Changes from 3.1 to 3.2?

Greetings,

I have just upgraded from GNU Radio 3.1 to 3.2. At the same time I
upgraded from Ubuntu 8.04 to Ubuntu 9.04. The hardware setup is a USRP1
with an LFTX daughterboard.

Under the previous setup (8.04+3.1), I could generate signals with an
amplitude up to approx 1V. Above that the signal was severely distorted,
which fits very well with what I’ve been able to read about the
hardware.

With the new setup (9.04+3.2), the same code generates signals with an
amplitude of approx 0.28V, and above that the signal again is severely
distorted.

This bugs me. The hardware should definitely still be able to produce 1V
amplitude. I have tried this with two different USRPs and two different
daughterboards and also with two host computers.

Before I do any more debugging and start producing a small enough
example, let me ask: Has there been any changes from 3.1 to 3.2 that
could explain this behaviour?

Regards
/Mikael Olofsson
Universitetslektor (Associate Professor)
Linköpings universitet

Mikael Olofsson wrote:

With the new setup (9.04+3.2), the same code generates signals with an
amplitude of approx 0.28V, and above that the signal again is severely
distorted.

The old Python code has a line that sets the gain to midpoint, but it is
conditionalized out, leaving it at the maximum. The current C++ db code
sets it to midpoint in the constructor.

.28V is roughly 10 dB down. The default is now -10 dB, which is the
midpoint between 0 dB and -20dB.

You need to send a call to set_gain to set it back to the maximum of 0
dB.

Matt