it’s great you could get a numerical value : the difference clearly appears in the measures.
It is troubling because there should be no difference : the BER is measured on the received bits. Whether the bits are AMR or SACCH is of no importance.
So what’s happening…?
1) the RxQual is computed differently during AMR calls : no.
2) or TEMS is not properly computing it properly : possible but weird
3) There is also another possibility : when using specific cyclic frequency hopping, the SACCH would always fall on the same frequency, leading to differences between the SACCH rxqual and the TCH rxqual… but i highly doubt your FH is set this way.
4) I’m sorry to consider this but perhaps DTX is enabled and you just misread your parameter settings (yes, i do have some trust issues…)
In the end, it looks like he problem comes from TEMS ?
Anyway, you just want to know which one to use… Well, you can try to do it the proper (& long) way :
drive the same road twice consecutively (to get same radio load conditions). 1st time with EFR, 2nd time with AMR. See if the problem also appears in EFR (it should not), then use the value from the AMR-call that is the closest to the one from the EFR-call. Indeed, the RXQUAL should not change whatever the codec in use.
My guts are telling me the RXQUAL SUB is better…
Thanks a lot for investigating this issue. That might become a good TAC ticket 🙂