[Discuss-gnuradio] An interesting application for my new USRP... some input requested

[John Ackermann N8UR]

Well, it didn't take long for me to think of a use for the USRP I 
acquired from Matt at the Dayton Hamvention.

About six months from now, there will be a ham radio frequency measuring 
test where the challenge will be to measure, as accurately as possible, 
the frequency of a signal broadcast in the HF band.  Actually, there 
will be three or four signals, on different frequency bands (near 3.58, 
7.08, 14.08, and 21.08 MHz).  The signal will be an unmodulated carrier 
(well, there will be some CW identification) and will be transmitted for 
about 10-15 minutes.

It seems to me that with two of the Basic RX boards, I could monitor all 
four frequencies simultaneously and save the data streams to disk for 
processing with a deep FFT to extract the average frequency of each 
signal over the run.  Since I'll know the frequency ahead of time to 
within a few kHz, I won't need to capture a wide swath -- probably 10kHz 
on each RX input will do.

Since Matt tells me the USRP can be clocked by an external reference, I 
should be able to extract very accurate frequency information -- in past 
years, I've been able to read the frequency to within a few milliHertz 
using my old method, and I'd like to be able to do at least as well, 
with the advantage of being able to capture all bands simultaneously, 
instead of being able to spend only a couple of minutes on each one 
sequentially.

I'll need to build up some front-end boards with input amplifiers and 
some bandpass filtering, as well as learn about the gnuradio software in 
general, so it's not too early for me to start thinking about this.

A couple of questions for the group, and maybe especially for Matt:

1.  Is it feasible to capture 4 10kHz wide HF streams using the I and Q 
inputs on two Basic RX daughterboards?

2.  If I clock the USRP from a highly stable external reference, is 
there any other source of possible frequency error in the system (since 
we're at HF, I shouldn't need to use a downconverter)?  In other words, 
will the measured FFT frequency have any error source other than errors 
in the external reference?

3.  Leaving propagation effects aside, what kind of sample 
rate/decimation would yield the maximum amount of frequency information 
for the FFT?  Again, consider about a 10kHz data stream on each of the 
four channels, and less than 15 minutes of data to deal with.

4.  Any special insight on the design of the preamp/filter boards?  I'm 
assuming I'll need one for each frequency band.  For filtering, should I 
use a bandpass filter, or would lowpass filters suffice? (I do have 
strong local AM broadcast stations nearby, so that may be a factor.)

This sounds like fun.  Any ideas/suggestions would be greatly appreciated.


John