Hello, I am a bit unsure how the demodulator recovers the original IQ modulation frequency of a DVB-T baseband signal. In the standard the frequency "fc" is (IMHO) not specified, however there could be read (ETSI EN 300 744 V1.5.1 (2004-06), Page 45) "A subsequent up-conversion process then gives the real signal s(t) centred on the frequency fc". So it should be correct to look for the "center frequency" of the signal and use this as the IQ modulation frequency. Now my problem: If the baseband signal becomes upconverted to the RF band (e.g. with a TUF-x mixer) in the transmitter and downconverted in the receiver, the information of the original baseband frequency is lost?! So how can I recover the IQ values from an unknown fc? BTW: Its just for my theoratical understanding, not related to any existing hardware :-) Thanks for reading Ekkehard

# Recovery of IQ modulation frequency (DVB-T)

Started by ●April 6, 2006

Reply by ●April 6, 20062006-04-06

Ekkehard said the following on 06/04/2006 12:48:> I am a bit unsure how the demodulator recovers the original IQ > modulation frequency of a DVB-T baseband signal. > In the standard the frequency "fc" is (IMHO) not specified, however > there could be read (ETSI EN 300 744 V1.5.1 (2004-06), Page 45) "A > subsequent up-conversion process then gives the real signal s(t) > centred on the frequency fc". So it should be correct to look for the > "center frequency" of the signal and use this as the IQ modulation > frequency. > Now my problem: If the baseband signal becomes upconverted to the RF > band (e.g. with a TUF-x mixer) in the transmitter and downconverted in > the receiver, the information of the original baseband frequency is > lost?!Why do you say that? If you upconvert by fc, and then downconvert by fc, you're back to where you started. The problem, obviously, is LO offset, as with any modulation scheme. The problem is particularly bad in OFDM, however, because any offset destroys orthogonality and kills all your data. So how can I recover the IQ values from an unknown fc? When you say "unknown", presumably you don't mean "completely unknown", I'm guessing you will have a nominal fc, e.g. 11.25MHz. The "unknown" bit is the LO offset, and there are plenty of algorithms to fix that. Search for "carrier frequency offset" + OFDM. -- Oli

Reply by ●April 6, 20062006-04-06

> Why do you say that? If you upconvert by fc, and then downconvert by fc, > you're back to where you started.Example: basic fc is 8MHz upconversion to 474MHz than downconverted to 36MHz and than? LO Offset is not the problem I want to discuss Thanks Ekkehard

Reply by ●April 6, 20062006-04-06

Ekkehard said the following on 06/04/2006 13:31:>> Why do you say that? If you upconvert by fc, and then downconvert by fc, >> you're back to where you started. > > Example: basic fc is 8MHz upconversion to 474MHz than downconverted to > 36MHz and than? >What do you mean by "basic fc"? Centre frequency at baseband is (by definition) 0Hz. Are you saying that your receiver IF is 36MHz? If that is the case, then to recover baseband you downconvert by a further 36MHz. Maybe I'm missing something here... ;) -- Oli

Reply by ●April 6, 20062006-04-06

> What do you mean by "basic fc"? Centre frequency at baseband is (by > definition) 0Hz.Page 46 (small copy at the bottom of the post) of the ETSI 300 744 says that not as clear. This is only one of the possibilities (point b below). There are several other possibilities too. That is something that leads me into this problem.> Are you saying that your receiver IF is 36MHz? If that is the case, > then to recover baseband you downconvert by a further 36MHz.Ok, lets say the receivers IF is 36MHz: Will a (additional) donwconversion recreate the baseband? What happned if the basic fc (as described above) was original 36MHz, or 8MHz or 0Mhz? Snipped from Page 46 D.2 Choice of "baseband" centre frequency [...] In particular, either of the following alternatives is recommended as they satisfy this requirement while leading to a simple implementation: a) assign the middle carrier to the half-way index q = N 2 , i.e. the half-sampling-frequency term; or b) assign the middle carrier to index q = 0 , i.e. the DC or zero-frequency term.

Reply by ●April 6, 20062006-04-06

Ekkehard said the following on 06/04/2006 14:29:>> What do you mean by "basic fc"? Centre frequency at baseband is (by >> definition) 0Hz. > > Page 46 (small copy at the bottom of the post) of the ETSI 300 744 says > that not as clear. This is only one of the possibilities (point b > below). There are several other possibilities too. That is something > that leads me into this problem.The title of Annex D.2 is misleading, I think. What they're talking about is how to map the baseband frequency content onto the IFFT coefficients; they describe two ways of doing this, whose relationship is equivalent to that given by the fftshift() function in MATLAB, as far as I can see. But this is just an implementation detail, not a parameter that alters the actual DVB signal. Regardless of which way you implement it, the resulting passband signal s(t) must always be the same, a set of modulated carriers centred around fc (RF), and these will always be in the same order, as is explicitly stated in Annex D.3 case (a).> Snipped from Page 46 > D.2 Choice of "baseband" centre frequency > [...] > In particular, either of the following alternatives is recommended as > they satisfy this requirement while leading to a simple implementation: > a) assign the middle carrier to the half-way index q = N 2 , i.e. the > half-sampling-frequency term; or > b) assign the middle carrier to index q = 0 , i.e. the DC or > zero-frequency term. >-- Oli

Reply by ●April 6, 20062006-04-06

>Regardless of which way you implement it, the >resulting passband signal s(t) must always be the same, a set of >modulated carriers centred around fc (RF)That was what I assumed, but becomes confused while reading :-( Ok, I will give it a try Thanks for talking Ekkehard

Reply by ●April 7, 20062006-04-07

"Oli Filth" <catch@olifilth.co.uk> wrote in message news:dD7Zf.16927$g76.1739@newsfe2-gui.ntli.net...> Ekkehard said the following on 06/04/2006 12:48: > > I am a bit unsure how the demodulator recovers the original IQ > > modulation frequency of a DVB-T baseband signal. > > In the standard the frequency "fc" is (IMHO) not specified, however > > there could be read (ETSI EN 300 744 V1.5.1 (2004-06), Page 45) "A > > subsequent up-conversion process then gives the real signal s(t) > > centred on the frequency fc". So it should be correct to look for the > > "center frequency" of the signal and use this as the IQ modulation > > frequency. > > Now my problem: If the baseband signal becomes upconverted to the RF > > band (e.g. with a TUF-x mixer) in the transmitter and downconverted in > > the receiver, the information of the original baseband frequency is > > lost?! > > Why do you say that? If you upconvert by fc, and then downconvert by fc, > you're back to where you started. > > The problem, obviously, is LO offset, as with any modulation scheme. > The problem is particularly bad in OFDM, however, because any offset > destroys orthogonality and kills all your data. > > > So how can I recover the IQ values from an unknown fc? > > When you say "unknown", presumably you don't mean "completely unknown", > I'm guessing you will have a nominal fc, e.g. 11.25MHz. The "unknown" > bit is the LO offset, and there are plenty of algorithms to fix that. > Search for "carrier frequency offset" + OFDM. > > > -- > OliUse a PLL. Tam