It is a convention in GNU Radio that floats and complex floats are within the range -1.0 … +1.0. If I recall correctly, the gr-osmosdr block only support complex float sample type, even though SoapySDR supports different formats.
Hmmm - the gps sim I’m using, if I throw a scopesink on the flow into the osmosdr sink, shows a range of about +/- 1024, and it works fine. I Tried a mult-by-const (0.0009) then I get nothing in the receiver (not a thorough test).
Pretty simple test shows Alexandru is right - so I don’t know how the gps thing is even working well Here’s a .1 variable amplitude / variable freq signal into an osmocom sink, over the air to a source and fft displayed:
The signal is nicely behaved and responds as expected, but crank the amplitude up to 1.1 and heck breaks out:
So scratching head over the gps simulation example.