Hello All I am struggling to model phase errors so that I can check if my PLL is working or not. I tried to use one function available online to model rayleigh fading but it does not give me phase rotation. Basically I would like to model phase rotation due to doppler at baseband. I dont want to model the expansion and compression due to doppler. I know what I am doing is wrong but If you could give me any hint and a starting point, that will be great. Thanks, Chintan
Phase rotation
Started by ●October 2, 2008
Reply by ●October 2, 20082008-10-02
On Oct 2, 8:30�am, "cpshah99" <cpsha...@rediffmail.com> wrote:> Hello All > > I am struggling to model phase errors so that I can check if my PLL is > working or not. > > I tried to use one function available online to model rayleigh fading but > it does not give me phase rotation. > > Basically I would like to model phase rotation due to doppler at baseband. > I dont want to model the expansion and compression due to doppler. > > I know what I am doing is wrong but If you could give me any hint and a > starting point, that will be great. > > Thanks, > > ChintanFor the simple case, just multiply your transmitted signal by exp(jwt) where w is the Doppler offset frequency (rad/sec). John
Reply by ●October 2, 20082008-10-02
On Oct 2, 8:38 am, John <sampson...@gmail.com> wrote:> > For the simple case, just multiply your transmitted signal by exp(jwt) > where w is the Doppler offset frequency (rad/sec). > > JohnThat's correct for baseband. For passband it's a bit more involved, what is needed is a frequency shift of a real-valued signal. It's sufficient to use a real-valued mixer and a filter to reject the extraneous images in the frequency domain. Julius
Reply by ●October 2, 20082008-10-02
On Oct 2, 8:50�am, julius <juli...@gmail.com> wrote:> On Oct 2, 8:38 am, John <sampson...@gmail.com> wrote: > > > > > For the simple case, just multiply your transmitted signal by exp(jwt) > > where w is the Doppler offset frequency (rad/sec). > > > John > > That's correct for baseband. �For passband it's a bit more involved, > what > is needed is a frequency shift of a real-valued signal. �It's > sufficient to use > a real-valued mixer and a filter to reject the extraneous images in > the > frequency domain. > > JuliusYou are correct, but I think the OP specified phase rotation at baseband anyway. John
Reply by ●October 2, 20082008-10-02
>On Oct 2, 8:50=A0am, julius <juli...@gmail.com> wrote: >> On Oct 2, 8:38 am, John <sampson...@gmail.com> wrote: >> >> >> >> > For the simple case, just multiply your transmitted signal byexp(jwt)>> > where w is the Doppler offset frequency (rad/sec). >> >> > John >> >> That's correct for baseband. =A0For passband it's a bit more involved, >> what >> is needed is a frequency shift of a real-valued signal. =A0It's >> sufficient to use >> a real-valued mixer and a filter to reject the extraneous images in >> the >> frequency domain. >> >> Julius > >You are correct, but I think the OP specified phase rotation at >baseband anyway. > >John >%%%% Hello Thanks a lot John and Julius. I would like to read more about this topic. Can u please suggest any starting point. One day julius had said, synchronisation is really tough topic...i have realised that....so I am going step by step....first phase synchronisation and then time synchronisation..... and in the abover eqn exp(jwt), how i put value of 't', i knw it is silly que but please tell me. Thanks a lot. Regards, Chintan
Reply by ●October 2, 20082008-10-02
On Oct 2, 11:31�am, "cpshah99" <cpsha...@rediffmail.com> wrote:> >On Oct 2, 8:50=A0am, julius <juli...@gmail.com> wrote: > >> On Oct 2, 8:38 am, John <sampson...@gmail.com> wrote: > > >> > For the simple case, just multiply your transmitted signal by > exp(jwt) > >> > where w is the Doppler offset frequency (rad/sec). > > >> > John > > >> That's correct for baseband. =A0For passband it's a bit more involved, > >> what > >> is needed is a frequency shift of a real-valued signal. =A0It's > >> sufficient to use > >> a real-valued mixer and a filter to reject the extraneous images in > >> the > >> frequency domain. > > >> Julius > > >You are correct, but I think the OP specified phase rotation at > >baseband anyway. > > >John > > %%%% > > Hello > > Thanks a lot John and Julius. > > I would like to read more about this topic. Can u please suggest any > starting point. > > One day julius had said, synchronisation is really tough topic...i have > realised that....so I am going step by step....first phase synchronisation > and then time synchronisation..... > > and in the abover eqn exp(jwt), how i put value of 't', i knw it is silly > que but please tell me. > > Thanks a lot. > > Regards, > > Chintant is a time vector. It goes from 0 to your last sample instant, with a spacing of one over your sample rate. For example, in Matlab you could do this: Fs=1e6; w=2*pi*1e3; t=[0:Fs-1]/Fs; osc=exp(j*w*t); John
