I am using a Xilinx ZCU102 board + ADRV9009 to transmit a CW RF signal via a cable, split between RX1 and RX2 on a second ZCU102 + ADRV9008-1 where I measure the phase difference between Rx1 and RX2 over 256 K captured samples. I am using IIO Oscilloscope and the latest SD card image on the Transmit setup. I have the 30.72 MHz ref connected as required. The 9009 is set to 600 MHz TRX LO and a 1 MHz single CW tone (actual 999390 Hz), 0dB, 90 degrees. Profile is 100 MHz bandwidth 122.88 MHz sample rate. I have different length cables between the splitter and RX1, RX2 to cause a frequency dependent phase difference. I am measuring the phase difference using cross correlation using a window of 198 samples, sliding this window along by 64 samples at a time and re-evaluating within my captured samples.
Using the 9009 as a source, I see +/- 9 degrees of phase variation appearing as a regular sine wave in time. Using an external Signal Generator set at 601 MHz, with the same 9008-1 receive setup, I see around +/-0.1 degrees.
Is it possible that the ADRV9009 is generating the 600 + 1MHz CW signal with a small amount of FM modulation?
I have not enabled any tracking, or applied any calibrations, just loaded up IIO Oscilloscope and entered the frequencies required.
Are there any settings required to generate a purer sine wave ? Should I use I/Q waveform mode instead ?
Many thanks for any suggestions,
After consultation and testing with an RF expert, we fixed this issue by inserting a suitable bandpass filter. The RX1 to RX2 phase measurement was then stable. The 9009 and 9008 development boards quite reasonably do not have RF filter components. I am surprised that there was sufficient out of band energy to cause this error, but am very happy to see the problem solved.