We increased mechanical downtilt by 2° on a 3.5 GHz 64T64R (n78) sector with a 2.7 m panel and UL BLER jumped from about 4% to 18% past about 1.2 km even as DL RSRP improved about 3 dB. Would you fix this with E‑tilt reshaping or a slight cross‑tilt on the 700 MHz anchor to pull PUSCH in, or is this more likely a vertical pattern/beamforming null that needs a narrower HPBW panel?
I’ve fixed this on a 3.5 GHz 64T with a 2.7 m panel by nudging e‑tilt up +0.5° and slightly reweighting the upper subarray, which pulled the “UL BLER past about 1.2 km” back under about 6% without giving up the ~+3 dB DL. Small caveat: cross‑tilting the 700 anchor did pull PUSCH, but it skewed HO borders more than it helped, so I’d try the e‑tilt reshape first, @maria-k.
Seen this on n78 64T with a 2.7 m panel; the 2° mechanical tweak can drop a vertical null right at about 1.2 km and raise ICI — “UL BLER past about 1.2 km” fits. Before cross‑tilting 700, try SRS‑driven UL elevation sweep/port reweighting and +2 dB on P0_PUSCH for edge UEs; if the null persists, back off mech by 0.5° and add a tiny digital up‑steer instead. @camille_ross76 is on the right track; also check your TDD pattern gives enough UL symbols, or you’re just stepping harder on the same hose.
@camille_ross76 makes sense on the null; if you’re SA, cross‑tilting 700 won’t pull NR PUSCH. We’ve clawed back that ‘18% BLER’ on a 3.5 GHz 64T64R after a 2° mech change by bumping n78 UL P0 +1 dB and SRS +2 dB for edge UEs, then a tiny 0.3° electrical uptilt on the upper rows.