BOOSTXL-AOA: Is this the expected performance from the BOOSTXL-AOA?

Hi Michael,
I have assigned an expert to get back to you with feedback.

Thanks, I should mention our concern is the large std after 1.5m. Don't know if this is typical or specific to our setup.

Can you specify which SDK you are using and if you have applied all the fixed in the known issue post?
e2e.ti.com/.../770218

HI Christin,

It's 2.40.00.32.  It seems most of the AOA related bugs in 2.30.00.28 might be fixed. Take a look at this new graph as well showing the angle measurements. It seems it reads well to about 1.5m.  The angle error reported here

is < 4 degrees at 2.8m. But TI does specify this to be in an antiechoic chamber.  Do you have any real world measurements that may indicate our measurements are good or whether we are making a mistake somewhere.

Thanks!

We have briefly tested the AoA calculation in chamber with attenuator however we did not see this huge variation.

Do you have to have rssi result when the distant are larger than 2.7m?

I've attached a graph showing the RSSI from -90 to 90 degrees to 2.6m.

Also, we the receiver was placed 1m above the floor and the transmitter was held by a person standing directly behind the board with transmitter and person in the same radial line from the receiver. Also, the planes of both the receiver and transmitter boards were parallel to the floor. Do you think having a person hold the receiver may cause such noise?

Also, attached a pic of the testing space and should also note there is a good amount of active Bluetooth and Wi-fi activity in this environment as well with windowed walls about 1.5m on two sides next to the receiver. 

Have you tried using only 1 array for the testing?

There could be a selection problem when using our OOB example, because we are comparing rssi value from array 1 and array 2. However, array 1 and 2 are receiving packets on different data channel, the rssi values is not really comparable.

We are working on reproducing the result but it might take us a while before getting back to you.

Hi Michael,

Would it be possible for you to collect the data in AOA_MODE_PAIR_ANGLES mode?

In test_rtls.py use:

passivenode1.rtls.aoa_set_params('AOA_PASSIVE', 'AOA_MODE_PAIR_ANGLES', 4, 4, 20)

instead of:

passivenode1.rtls.aoa_set_params('AOA_PASSIVE', 'AOA_MODE_ANGLE', 4, 4, 20)

Best Regards,

R.M

Best Regards,

R.M

Hi RM,

Sure, we'll give it a shot. What does that do?

We did give that a shot. Thought maybe switching between arrays might shuffle things a bit, but it did actually seem to perform better with both. But I'll try again, maybe I screwed it up.

thanks!

Hi

AOA_MODE_PAIR_ANGLES will output the delta phase measurements (compensated) between the antennas for both arrays before it try to guess which array to trust based on rssi. With that data we can see if the problem you are seeing is in the array selection or if it comes from the actual phase measurements.

Best Regards,
R.M

Hi Christin,

Average wise seems stable using one array, but the variance is shown as in the attached picture at 2.5m. Is this expected?

thanks

Hi, 

Is there any progress for how to solving big aoa-errors?  

Hi,

Sorry for the delayed response.
Were you able to run the test using pair angle mode? 

If yes, can you attach the result?