[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] An interesting application for my new USRP... som

From: John Ackermann N8UR
Subject: Re: [Discuss-gnuradio] An interesting application for my new USRP... some input requested
Date: Thu, 26 May 2005 15:03:12 -0400
User-agent: Mozilla Thunderbird 1.0 (Windows/20041206)

I'm confused by the need to insert a reference signal and resample from that. I was thinking of just programming the four DACs (through two Basic RX boards) to each look at a, say, 10kHz wide chunk of spectrum around the nominal frequency, and then do a deep FFT of each of those streams. I'm confused by the need to inject another signal and resample. I'm sure I'm missing something, but I don't know what...


PS -- it may be you're describing what I did in the last couple of FMTs -- insert a known frequency from a signal generator within 100Hz or so of the unknown, then using FFT measure the delta between the two. That works very well, but requires me to measure only one signal at a time. I was hoping that with the USRP I'd be able to measure all four bands simulateously.

n4hy wrote:
There are no other clock sources of error in your analysis system since the clocking is all driven by the 64 Mhz stabilized oscillator. No sound card oscillators are involved. This will be a straight digital transfer to the analysis program(s) and all clocks from one reference.

I would do this as follows: Insert a known tamed oscillator in LSB and have the measured tone in USB and use "arithmetic" to back out the offset (phase and frequency) using the known source. Since this is a frequency measuring test, a little phase modulation makes little difference. Using the USRP or SDR-1000, etc. the known reference is the key and you are doing that. In the SDR-1000 case, using the tamed and controlled 8640B, put it in LSB, the signal of interesting in the USB (say) and use the console to record an IQ signal. The analysis and the computation is then easy. The digitally recorded signal is resampled based on the reference signal. Then the frequency of the unknown signal is computed using a large FFT and the known new sample rate. You could do it this way with a known external reference and a combiner and a digital recording program. You record the known reference and the unknown signal in the same passband. Again, I would use my tamed 8640B as the reference. Then you can do it with analog receiver and a digital

reply via email to

[Prev in Thread] Current Thread [Next in Thread]