AoA: IQ samples sampling time and their correction

[Ales]

Hi!

There is two things confusing to me about taking IQ samples in AoA application.

1. It is clear to me that that the time between switching of antennae (TSWITCHSPACING) is dependent of sample slot and switch slot which can be 1 us or 2 us. So if we take sample slot of 1 us for example, then my understanding is that I will get one IQ value for every 2 us. That means I will have 8 IQ samples for reference period and then one IQ sample every 2 us in the interval after reference period. But after reading a couple of problems related to AoA on this forum by other engineers I saw that they are getting far more IQ samples (For example in https://devzone.nordicsemi.com/f/nordic-q-a/59322/how-to-get-right-digree-nrf52811-direction-finding/240869#240869.  They have 125 ns time period of each IQ sample). So my question is: are IQ samples taken continuously in sample slot and after the switch happen and saved to RAM (meaning more than one IQ sample in one sample period) or is taken exactly one IQ sample in one sample period and saved to RAM (so I have as many IQ samples in RAM as there are sample slots in entire CTE + samples taken in reference period)? Also this question and answers: https://devzone.nordicsemi.com/f/nordic-q-a/55626/dfe-iq-sampling-and-antenna-switching-slots-confusion contradict my understanding completely. Is sampling time independent of duration of sample slot?

2. Second question is related comparing phase information of IQ samples between antennae. For phase comparing between antennae I need phase information for each antenna in exactly the same time taken. Since one IQ sample in CTE is delayed from previous IQ sample depending of the size of sample slot and switch slot (2 us or 4 us) then I don't get samples taken at exactly the same time. So do I need to do the correction for each sample taken (except the reference one) to make it appear as they are taken at the same time (
from taken phase subtract a phase difference that RF wave would make in 2 us or 4 us period of samples taken) or is this correction done beforehand and values read from RAM appear as they would be taken at the same time?

Hope my questions are understood. Any help would be appreciated.

[Øyvind]

Hello, 

My apologies for the late reply. 

This should be covered by the nWP036 - Direction Finding white paper. Please point out what is unclear in the document, then I will need to discuss with our AoA experts.

Thank you.

Kind regards,
Øyvind

[Piotr]

Hi Jason and Ales.

Your assumption is correct. Radio peripheral takse samples one after another for antennas configured by antenna switching pattern. To evaluate angles according to method described in whitepaper nWP036 you have to apply correction to received IQ samples. You have to correct phase by angle related with the delay between first and particular sample. 

So e.g. samples spacing is 2us

- sample 1 -> 0 correction

- sample 2 -> 2us correction

- samples 3 -> 4us correction etc.

Then all IQ samples should have values as if they were taken in the same instant.

[Jason Lee]

Hi Piotr.

 

Thank you so much. That's what I need to know.

 

BTW, would you tell me more specific way that how to apply correction to received IQ samples?

For now, I'm calculate average phase diff. in ref. period. And correction with that value.

Is there any guide or reference for this work?

 

You answer relieved me. I genuinely appreciate that.

 

I wish you have a nice day. thank you.

[Piotr]

Hi Jason,

I do not know about reference approach for phase correction, maybe Oyvind has heard something about it.

Your approach seems to be ok. Reference period is aimed to be used for such evaluation. As a hint I can add (I believer you know about it, but just to remind it) that there is a frequency drift (allowed by BT Core Spec.) during whole PDU (including CTE) reception period. So the error evaluated from IQ samples taken in reference period may grow in time.

BR Piotr 

[Øyvind]

Hello all, 

---

Last week, NCS v1.6.0 release candidate 2 was released. In this a new version of our Direction finding examples using AoA/AoD in the Bluetooth Direction finding Beacon and Locator samples in the latest nRFConnect SDK. These new examples are similar to the previous versions, but are now compliant with the BLE 5.1 Direction Finding specification, while the previous ones were a proprietary AoA solution.

---

I am not aware of a reference approach at this moment, but I hope to get a better overview when the official NCS v1.6.0 tag is released.

Kind regards,
Øyvind

[Jason Lee]

Thank you, Piotr.

I think I good enough for now.

Your answer is so helpful.

Have a nice day.

[Øyvind]

Hello, needed to check what the reference approach it, and currently this is something customers need to implement by calculating angle based on the nWP036 paper. The samples I linked to below will receive IQ samples only. 

Kind regards,
Øyvind