Re: [Discuss-gnuradio] only zeros in dual-channel-mode

[Eric Blossom]

On Fri, Dec 28, 2007 at 04:50:36AM -0800, wolfgang buesser wrote:
> 
> I am trying to build a network analyzer using usrp and LF RX/TX- so i need
> one TX and two RX-channels. However I only get noise lcose to zero on the
> secon channel:
> 
> Here is what I am doing in the RX_path (in brief). Maybe somebody can spot
> the mistake:
> 
>         self.rx = usrp.source_c(0,decim_rate=128,
> fpga_filename="std_2rxhb_2tx.rbf")
>         self.rx.set_nchannels(2)  
>         self.di=gr.deinterleave(gr.sizeof_gr_complex) 
>         self.dst_r = gr.file_sink(gr.sizeof_gr_complex, "test_r.dat") 
>         self.dst_t = gr.file_sink(gr.sizeof_gr_complex, "test_t.dat") 
> 
>         self.rx_r_subdev = usrp.selected_subdev(self.rx, (0,0))
>         self.rx_r_subdev = usrp.selected_subdev(self.rx, (1,0))

Looks like a typo here ^  you're setting rx_r_subdev twice

Also, how many LF_RX's do you have?  Two? or are you using one, and
treating the two inputs as separate channels.  If you're using two,
then the lines above are OK, otherwise they should be

  (0, 0)
  (0, 1)

> 
>         self.rx.set_mux(gru.hexint(0xf0f0f0f0)) # expecting to get port A of
> side A on both channels

OK, but I'm not sure I see why you'd want to feed the same A/D to both DDCs...

>         rrx_r = self.rx.tune(0, self.rx_r_subdev, 10e6+1e3) 
>         rrx_t = self.rx.tune(0, self.rx_t_subdev,  10e6+1e3) 

One of these should be tune(1, ...)

Note that in general tune won't work correctly for more than one
channel. Since you're using LF_RX, you're probably OK.

>         self.head_r = gr.head(gr.sizeof_gr_complex, 1000) 
>         self.head_t = gr.head(gr.sizeof_gr_complex, 1000) 
>         self.connect(self.rx,self.di) 
>         self.connect((self.di,0),self.head_r, self.dst_r) 
>         self.connect((self.di,1),self.head_t, self.dst_t) 
>         self.connect (self.siggen, self.tx)
> 
> I get nice data in test_r.dat but only noise close to zero in test_t.dat

Eric