AoA IQ samples analysis

[AllenWang]

Hi, 

I have run up the connectionless rx aoa sample based on NCS2.1.2. and get the cte IQ samples. just as below:

as the white paper's guidance. the pre-8samples is reference samples. and from the index of 8, IQ sampled each 4us.  totally 160us  for 45samples.  it's seems to normally runed.

but as my understanding about the samples above, from the index of 8. the IQ sampled according antten's switchpattern. and loop from switchpattern[2] to swichpattern[9]. there are 10 antennas and petterns I used.

that means

IQ[8]----switchpettern[2]

IQ[9]----switchpettern[3]

IQ[10]----switchpettern[4]

.......

IQ[15]----switchpettern[9]

IQ[16]----switchpettern[2]

IQ[17]----switchpettern[3]

.......

IQ[23]----switchpettern[9]

........

 

thus,  the phase diff between(IQ[8] and IQ[9]) should equal with (IQ[16] and IQ[17]).  I am wonder if i got  the right understanding?

the phase for each IQ sample is.   p = atan2(Q/I),   (is this correct or not).

 

beside. the IQ samples's amptitude for each should be the same.  ( I[i]*I[i] +Q[i]*Q[i]) =. (I[j]*I[j]+Q[j]*Q[j]).  but the samples I got above does not satisfy this form. why and  how to understanding it.?

rgds

[Saleh]

Hi Allen,

I think the corresponding antenna pattern and the IQ sample index is not as the above. I'm assuming you're using the latest version of the Zephyr, then the effective antenna pattern written to the nRF52833 register would be:

{PDU_ANTENNA, patterns[0], patterns[1], ..., patterns[N], patterns[0]}

PDU_ANTENNA is declared in  as dfe-pdu-antenna in the the overlay file: github.com/.../nrf52833dk_nrf52833.overlay

patterns is the antenna pattern declared at the sample application source code: https://github.com/zephyrproject-rtos/zephyr/blob/main/samples/bluetooth/direction_finding_connectionless_rx/src/main.c#L48

And I suppose by "switchpettern" you're referring to the the pattern provided by the application, so the correlation between samples and antenna is:

IQ[0-7] --> ant_patterns[0]

IQ[8] --> ant_patterns[1]

IQ[9] --> ant_patterns[2]

...

IQ[8+N] --> ant_patterns[N]

IQ[8+N+1] --> ant_patterns[0]

IQ[8+N+2] --> ant_patterns[1]

...

For your reference:

How nRF52833 handles the effective antenna pattern: https://infocenter.nordicsemi.com/pdf/nRF52833_PS_v1.5.pdf (Switch pattern configuration Page 293)

How Zephyr writes the antenna pattern to the nRF52833 register: https://github.com/zephyrproject-rtos/zephyr/blob/main/subsys/bluetooth/controller/ll_sw/nordic/hal/nrf5/radio/radio_df.c#L93 (latest version of Zephyr on time of writing this answer is commit number acde5fe)

AllenWang said:

the phase diff between(IQ[8] and IQ[9]) should equal with (IQ[16] and IQ[17]).  I am wonder if i got  the right understanding?

By assuming the correct indexes are used, in the reality the phase differences will not be exactly the same. Because the environmental noises and errors of ADC capturing can affect on the results. But if everything was ideal, then the above assumption is correct.

AllenWang said:

the phase for each IQ sample is.   p = atan2(Q/I),   (is this correct or not).

I think so!

[AllenWang]

Hi, Saleh:

       it's most excited to get the reply from you, and it's very useful for me to promote my project implemetation.

      for first about antenna array:  I have used oscillograph to catch the antenna gpio wave. which I get is totally matched as your guidance.  so this is fixed.

      and for the IQ samples format, I think I made the right understanding as you said. but still I got another confusion On the IQ samples which get from the reference period and sample slot。

      1. from the White Paper， I got the information that the reference period samples is used to calculate the frequency offset t between the transmitter and receiver。But I am wondering how to calculate this？could you give some directly method according to your experience.

as my understanding about the IQ sampleing。within ideally environment， the raido signal send from transmitter during the CTE period keeps  as a stable frequency of sin/cos wave.  in the first guard time(for adjustment).  so the difference of the IQ samples we get  from the reference and sample slot is the sample frequency . and sample point is difference.

 actually I think the deep reason is  I cannot understanding the frequency offset between transmitter and receiver and influence of the offset would lead. could you please give me a clear words about this point.

     2. and the IQ wave when I use only on antenna, I got the same problem with this. but it seems still  not got a good solution.  AoA IQ data seem NOT uncorrect, when used one antenna? 

 Thank you,Saleh 

Have a nice day. 

[Saleh]

Hi Allen,

AllenWang said:

1. from the White Paper， I got the information that the reference period samples is used to calculate the frequency offset t between the transmitter and receiver。But I am wondering how to calculate this？

You can use the reference period as a reference for the received signal, and calculate the phase difference between each point and this reference part. Actually, we don't make any assumption about the absolute signal phase and everything is measured compared to this reference point. You know the CTE signal frequency (250KHz or 500KHz) and the time of each sample is taken, so you can calculate the phase difference of each sample with the reference one.

AllenWang said:

 2. and the IQ wave when I use only on antenna, I got the same problem with this. but it seems still  not got a good solution.  AoA IQ data seem NOT uncorrect, when used one antenna? 

If the transmitter sends CTE on 2M PHY, then the captured CTE signal will be 500KHz, if 1M PHY is used, then would be 250 KHz. It means if you use a single antenna at the receiver side, by visualizing the received IQ samples, you should be able to see a sine wave (If it's set to 1M PHY, then you can see this better). Please take a look to the images sent on this question.

Also you can change the sampling rate to the non standard values in order to capture the data faster and see the CTE signal smoothly. For example this line of code can be changed to RADIO_DFECTRL1_TSAMPLESPACING_125ns

It would be a good idea to capture and draw the IQ samples using a single antenna, just to make sure you are receiving the IQ samples properly (it should be a sine wave).