High LO leakage in xcvr 2450 with new UHD?

GNU Radio
1

Hi All -

I’ve been using the new UHD driver (003.004.002) stable release w/ XCVR
boards on the USRP N200.

I’m using the xcvr as a transceiver - it receives a packet, then waits a
while before sending a packet. However, I can see some spikes in the
preamble detection stage which seemingly correspond to the time around
which the packet is transmitted. I know XCVR is not full duplex, so I’m
guessing it’s some kind of LO leakage, perhaps?

On replacing the UHD driver with an old one (003.004.000), the issue
seems
to have disappeared.

Any ideas? Attached is a snapshot which shows the smaller spikes
(corresponding to tx) next to the bigger ones (corresponds to the rx).

Thanks,
Apurv

2

On 07/14/2012 12:23 PM, Apurv Bhartia wrote:

Its just the the LO is directly in the center of the passband. I
recommend tuning with an LO offset:
http://files.ettus.com/uhd_docs/manual/html/general.html#tuning-notes

On replacing the UHD driver with an old one (003.004.000), the issue seems
to have disappeared.

Any ideas? Attached is a snapshot which shows the smaller spikes
(corresponding to tx) next to the bigger ones (corresponds to the rx).

This LO offset behavior was automatic for 003.004.002, however that was
causing more trouble than it was worth. For applications that tuned both
RX and TX. Because the setting of one side would stop on the other (due
to shared LO), and you would end up with incorrectly tuned CORDICs in
the DSP.

-josh