0

I get an IF signal from microwave circuit which looks like this:

IF Signal

As we can see there is a distinct unwanted Amplitude Modulation which is believed to be due to coupling between antennae. This signal will also have a FM modulation.

I want to remove this AM modulation so that only FM modulation remains. I want the carrier signal NOT the AM modulation. Something like how a Automated Gain Control (AGC) would work.

Marcus Müller
  • 34,677
  • 4
  • 53
  • 94
JBuch
  • 73
  • 1
  • 2
  • 7
  • 1
    Have you tried anything? – akalikin Aug 21 '15 at 08:55
  • Yes. I tried to filter the lower freqs but that dose not removes the AM modulation. I also tried the FreqDemod module but as far as i tried it extracts the signal, in my case signal is noise actually and carrier is of interest. – JBuch Aug 21 '15 at 09:06

1 Answers1

0

I tried to filter the lower freqs but that dose not removes the AM modulation.

Well, then you need to adjust your filter. Every kind of AGC is but a high pass filter in effect.

I want to remove this AM modulation so that only FM modulation remains.

Seriously, don't do that. FM demodulation doesn't need constant envelope. As a FM demodulator doesn't care the least about amplitude, it will not care about what you consider to be AM.

EDIT I though you'd probably already tried, but what about the AGC, AGC2 and AGC3 block in GNU Radio?

Marcus Müller
  • 34,677
  • 4
  • 53
  • 94
  • @ Marcus Muller I tried to adjust my filter and it does improve the results a bit but the AM modulation is similar to a Triangular wave which has a large number of frequencies [link](https://upload.wikimedia.org/wikipedia/commons/thumb/5/5b/Window_function_and_frequency_response_-_Triangular.svg/480px-Window_function_and_frequency_response_-_Triangular.svg.png) – JBuch Aug 27 '15 at 07:05
  • @ Marcus Muller Also I need the power spectral density which is why i want the amplitude to be a constant. PS sorry for the delay. – JBuch Aug 27 '15 at 07:19
  • @superbuch: But if you correct the amplitude, you will change the PSD! That doesn't sound right. By the way, for coupling, this sure looks *very* regular... I'd rather assume periodic behaviour of amplifier / mixer circuits, TX/RX clock drift etc. – Marcus Müller Aug 27 '15 at 08:58
  • Muller I have been trying to install GNU radio on windows but i am till now unsuccessful. I found this [link](http://labrosa.ee.columbia.edu/matlab/tf_agc/) which i am now trying to convert to python. Will update soon. Also since I want only the change in freq. i want the specgram to look pretty. – JBuch Aug 28 '15 at 05:52
  • @superbuch: Looking pretty is usually not reall what you want in science or engineering, that's arts ;) no, I can understand your desire. I don't really trust the link you referred to, because once you not only normalize energy over time (AGC) but also in individual frequency bands, it's very hard to argue that the PSD you're looking at has much to do with your original signal. Better do something like original signal -> Energy (==samples squared) -> low pass -> 1/x -> array of inverse energy and multiply that with your input signal – Marcus Müller Aug 28 '15 at 06:47