RAvirani

It's definitely a big step down. Hopefully Sprint/Samsung get this resolved soon. Have we seen similar problems in Ericsson markets?

How do you know they are using TDD config 5?

I've personally noticed that Samsung's M-MIMO equipment broadcasts LTE in TM7-SISO mode, which drops LTE performance significantly. I'm not sure why this is, but I'm sure once they get 4x4 working properly, we'll see a big uptick in speeds in their markets (from ENDC).

I can echo this experience throughout most of Seattle. 3x3 L800 is hard to find, but it always works when I do encounter it.

Yes. Both antennas are 2x low/4x high.

Size/weight/structural limitations are the most common reason. Carriers try to deploy bigger/heavier higher-gain antennas where possible and resort to smaller/lighter lower-gain antennas when necessary. Carriers also have different setups for different beamwidths (e.g. in my market, Verizon uses different antennas for sites with 90° sector spacing than for sites with 120° sector spacing). Lastly, carriers may use different antennas that are the same size due to their different gain characteristics. For example, Sprint stocks two different 72" octoport dualband antennas. Their 800/1900 gain figures are 14.7/18.0 for the RFS variety and 15.0/17.2 for the Commscope variety.

That looks like a T-Mobile 600-ready setup. Dual Commscope antennas.

Hi Robert, To follow up, over the last two weeks (since 8/5) I've had 276 Ultimate Coverage Map page loads. I can report back in another two weeks with more data, if you'd like.

QQualMin is one of the cell reselection parameters included in SIB1/SIB3. It is the minimum RSRQ for a UE to connect to a particular LTE carrier. When correctly configured, it can kick UEs off an LTE carrier at higher signal strength (RSRP) if the carrier is congested (noisy with a low SNR/RSRQ), while allowing UEs to remain connected at a lower signal strength when the carrier is less congested (SNR/RSRQ are higher). Like the QRxLevMin, this value also has a corresponding offset value (called the QQualMinOffset) which is taken into account during inter-PLMN cell reselection (e.g. to allow you to connect to a weak native carrier over a strong roaming carrier).

CA is disabled when HPUE is active. HPUE devices should be able to use L2500 at signal levels down to -128. If the you often wind up in scenarios where the uplink is too congested, then the Qqualmin value is likely set too low.

I was in Fort Lauterdale and Palm Beach most of last week and wanted to quickly share my experience using AT&T down there. This was undoubtedly the worst non-rural cellular experience I have had in the US in a long time. This was very unexpected because of the amount of spectrum AT&T has in the area. They are broadcasting 10x10 L700 (5110/5780), 10x10 L850 (2501), 10x10 L850 (2600), 10x10 L1900 (750), 5x5 L2100 (2175) and 10x10 L2300 (9820). Despite all of this spectrum, I found that probably 80% of the time I was either on weak L700, weak L850 or UMTS/HSPA, even outdoors. This was in both urban and suburban settings. There were two restaurants I ate in which had no service at all, not even voice. By contrast I never lost LTE on my Verizon or Sprint devices. It seems that AT&T's poor coverage is due to a combination of poor site spacing and seemingly poor quality equipment. The AT&T and Sprint sites that served my hotel were atop the same building with the same sector alignment, but I found that AT&T would very quickly drop to weak lowband LTE while Sprint was able to maintain usable L1900 in the lobby. All of AT&T's sites I saw used three relatively short antennas which I assume were not very high gain. I believe radios were manufactured by Ericsson. For reference, the Qrxlevmin on all of AT&T's bands except WCS is their usual national standard: -122. Sprint's were -122 on 800/1900 and -128 on 2500. Overall, I was very unimpressed with AT&T's performance and would not recommend them to anyone living in the area.

Inter-eNB carrier aggregation is possible and standardized (in the more recent releases) although it requires a unified backend core, so it won't be of much help until the transitional period has ended.

The antennas T-Mobile is using are different antennas from the same family. The models are: Commscope FFHH-65C-R3 Commscope FFHH-65B-R3 Commscope FHH-65C-R3 (usually a dual-antenna setup)

No. The VoLTE enabled flag still exists and is used on the network.

Interesting. I'm mostly curious about the rounder ones. Their 2500 coverage is not great but their 800/1900 range/propagation is incredible. They are much better than the old PCS-only antennas and even put the brand new Verizon Amphenol panels to shame.

@lilotimz do you know the manufacturers/models of the 10-port triband antennas Sprint has been deploying? This photo shows one of them and there is another more boxy-looking one I've seen as well.

I'm out of Verizon coverage at my house in the suburbs of Seattle .

I'm with your first point. In Seattle, many macros already have 120 MHz of LTE on air. M-MIMO antennas are going to be a sidegrade at best as far as capacity is concerned. I think the eventual goal is to run 60 MHz NR, 40 MHz NR/LTE DSS and 20 MHz LTE simultaneously.

M-MIMO equipment has a 120 MHz bandwidth limit. That has to be split between LTE and NR.

I haven't really set up bulk logging/high-level usage stats on that particular part of the site. How about I record usage for the next week or two and give you a good usage estimate then?

LTE > HSPA+ (4G) > HSPA (4G) > WCDMA/UMTS (3G/4G).

Leaked maps could be a big issue. I'm also curious if map markets are rendered individually on the client side or if they're rendered as tiles on the server side. If they're done client-side, performance with large maps (e.g. the NV sites map or Clear sites map) will be terrible. Can you link me to one of these maps?

As Google is turning fusiontables down soon, I've been looking for ways to preserve our maps. Playing around with the Google Maps API, it looks like it's possible to take data from a MySQL table and display it in the form of pins on a Google Map via PHP. This is virtually identical to what fusiontables did. If map usage stays low enough to be covered by the free tier, this is something I could look into doing on a bigger scale. If anyone else has other ideas for fusiontables alternatives, share them here. The shutdown date is approaching quickly.

Not having L800 will prevent the combined company from dedicating all of their 600 spectrum to NR in many markets. Running 600 RRHs in NR+LTE split mode reduces the maximum Tx/Rx diversity to 2x2 for both air interfaces. This is a big hit to lowband 5G performance.

I'm going to have to agree here. No hostility, but I always find it incredibly difficult to understand mattp's posts.