Re: [Discuss-gnuradio] why two switches and two Rx in RFX series daughter boards?

[zhifeng chen]

Hi Matt,
Thanks a lot! So, you means I just need connect between Tx/Rx and leave Rx2 unused for RFX 900. But I want to implement FDD (frequency division duplex), it should use Rx2. Do I need to make change to hardware? how?

Thank you!

2007/7/6, Matt Ettus <address@hidden>:

zhifeng chen wrote:
> I am wondering why there are two switches in the RFX series daughter
> from schematics. It seems their purposes is different: one for switch
> between Tx and Rx, another switch between Rx and Rx2. So, my questions as:
> 1. we may use one tx and rx enough, why need to use two Rx?

The RX2 port is for those people using external amplifiers who need RX
and TX on separate ports.

> 2.I follow the python code into usrp.pick_rx_subdevice (u) and the
> return value is side and subdev. Since there are two rx there, how may
> I know subdev refer to rx or rx2?

The subdev is not used on this board.  The select rx antenna function is
used to choose between the 2 antennas.

> 3. if we call set_auto_tr(), it refer to which switch?

It controls all switches.

>
> I also test usrp_fft.py and find both Tx/Rx and Rx2 have signal input
> (by using the same usrp_siggen.py), while input from Tx/Rx is stronger
> than Rx2. Since Tx/Rx is followed by a filter, it should be less than
> Rx2, so, what happens here?

The RX2 signal is weaker by about 25dB because it is disconnected.  The
switches don't have infinite blocking.

>
> I use two sets of USRP with RFX900 to test a FDD( frequency devision
> duplex) by cable connection. According to my experience, the cable
> loss should be less than 0.5-1 db, right? (which come with RFX900) So,
> If I set the same gain for two RFX900 tx, say -a 100, the received
> signal is much smaller than the coupled signal from itself, whatever
> the different frenquency is. So, why there are both input signal? it
> is coupled signal or intentionally designed? I think it is intentional
> designed, and since the tx signal is routed to Rx2 directly, so it is
> much stronger than come from its peer end, is it right?
>

You should NEVER connect the TX output to an RX input without a lot of
attenuation in between.  You should also not be using RX2 in your situation.

Matt