Messy Constellation from the ZCU111 MVDR Beamforming example

Hi all, I ran the design at https://github.com/mathworks/FPGA-Adaptive-Beamforming-and-Radar-Examples on ZCU111. Everything is fine except that the constellation is messy. The weird thing is the graph is constantly messy but at some moment, very shortly, it shows good result---yellow dots tightly around four red x. The only difference between my experiment and the author's is that I use VFE-100 instead of XM500 as daughterboard. The VFE-100 has baluned single-ended I/Os while XM-500 has unbaluned differential I/Os(used in this experiment). Could that be the reason? Or it is something with the clock?

Thanks

Answer 1

[UDAYA PEDDIRAJU]

Hello Yang,

As per my understanding, the issue you’re facing is with the IQ samples distribution with respect to the ideal constellation points. As per the repository, you must use “XM500” daughterboard. The issue you’re experiencing with the messy constellation could indeed be due to the difference in the daughterboards you’re using. The “VFE-100” and “XM500” have different I/O configurations and it could affect the signal integrity.

You can follow the below steps in case you’re unable to arrange the same setup as per repository including the peripherals:

1. Signal Integrity: Ensure that the signals are being properly conditioned and there are no issues with signal reflection, especially given the single-ended nature of the “VFE-100”.
2. Clock Stability: Verify that the clock signal is stable and clean. Any noise or jitter could potentially impact the constellation.
3. Board Configuration: Make sure that the “ZCU111” board is properly configured to work with the “VFE-100”.

I hope this helps!!