5g home internet and gaming RSRP/SNIR Help

With cellular metrics as you have it is clear the n41 cellular signal is pretty solid. The source of the problem is not your connection to the tower/cell. OK so the receive power could be improved but -90 is still higher than ours so actually that is still in the good range. An external 4X4 MIMO antenna might push it up to the upper end of the range or into the excellent range but at a cost of ~$400 or just a little more. Waveform has some offerings and excellent guides that provide info on how to make the connections etc… and where to place the antennas. Be aware T-Mobile does not support the use of external antennas and if you damage the gateway in your endeavors that's on you. The connectors in the gateway on the board are delicate and very easy to break so extreme care is warranted.

If you are in an urban location, higher home density, the behavior is probably more related to bandwidth throttling to distribute the traffic flows more balanced for all users on the cell. More than likely your traffic reached a level where it triggers the throttling. The 5G cellular solution is pretty good but not the most gamer friendly approach. In an area where there are more subscribers on the cells the bandwidth throttling seems to be a bigger issue. 

Helpful chart that explains the cellular metrics:

If your client(s) have 802.11ax wireless network adapters there could be a local wireless explanation which is not common but in some locations where there is a higher density of wireless signals and radar emissions it can come into play. The use of 802.11ax WIFI6 support DFS which may leverage extra bands that can be useful but also lead to channel switching. The operation of the master, i.e. the gateway to the slave i.e. the client adapter establishes a communication between the two where the master has to monitor for radar use on the DFS bands and notify the client/slave there is a channel transition necessary. There are other conditions that can lead to false positives which also elicit a channel transition so it is not a perfect solution. A message in effect is sent from the master to notify the slave of this impending channel transition and then the transition takes place. The only way you might be able to see if this is taking place is in the system logs if no message pops. A transition should be pretty quick but there can be some delay which is noticeable by real time sessions. With clients that cannot use the DFS bands this should not be a problem. A client with 802.11ac or earlier would not be subject to such behavior as they would not support the extra channels. 

If you are interested: (Warning networking geek alert!)

http://wifinigel.blogspot.com/2018/05/the-5ghz-problem-for-wi-fi-networks-dfs.html

I doubt the DFS is related just passing on information. The behavior is probably due to bandwidth throttling to balance subscriber loads during peak hours. Just my speculation as I have no hard data to support the theory but the practice does take place so the probability is there.