nRF52840 - Unreliable RSSI

[Igor]

Problem:
We have a huge variability in received RSSI (see picture) despite both devices being stationary in test setup. For our use case, accurate RSSI is critical. 
nRF52840 datasheet shows RSSI accuracy +/-2 dB. 

Question:
Why do we have such a huge variability of RSSI?
What can we do about it?

Test hardware setup:
beacon GAP advertiser (stationary)
beacon GAP receiver (stationary) with filter for our beacons
both devices use the same nRF module.

Envirioment:
nRF52840, Raytec module
nRF5 SDK 17.0.2
S140

[Håkon Alseth]

Hi,

 

Test hardware setup:
beacon GAP advertiser (stationary)
beacon GAP receiver (stationary) with filter for our beacons
both devices use the same nRF module.

Is this a radiated measurement?

If yes, then this is expected to vary. See other threads (search for "rssi radiated" or similar on this forum), for instance this:

 RE: RSSI Instability: Seeking Insights on nrf52840 

 

Kind regards,

Håkon

[Igor]

These measurements are done on RX side.

Please check the attached video of specral analyzer with lab-grade log-periodic RX antenna.

Despite having the same nRF transmitter, we are unable to replicate this RX variability with our lab-grade antenna. 

So, the question remains, why do we get this RX RSSI variability? And mostly, what can we do to mitigate it?

[Håkon Alseth]

Hi,

 

Can you please confirm how you are measuring the RSSI?

If it is conducted, ie. connected via a coaxial cable, then you should see little variation.

If radiated, ie. antenna to antenna, you will see high variation.

 

Kind regards,

Håkon

[Igor]

The spectrum analyzer is connected to RX antenna, i.e. signal goes through the air. 
TX - RX air distance is cca 30 cm in the test setup (in practice it will be meters).
Can we do anything to minimize this RX RSSI variability of the nRF?

[Håkon Alseth]

Hi,

 

It is expected that RSSI varies when measuring radiated.

Igor said:

Can we do anything to minimize this RX RSSI variability of the nRF?

Just to avoid any confusion here:

It is not from the nRF radio itself, it is due to reflections and other physical parameters, as listed in the initial link.

If you connect a coaxial cable between the transmitter and receiver, and place it in a shielded box; you will get accurate measurements.

 

Kind regards,

Håkon