Reply by ●October 2, 20082008-10-02
>On Oct 2, 11:31=A0am, "cpshah99" <cpsha...@rediffmail.com> wrote: >> >On Oct 2, 8:50=3DA0am, julius <juli...@gmail.com> wrote: >> >> On Oct 2, 8:38 am, John <sampson...@gmail.com> wrote: >> >> >> > For the simple case, just multiply your transmitted signal by >> exp(jwt) >> >> > where w is the Doppler offset frequency (rad/sec). >> >> >> > John >> >> >> That's correct for baseband. =3DA0For passband it's a bit moreinvolve=>d, >> >> what >> >> is needed is a frequency shift of a real-valued signal. =3DA0It's >> >> sufficient to use >> >> a real-valued mixer and a filter to reject the extraneous images in >> >> the >> >> frequency domain. >> >> >> Julius >> >> >You are correct, but I think the OP specified phase rotation at >> >baseband anyway. >> >> >John >> >> %%%% >> >> Hello >> >> Thanks a lot John and Julius. >> >> I would like to read more about this topic. Can u please suggest any >> starting point. >> >> One day julius had said, synchronisation is really tough topic...ihave>> realised that....so I am going step by step....first phasesynchronisatio=>n >> and then time synchronisation..... >> >> and in the abover eqn exp(jwt), how i put value of 't', i knw it issilly>> que but please tell me. >> >> Thanks a lot. >> >> Regards, >> >> Chintan > >t is a time vector. It goes from 0 to your last sample instant, with a >spacing of one over your sample rate. > >For example, in Matlab you could do this: > >Fs=3D1e6; >w=3D2*pi*1e3; >t=3D[0:Fs-1]/Fs; >osc=3Dexp(j*w*t); > >John >%%%% Hi John Thanks a lot. But I am just using one sample per symbol. so if my tx signal is 100 symbols then t=[0:length(msg-1)]./length(msg) is that correct?
Reply by ●October 2, 20082008-10-02
On Oct 2, 12:27�pm, "cpshah99" <cpsha...@rediffmail.com> wrote:> >On Oct 2, 11:31=A0am, "cpshah99" <cpsha...@rediffmail.com> wrote: > >> >On Oct 2, 8:50=3DA0am, julius <juli...@gmail.com> wrote: > >> >> On Oct 2, 8:38 am, John <sampson...@gmail.com> wrote: > > >> >> > For the simple case, just multiply your transmitted signal by > >> exp(jwt) > >> >> > where w is the Doppler offset frequency (rad/sec). > > >> >> > John > > >> >> That's correct for baseband. =3DA0For passband it's a bit more > involve= > >d, > >> >> what > >> >> is needed is a frequency shift of a real-valued signal. =3DA0It's > >> >> sufficient to use > >> >> a real-valued mixer and a filter to reject the extraneous images in > >> >> the > >> >> frequency domain. > > >> >> Julius > > >> >You are correct, but I think the OP specified phase rotation at > >> >baseband anyway. > > >> >John > > >> %%%% > > >> Hello > > >> Thanks a lot John and Julius. > > >> I would like to read more about this topic. Can u please suggest any > >> starting point. > > >> One day julius had said, synchronisation is really tough topic...i > have > >> realised that....so I am going step by step....first phase > synchronisatio= > >n > >> and then time synchronisation..... > > >> and in the abover eqn exp(jwt), how i put value of 't', i knw it is > silly > >> que but please tell me. > > >> Thanks a lot. > > >> Regards, > > >> Chintan > > >t is a time vector. It goes from 0 to your last sample instant, with a > >spacing of one over your sample rate. > > >For example, in Matlab you could do this: > > >Fs=3D1e6; > >w=3D2*pi*1e3; > >t=3D[0:Fs-1]/Fs; > >osc=3Dexp(j*w*t); > > >John > > %%%% > > Hi John > > Thanks a lot. > > But I am just using one sample per symbol. > > so if my tx signal is 100 symbols then t=[0:length(msg-1)]./length(msg) > > is that correct?Not quite. the -1 should be outside the parens (typo?). The denominator should equal the sample rate ( same as symbol rate in your case). John
Reply by ●October 2, 20082008-10-02
