On 05/31/2021 08:30 AM, Alexey Silyuk wrote:
HI everybody, I am new at mailing list.
These days I am writing an emulator for emulating tx/rx streams based
on uhd 3.15LTS
I am using device usrp2 device from uhd/lib.
As I saw in examples (rx_samples_to_file), I am creating a mock
streamer ( taken from uhd tests ).
and now i want to push samples inside the buffer.
I see that i can define type of samples, if am using 'short' type of
samples, I am creating uint32_t variable, 16 MSB i put real part of
sample, and 16 LSB i put imaginary part of sample (i am using little
endian)
After that i am calling usrp->recv() and i get samples as i defined (
spp is 2000)
(0,0)
(2,2)
....
(1999,1999)
as expected, according to data I push to the mock buffer.
now while i want to use type 'float' i am using for example
float real = 1.5
float imag = 2.5
and trying to push sample to the mock buffer, i am doing:
buffer[0] = (uint32_t)real;
buffer[1] = (uint32_t)imag;
and expecting to get after calling usrp->recv()
(1.5, 2.5)
but i get incorrect result (convertor settings: cpu_format "fc32",
otw_format "sf16" )
i am printing real and imag parameters as bitset<32> and see in real
part, in 16 LSB part of result, my 16 MSB bits from sample value 1.5,
and in imagine part zeros, in next sample i see in 16 LSB part of
result, 16 MSB bits of sample value 2.5. I mean one sample i push to
buffer uses whole sample after calling recv()
P.S i set scale factor to 1.0
What am I doing wrong to use the type 'float' and 'double' for mock
samples?
I am new at usrp so I don't understand what samples I should use, will
be glad to get references.
Thanks
Your OTW format is still SC16 -- normally only the *HOST* side uses
floating-point and no actual USRP devices that I'm aware of can
support floating-point *ON THE WIRE* (OTW).
_______________________________________________
USRP-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
