#19832

Rainer,

?

Thank you for your reply, and sorry I can't get to everyone's reply timely.

?

On Tue, Apr 1, 2025 at 04:22 AM, Rainer Hantsch wrote:

I installed many WiFi networks in my life, but I would never go above 30-40 clients on one Access Point (AP). More clients may work, but with huge impact on throughput, but it is also possible that the accesspoints cannot handle that number of MAC addresses.
I would also never install wireless repeaters (only repeating over WiFi). Every intermediate repeater reduces the throughput, so when you have 3~4 cascaded ones the construct can become very unstable ...

Thanks. I'm unfortunately well aware of the MAC address limits. The Ubiquiti APs I use advertise being able to support 200 to 350 clients each. I tested this, but this is not reality. When I disabled all but one AP, 120 clients on the 2.4 GHz band is the maximum the lone AP could handle. Subsequent connections were rejected, with a smartphone located near the AP getting an authentication error when trying to connect. I filed a case with Ubiquiti and they claim they do this because airtime is likely to become the bottleneck with this many clients. It may be true, but their specs are still false advertising. I don't have nearly enough 5 GHz clients to see if the same MAC address limit applies to that band. I could easily test whether the airtime is the bottleneck by placing 2 APs next to each other on the same channel, and seeing if I can get more than 120 clients to connect, between the two of them. I haven't taken the time, but I strongly suspect I would get >120.

?

While the clients are numerous, most of them are lightbulbs, which are very low bandwidth, and support only 2.4 GHz channels. For example, my porch light Wifi bulb, which is connected to Wifi 24/7, averages 135 bps for the last 24 hours in the Unifi controller. A good chunk of that traffic is likely from smokeping that I have been running to track packet loss and latency. I disable/enable smokeping periodically for testing purposes. It significantly increases traffic as it will be pinging all devices, but is not the root cause of the Wifi problems.

?

I am not using repeaters, but I'm using a mesh network, because I only have Ethernet plugs in 2 indoor locations on the same bottom floor, when there are 17 indoor locations to be covered, and many more outdoors (I haven't kept a count).

?

I know that there are standards like 802.11r.? Such routers/APs support mesh radio (so they transparently hand over moving clients.? But a warning, I (still) have a few WiFi cameras in use that do not support that. Even worse, the cameras do not connect to this Mesh-WLAN and I had to define a separate VLAN without "r" especially for them.?? I doubt that every little smarthome device will smoothly support Mesh WLANs.

My Ubiquiti APs support 802.11r, and they use it for meshing. The 802.11r feature can be enabled/disabled per-SSID. I was surprised to find that my Samsung Galaxy S22 Ultra does not support this. I discovered by turning off auto-reconnect on the SSID on my phone. It dropped out of Wifi every few minutes. I turned off "Fast roaming", which is Ubiquiti's name for 802.11r on the SSID, and the phone stopped dropping out. You are right that the IoT devices are not going to support this, especially as there are several brands/models of them. I could separate SSIDs, but this gets messy, uses additional airtime, and Ubiquiti has a hard limit on the number of SSIDs on any single AP per radio.

So I would start with having a close look on number and location of APs to not have too many devices connected to a particular AP. In worst case you need to add more APs and lower their signal level to make radio cells smaller.

And use copper backbones (connect every AP to the LAN switch by an individual LAN cable). Copper is LOTS faster than WiFi, and this way you also eliminate concatenated APs/Repeaters.

Currently, I have 2 wired APs, and 7 meshed. Most of the mesh links are 1 hop. There are 2 mesh links that are 2 hops.

