On 2022-04-11 07:07, Tobias Kronauer wrote:
thanks for your prompt reply! Indeed, my description was lacking a bit
of details.
In short:
- We transmit with 50MHz sample rate and 400MHZ analog bandwidth
(according to RF config setup)
- We generate a signal that consists of sections with pure (complex)
tones of different frequencies with constant amplitude, ranging from
-25MHz to +25MHz frequency in baseband (-25, -19.5, -14, -8, -3, 3, 8,
14, 19.5, 25).
- We receive the signal on the same USRP and plot the time domain of
the TX and RX signal. (first figure row)
- As visible, already the 8MHz tone is attenuated by around 3dB
compared to the center frequency. Higher frequencies are more
attenuated. This is also visible in the spectrogramm.
We know and expect that the passband of the X410 is not totally flat.
However, we do not expect the passband to be as severely shaped as we
measured. We know we cannot expect the edges to be not attenuated, but
actually we would have expected at least say -20..20MHz to be
relatively flat (for 50MHz sampling rate). Hence we wonder:
- if the problem can be reproduced on your (or other's side)?
- if the seen behaviour is correct and more flatness cannot be expected?
Thank you.
I updated the plot as well.
How are you connecting your TX and RX? Antenna or direct-wired? If
direct-wired, make sure that your cable has at least 30dB of attenuation
in it and that your TX
power levels aren't producing more than about 0dBm at the output.
With a master clock rate of 250MHz, 50Mhz delivered bandwidth requires
an odd decimation rate. This has historically caused a passband
response with a half-band
"shape" that is like a "hump" as you describe. I don't know whether
this applies to X410 or not.
It REALLY would be helpful for you to plot the RX response using a
normal FFT plot--with the gain turned up and a terminator on the RX
input, you should be able to see
the instrumental response of the device. The "uhd_fft" example
code with Gnu Radio produces nice plots for this sort of thing.
_______________________________________________
USRP-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
