Re: Complex to 2x real - how?

On Oct 10, 1:31 pm, Randy Yates <ya...@[EMAIL PROTECTED]
> wrote:

> ? If the ****tion of the spectrum that was from 0 to Fs / 2 is reversed,
> or placed in a contrary order, with that from Fs / 2 to Fs, then how
> is that not inverting?

It would be (partially) inverting if that is what were to happen.

But that is NOT what happens in the described method.

In the described method, a constant offset is added to the frequency,
so what  was  negative is now slightly positive, and what was positive
is now even more positive.  There's no flipping, only a constant ****ft
upwards in frequency.

Of course if you mix up the I & Q, you can get  it to offset *and*
invert,  but as a contiguous region.

You could find a way to offset and invert only the negative
frequencies as you've proposed, but that would make a very messy
signal.  What would you do with it then?

IQ centered at zero and real-only centered at FS/4 are both very
practical ways to represent a bandwidth.  Which is preferable depends
on where you'd like to have simplicity and where you will tolerate
complexity in order to pay for it.