Question: Why does the spectrum peak some +20 dB over the noise floor at the center frequency?
Background: I've been using a USRP 2944R for a while, and I am struggling to eliminate the erroneous peak that is generated on the magnitude spectrum at the center frequency.
For example, look at the "Magnitude Spectrum" display at this link:
I see a similar peak a the center frequency, regardless of if I use this particular example or one of the other USRP examples included in LabVIEW. Changing the center frequency has no effect. I'm curious to know if anyone can explain the reason why this peak exists even though there is no signal being injected into the USRP at this frequency.
Solved! Go to Solution.
There might be a better forum for this question... I do not know much about the USRP, but what you are seeing looks like LO leakage. This is a common problem for most RF Receivers. Rejecting LO leakage 100% is very difficult... so in some of the RF input applications I have worked on I worked around the problem by simply using a wider input bandwidth and offsetting the LO frequency away from the signal of interest. For example, if you care about a 1.5GHz center frequency with a 100kHz bandwidth, then I would shift the LO to 1.5GHz+150kHz, and then expand the input bandwidth to at least 200kHz, and throw away the data that is out of range.
Thank you Logan_K. I had been using the offset LO technique, assuming that it was a Band-Aid and not a legitimate work around.
Recommendation for a better forum to post?
There appear to be both USRP and RF Devices forums that would be good for this sort of question: