Hi, May I ask you for help, I have quite simple(I assume) question: how to reconstruct original waveform after IQ down-conversion. Let's assume I've got IQ samples of an RF signal down-converted to be zero centered. I know RF signal is just an AM wave. So, examining spectra calculated from these IQ samples I see it is what I want: an AM with Fs = 0Hz. Hovewer, waveform is just harmonic oscillation with almost constant amplitude, even when I feed 100% depth AM in. I tried to shift it up by multiplying to e^(jwt) - spectra is moved but time domain signel shape is still almost the same. Did I miss anything important? What does this IQ down-conwersion do with AM signal? Thanks in advance, Igor.
How to reconstruct original waveform from IQ
Started by ●November 12, 2008
Reply by ●November 12, 20082008-11-12
On Nov 12, 11:27�am, "ivli" <i...@pochta.ru> wrote:> Hi, > > May I ask you for help, I have quite simple(I assume) question: how to > reconstruct original waveform after IQ down-conversion. > Let's assume I've got IQ samples of an RF signal down-converted to be zero > centered. I know RF signal is just an AM wave. So, examining spectra > calculated from these IQ samples I see it is what I want: an AM with Fs = > 0Hz. Hovewer, waveform is just harmonic oscillation with almost constant > amplitude, even when I feed 100% depth AM in. I tried to shift it up by > multiplying to e^(jwt) - spectra is moved but time domain signel shape is > still almost the same. Did I miss anything important? What does this IQ > down-conwersion do with AM signal? > > Thanks in advance, > Igor.Just add I+Q
Reply by ●November 12, 20082008-11-12
Sorry, do you mean sqrt(I^2 + Q^2)? If so, it's what I did, alas... Igor.>On Nov 12, 11:27=A0am, "ivli" <i...@pochta.ru> wrote: >> Hi, >> >> May I ask you for help, I have quite simple(I assume) question: how to >> reconstruct original waveform after IQ down-conversion. >> Let's assume I've got IQ samples of an RF signal down-converted to bezer=>o >> centered. I know RF signal is just an AM wave. So, examining spectra >> calculated from these IQ samples I see it is what I want: an AM with Fs=>=3D >> 0Hz. Hovewer, waveform is just harmonic oscillation with almostconstant>> amplitude, even when I feed 100% depth AM in. I tried to shift it upby>> multiplying to e^(jwt) - spectra is moved but time domain signel shapeis>> still almost the same. Did I miss anything important? What does thisIQ>> down-conwersion do with AM signal? >> >> Thanks in advance, >> Igor. > >Just add I+Q >
Reply by ●November 12, 20082008-11-12
On Nov 12, 8:29 am, "ivli" <i...@pochta.ru> wrote:> Sorry, do you mean sqrt(I^2 + Q^2)? If so, it's what I did, alas... > > Igor. > >That should work if you truly have an I/Q signal. Explain how you produced your I/Q signal and centered it at zero. Is your downconverter using a complex mixer? Darol Klawetter
Reply by ●November 12, 20082008-11-12
On Nov 12, 8:37�am, Alfreeeeed <Alfredo.Tad...@gmail.com> wrote:> On Nov 12, 11:27�am, "ivli" <i...@pochta.ru> wrote: > > > Hi, > > > May I ask you for help, I have quite simple(I assume) question: how to > > reconstruct original waveform after IQ down-conversion. > > Let's assume I've got IQ samples of an RF signal down-converted to be zero > > centered. I know RF signal is just an AM wave. So, examining spectra > > calculated from these IQ samples I see it is what I want: an AM with Fs = > > 0Hz. Hovewer, waveform is just harmonic oscillation with almost constant > > amplitude, even when I feed 100% depth AM in. I tried to shift it up by > > multiplying to e^(jwt) - spectra is moved but time domain signel shape is > > still almost the same. Did I miss anything important? What does this IQ > > down-conwersion do with AM signal? > > > Thanks in advance, > > Igor. > > Just add I+QNot for Zero-centered I/Q representation. Dirk
Reply by ●November 13, 20082008-11-13
