My understanding of how demodulation works is that a high pass filter is applied to the waveform in order to remove the normal lower frequency components (eg, imbalance, misalignment, etc). This leaves us with a high frequency waveform which is modulated by the lower frequencies. We then make an FFT from the lower modulating frequency so that we can read it as a spectrum. I can see how this will produce fundamental frequencies in the demod spectrum, but I'm struggling to visualise how this can produce harmonics. Can anyone shed any light on this please?

In my favorite flavor there is also a rectification stage and a filtering sometimes the antialiasing filter is used. In my world this procedure do not make any multiples part from those that exist. What You See Is What You Get and that exist in the input signal but you don´t see it due to scaling by unbalance and such, possibly hidden by the random noise that demod procedure may take away. So in my view it´s not adding, possibly taking away pending your input signal. Compare to a similar range acc raw signal. Or? Olov

How does rectification not create harmonics not present in the original signal?

Al - If the envelope is periodic, but not sinusoidal, then it contains harmonics

So it´s pending the filtering?
I can´t say I have found harmonics in demod IRL
that has I can relate to being created by demod function or maybe my bearing faults and GBX´s all are so old so they only generate sinus? Olov

Rectify a sine wave and you get a harmonic function, which is not periodic, i.e. harmonics.

In fact the derivative at the zero crossing (after rectification, zero touching) does not exist after rectification.

From my point of view , more than one signals can modulate a given high frequency signal. Harmonics of low freq signal are already present in the original signal . That is why we are getting harmonics after demodulation in FFT in addition to the low freq signal.

Right, I accept the sinus case. I use this mostly to hunt bearing problems and that signal is normally nothing like a "continuous" sinus, more like a transient, maybe looking like a single cycle sine wave at regular interval as filtered raw signal. At least in theory. It will generate/contain harmonics in it´s natural state, how much more will it be when rectified? I am just trying to figure out why the demod´s I see normally are not crowded by harmonics that they maybe should be. Are yours that? Olov

Bill,

What you say is true, but it does not apply. You are looking for the amplitude envelope after removing the carrier frwquency, and therefore it is still a sine wave without harmonics. The demodulation process does not introduce harmonics on the low frequency "modulating" signal. As Olov says, what you see is what is there, or did he say that?

Walt

The usual purpose of demodulation is to measure lower frequency variations in signals of higher frequency. The most essential step in demodulation is rectification. After rectification, the higher frequency signals are usually removed by some low pass filter process which may or may not be an integral part of the demodulation process. Prior to demodulation, the signal is usually filtered to remove the raw low frequency content.

Enveloping article:
http://www.aptgroup.com.au/envelope.htm
http://www.aptgroup.com.au/envelope_faq.htm

Walt, yes that was what I have seen so far and tried to say. Olov

Thanks everyone for your inputs. I think I've got it now.