Emitting Power Issue of USRP N210

Dear all,

I have questions about the transmitting power of USRP N210 with WBX
daughter board.

I made a simple DBPSK modulation flow graph(please see fig.0),
basically,

File Source -> Packet Encoder -> PSK Mod -> Multiply Const -> USRP Sink

|

-> FFT Sink A

USRP Source -> PSK Demod -> Packet Decoder -> File Sink

                         |

                         -> FFT Sink B

Between USRP Src and USRP snk, there is a cable connection with a 30dB
attenuator. (please see fig.1)

Since I want to know the emitting power, I use a power spectrum analyzer
connected with the TX/RX port of N210(please see fig.2).

First I run the GRC (GRC inside computer -> USRP N210 TX/RX Port ->
cable
with 30dB attenuator -> power spectrum analyzer), I could read the
emitting
power from screen of power spectrum analyzer.

I think the transmitted power = Amplitude ^2.

So for example, if I get an amplitude of -17dBm reading from power
spectrum
analyzer, then the power emitted should be of around -34dBm.

I start the flow graph again, from the FFT Sink A connecting with USRP
Sink, I see a much lower power, about -80dBm.

If I set a 5dB gain inside the USRP Sink, I think the transmit power
should
be -80dBm + 5dB. However, this -75dBm is not equal to what I measured
previously from power spectrum analyzer i.e.-34dBm.Why?

My question :

What do I really read from the software FFT Sink A?

I see a unit of dB in software FFT Sink, what is reference to this dB
value?

Any suggestion for how to correctly measure the emitting power from
TX/RX
Port of USRP N210 ?

I am really appreciated if any one could answer these questions!

Thanks in advance!

Weihua