digital I/Q downconversion

[shameem]

hi,
i want to generate I and Q samples from the real valued samples taken by adc in the front end of digital radio hardware. In some papers, the real valued samples are multiplied by a cos and a sin waves while in some other papers, they multiply it with cos and -sin waves.
i want to know what is the difference b/w both the operations?

 

[LOSTINCITY]

in my openion there is no diffrents between these.
the second situation just a 180 degree phase shift of LO

 

[shameem]

hi,
but my problem is that for the downconversion process, should i multiply the samples with a cos wave and a +sin wave or -sin wave.

 

[m_llaa]

Hi
It is not important at all.
The differene between +sin and -sin is depend on definition. in some refrence they use +sin and in some -sin
but for your problem, it depent on u. catch on 1 definition and go to end just by this definition

 

[zorro]

Hi shameem,

The results are different. But both results are useful.
If you use +, you are using a local carrier with positive frequency:
cos()+j*sin()=exp[j*()]
If you use -, you are using a local carrier with negative frequency:
cos()-j*sin()=exp[-j*()]
In the first case, the negative part of the passband spectrum of the real-valued signal goes to baseband. In the second case, the positive part does that. Both parts are symmetrical with respect of the other.
Between the two cases, there is an inversion of the baseband spectrum. The results are conjugated in time-domain one of the other.
For instance, imagine that detecting PSK there is a +90 degrees phase change between two consecutive symbols. Using -sin() you get +90 degrees phase change in baseband, but using +sin() you get -90 degrees change in baseband.
Regards

Z

 

[shameem]

hi,
Actually i have real valued samples of a digital radio signal whose if is 38.912 mhz and is undersampled at 8.192 mhz sampling frequency.
In the undersamplig process, spectrum inversion will occur and one alias of the spectrum comes down at 2.048 mhz. Now i want to do digital i/q downconversion of this signal to baseband.
For this, should i multiply the real valued samples with a cos and -sin waves of the frequency 2.048 mhz or cos and +sin waves of the frequency 2.048 mhz so that after fft, i get the spectrum spectrally uninverted.

thanks
regards
shameem