Lucien, si-list, I have followed the responses to Lucien's question with interest. I do understand the problems he is facing, text books are often ambiguous. The term "modeled" is indeed a bad choice. First note that we only know of functions or signals in the time-domain. Although one often use the term frequency domain, one should be aware of the fact that sinoids are time-domain functions. However, one of their interesting and very useful engineering attributes is their frequency. So engineers and scientists use it all the time in their work. Time domain signals are but an extension of vectors, albeit that continuous signals are vectors of an infinite dimensional nature. Consider first a 2-dimensional vector V in a rectangular coordinate system. We can decompose V into two components along the 1x and 1y unit basis vectors by means of the scalar or dot product: V=a1*1x + b1*1y, where a1 and b1 are the elements of the vector under the rectangular basis. Now if we take another coordinate system (e.g. cylindrical) then the same vector is written V=a2*1r+b2*1phy. Here 1r and 1phy are the cylindrical unit vectors (radius and angle). Using linear algebra we can find out how the unit vectors are related (and transform) so that we can find a relation between the vector elemens (a1, b1) and (a2,b2). So the same vector has two different representations, but both are equal and dependent on the chosen basis. You might wonder were this discussing is heading for. Actually it is quiet simple, but it requires a bit of abstract thinking. As vectors need unit basis vectors for a proper representation one can also state that a signal S needs (possibly an infinite number of) basis functions for a proper representation. For example we can state that a signal S uses sinoids as basis functions and the linear sum of weigthed (amplitude) sinuoids (frequency & phaze) is just the signal S, just as the sum of the basis vectors make up the vector. There exist many useful basis functions: delta functions, polynomials, bessel functions, pulse functions, triangular .... Obvious basis functions need to fullfil some fundamental requirements (orthogonality) to be useful as a basis function. Not all functions have the same capabilities, for example, sinoids cannot be used to represent signals with discontinuities as sinusoids have continuous derivatives, because even an infinite sum of sinusoids cannot represent a jump. This leads to Gibs phenomena. In science and engineering is the choice of the basis functions important because it can simplify the solution of a problem. The answer to your question: "Is a square wave physically composed of harmonic sinusoids?" is negative. It depends on the chosen basis functions used for the signal representation! Physics is based on isomorphism (translation=same shape), that is, we have models for the "real world" that behave the same as the real world (have the same shape --> are isomorph). Take for example some voltage in an electric circuit. We can represent it as an infinte set of real numbers from simulations and compare this to a finite set of measured real numbers. Often both sets agree well. Because we can use whatever basis to represent the simulation or measured set we cannot state that the signal is physically composed of sinoids or other basis functions. If one uses a frequency domain measurement tool then one projects the measured signal onto sinoid basis functions (one-by-one) to extract its amplitude (and possible phase). The electrical circuits used for that are narrow band filters that just ring to a specific basis function frequency. In the time-domain one uses a gate- or sampling function to approximate a delta function. The gate function projects the signal onto the delta function basis functions to extract its amplitude, which is easily displayed on a screen. It does not make sense to state that a "real world" signal is physically composed of sinuoids. First we need to state which basis functions one uses, which allows us to extract a set of real numbers (by projecting). We can use whatever basis functions and extract a set of real numbers, that multiplied by the basis functions, and summed represent the signal. All representations give equal results. No basis funtion is better or more fundamental than others. I hope that this answers somewhat your question. It is about perception. Currently I am at my desk looking at a picture of my 8 month old son. If I change position he looks different, even though it is the same little guy. It is all about how you look at things. The same applies to the frequency and time domain representation of square waveforms, your looking from a different perspective at the same thing! See you, Jan Vercammen Agfa-Gevaert Belgium lucien_op@xxxxxxxxx@freelists.org on 11/26/2001 10:59:32 PM Please respond to lucien_op@xxxxxxxxx Sent by: si-list-bounce@xxxxxxxxxxxxx To: si-list@xxxxxxxxxxxxx cc: Subject: [SI-LIST] Square wave harmonics I'm an undergrad at UW, Seattle. I have a question concerning square waves. My signals text book says, a square wave is modeled by a sum of harmonically-related sinusoids (the Fourier series). Mark the word "modeled." Another source uses the term "represented." Recently, I've been told at my workplace by several senior engineers that a physical square wave is PHYSICALLY composed of harmonics. In other words, they say that the Fourier series is not just a mathematical tool describing square waves, but is indeed an accurate description of the physical square wave. They tell me all physical square waves contain harmonics. The two ideas above seem in conflict. My undergraduate brain is growing frustrated, and all I can conclude with certainty is that a square wave BEHAVES AS a Fourier series, regardless of how it is created. I know from reading HP manuals for signal/pulse generators that these devices do not build square waves by adding sinusoids. So in my mind, it doesn't seem possible that these square waves can contain harmonics. As for how a spectrum analyzer gives Fourier Coefficients I have no idea. I don't know if it just calculates and displays the Fourier coefficients, or if it actually detects physical harmonics within a signal and displays their magnitudes. Can anyone give me the low-down on square waves! Basically, my question is: In our physical reality, do square waves contain harmonics? Or does the idea of square wave harmonics only exist on paper as a mathematical model, used to PREDICT the natural behavior of the square wave? Lucien Opperman Seattle ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field or to administer your membership from a web page, go to: //www.freelists.org/webpage/si-list For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List archives are viewable at: //www.freelists.org/archives/si-list or at our remote archives: http://groups.yahoo.com/group/si-list/messages Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field or to administer your membership from a web page, go to: //www.freelists.org/webpage/si-list For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List archives are viewable at: //www.freelists.org/archives/si-list or at our remote archives: http://groups.yahoo.com/group/si-list/messages Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu