[Discuss-gnuradio] a problem with digital-ber/benchmark.rx

[Ran Tian]

Hi all,
I am working on BER measurement. I'm using benchmark_tx and benchmark
_rx in digital-ber folder. At the transmitter part i used 2.4MHz and other things set up as default. More info as follow:
/*--------------------------------------------------


>>>gr_fir_ccf:using SSE
Modulation: 250k bits/sec
TX IF rate: 500k samples/sec
USRP interpolation: 256
DAC amplitude: 2000
Center frequency: 2.4G
TX d'board:B: FLEX 2400 Tx
----------------------------------------------------------*/


At the receiver part:
/*-------------------------------------------------------
USRP decimation rate 8
Center frequency: 2.4G
RX d'board: A: DBS Rx
IF sample rate: 8M
Symbol rate: 250k
Samples/symbol: 32


RRC bandwidth: 0.35
>>> gr_fir_ccf: using SSE
Costas alpha: 0.05
Costas beta: 0.00025
Costas max: 0.05
MM gain mu: 0.001
MM gain omega: 1e-06
MM omega limit: 0.0001

*** SNR estimator is inaccurate below about 7dB


*** BER estimator is inaccurate above about 10%
------------------------------------------------------------------------*/

Now is my prob: 1. my USRP overrun all the time durning the measurement which means i received some "uO" at the beginning part, such as:


/*--------------------------------------------------------------------
uOuOuOuOuOFreq. Offset: -17279 Hz  Timing Offset:  -0.4 ppm  Estimated SNR:  6.6 dB  BER: 0.0362216
-------------------------------------------------------------------*/


and of course as u c the SNR always below 6.6dB which means the estimator would be inaccurate. So i reset up the transmit amplitude, then SNR became over 7dB. However, the BER still around 0.06. So whats the prob?



2. I found that even i did not run the transmit part, i still got 5.5dB SNR and 17% BER, I donno how that happened. Does it make any sense?

I'm just a beginner on gnuradio and USRP , so Anyone could help me with it? 
thx all.