Official bladeRF forum

I've attached some time domain plots of the raw samples received when looping back a 100 KHz sine through an external cable and 30 dB attenuator. (in-phase red, quadrature yellow, envelope blue). In the first image, TX and RX are both tuned to 176 MHz, while in the second Tx is 176 MHz, Rx is 166 MHz, and the signal is brought digitally to baseband. It appear that clipping occurs near 1300 and -1300 for the quadrature component in the first image, which is around 63% of the ADC range of [-2048, 2047]. In the second image, the envelope clips, but the IQ components show even stranger behavior. If I dial down the Rx gain by 1 dB, the signals are fine, with the envelope spanning the range [-1300, 1300].

No matter what Tx/Rx gain combination I try, I am never able to obtain good IQ samples past 1400. Is there a problem in my setup? If so, what can I do to fully exploit the dynamic range of the ADCs? Otherwise, how would you explain this limitation?




Sample rate: Tx, Rx 40 Msps
Bandwidth: Tx, Rx 35 Mhz
Manual gain control: Tx 60 dB, Rx 18 dB

Configuration details:
Host: MacBook Air 7,2 macOS 10.14.5, libbladeRF 2.2.0 (Same problem on Windows 10). Samples obtained from C++ application (Same problem on cli)
FPGA image: 0.11.0 Built from dev-fixes March 30, 2019 commit. (Same problem on hosted 2018-12-26 - v0.10.2)
FX3 Firmware: 2.3.2

Is this problem still persisting? Please try commit 994b19a73ea , there a few important commits that were recently pushed that might help.

Does this behavior present itself if you use something like gqrx? What is the CW's input power?