Phase difference between the two RX ports of bladerf 2.0

[huiyellow]

I’m working on an SFCW radar project using the bladeRF 2.0 xA4, and I need to measure the channel response at different frequency steps.

In my recent experiment, I encountered some unexpected results. I connected Tx1 to Rx1 using a 30 cm cable with a -20 dB attenuator, and similarly connected Tx2 to Rx2 with the same length of cable and attenuator. I tuned the LOs of the tx and rx channels to 2.4 GHz, then used GNURadio to send a 1MHz sinusoid. The samples received from the two Rx ports were divided to extract the phase difference. Since the cables are of the same length and phase-matched, I expected to observe a zero phase difference. However, the observed phase difference was much larger than zero.

Next, I changed the cable between Tx1 and Rx1 from 30 cm to 60 cm and tuned the frequency of the LOs from 2 GHz to 3 GHz in 40 MHz steps. According to theory, the phase change for each step should be consistent and can be calculated using the formula 40MHz x 30cm / C. However, the phase changes I recorded between each step were random.

Could you help me understand the possible reasons behind these discrepancies?

[Patrick893]

It sounds like you're facing some intriguing challenges with your phase measurements. The unexpected phase differences could stem from several factors, such as imperfect phase matching or variations in cable characteristics. Additionally, the frequency response of the bladeRF system and potential LO phase noise might be contributing to the inconsistencies. It might be helpful to calibrate your setup and test with knownslope ball reference signals to isolate the issue further. Good luck with your SFCW radar project!

[huiyellow]

It sounds like you're facing some intriguing challenges with your phase measurements. The unexpected phase differences could stem from several factors, such as imperfect phase matching or variations in cable characteristics. Additionally, the frequency response of the bladeRF system and potential LO phase noise might be contributing to the inconsistencies. It might be helpful to calibrate your setup and test with knownslope ball reference signals to isolate the issue further. Good luck with your SFCW radar project!

Thanks Patrick893, for your suggestion

I think I might find the problem. I just ignored this issue and then start testing my implementation. Interestingly, the radar works. I can get the range profile and the accuracy is pretty good. When the Tx and Rx are connected with real antennas, and pointed to a mental target, the phase change of each frequency step changes almost the same amount.

Maybe the unexpected results I got by directly connect Tx and Rx together is due to the internal coupling?

Regards,
Hui