Arcadyan KVD21 5G which connectors for 5G external antenna

Thanks a bunch, iTinkeralot,

I somehow missed the 2x2 vs 4x4 perspective. I know that I am about 1-1.5 niles straight line from my tower so I thought a 2x2 would be fine. I did not consider the bandwidth perspective. 

 I will attach a screenshots from Ookla results. In the mornings and late evenings, results are always better. 

I will shop for a 4x4 antenna probably from Proxicast. 

Here is a Ookla screenshot with the Jitter. Geez, If I got these results all the time I wouldn't complain. But, by 8-9 am, downloads will drop to 30-45 Mbps.

Some of my T-Mobile metrics show my low SINR, which seems to go up/down based on the time of day. Late night and early mornings my SINR will go up to 9. It would be helpful to have a guide available in the form of a checklist for customers - here are steps you can take to improve your SINR (and things that impact the metrics). 

If you can get the cellular metrics in more detail understanding just what is going on might help a bit more. The speeds in the history are pretty inconsistent. Do you have the metrics for both the 4G & 5G signals such as the band and the bandwidth in MHz for them? I suspect in some of the cells if the bandwidth is less than 20 MHz and user load/demand is higher that contributes to less bandwidth per user session. Then it is more critical to have a clean signal to prevent packet loss that will result in retransmissions. 
Use cellmapper.net to confirm the locations of other towers/cells in the area as you consider antenna placement in relation to the tower you are using. Here I know there is noise from other RF sources towers the west/northwest. I have my gateway in a window with northeast exposure to get the signal from the tower that is due north just because the signal is cleaner than using the north facing window. Results are better so I know using the house as a shield can help. The other obscure issue can also be with reflected signals. Cellular signals seem to be subject to multiple external factors so it is tricky at times to avoid issues. 

I am using Cellmapper and did locate my n41 tower and b66 LTE tower. I just realized that maybe I am bouncing back-forth between n41 and b66? With the T-Mobile app I am unaware of how I can get more cell metrics, unless its through a 3rd party app? I will attach a few more screenshots which may be helpful. ¡Gracias! 

Here is the n41 map, which shows the towers farther north from my location. Is there a definitive way to tell which signal I am receiving or tied to? My T-Mobile app shows I can either be on n41 5G or b66 LTE.

Expand the details for the specific cells on the tower. On the left select details. There is a bit more you can see once you have the PCI for the cell. 

With the speed test results you last posted I agree it would be no problem but I an guessing it is somewhat due to user load and the backhaul routers. The signal noise is probably due to the volume of RF in the area. I would guess if you look for ATT and Verizon towers you would see lots of cells in that area. The noise could be from some type of industrial facility or emergency responder transmissions and such. 
You can set a preferred source vs allowing Speedtest.net to just go out and determine an "optimal" target to use for the testing. Then you have a reference so to speak. If you have not gone into the Ookia advanced settings and set them I suggest you do. The more you see the more you know. Also run fast.com tests as well. Free tools. 

I did not ask but are you using a MIMO panel antenna or a MIMO log periodic antenna? The log periodic is a more narrow directional and good for line of sight which might help improve gain and prevent unwanted noise. It depends upon the location and possible obstructions due to trees, buildings and such. Lots to consider. 

If you haven't reviewed Waveform's guide for the T-Mobile Arcadyan gateway you really should. They have a very good guide.