I have a simple question, but sadly I'm kind of "noob" in signals theory. A signal having 4 harmonics at the following frequencies: 1 kHz, 2 kHz, 3.5kHz and 4.2 kHz. (How can a signal have harmonics "without" a fundamental frequency?) Find the minimum sampling frequency so that the signal can be completely recovered from its sampled version.

I think I should fint $f_s = 2 f_{max}$ but I do not know how to find $f_{max}$.

Please somebody help me.

  • $\begingroup$ Do you need additional answers? $\endgroup$ – Laurent Duval Dec 30 '18 at 22:01

This exercise is a verbatim copy from 5.C.1. in document Lab 5: SAMPLING OF SIGNALS, dating from 27/11/2017. This document apparently contains a "lecture" part, that "could" help you finding the answer. Unfortunately, some assertions in this document belong to folk signal processing, such as (5.4):

The sampling frequency ($f_p$) must be greater than twice the highest frequency of $x(t)$: $f_s > 2 f_{\max}$

There is a typo in the text, as $f_p$ and $f_s$ refer to the same sampling frequency concept. The "must" part is not true. Anyway, there are three levels of possible answers. To start with, an harmonic is a common name for a sinusoidal signal defined by a frequency, an amplitude and a phase. However, a signal with harmonics, i.e. a signal containing multiples of a fundamental frequency $f_0$, i.e. $kf_0$, $k\in \mathbb{N}^*$ can exist without the fundamental at $k=1$ (if I remember well). You can read (and hear) more at Physics of music, notes: The missing fundamental or The Well-Tempered Timpani, In Search of the Missing Fundamental: The Missing Fundamental.

Now, let us proceed with the answers.

  1. The most probably expected answer is twice the maximum frequency among the four harmonics, which is the basic Nyquist
  2. A more involved form considers the frequency span, between 1 kHz and 4.2 kHz, as you can further reduce the rate using an $f_{\min}$-$f_{\max}$ diagram, eluded to in Confusion regarding Nyquist Sampling Theorem or here, from Vaughan et al., 1991, The theory of bandpass sampling: Vaughan, The theory of bandpass sampling
  3. The last I can think of further use the fact than only four harmonics are considered, thus the signal is sparse in the Fourier domain, and compressive sensing can offer even lower rate sampling.
  • 1
    $\begingroup$ missing fundamental is (was) a hot topic in audio engineering, especially for commerical electronics such as TV sets, where due to space constraints loudspeaker enclosures are small and cannot produce low frequencies (50-150 Hz) adequately. As a result, bandwidth extension techniques using psychoacoustic phantoms were proposed to compansate for the physical loss of the missing fundamentals. Philips labs had a number of algorithms and devices (that I know) for this. @MattL. should also know about ;-) . May be he could share some of that expertise with us here on DSP.SE ? $\endgroup$ – Fat32 Nov 29 '18 at 22:32

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.