--- In [email protected], k5nwa <[EMAIL PROTECTED]> wrote:
>
> For example you can have an amplifier and band pass filter followed by 
> one very high speed A/D converter, the rest of the functions of the 
> radio are implemented in software, this is a Software Defined Radio,
> but there is no I and Q signals going into the computer, you can even
> do it without having I and Q signals in the software, but you won't
> have image rejection but it will still work as a radio.

 Cecil, 

   that is _almost_ correct, but not completely. When you use a high
speed ADC to digitize directly the RF (as in SDR-IQ, SDR-14, Perseus),
then you have, in the firmware of the radio, to downconvert and
downsample the digitized data. To do this, a NCO (Numerically
Controlled Oscillator) is used, with a half-complex mixer, all
implemented in firmware, which is akin to software. The NCO generates
two signals at 90 degrees each other, and the output of the
half-complex mixer (which, implementation-wise, is just a couple of
multiplications) is an I/Q pair, which, after downsampling, is sent to
the PC, usually via the USB port. 

In this kind of implementation the I/Q pair is generated _after_ the
conversion from analog to digital, so there are no problems of
unbalancing and poor image rejection. Being all done numerically,
everything is balanced (apart from truncation errors, but those are
orders of magnitude lower).

At the end of the story, you *always* need an I/Q pair, as what you
have is a portion of the spectrum centered at zero Hz, and you have to
select either the positive part of it, or the negative part when you
perform the demodulation. And that is possible only if your signal is
analytic, i.e. is composed of two quadrature components, I and Q.

73  Alberto  I2PHD


Reply via email to