>On Oct 2, 12:27=A0pm, "cpshah99" <cpsha...@rediffmail.com> wrote: >> >On Oct 2, 11:31=3DA0am, "cpshah99" <cpsha...@rediffmail.com> wrote: >> >> >On Oct 2, 8:50=3D3DA0am, julius <juli...@gmail.com> wrote: >> >> >> On Oct 2, 8:38 am, John <sampson...@gmail.com> wrote: >> >> >> >> > For the simple case, just multiply your transmitted signal by >> >> exp(jwt) >> >> >> > where w is the Doppler offset frequency (rad/sec). >> >> >> >> > John >> >> >> >> That's correct for baseband. =3D3DA0For passband it's a bit more >> involve=3D >> >d, >> >> >> what >> >> >> is needed is a frequency shift of a real-valued signal.=3D3DA0It's>> >> >> sufficient to use >> >> >> a real-valued mixer and a filter to reject the extraneous imagesin>> >> >> the >> >> >> frequency domain. >> >> >> >> Julius >> >> >> >You are correct, but I think the OP specified phase rotation at >> >> >baseband anyway. >> >> >> >John >> >> >> %%%% >> >> >> Hello >> >> >> Thanks a lot John and Julius. >> >> >> I would like to read more about this topic. Can u please suggestany>> >> starting point. >> >> >> One day julius had said, synchronisation is really tough topic...i >> have >> >> realised that....so I am going step by step....first phase >> synchronisatio=3D >> >n >> >> and then time synchronisation..... >> >> >> and in the abover eqn exp(jwt), how i put value of 't', i knw it is >> silly >> >> que but please tell me. >> >> >> Thanks a lot. >> >> >> Regards, >> >> >> Chintan >> >> >t is a time vector. It goes from 0 to your last sample instant, witha>> >spacing of one over your sample rate. >> >> >For example, in Matlab you could do this: >> >> >Fs=3D3D1e6; >> >w=3D3D2*pi*1e3; >> >t=3D3D[0:Fs-1]/Fs; >> >osc=3D3Dexp(j*w*t); >> >> >John >> >> %%%% >> >> Hi John >> >> Thanks a lot. >> >> But I am just using one sample per symbol. >> >> so if my tx signal is 100 symbols thent=3D[0:length(msg-1)]./length(msg)>> >> is that correct? > >Not quite. the -1 should be outside the parens (typo?). The >denominator should equal the sample rate ( same as symbol rate in your >case). > >John >%%%% Sorry for the typo. for my qpsk modulation, is this correct: msg=(sign(randn(1,100))+j*sign(randn(1,100)))/sqrt(2); % qpsk symbols t=[0:length(msg)-1]./100; op=msg.*exp(2*pi*1*t); % doppler of 1 Hz and how do I do for multipath channels, i.e. multipath and also all the path has got phase rotation. I can apply individual phase rotation and after adding proper delay, add them together.
Reply by ●October 2, 20082008-10-02
On Oct 2, 1:10�pm, "cpshah99" <cpsha...@rediffmail.com> wrote:> >On Oct 2, 12:27=A0pm, "cpshah99" <cpsha...@rediffmail.com> wrote: > >> >On Oct 2, 11:31=3DA0am, "cpshah99" <cpsha...@rediffmail.com> wrote: > >> >> >On Oct 2, 8:50=3D3DA0am, julius <juli...@gmail.com> wrote: > >> >> >> On Oct 2, 8:38 am, John <sampson...@gmail.com> wrote: > > >> >> >> > For the simple case, just multiply your transmitted signal by > >> >> exp(jwt) > >> >> >> > where w is the Doppler offset frequency (rad/sec). > > >> >> >> > John > > >> >> >> That's correct for baseband. =3D3DA0For passband it's a bit more > >> involve=3D > >> >d, > >> >> >> what > >> >> >> is needed is a frequency shift of a real-valued signal. > =3D3DA0It's > >> >> >> sufficient to use > >> >> >> a real-valued mixer and a filter to reject the extraneous images > in > >> >> >> the > >> >> >> frequency domain. > > >> >> >> Julius > > >> >> >You are correct, but I think the OP specified phase rotation at > >> >> >baseband anyway. > > >> >> >John > > >> >> %%%% > > >> >> Hello > > >> >> Thanks a lot John and Julius. > > >> >> I would like to read more about this topic. Can u please suggest > any > >> >> starting point. > > >> >> One day julius had said, synchronisation is really tough topic...i > >> have > >> >> realised that....so I am going step by step....first phase > >> synchronisatio=3D > >> >n > >> >> and then time synchronisation..... > > >> >> and in the abover eqn exp(jwt), how i put value of 't', i knw it is > >> silly > >> >> que but please tell me. > > >> >> Thanks a lot. > > >> >> Regards, > > >> >> Chintan > > >> >t is a time vector. It goes from 0 to your last sample instant, with > a > >> >spacing of one over your sample rate. > > >> >For example, in Matlab you could do this: > > >> >Fs=3D3D1e6; > >> >w=3D3D2*pi*1e3; > >> >t=3D3D[0:Fs-1]/Fs; > >> >osc=3D3Dexp(j*w*t); > > >> >John > > >> %%%% > > >> Hi John > > >> Thanks a lot. > > >> But I am just using one sample per symbol. > > >> so if my tx signal is 100 symbols then > > t=3D[0:length(msg-1)]./length(msg) > > > > >> is that correct? > > >Not quite. the -1 should be outside the parens (typo?). The > >denominator should equal the sample rate ( same as symbol rate in your > >case). > > >John > > %%%% > > Sorry for the typo. for my qpsk modulation, is this correct: > > msg=(sign(randn(1,100))+j*sign(randn(1,100)))/sqrt(2); % qpsk symbols > t=[0:length(msg)-1]./100; > op=msg.*exp(2*pi*1*t); % doppler of 1 Hz > > and how do I do for multipath channels, i.e. multipath and also all the > path has got phase rotation. > > I can apply individual phase rotation and after adding proper delay, add > them together.you're missing a j in there somewhere.
