If there is no radio resources limitation, do parameters comparison with normal site and seems the main issue here due to DL grant which is not full 1600.
How much grant UE has in this case?
As per experience, TX issue could be one of the reasons, like TX port limited. Check how many bw 1Gbps or 10Gbps, is it the same as the normal site?
Faulty base band also can be a factor.
RIM configurations also can be a factor as well in this case need to check.
Also to rule out radio please check from same throughput testing these indicator:
CQI
DL BLER%
DL MCS Avg
DL 16QAM%
DL 256QAM%
DL 64QAM%
DL QPSK%
Rank 1
Rank 2
Rank 3
Rank 4
Also maybe you are checking backhaul from gNB to RTN which maybe 10G but from RTN-IIB connectivity has limitation of 1G, for this you need to involve TXN / Transport team.
Is there any packet loss? you can check rate from OSS.
Also check SCTP retransmissions of all SCTP links of the site // reflect to transport congestion.
Also explain the genre of your problem is it a UE or area specific, more importantly issue is with a moving ue(like your drive test results) or static (more of a complaint based)?
Reason is NSA some parameter setting will favour static UE instead of moving UE and vice versa due to different channel estimation algorithm / settings
NRDUCellPdcch.UlMaxCcePct: If this parameter is set to a too small value, there is a high probability that uplink services are restricted. If this parameter is set to a too large value, there is a high probability that downlink services are restricted.
UL_DL_CCE_RATIO_ADAPT_SW: After this function is enabled, the ratio of available uplink CCEs to available downlink CCEs is adaptively adjusted based on factors such as required uplink and downlink CCEs and CCE usage, and the setting of the NRDUCellPdcch.UlMaxCcePct parameter no longer takes effect. Adaptive configuration is enabled by default.