Greetings,
I am experiencing difficulties with unwanted transmitter harmonics in
the FLEX 400 daughtercard. I am attempting to transmit a signal at 460
MHz and simultaneously listen for a responding signal at exactly twice
that frequency, 920 MHz, with a FLEX 900. My USRP is configured to send
and receive a 320 kSa/s complex signal both ways, with an interpolation
of 400 and a decimation of 200.
GNU Radio appears to be working as expected, but the transmitter on the
FLEX 400 is producing a harmonic on the frequency I’m trying to listen
to. My setup is as follows:
TX Signal USRP LPF
320 kSa/s --> usrp_sink_c --> Interp 400 --> FLEX 400 --> 700 MHz
–> TX Antenna
10 kHz wide
RX Signal USRP
320 kSa/s <-- usrp_source_c <-- Decim 200 <-- FLEX 900 <-- 50 Ohm
Terminator
10 kHz wide
I tested the USRP with a spectrum analyzer and a near-field probe.
Attaching a low-pass filter, using a narrow-band antenna, and closing
the USRP’s case is sufficient to stop the harmonic from radiating into
the environment, but that doesn’t do anything about its presence inside
the enclosure itself. I think the harmonic might be directly coupling
with the receiver daughtercard, but I need to do more testing to be
sure.
The power level of the harmonic (as seen by the receiver) is roughly 25
dB above the noise floor. Strangely enough, the harmonic gets stronger
if I lower the transmitter’s gain (by applying a gain in software of
1000 instead of 10000). Since I’m using maximum interpolation, could
this be an artifact of the interpolation process?
Is there any way I can increase the electromagnetic compatibility of the
USRP daughtercards—besides just putting them in separate USRPs? Is there
any additional shielding or grounding that might help? Is it possible to
build an extension cable so I can put the FLEX 400 in its own enclosure,
or would this cause too many reflection/timing issues?
This problem is not unique to the USRP—even the very expensive signal
generators I’ve used seem to have this second harmonic. I’m going to
experiment with this a bit more and see if I can find a solution.
Thanks,
Colin S.
Mo S&T Electrical Engineering