It would cost a minimum of $5K to add 5 copper backbones, and in practice likely to be done outdoors, which would be unsightly. Running indoors can't be done inside the walls. It could potentially be done with piercing and many raceways, but I'm not even certain it can. I know the wireless backhaul is much slower, but I don't believe it is the bottleneck here. When I use iperf3 (Magic iperf app) on my Android phone, on a 5 GHz SSID, there is no point on my property where I get less than about 30 Mbps, and that is one wall away from the location of the 2 hop meshed APs, outdoors. That kind of bandwidth is more than sufficient to carry the data for 300 2.4 GHz lightbulbs. Near the 2 wired APs, iperf3 reports about 600-800 Mbps. In many rooms near the single hops, I get 150 - 500 Mbps depending on distance/walls. All of that is much slower than the 10 Gbps copper LAN between the office and home theater, and I would much prefer a copper backbone for all 9 APs, but while I may want to do i eventually, I don't believe it will solve the issue with the large number of 2.4 GHz clients. For the record, this is how the mesh network looks like today, with the vast majority of the Wiz lightbulbs turned off by relay. The Office and Home theater APs are the 2 wired ones. All the other APs are meshed. The double-meshed ones are Bar AP and Master bathroom. Master bathroom has to be double-meshed, as it is in practice out of range of either wired AP - it can technically connect, but will get < 1 Mbps throughput. Bar AP could be changed to single hop, but gets a better signal strength from another mesh AP. The average ping times for those 2-hop APs is under 10ms, so I don't believe the 2-hop are adding too much latency. They are definitely reducing the bandwidth, though.

I have several cases opened with Ubiquiti about problems related to those Tx retries and incorrect channels displayed for some mesh APs. And others related to signal strength/latency, not GUI. Note that all APs have DHCP reservations. These are not dynamic leases. Same for every client device on my IP network, whether Wifi or Wired. I'm using a /22.

?

#19838

Don,

?

Thanks for your reply. Response inline.

?

On Tue, Apr 1, 2025 at 06:01 AM, Don Jarvis - NNA6DJ wrote:

You are living in a RF hell due to having way, way too many wireless devices. As an amateur radio operator and having deployed some large outdoor wifi networks (2000+ plus devices), I have to fight RF interference on an ongoing basis. Some comments:

?

Start reducing the number of wireless devices strategically as unfortunately you are the maker and creator of your own RF hell. Change out the wireless light bulbs with regular ones in areas that will have little impact like in bathrooms, bedrooms, closets.

Actually, due to macular degeneration, I really want to be able to set at least the color temperature dynamically, and dim as needed. Some tasks and locations require daylight, and others don't. Bathrooms and bedrooms are definitely included. There is also the neat party mode effect I enjoy. Closets and laundry could use regular bulbs with fixed color temperature, but there aren't many of them. We are talking about 10 bulbs total that I could convert back to regular. I still have all the regular LEDs that are all 2700K, and many have lower lumen than I need.

2 weeks ago, I had lights replaced in my garage. 7 fixtures with 14 fluorescent tubes were replaced with 19 fixtures with 38 LED tubes. Thank got for Costco and my handyman. The lights are very bright, and I can finally see all the dark cables in my dark storage cabinets. My husband on the other hand now has to wear sunglasses in the garage if staying there too long to avoid getting a headache. I could probably grow plants there now. I don't believe these tubes are dimmable, unfortunately. It's not in the manual. I just sent an e-mail to FeiT. I would like to turn the tubes down a bit as most of the time, they don't need to be this bright all the time even for me. Unfortunately, smart LED tubes are still very rare and extremely expensive, many in fixtures with poor designs such as non-replaceable tubes, smart protocols that require cloud to operate, etc, none of which I would touch. The tubes are on a vacancy switch in the garage and are not on 24/7 - only up to 30 minutes unless we are in the garage and there is motion.

Change the manufacture on your bulbs and buy better quality. In my radio shack, I once changed all the light bulbs to led tubes. Ist few worked great, then I purchased another brand and my RF noise levels rose from an S3 signal level to +20 over S9. Had to turn off the lights in the room to even be able to use my HF radio. Fix was to buy better quality bulbs.

There aren't many manufacturer of smart bulbs I would trust, much less suitable to use cloud-free. The Philips Wiz are the main ones I zero'ed in for that reason. There are lots of fly-by-night Chinese makers of smart bulbs that will stop working the moment they shut down their cloud servers. I do have the existing non-smart bulbs from FeIT, but that doesn't resolve the problem. I really need smart bulbs at least to control brightness.

Solar inverter and panels are well known for having RFI issues. Research your manufacturer to see if there are any known problems or defects in the solar panels or inverters.

?

The micro-inverters are 28 Enphase M215 and 42 IQ8+. The panels are of different vintage/wattage/manufacturers. There are 3 brands - Sharp, Jinko and Solarever, all installed on different dates. The system is 23.2 kW DC.

