IQ Samples in Reference Period - AOA

[SRINATH]

Hi,

Iam working on direction finding using AOA technology. I have nrf52833 board integrated with Core-hw 4*4 antenna array which acts as locator. The slot duration I have set is 2 us .That is 2 us  switching and 2 us sampling period. Iam trying to do phase compensation with the average phase difference in reference period. From this attached context (https://devzone.nordicsemi.com/f/nordic-q-a/75446/aoa-iq-samples-sampling-time-and-their-correction) My understanding is that the phase angles in the reference period (i.e for first 8 samples) can be different, but its phase difference should be regular or equal. In my case the phase difference calculated in reference period is in the range of -102 to 107. Can you suggest what is the issue for getting the reference phase differences in this range or unequal values.

How can it be corrected?

[Simonr]

Hi

What kind of environment are you testing in here, and is this the phase difference from one antenna that differs this much, or is it between all the antennas you're using? You need to measure the phase on all antennas against the reference slot. The reference phase should not be static, as the frequency on transmitter and receiver are not the same. The phase should and will drift in a linear fashion. 

Is the antenna array you've designed yourself or is it one bought by a third party for example? By default our sample projects use the Insight SiP AoA antenna array with 12 antennas, that work as expected in our experience.

Best regards,

Simon

[SRINATH]

1. It is the phase difference obtained from reference antenna since antenna switching will not happen during reference period.
2. we have used CoreHW 4*4 antenna array with nrf52833

We agree that reference phase will not be static but phase difference should be approximately equal as of our understanding. we calculated phase difference for first 8 IQ samples (taken during reference period) and it varies a lot. what could be the cause?

We have attached our IQ sample for your reference

L2 (2).csv

[Simonr]

Hi

An unstable reference period is usually expected in a noisy environment. The results need to be filtered, and you should drop the ones with large differences form the mean value. It would be best to do initial testing in a noise-free environment (outside for example) to ensure that this is not a HW issue.

You can also try shortening the slots as much is the RF switch allows to get better results.

Best regards,

Simon