iPhone on Omada APs

[feelyat]

I have had a good experience with Omada APs. I used three EAP225s in my home with reasonable success for years. Earlier this year, I decided to upgrade to WiFi 6 APs, and went with the EAP610. I know, not a huge upgrade, but I mainly wanted to try WiFi 6 since my devices were mostly compatible.

With the new APs in the same places as the old ones, I get decent speeds, but there are huge problems with roaming with my family's iPhones. Especially when roaming between floors, they seem to have long pauses (1-2minutes) where they have no connectivity, then finally wake up and connect to the closest AP. Other devices - laptops, streamers, TVs - have no such issue, but it seems like Apple's phones are trying to hang on to weak signals forever. Once connected, I get the expected speeds; it's just that big delay that makes things weird.

Is this a known thing? Are there settings I should tweak?

 

[degrub]

yes, you can try turning down the power on the 2.4 and 5 GHz radios as a start. There is at least one setting you can try (i don't remember, sorry) if the AP supports "roaming" IEEE std 802.11 k ? v ? support. i believe there is a setting to knock off a client at certain received signal level, but ultimately the client device determines when to detach. Turning power down may be the easiest, particularly 2.4 GHz as it penetrates more obstacles (walls, etc) and reaches further.

 

[Tech9]

For 802.11k/v/r support and clients tracking - Omada SDN Controller.

 

[tgl]

I'm wondering what's different in your configuration of the 610s vs the old 225s. The other replies already mentioned likely sources of problems, but those would apply just as much to the 225s if they were set up the same way (same power levels, with or without a central controller, etc).

 

[sfx2000]

Are all the AP's the same SSID across the bands?

Apple roams just fine, but if each AP has their own SSID, then Apple will camp on the SSID until it's lost...

 

[feelyat]

yes, you can try turning down the power on the 2.4 and 5 GHz radios as a start. There is at least one setting you can try (i don't remember, sorry) if the AP supports "roaming" IEEE std 802.11 k ? v ? support. i believe there is a setting to knock off a client at certain received signal level, but ultimately the client device determines when to detach. Turning power down may be the easiest, particularly 2.4 GHz as it penetrates more obstacles (walls, etc) and reaches further.

Thanks. Any reason why that would have to change from the setting on the old devices? These same phones had no problem roaming freely with the WiFi 5 APs, but are really struggling with 6.

There are all kinds of advanced settings (Force-Disassociation is the one I think you are referring to). I've tried most of them in various combinations, with little impact.

I'm wondering what's different in your configuration of the 610s vs the old 225s. The other replies already mentioned likely sources of problems, but those would apply just as much to the 225s if they were set up the same way (same power levels, with or without a central controller, etc).

Exactly. The only difference is AX vs AC. Might the phones affinity for a weak signal be different with the different WiFi standards?

 

[feelyat]

Are all the AP's the same SSID across the bands?

Apple roams just fine, but if each AP has their own SSID, then Apple will camp on the SSID until it's lost...

Yes. Omada’s SDN sets up all APs with identical SSIDs, and roaming happens transparently across them all. I have the same SSIDs across all of the APs.

 

[Tech9]

Perhaps the new APs being a budget model don’t have dynamic power adjustment for roaming assistance.

 

[degrub]

Try turning off MIMO support in the APs
Next try turning off the 2.4 GHz radio.

What generation are the radios in the i-Thingys ?

 

[tgl]

Apple has a pretty detailed writeup about how their devices do roaming here. It's worth some study, but the takeaway points are that (1) an iDevice won't consider roaming until the signal from the AP it's connected to falls below a preset threshold, and (2) once it does consider roaming, that will happen faster and more smoothly if the APs support 802.11k/v and possibly 802.11r.

My guess is that you need to tweak the APs' Tx power settings to encourage roaming --- too much overlap of their coverage areas is bad, and you typically want 2.4G Tx power at least 6dB less than 5G Tx power. Be sure k/v support is turned on too. I have no experience with Omada so I can't say exactly what they call that or whether it requires continuous controller support.

I do notice that the Apple doc mentions a preference for AX over AC, but I don't see how that relates to your results, because (IIUC) you've transitioned from an all-AC to an all-AX radio environment. So really the roaming behavior should be the same as before. I wonder whether Omada SDN has done something different behind your back.

