Re: [Discuss-gnuradio] Questions for 150/400MHz beacon RX development

[Eric Blossom]

On Wed, Jan 10, 2007 at 02:52:54AM +0900, Mamoru Yamamoto wrote:
> Thank you for your kind help to my questions.
> I am developing a satellite beacon receiver of 
> 150/400MHz.  Now I am receiving both signals by 
> using Basic RX (150MHz) and FLEX400 (400MHz) boards.
> I am happy so far.
> 
> I have some more questions for further development.
> 
> (1) File format
> I record signals into files.  I want to read them 
> for analysis.  What is the data format written/read 
> by file_sink() / file_source()?  I am recording 
> complex signals in the files.

They are binary files in "native endian".
If you're writing gr_complex, the file will contain interleaved
real, imaginary values where the real and imaginary components are
32-bit binary floats.

FYI, there's code to read these files into Octave or Matlab in 
gnuradio-core/src/utils/read_*_binary.m

> (2) "O" "U" "u" "a" 
> These characters appear on the screen when I run 
> my program.  I understand they appear when data flow 
> from USRP to PC is stopped or something.  I am curious 
> what are the meaning of these characters.  Does faster 
> PC help reducing such data flow troubles?

"u" = USRP
"a" = audio (sound card)
"O" = overrun (PC not keeping up with received data from usrp or audio card)
"U" = underrun (PC not providing data quickly enough)

Yes, a faster machine will generally cure this problem.  This assumes
that you're not asking the USB to do something that it can't.  In our
best case, we sustain 32MB/s across the USB.  I suggest avoiding intel
celeron or other "cache crippled" parts.


> (3) Signal coherence .. (a)
> I need to keep signal coherence between 150MHz and 400MHz.
> I should run the FLEX400 board at the same timing to the 
> USRP main board.  I found the following information URL
> http://comsec.com/wiki?USRPClockingNotes
> Is this instruction still valid?  I am using USRP ver4.

I believe this is still valid.  Matt, can you confirm?

> (4) Signal coherence .. (b)
> I have one more question on the signal coherence.  In my 
> program I keep the same sample speed and the same FIR filter 
> for both 150MHz and 400MHz. (Sure I am using a "deinterleaver" 
> to separate the signals.)  Can you guarantee signal coherence 
> in the GNUradio data flow?

Yes you can.  The important thing to keep in mind is that if there are
different filters in the two paths downstream from the deinterleaver,
they should have the same group delay through them.  In the common
case, you can ensure this by using the same number of taps in each.
It sounds like you are doing this, so you should be OK.

Good luck!
Eric