Please double check your values : the rxqual is directly linked to the BER. Bad BER = bad RXQUAL.
What is the RxQual_Full?
The SQI also should be taking into account the FER… So bad FER = bad SQI.
Now the thing is : we could assume that the voice is poorly encoded by the TRX, or that the called part is sending “poor” speech frames. But in this case, the BER and FER would be good in the TEMS.
There is something wrong in your DT analysis… Are you looking at instantaneous values, or do you make an average over 5 consecutive seconds