Re: [Discuss-gnuradio] halfband in CVS

[Eric Blossom]

On Tue, Oct 25, 2005 at 12:16:09AM -0400, Lamar Owen wrote:
> On Monday 24 October 2005 22:19, Eric Blossom wrote:
> > Matt's been hard at work on the halfband filter for the FPGA, and it's
> > in pretty good shape.  As long as you're not using a decimation factor
> > of 10, everything looks good.  The FPGA code along with the relevant
> > host side mods are checked in.
> 
> Very cool.
> 
> > Also, there's some work-in-progress to support inbound 8-bit samples
> > across the USB.  The interfaces are in (u.set_format(...)), but it's
> > not all sorted out yet.
> 
> Since it's not all sorted out, any way possible that an arbitrary 8-bit 
> window 
> out of the 12 bit gamut be available?  That is, will the API allow me to set 
> the LSB of the 8 bit samples to be at any of bits 0, 1, 2, 3, or 4?  Kindof a 
> crude gain control and noise reduction in one fell swoop.  This is extremely 
> cool; I can double my pulsar system's bandwidth easily with this.

Yes.  We plan to add a barrel shifter.  The shift count is a field in
the format control word.  We'll also support real samples only for
those who want it.  This probably won't make it into 2.6, since I
don't want to hold it up.  

Just to be clear, I'm talking about the output of the DDC, not the raw
samples from the converters.

> I will attempt to try this out tomorrow on the PARI pulsar system, since it 
> is 
> currently tracking CVS, and the FFT curves are well characterized against the 
> passband of the frontend there, and I know the hot-cold difference, passband 
> shape, etc really well at this point. 

Good.  Let us know how it goes.

Eric