Even if the two boards have slightly different frequencies, this should
not impact decoding of the receive signal as the received signal jumps
between the I and Q channels (depending of course on the packet lengths
and assuming that you have a packet decoder on I and Q separately).
More seriously using separate uncorrelated LO signals for transmit and
receive significantly degrades receiver sensivity. The transmit signal
is typically 30dBm+ and the same antenna or nearby antenna is used to
get the receive signal - the received signal has this huge transmit
signal along with a -60dBm backscattered signal from the tag that is 50
to 200kHz away from carrier. If the same LO is used for transmit and
receive, then at the dowconversion mixer, there is a high degree of
correlation between signals at the LO port and the RF port (esp. the
transmitter leakage), and much of the transmitter noise shows up as a DC
offset. If separate LO’s then the noise at the two ports are large
uncorrelated contributing to the baseband noise. You can expect 5db to
20db difference in SNR between using same LO’s vs separate LO’s
— On Tue, 4/19/11, Colby B. [email protected] wrote:
From: Colby B. [email protected]
Subject: Re: [Discuss-gnuradio] Odd use of LO phase lock feature on USRP
for RFID application
To: [email protected]
Cc: “Matt E.” [email protected], “GNU Radio D.ion”
Date: Tuesday, April 19, 2011, 5:00 PM
The two boards should have different clocks, so there should be some
frequency offset. Even in typical SISO systems, you use a PLL block to
deal with this since you can’t access the other LO because its
physically somewhere else.
While receiving, the transmitter is still running at full power to run
the RFID tag. The transmitters carrier is down converted by the receiver
board. Unless I have a misunderstanding, and the two daughter boards
share the same clock there should be some frequency offset.
On Tue, Apr 19, 2011 at 12:44 PM, [email protected] wrote:
On Tue, 19 Apr 2011 11:41:46 -0700, Matt E. [email protected] wrote:
On 04/19/2011 11:38 AM, Colby B. wrote:
In RFID applications, a reader receives (backscatter from RFID tag) and
transmits (constant tone) at the same frequency. With commercial
readers, a single LO will be shared by the RX and TX chain. However, in
the USRP case, two separate daughter boards are used so different LOs
are in use for the RX and TX chain. So you should end up with some
frequency offset in RX chain due to mismatched clocks.
Is it possible to lock the LOs of a TX daughter board and a RX daughter
board, as you would for a traditional MIMO 2 TX or 2 RX setup? There
appears to numerous discussions and examples of the latter. I’m thinking
it would be possible. But I’m more of a systems guy and less of a RF
hardware guy, so any comments would be appreciated.
As long as you set them to the same frequency, they’re already locked.
No need to do anything different.
Discuss-gnuradio mailing list
True, for a SISO system with TDD(not FDD) theres no problem for your
kind of application.
-----Inline Attachment Follows-----