 

[tgl]

Oh ... it occurred to me to look up the tech specs for your APs, and I see there is a difference in max Tx power:

EAP225:
• FCC:
≤24 dBm(2.4 GHz)
≤22 dBm(5 GHz)

EAP610:
• FCC:
<25 dBm (2.4 GHz)
<25 dBm (5 GHz)

Now I'm wondering if the "configuration difference" is that both sets of APs were set at max Tx power, and the extra 3dB from the 610s is the root of your problems.

Running an AP at max power is almost always a beginner's mistake, and here's the reason: wifi is a bidirectional medium. It does no good if devices can "hear" the AP but the AP can't hear the devices. But typical Tx power for portable devices such as phones is way less than the legal limit for APs, often no more than 15dBm or so. It's okay for the AP's Tx power to be somewhat more than the client's, but not 10dB more. That leads to situations where the client still thinks it can hear the AP just fine --- so it won't roam --- but the AP can't pick up the client's signal so you get no service.

I'd start by turning the 5GHz Tx power down to around 18dBm on each AP, then walk around the house and see where you get service and where not. Adjust up if there's a clear dead zone midway between APs; if not, maybe you can go down some more. Keep the 2.4G Tx power 6dB less than whatever 5GHz is set to. It helps if you have an actual readout on your test client of received signal strength (RSSI) so that you can see where it drops to the roam threshold. I don't believe iOS exposes that natively anywhere, but you can use Apple's free AirPort Utility as a bare-bones wifi scanner. (Enable "Wi-Fi Scanner" in its preferences, then hit the "Wi-Fi Scan" button in the app itself.)

 

[feelyat]

Oh ... it occurred to me to look up the tech specs for your APs, and I see there is a difference in max Tx power:

EAP225:
• FCC:
≤24 dBm(2.4 GHz)
≤22 dBm(5 GHz)

EAP610:
• FCC:
<25 dBm (2.4 GHz)
<25 dBm (5 GHz)

Now I'm wondering if the "configuration difference" is that both sets of APs were set at max Tx power, and the extra 3dB from the 610s is the root of your problems.

Not max, but "auto", which may amount to the same thing. There is a WLAN Optimization feature of Omada SDN that seems to tweak power levels. I did try that as a diagnostic measure, but it didn't seem to have much impact.

The difference in max Tx power is illuminating. I did try setting one of my APs (the central one, with the most overlap) to a lower power setting on both 2.4Ghz and 5GHz, and the roaming got significantly better. Perhaps, as you say, I was on the edge before, and that extra 3dBm is enough to throw everything out of whack.

Running an AP at max power is almost always a beginner's mistake, and here's the reason: wifi is a bidirectional medium. It does no good if devices can "hear" the AP but the AP can't hear the devices. But typical Tx power for portable devices such as phones is way less than the legal limit for APs, often no more than 15dBm or so. It's okay for the AP's Tx power to be somewhat more than the client's, but not 10dB more. That leads to situations where the client still thinks it can hear the AP just fine --- so it won't roam --- but the AP can't pick up the client's signal so you get no service.

Thanks, that's a very clear explanation. My issue is that the upstairs AP does double duty handling the backyard, which it does well, as well as being the mesh connection for the detached garage, where I have IP cameras living. I think I'm going to need to keep that particular AP at a fairly high power level, but reduce the levels for the rest of the house.

I'd start by turning the 5GHz Tx power down to around 18dBm on each AP, then walk around the house and see where you get service and where not. Adjust up if there's a clear dead zone midway between APs; if not, maybe you can go down some more. Keep the 2.4G Tx power 6dB less than whatever 5GHz is set to. It helps if you have an actual readout on your test client of received signal strength (RSSI) so that you can see where it drops to the roam threshold.

The only tool I have that does that is NetSpot on my mac laptop. It also shows me a graph over time so I can see how the RSSI changes over time. Unfortunately, that's not on the phone, but I know Apple's roam thresholds for phones and laptops are different, so it could still give some insight. If I don't get anywhere with the power level change and NetSpot for monitoring, I'll try the Airport Utility as you suggested.

Thanks!

