Scribd
Upload
13 views2 pages

Understanding Digital Frequency Ω Explained

The document defines and compares digital frequency Ω to angular frequency ω. Digital frequency represents the progression of an underlying physical signal's oscillation from one discrete digital sample to the next. It is calculated as Ω = ωT, where ω is the angular frequency and T is the sampling period. Using digital frequency allows signal processing techniques to be applied uniformly regardless of sampling rate. Digital frequency normalizes the physical signal's frequency to the sampling rate.

Uploaded by

john al
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
13 views2 pages

Understanding Digital Frequency Ω Explained

The document defines and compares digital frequency Ω to angular frequency ω. Digital frequency represents the progression of an underlying physical signal's oscillation from one discrete digital sample to the next. It is calculated as Ω = ωT, where ω is the angular frequency and T is the sampling period. Using digital frequency allows signal processing techniques to be applied uniformly regardless of sampling rate. Digital frequency normalizes the physical signal's frequency to the sampling rate.

Uploaded by

john al
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

The Big Ω: What the heck is digital frequency anyway?

This tech note concerns the definition and usage of the digital frequency Ω and compares/contrasts
to the angular frequency ω.

A signal x(t) representing some physical quantity (river height vs. time, electrode voltage vs.
time, strain due to gravity waves, whatever) is sampled at regular intervals spaced T apart. We
call T the sampling period . Thus, the corresponding sampling frequency is given by fs = 1/T .
The discrete times at which we sample the signal are given by
n
tn = nT = where n = 0, 1, 2, . . .
fs

So we can think of n as the integer index keeping track of time.

Figure 1: An oscillating signal x(t) bobs up and down with frequency fo signal period To . The
digital sequence x[n] is sampled at discrete times tn = nT , where T is the sampling period.

Now if our underlying signal is given by:

x(t) = A cos(ωt + φ)

1
and we sample it at discrete times tn , our digital sequence is given by:

x[n] = A cos(ωtn + φ) (1)


= A cos(ω(nT ) + φ) (2)
= A cos((ωT )n + φ) (3)
x[n] = A cos(Ωn + φ) (4)
(5)

Here we have defined the digital frequency Ω = ωT . Note that our digital sequence sequence
x[n] still traces out an oscillating signal as integer time index n advances.

Note also that the digital frequency Ω is the angular frequency ω (rad/s) times the sampling
period T (s/sample), so the units digital frequency are rad/sample. One nice interpretation of
Ω therefore, is how many radians the underlying oscillation progressed through from one digital
sample to the next. For example, if Ω = π/10 rad/sample, the oscillation has advanced only a
“little bit”, namely π/10/(2π) = 1/20 of one full oscillation. By contrast if Ω = π rad/sample, then
the physical oscillation has already undergone one-half of a full oscillation. We’ll see later that this
latter case has profound consequences in signal processing land.

One major reason for using Ω instead of ω is that signal processing techniques fundamentally
take in a sequence of numbers x[n] and operate on them in the same way regardless of the sampling
period. By generalizing sequences to think of time as an integer index, we can generalize signal
processing techniques. For example, our river rock data may be sampled at fs = 40 Hz, while our
ECG data was sampled at fs = 360 Hz. If we want to apply, say, a moving mean filter, the math
will work out more nicely if we use Ω instead of ω.

In essence, we are simply normalizing the frequency of the physical signal to how fast it was
sampled. Specifically:

Ω = ωT (6)
ω
= (7)
fs
2πf
= (8)
fs
f
Ω = 2π (9)
fs
(10)

So the digital frequency is just 2π times the frequency of the actual physical signal f normalized
to the sampling rate fs .

You might also like