[Prev][Next][Index][Thread]

Square wave excitation

Steve;

>Actually, depending on the actual waveform (duty cycle, waveshape,
>etc...), you could have a fairly complicated Fourier spectrum, not
>necessarily just the odd harmonics.

Well now, given an arbitrary waveform, we should be able to have whatever
harmonics we wish, but Chip was specifically referring to a square wave,
which defines the wave shape, and suggests a a duty cycle of 50%, in the
absence of a mention of such. Under these circumstances, we're
looking at only the odd harmonics, as I suggested. We can try making
things more detailed by discussing risetimes, distortion, and non-50%
duty cycles, but I do think that going beyond the harmonics for a normal
square wave will unnecessarily complicate the issue at this time.

As far as their interference in the tuning process goes, they will of
course get in the way, but if it's the only signal generator available,
it'll be better than nothing, and will certainly help. When there's an
unwanted signal messing your data, but the signal is predictable, you
look for methods to compensate and correct the data for it.

>process. Since you would see lots of "mini-resonances" at nominal
>oscillator frequencies lower than the primary resonant frequency as the
>harmonics present in the square wave come into resonance, would the
>fundamental resonance still be clearly visible over and above all the
>harmonic resonances?

There's no reason why not, if you're careful with your data taking. The
fundamental component of the square wave will be significantly larger than
the third harmonic contribution. If you're merely looking for a resonant
peak, it's a simple choice of picking out the biggest one. Also, as you
sweep the frequency upward, I'd expect it to be the last one you see,
as I think you suggested. Since even the quantity of unwanted signal is
known to a fair degree of accuracy, it only takes awareness of its
presence and some thought to deal with it adequately.

There are situations when it's tricky to pick out meaningful data from a
lousy signal, but this one looks like it should be a peice of cake.

Wes B.