Re: Complex to 2x real - how?

cs_posting@[EMAIL PROTECTED]
 wrote:
> On Oct 11, 12:46 pm, System Alchemist <dotter@[EMAIL PROTECTED]
>
> wrote:
> 
> 
>>I'm looking at the GC1012 wideband downconverter with a view to
>>implementing some of the functionality.  The output formatter
>>has the option to invert and ****ft the complex output spectrum,
>>both these operations I'm comfortable with.
> 
> The pictures of the choices in the data sheet should provide a lot of
> clarity.
 >
> 
>>It also has the option to create a ****fted real output at twice the
>>sample rate so that no information is lost, the output spectrum being
>>reflected about 0Hz.  I'm pretty sure that this is achieveable without
>>interpolation and additional filtering but struggling to understand
>>how to accomplish this.
> 
> 
> It is either accomplished by interpolation or by keeping a 2x sampling
> rate in the first place.

Given the architecture of the device and various FPGA implmentations, 
interpolation seems unlikely and more likely to be some cleverness
though I wonder if this is in the implementation space rather than
signal processing.

> ****fting is accomplished by multiplying by a complex "local
> oscillator" at 1/4 of the output sample rate (which is fairly trivial
> given the actual samples that the LO will generate)
> 
> Note that ****fting and "flipping" (spectral inversion) are independent
> choices - you can have neither, either, or both.

****fting and flipping appear well known and straight forward to implement.