of the analog signal, the
only strategy by which in-band signal corruption by
aliasing may be prevented is by low-pass, anti-aliasing
filtering prior to the A/D converter. Such filtering must be
carefully set both to minimize attenuation of pass-band
signals of interest and maximize attenuation of out-ofband
aliasing frequencies. Further, the sharpness of the
applied filter – as with those offered by Precision Filters –
has a direct effect on both the quality of in-band signal
fidelity as well as the required sampling frequency. \\
Douglas Firth is president of Precision Filters
5
/www.pfinc.com
/AEROSPACETESTINGINTERNATIONAL.COM
link