Throughput issues cnPilot e600s

Products cnMaestro
1

If there is a better forum for this request, please let me know.

We just took over a church client with 12 cnPilot e600s. All APs have home runs to a PoE switch in the rack - no meshing. There is a Sonicwall at the edge and 2 Unifi switches. They have gigabit fiber service, and throughput measured for a wired client averages about 800Mbs down, 640Mbps up. Over wifi, though, I’m seeing 180Mbps down when I’m standing directly under an AP with a fresh connection. Connecting my laptop to the ethernet feed for that AP gives me normal wired speeds. My testing was done during the week, when there are only a handful of workstations in use and single-digits wifi users. the low speeds are experienced with each of the 4 or 5 APs I tested. I also made sure that all APs were on the latest firmware and had fresh reboots.

We manage about 50 Unifi networks, but this is the only Cambium setup. I’m looking around in cnMaestro and don’t really see anything that jumps out at me as far as misconfiguration or alarms. The layout and configuration is pretty wildly different than the way Unifi does things, so I could easily be missing something obvious.

There are 5 SSIDs altogether, with all SSIDs currently being broadcast by all APs. This is on the list of things to reconfigure, since we could separate everything into two groups based on their location & structure. I didn’t want to start down that path until I got a handle on the throughput issue, though.

Can you suggest where I should start to troubleshoot this low throughput?

2

Replying to my own post - I went through and did RF scans on each access point and manually set the 2.4GHz and 5GHz channels for all. They were previously all set to “auto”. While I didn’t see any obvious mistakes by the auto setup in what channels were chosen, there were several APs where other choices offered a bit lower interference, so I changed them as appropriate. Where the auto choices were fine, I made those same channels manual choices so they wouldn’t change for now.

Unfortunately, this hasn’t helped throughput at all. Latest measurement 180Mbps download & 230Mbps Upload approx 10’ from an overhead AP.

3

What channel widths are you using for 2.4 and 5GHz? What band is your test laptop connecting on? I assume you’re using at minimum an AC wifi adapter in your test laptop? Have you updated your wifi adapters drivers to the newest stable revision?

4

Thanks for responding. We’re using standard 20 & 40 channel widths. My field laptop is a 3 year old Dell XPS, wifi 5 but not 6. Latest drivers according to Dell Command. 2.4GHz tops out at somewhere around 120Mbps, so my measured result of 200 means I had to have been connected to the 5GHz band. Using this same laptop to test the wifi in my shop just now (my service is 300/300) I’m getting 330 down & 240 up, so I know it’s not the wifi adapter limiting things. I also tested from a new Lenovo laptop of the clients and a 2-year old Mac of the clients as well with very similar results. Whatever this is, it’s with the APs or their configuration. I’ll be darned if I can find it, though.

5

Please try increasing the channel width on the AP to 80MHz and re-run your tests.

You need to identify who makes the wireless card in your test laptop and get drivers directly from the manufacturer of the card. e.g. if it’s an Intel NIC, then search google for “Intel wireless drivers” and then download and install the newest stable drivers. This ensures best performance of the wireless card.

6

I appreciate your response, but the individual drivers in my laptop aren’t at the top of my suspect list since I got similar performance using 3 different computers, all with different wifi cards; AND got better performance with the same laptop on my own wifi network. Additionally, I’m happy to TEST using a 80MHz channel width but this is likely a poor choice for an ongoing setup. This is a church, so you might have 5 or 10 simultaneous users during the week, but 400 on Sunday. Additionally having 12 APs means we have to be cognizant of the interference with of one AP iwth it’s neighbors. Using 80MHz channel widths makes this job harder, not to mention its impact on neighboring networks (which thankfully are few in this location, but still…).

7

Mark,

I’ve just been through a similar situation with an outdoor Wi-Fi network in a campground. E600 is a great AP. Having 12 of them in close proximity causes some peculiar issues. When the integrator installed the 30 AP’s in the campground, they used auto RF and auto channel (the defaults) as well.

In a church setting, your primary wifi client are phones. There might be some tablets, but normally 95% or more phones. Smart watches as well now too. Phones and watches are affected by the body of the person holding the phone or wearing the watch.

All of these are very low gain devices. When the church is full, your enemy is adjacent channel interference. Wi-Fi is not synchronized, so you have no control when a device transmits or when an AP transmits. If you are using multiple channels, one device transmitting on one channel is causing interference to devices on other channels. Wi-Fi has a mechanism for this…called RTS/CTS (request to send, clear to send).

If you have 12 AP’s, I would take groups of 4 or maybe 6 AP’s and use a single 80 MHz channel for each group. That way, all the Wi-Fi clients are on the same channel, and are listening to each other, and will be much more courteous to each other, enforced by RTS/CTS, and only transmit when they have a clear channel. The reduces adjacent channel interference. This sounds counter-intuitive, but it can increase the volume of traffic that can pass during busy hour. I’d also use a hotspot controller, like in cnMaestro, and limit each connection to about 25 Mb/s DL and UL, band steer to 5GHz for all devices that can use 5 GHz, and set the roaming threshold high - 22 dB SNR or maybe a bit higher to prevent a client from ‘sticking’ to an AP long after it should have moved to a better one. The RF communication will happen at the max modulation level, and thus the quickest it can, leaving room for other devices, rather than a few great signal quality devices monopolizing the airtime. When you configure an AP group in cnMaestro, you can use auto channel but limit the channel choices to 1 channel in 2.4 GHz (I’d use ch 6 @ 20 MHz and use 2 channels that are very non overlapping 80 MHz channels in 5 GHz).

You can discourage the use of the 2.4 GHz channel by shutting it off, or if you don’t want to do that, use 12 Mb/s as the minimum data rate. That way, a client device is not going to associate to the 2.4 GHz channel and stay down at a low modulation level. Most Wi-Fi clients will choose 5 GHz over 2.4 GHz in this configuration.
For power levels, this is a bit of a guessing game, since I don’t have your floor plan. I would not use automatic, but manually set the TX power to +30 dBm or less for 5GHz, and +20 dBm for 2.4 GHz on every AP.
Make sure beacons are 100ms. Use as few SSID’s as you can. Each SSID transmits its own beacon, so the fewer of those, the less RF transmitted. I would say staff SSID, and guest SSID.

I hope this helps,

8

I’ve narrowed this problem down enough that I need to quit spending time on it. The original installer says “200Mbps is the real-world limit of the e600”. I disagree by reading the specs, but do agree that these things are getting long in the tooth - last sale date was 7/25, so in my mind they shouldn’t have been chosen for this installation which was done in 9/24. The spec throughput on those radios is ~800Mbps, so I would think a real-world factor of 50% should be achievable, but apparently not.

Shutting off all APs but one, and then changing the channel width on that single unit to 80MHz bumps up the speed to 300Mbps. Changing back to 40MHz channel width drops it again to 200Mbps. I’m ready to concede that this is all these things are capable of. They may be great at other things, but raw throughput isn’t one of their strengths. I kind of want to dig out a Unifi AP-Pro (similar age and spec) from my junk pile and see what it does, but as I stated, I can’t justify any more time on this.