Weird Behavior of BPSK on RFX 2400

GNU Radio
#1

Hi guys,
I am trying to analyze the functionality of each of the blocks in the
dbpsk
mod/demod when I noticed the following behavior.

The high level modulator flow control is as | Symbols |-> | RRC filter
|
-> |amplifier | -> channel
and that for the demodulator is channel -> | Pre-scaler|
->
|agc| -> |RRC | -> |clock recovery| -> |slicer|

My understanding is that the real part of the output of the clock
recovery
must be close to that of the symbols or some scaled version of it(+a, -a
in
case of bpsk).

However, my observations (in the link below) show that the output of the
clock recovery depends a lot on the bit pattern, i.e., for the bit
pattern
1010 the peaks of the 1s and the 0s are way smaller than those of the
bit
pattern 1111110000.

  http://nms.csail.mit.edu/~gshyam/data/clock.pdf

Since we are using a RRC I would doubt it is due to ISI . and since the
radios are close to each other and there is a direct LOS I donot think
this
is due to multi-path (or is it?). Is this behavior due to some
component on
RFX 2400/USRP or something else?

thanks
Shyam

View this message in context:
http://www.nabble.com/Weird-Behavior-of-BPSK-on-RFX-2400-tf4264711.html#a12137123
Sent from the GnuRadio mailing list archive at Nabble.com.

#2

In the clock recovery, the useful term in the output of the M&M error
detector becomes small for long sequences of identical symbols. This
affects the sampling instant, and I suspect is related to what you are
seeing. You can modify the gain_mu value to get these and similar
effects.

-TT