Hi Martin,
Okay, thank you - so there does appear to be some variation coming from somewhere that I guess is not due to clock changes. My test was the following:
1.) Transmit with a single antenna and receive using uhd_fft. The received peak tone is very stable in magnitude (.01 dB or even less)
2.) Transmit with multiple antennas, and the peak tone has similar variation as above.
3.) Transmit with multiple antennas, but I call tb.start(), then tb.stop(), then tb.wait() in every iteration of my while loop, and I see that the peak tone changes by 1-2 dB. Maybe I'm missing something very simple. I figured it was because that the random but constant offset changes with every call to start(), so I thought it was due to maybe something happening in start() that I didn't know about, but maybe I'm just hallucinating.
And I am using the SBX-40 400-4400 MHz Rx/Tx daughterboards with the USRP N210s. Thanks again.
--