We have a tower with a 3000 AP using the Cambium antenna. It is running 4.4.2. All SMs are running 4.4.2. We are mixed mode with the SMs, some are Force 200's and others are Force 300-25's.
What we are seeing is every Froce 300-25 SM has very high latency when pinging the radio and every Force 200 has normal ping times. The latency is high enough on our 300-25's that it is tripping our pre-set alert limits of 30ms. We are pinging the radios from a poller box located at the same tower as the AP so the only wireless link the ping is going over is the 3000 AP to SM link.
Here is what I am talking about. Here are two 24 hour graphs of Force 300-25's. The red lines denote ping times above 30ms:
On our 3000 AP with 100% AC clients (all Force 300-25's and 300-16's), we do NOT see these latency issues. It seems to be related to having old clients on a new AP? But, it is bad enough that we are getting complaints from the Force 300 customers on this AP.
My assumption is it is related to the secondary 2000 upload panel. The one that is currently not able to be used until 4.5+ from what i heard
I know the legacy gear had an option to chose to upload to that or the main panel, whereas 300's do not. I think the 300s are limping to the main panel and waiting for the other panel
We purchased a 3000L and Force 300-25 to test whether we could use Cambium for a new site deployment. Used firmware v4.6.0.1. 3000L was only AP in operation at site. Used RF Elements 20 degree asymmetrical horn. We tested equidistant to an Ubiquiti LTU AP pointed at us from another direction, using same horn. Unfortunately latency kept spiking to 120ms+. We tried both low-noise DFS channels (not the one used by LTU) and non-DFS channels (both bottom end ~5200 and top end ~5750) with full gain, but the problem persisted regardless of SNR. Throughput was solid 100Mbps download, and upload was 5Mbps to 20Mbps on most channels at a distance of 6km, which is fantastic, but latency kept spiking dramatically. We tested an LTU-LR from the same spot, equidistant to the LTU AP coming from the other direction, and saw that latency stayed below 17ms. Throughput was marginally higher on LTU (130/15). For both Cambium and LTU we tested latency at 200ms intervals using MTR. Just wanted to share this because in the future we would really prefer to use Cambium, and were prepared to invest a lot in EPMP line, but at the moment the jitter and latency spikes on make the EPMP line unsuitable for our particular market, where gaming customers are sensitive to latency. The majority of households we service have children who play multiplayer online games (maybe 90%), and we would not want to offer service with this sort of jitter because the children would prompt the family to switch to a competitor if they experienced latency spikes of this sort. I imagine other WISPs in competitive markets see these latency issues and similarly feel stuck with Ubiquiti until it gets fixed in EPMP. Aside from that we really like the EPMP hardware and software a lot. I am hopeful this issue gets fixed in future firmware or with EPMP 4000 line so that we can finally switch to Cambium.
I am sorry that your tests were unsuccessful, but am pleased that you posted this, I am one of many that invested heavily out of the gate for 3 tower sites with epmp lineup and have high latency and jitter between the SM and sector.
This is my latency - this if from my Laptop to Google - and it’s a reasonably flat ~40ms to google - with no spikes or excessively high latency. On of our son’s is a very hard core gamer, and other than our router occasionally needing a power cycle, he has no issues.
So, this is from my Lenovo laptop, wirelessly connected to my home router, which is wired to my SM (ePMP Force) which is 6.3km from the ePMP AP. That’s backhauled by a Force300csm<~~>Force300csm backhaul which is in TDD 75%/25%, and then through an AF24 link to our fiber.