The device I'm working on is UHF receiver. Actually, it was almost ready a couple years ago but, due to some rason put on hold. Those time I wasn't on board, so I can miss some details. It designed as such: two IF, ADC on low IF, then IQ down-conwersion and a set of demodulators AM/FM/SSB. All this stuff is controleed from PC. What I am trying to do is to predict modulation type in order to use correct demodulator automatically, and for this I need real signal, there is no interface to get it stright from ADC, but, IQ are avaliable. Is there a way to achieve this?>On Nov 12, 8:29 am, "ivli" <i...@pochta.ru> wrote: >> Sorry, do you mean sqrt(I^2 + Q^2)? If so, it's what I did, alas... >> >> Igor. >> >> > >That should work if you truly have an I/Q signal. Explain how you >produced your I/Q signal and centered it at zero. Is your >downconverter using a complex mixer? > >Darol Klawetter >
Reply by ●November 13, 20082008-11-13
On Wed, 12 Nov 2008 07:27:40 -0600, ivli wrote:> Hi, > > May I ask you for help, I have quite simple(I assume) question: how to > reconstruct original waveform after IQ down-conversion. Let's assume > I've got IQ samples of an RF signal down-converted to be zero centered. > I know RF signal is just an AM wave. So, examining spectra calculated > from these IQ samples I see it is what I want: an AM with Fs = 0Hz. > Hovewer, waveform is just harmonic oscillation with almost constant > amplitude, even when I feed 100% depth AM in. I tried to shift it up by > multiplying to e^(jwt) - spectra is moved but time domain signel shape > is still almost the same. Did I miss anything important? What does this > IQ down-conwersion do with AM signal? > > Thanks in advance, > Igor.I think that instead of asking "how do I demodulate this?" you should be asking "did I modulate this correctly?" The best way to answer that probably isn't by asking here. It's probably to look at the waveforms you're generating at each step and see if they're what you expect. An AM signal is so simple you can do the math on the back of an envelope with a blunt pencil, so you should also seriously consider working out the solution symbolically, to make double sure. -- Tim Wescott Control systems and communications consulting http://www.wescottdesign.com Need to learn how to apply control theory in your embedded system? "Applied Control Theory for Embedded Systems" by Tim Wescott Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html
Reply by ●November 13, 20082008-11-13
On Nov 13, 6:35 am, "ivli" <i...@pochta.ru> wrote:> ... > What I am trying to do is to predict > modulation type in order to use correct demodulator automatically, and for > this I need real signal, there is no interface to get it stright from ADC, > but, IQ are avaliable. Is there a way to achieve this? > ...To convert your complex sample stream to real: 1) Upsample to a higher sampling frequency, perhaps by inserting zero samples between original samples. 2) Lowpass filter to remove signal images not centered at zero frequency. (An efficient implementation will not perform calculations with the zero valued samples.) 3) Frequency shift (complex modulate) the image at DC to the desired center frequency. 4) Take the real part. (This means that an effecient version of 3) need not calculate the imaginary component.) The exact filter response and frequency shift will depend on the parameters of the downconvertor in your system and perhaps the signals you are expecting to determine the modulation of. Dale B. Dalrymple http://dbdimages.com
Reply by ●November 13, 20082008-11-13
On Nov 13, 9:35 am, "ivli" <i...@pochta.ru> wrote:> What I am trying to do is to predict > modulation type in order to use correct demodulator automatically, and for > this I need real signal, there is no interface to get it stright from ADC, > but, IQ are avaliable. Is there a way to achieve this?Maybe you should re-think your algorithm to work with complex (IQ) zero-if data. If not, I see dbd has already described how to do the conversion from complex zero-if to real non-zero if.
Reply by ●November 14, 20082008-11-14
Thank you all, I think I'd better modify existing algorythm to work with zero centered IQ, at least it looks more efficient. Igor.>On Nov 13, 9:35 am, "ivli" <i...@pochta.ru> wrote: > >> What I am trying to do is to predict >> modulation type in order to use correct demodulator automatically, andfor>> this I need real signal, there is no interface to get it stright fromADC,>> but, IQ are avaliable. Is there a way to achieve this? > >Maybe you should re-think your algorithm to work with complex (IQ) >zero-if data. > >If not, I see dbd has already described how to do the conversion from >complex zero-if to real non-zero if. >