 

[degrub]

"
My issue is that the upstairs AP does double duty handling the backyard, which it does well, as well as being the mesh connection for the detached garage, where I have IP cameras living. I think I'm going to need to keep that particular AP at a fairly high power level, but reduce the levels for the rest of the house.
"
In case that approach doesn't work well, consider a point to point external, under eave AP pair, one on the house eave, one on the garage eve with line of sight. Then place the AP for the garage and outside areas inside/near the garage. Then you should be able to reduce the power on the upstairs AP until it just maintains the AiMesh connection. Or maybe use unused coax run with a pair of moca modems to do the ethernet backhaul.

 

[tgl]

Yeah, it sounds like you may be asking that upstairs AP to do too much. I'd focus first on getting the right power levels to serve the house well, and then see if you need to add equipment to serve the yard and garage. I second @degrub's thought of PtP wifi to extend your ethernet to the garage. (Ubiquiti has a ton of options for that in their UISP product line, and you'd probably do fine with even the bottom-end choices.) Another idea perhaps is ethernet-over-powerline. Neither of those will give you gigabit rates to the garage, but you don't need that for a few cameras.

 

[feelyat]

Yeah, it sounds like you may be asking that upstairs AP to do too much. I'd focus first on getting the right power levels to serve the house well, and then see if you need to add equipment to serve the yard and garage. I second @degrub's thought of PtP wifi to extend your ethernet to the garage. (Ubiquiti has a ton of options for that in their UISP product line, and you'd probably do fine with even the bottom-end choices.) Another idea perhaps is ethernet-over-powerline. Neither of those will give you gigabit rates to the garage, but you don't need that for a few cameras.

Before I attempted the Omada mesh mode, I used a dedicated wifi AP (non-omada, non-mesh) to bridge the garage network. It wasn't exactly PtP; it was just a client to the house WiFi, and worked well enough. The mesh approach is just so much easier to set up and configure, and comes with extra WiFi coverage in the garage for free.

Curiously, I just tried reconfiguring the upstairs AP from max power to 18dBm and the mesh connection not only didn't die, it actually improved. (-70dBm to -68dBm). That seems odd to me, and it might be a phantom, but connectivity to the IP cameras is strong as ever, so who am I to complain?

 

[degrub]

sometimes you get RF reflections that overload client and AP radio receivers.
Consumer gear is basically designed to put as much power out as is legal from one device to cover the entire house with one device.
When it doesn't fit the foot, then you want a more SMB approach with several lower power APs and a central router/controller.

 

[tgl]

Curiously, I just tried reconfiguring the upstairs AP from max power to 18dBm and the mesh connection not only didn't die, it actually improved. (-70dBm to -68dBm).

Could just be a mirage: RSSI numbers tend to bounce around a little. But assuming it's real --- what are you measuring there, the AP's RSSI for the garage end, or the garage device's RSSI for the AP? What is the garage device anyway? @degrub's idea of reflections is plausible. I'm also wondering if the garage device saw an RSSI drop, figured the AP is now further away, and boosted its own signal to compensate.

 

[feelyat]

Could just be a mirage: RSSI numbers tend to bounce around a little. But assuming it's real --- what are you measuring there, the AP's RSSI for the garage end, or the garage device's RSSI for the AP? What is the garage device anyway? @degrub's idea of reflections is plausible. I'm also wondering if the garage device saw an RSSI drop, figured the AP is now further away, and boosted its own signal to compensate.

The garage device is one of the EAP225s I had left over from the upgrade. The measurement is reported in the SDN console in both directions - downlink from the wired AP and uplink from the mesh one (in the garage). Both appear to have the same number, so I'm guessing that's a reported "mesh signal" that's part of the Omada mesh protocols.

Lowering the power level on all of the APs appears to have made everything work much better, and I'm not noticing any real dead spots, including the garage and the backyard. I seriously think the higher Tx power in the upgraded APs was causing all kinds of issues, and backing it down a bit (I tried your suggestion of 18) has made the roaming more reliable and consistent.

 

[Tech9]

Lowering the power level on all of the APs appears to have made everything work much better

Happy to hear you found the solution.

I run a system with 4x APs at minimum power in Europe, works very well.