Re: Receive amplitude consistency

[Marcus D. Leech]

On 04/21/2021 11:03 AM, Mike wrote:
> Marcus,
>
> Thank you for your response.
>
> The variation is on the order of 20%, although it changes from run to run.
>
> Power level from the generator is 10 mV through 40 dB of attenuation.
>
> The B200 is set for maximum gain (minimum attenuation) and is using sc8
> format with peak=0.1.
>
>
> Thanks!
That's about 0.8dB variability.  Which is a bit more than I would expect 
with the way analog RF hardware tends to have
  repeatability issues, particularly as you go up in frequency.

What happens if you use 16-bit samples?   What happens if you lower the 
sig-gen output by a further 10dB?

The AD9361 chip goes through some calibration exercises (for gain and 
other things) every time it starts up, and with a strong
  (-30dBm at the input is somewhat strong) narrowband signal on the 
input, that calibration might get "confused".

I've copied the usrp-users mailing list, where this discussion is 
probably more appropriate.

> On 4/21/2021 10:42 AM, Marcus D Leech wrote:
> > What is the magnitude of the inconsistency?
> >
> > If you replace the signal generator with a 50 ohm terminator and increase the 
> > gain to near maximum, do you still see the inconsistency?
> >
> > What power level are you injecting from the signal generator.
> >
> > Sent from my iPhone
> >
> > > On Apr 21, 2021, at 12:27 AM, Mike <mike.nelson@rdss.com> wrote:
> > >
> > > Hello,
> > >
> > > I'm hoping someone can shed some light on an apparent inconsistency
> > > issue I'm seeing with received I/Q amplitude.
> > >
> > > I have a B200 directly connected to an arbitrary waveform generator
> > > through an attenuator.  The generator is programmed to repeatedly send a
> > > fixed sequence of AM pulses at a set power level.  On the host (Ubuntu
> > > 20.04, UHD 3.15.0), a C++ program simply gathers samples (48 MSPS, sc8
> > > OTW and CPU) and computes the average amplitude as I^2 + Q^2.
> > >
> > > During a run of the host software, the reported amplitudes are
> > > consistent from sequence to sequence.
> > >
> > > However, if the host software is stopped and immediately restarted,
> > > without changing the generator, the reported amplitudes are again
> > > consistent but at a different level than the previous invocation.
> > >
> > > Is there some internal adjustment or variable setting performed during
> > > USRP creation and setup that would cause the received I/Q sample
> > > amplitudes to vary with each startup?
> > >
> > >
> > > Thanks!