Unexpectedly large variation in raw IQ values from nRF5340 (±500 to ±1000) for MCPD ranging

[TG3112]

Hi Nordic team,

I’m working with raw IQ data from the nRF5340 using the Distance Measurement (nrf_dm) library.

I noticed that most IQ samples (both I_local, Q_local, I_remote, Q_remote) are in a typical range of about ±150–200, but occasionally, some tones show much higher magnitudes. Sometimes up to ±700 or even ±1000, even when the link quality (quality = 0) and distance are normal (e.g., 1–3 m).

These high-magnitude IQ samples are rare (a few out of 100k) and do not seem correlated with distance or quality flags.

My questions are:

Are these large IQ amplitudes expected due to internal AGC or ADC scaling effects?

Can they be considered valid measurements, or do they indicate over-saturation or invalid tones?

Is there a known nominal amplitude range for IQ values on the nRF5340 ?

Would it be safe to clip or discard samples where any |IQ| exceeds a certain threshold (e.g., 600)?

Thanks a lot for your clarification.

[Amanda Hsieh]

Hi, 

I assume you refer to the i_local, q_local, i_remote, and q_remote elements here, correct? These are IQ measurements where I means "In-phase" and Q means "quadrature" components. It could be anything between -2048 and 2048 as it is a 12 bit value. Won't see numbers outside 1200-ish though. 

Regards,
Amanda H.