On Tue, May 3, 2011 at 2:07 PM, Mantas Juskauskas
<address@hidden> wrote:
Hello,
I am recently working with gnuradio's tx_samples.cc and USRP2. I have input files that are .sc type, which are written and saved on Matlab. I have a question though.
I've noticed that signals in the input files before being saved to .sc type are being scaled, for example:
signal = sin(A);
scale = 10000;
new_signal = signal*scale;
frwite(...)
Why do we need to use this scaling factor?
Thanks!
This is a result of the change from float to fixed point representation as the signals are sent over the wire. When using the tx_samples program, it's assuming shorts (16 bit I&Q), but this means that if you send a 1.0, it becomes a 1 out of a scale of +-32k. The copier here doesn't do any scaling (unlike the tx_32fc transmit function, which does the scaling for you).
Tom