Hi all, After seaching the comp.dsp archive for FM demod I did not find anything on the case where there is a frequency offset between the transmitter and receiver. Is there a reference that explains how the frequency offset between the transmitter and reciever effects the performace of the demodulator? I assume that a digital PLL implementation would be required or would the method in the Frerking book work if there is a frequency offset. A second question is could anyone explain how to use the hilbert transform to demodulate an FM signal. I hear that it is possible but I am unsure how to use the hilbert transform to demodulate an FM signal. Thanks, Jon
Digital FM Demodulation
Started by ●February 3, 2004
Reply by ●February 4, 20042004-02-04
Jon wrote:> Hi all, > After seaching the comp.dsp archive for FM demod I did not find > anything on the case where there is a frequency offset between the > transmitter and receiver. Is there a reference that explains how the > frequency offset between the transmitter and reciever effects the > performace of the demodulator? I assume that a digital PLL > implementation would be required or would the method in the Frerking > book work if there is a frequency offset. > > A second question is could anyone explain how to use the hilbert > transform to demodulate an FM signal. I hear that it is possible but > I am unsure how to use the hilbert transform to demodulate an FM > signal. > > Thanks, > > JonHow does a frequency offset differ from a DC modulation component? Narrow-band FM is essentially AM with a 80 degree carrier shift. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●February 4, 20042004-02-04
> How does a frequency offset differ from a DC modulation component?The whole truth does not lie in the answer to your question! (..neither in my contribution to the thread) Adding a message signal to this unintended DC offset often make the demodulator operate in a non-linear area. Depending on which demodulation scheme that is used you can experience a distortion on the either the upper or lower part of the message signal with respect to your reference. The extend of this problem is ofcourse releated to the non-linearity of your receiver. Regards Peter H "Jerry Avins" <jya@ieee.org> wrote in message news:402075a5$0$8381$61fed72c@news.rcn.com...> Jon wrote: > > > Hi all, > > After seaching the comp.dsp archive for FM demod I did not find > > anything on the case where there is a frequency offset between the > > transmitter and receiver. Is there a reference that explains how the > > frequency offset between the transmitter and reciever effects the > > performace of the demodulator? I assume that a digital PLL > > implementation would be required or would the method in the Frerking > > book work if there is a frequency offset. > > > > A second question is could anyone explain how to use the hilbert > > transform to demodulate an FM signal. I hear that it is possible but > > I am unsure how to use the hilbert transform to demodulate an FM > > signal. > > > > Thanks, > > > > Jon >> > Narrow-band FM is essentially AM with a 80 degree carrier shift. > > Jerry > -- > Engineering is the art of making what you want from things you can get. > ����������������������������������������������������������������������� >
Reply by ●February 4, 20042004-02-04
On 3 Feb 2004 19:22:04 -0800, jon8spam@yahoo.com (Jon) wrote: (snipped)> >A second question is could anyone explain how to use the hilbert >transform to demodulate an FM signal. I hear that it is possible but >I am unsure how to use the hilbert transform to demodulate an FM >signal. > >Thanks, > >JonHi, well, the idea is that you perform the Hilbert transform on a real sequence, I(n), to obtain a "quadrature" sequence Q(n). Now Q(n)'s spectral components are all shifted in phase by 90 deg relative to I(n)'s spec components. Creating a complex sequence, C(n) = I(n)+jQ(n) allows you to measure the instantaneous phase of C(n) using arctangent. In theory, the instantaneous frequency (your demodulated FM) is then computed by taking the derivative of the real-valued instantaneous phase. There are all sorts of practical issues to worry about in this process: * how to efficiently perform arctangents * how to accurately perform differentiation * is auto gain control (AGC) being used * what happens if the magnitude of C(n) is fluctuating with time. Frerking's FM demod (Figure 6.20) in his 1st edition gets around some of these problems. His equations contain a couple of "typos" which I imagine have been corrected in later editions. (His scheme has been described many times here on comp.dsp over recent years.) I have a chapter on the Hilbert transform, and discuss FM demodulation, in the 2nd edition of my DSP book - which I hope'll be available in late March-early April. Good Luck, [-Rick-]
