MSK de-/modulation

Hi,

I figured out by now that the problem is not on how I decode MSK
within GNURadio, but that the signal I receive from the USRP is very
strange. To verify what my test code does, I loaded the received
message in octave and plotted the I and Q phases. A sample plot is
added to this email. I sampled the signal at a decimation rate of 8
which should give me 8 samples/symbol for the I and Q channel.

As one can see from the plot, this is not the case and there are a lot
of anomalies:

  • First two green peaks are exactly 6 samples from each other, same
    for the two red peaks just after that or the first red transition from
    low to high.
  • At around 30, the red signal has a bump in it. It could be decoded
    as two 0 or only one.
  • Just before 40, the green signal has a very wide swing, again
    questionable if it is two or one 0
  • Before 60, red swings towards low, though stops and goes back up.
    Again questionable if we should decode that as a 0 or did the signal
    just over-swing into the negative area.

Could it be a problem with the RFX2400 daughter boards? Any
suggestions on how to deal with this problem are appreciated.

Thomas