envelope and demodulation analysis

[Eng P&D]

hi, I'm traying to filter a signal with some methods like envelope analysis and demodulation, cause I´m interested in analise a range of low frequencies and their time-power line. if someone have an ideia about what are this methods and a basic 'how to implement'

[johnsold]

In a radio receiver the detection process recovers the envelope of an amplitude modulated waveform. So as a brute force method you could emulate that process. Start by taking the absolute value of the "rf" samples. Then use a low pass filter. As a starting point use a filter cutoff frequency of 10 times the highest frequency component in the envelope or the geometric mean between the envelope frequency and the carrier frequency, whichever is lower. If the carrier frequency is several orders of magnitude higher than the envelope frequency, a single pole filter may be adequate. The output of the filter should represent the envelope.

Lynn

[Eng P&D]

ooo, i think it's a litlle confuse to me. maybe i must to study some filter methods first. Do you have a litlle example implemented? thanks...

[rpursley8]

Sounds like you want to look at a signal extracted from an AM modulated signal at some carrier frequency.  This really isn't considered to be filtering as such.  What kind of signal are you starting with?

There are several ways to do this depending on what you need to do with the signal once you demodulate it.  The way already described is the easiest and probably will give you what information you need.

[johnsold]

Here is a simple Envelope Detector.VI which implements the method I suggested. It will not let you know if mail has been delivered to your postal box, however.

Lynn

[rpursley8]

That was cheating a little bit adding that offset to the modulation signal.

I think he/she will have to know the carrier frequency to really demodulate it.

[tbob]

"Mr." Lynn's software detector is right on.  There is no need to know the carrier frequency.  In a real AM radio, the carrier gets downconverted to an intermediate frequency before it is demodulated, so the detector stage doesn't have to know the carrier (or even IF) frequency.  It is the shape of the modulated envelope that is sought, independent of the frequency.  The shape is the same as the original modulating signal.

[rpursley8]

That is incorrect.  In an AM radio when you tune to the station you are changing the frequency that is used to downconvert to the IF.  So you know you have the right frequency when you hear the station.  "Mr" Lynn added the offset so the whole signal would be positive.  This assured that taking the absolute value of the signal didn't change the modulation.  You can't count on that in a general application and taking the absolute value doubles the carrier frequency and the modulation frequency.  But as long as the modulation index is low enough so the envelope of the signal doesn't reach zero, then "Mr" Lynn's approach will work.

BTW, the he/she was referring to the original poster not "Mr" Lynn.

Message Edited by rpursley8 on 02-04-2006 10:34 PM

[Eng P&D]

well, I don't understand well the envelope, but I'm now studying dsp to get more help on it.

my research is this: i'm trying to detect a noise ( low frequency ) to qualificate some compressors in time domain espectrum.

and some of the methods to do this are envelope and demodulation.

think I have to learn a litlle more to discuss this here.

but thanks for help.

[tbob]

Well of course you are changing the frequency when you tune an AM radio.  What I am saying is the the process of AM detection takes place without regard to what the carrier frequency is.  A detector is just a diode.  Doesn't matter what the frequency is to the diode.  A capacitor is added to smooth resulting envelope.  A resistor is added to develop a voltage.  Again it doesn't matter to the capacitor or resistor what the carrier frequency is.  All this means that the software detector does not need to know what the incoming carrier frequency is.

Also, taking the absolute value may double the carrier frequency but it does not double the modulating frequency.

Look at the attached picture.  Imagine flipping all negative values to positive.  That is what the absolute value does.  Carrier frequency is doubled.  Modulating envelope remains the same.  Now imagine converting all negative values to zero.  That is what a diode does.  Carrier frequency remains the same, so does modulating envelope.  Whether the carrier is doubled or not (absolute value or diode), the output of the detector is still the same, hence it does not rely upon carrier frequency.

Message Edited by tbob on 02-06-2006 09:41 AM