Have a aerial (done with a drone and infrared camera) temperature study performed on your solar panels for poor/defective connectors in the panels. They are noise generators.

?

Thanks. I had not heard of this.

Convert your mesh wifi access points to wired if at all possible. Mesh AP’s significantly increase the traffic load on the wifi systems. Check your channel frequency assignments to isolate and separate spatially. On 2.4G there are only 3 clear channels and on 5G only 8 or so.

?

I'm considering converting some to wired, but it will cost many thousands and be unsightly. The clients are nearly all 2.4 GHz, but due to many being out of range of the wired APs, their traffic is being carried over 5 GHz mesh. I use two separate 80 MHz channels for the 2 mesh chains, channels 36 and 149, from each of the 2 wired APs. I don't believe mesh is the bottleneck here. I know about the 3 non-overlapping 2.4 GHz channels. I believe I have spatially isolated already - ie. using channel 1 on APs on opposite sides/different floors of the house, same for channels 6 and 11. There are 9 APs, though, so 3 APs on each of these channels, and some may overlap.

Get rid of wall warts (i.e. plug in power supplies) wherever possible. Most are VERY noisy.

?

That would require replacing a lot of equipment. Even my TP-Link SX105 10gig ethernet switches use wall warts. Same for some specialized audio gear, KVM switch, some monitors, cable modem, etc. There has got to be at least 20-30 wall warts in use in the house. I have a drawer in the garage full of them with probably about 50 more unused. Just finished putting large volt/amp ratings labels on all of them last week-end now that I can finally see again in my garage. The portable Brother p-touch label printer came in very handy. Battery-powered, no wall wart.

Look at your wifi network diagnostics on the wifi side to determine performance of each AP and find if any are being overloaded.

?

Currently, with 89 Wifi clients connected in total, the 9 APs are reading 0.03 to 0.72 load average for the last 15 minutes. I just flipped a whole bunch of Z-wave smart switches in Home assistant to power more bulbs. The total client count for all APs went up to 145.? The 1-minute load average is between 0.02 and 0.61 . I would say there is no noticeable change with the increased client count. I'm going to walk around press all the non-smart switches to turn on the remaining clients. I needed to get my steps in for today. The client count went up to 252. The 1 minute load average is between 0.02 and 0.82. 5 minute is 0.01 to 0.70 . The older NanoHD units are getting higher load average - especially the one in the dining room with 50 clients. The U6-Lite with 27 clients is in the same range. The three U6-LR on the other hand all have fairly low load. Even the single hop meshed U6-LR in the master bedroom with 76 clients only has a 0.29 load average. The other two are both at 0.02 .

I have used many Ubquiti APs and they are nice products. Great price but lower performance. Consider buying top of the line AP’s which can handle more devices and have better performance.

?

Yes, that's under consideration, but I'm really not sure which other brands/models I should try. I think TP-Link Omada is one. I may not be able to spring for much more expensive gear like Ruckus or Cisco APs. They all need to be able to mesh due to the current lack of Ethernet wiring situation.

An EMI detector will be of little use as all it will tell you in the RF spectrum is saturated which you already know. Start cleaning up the RF environment.

Just a boolean "the RF environment is saturated" is not overly actionable information. There are several possible causes to my problems besides EMI, such as thick building materials, device firmware problems, defective radios, etc.? And it could well be a combination of all of the above.

?

I was hoping the tinySA could give me information for each room, near each device or group of devices, for each frequency band, to see if any might be malfunctioning WRT its/their radio, and possibly remove/replace/relocate the offending devices. Isn't this what an EMI detector is for ? Or at least can be used for ?

?

#19839

On Tue, Apr 1, 2025 at 06:21 AM, Mike N2MS wrote:

This power cycling method is a good way to isolate the interference and determine if it is coming from your property or outside. I would also get a DSP Receiver such as:

Even if you cannot locate the noise you have a radio to listen to.

Thanks. My guess is that it's from my property, based on low density of building/population around here. 2 houses down is a farm. I could be wrong, of course. For radio reception, I already have an antenna on my roof which does a very good job, getting over 100 ATSC stations, as well as FM, and combined with my Marantz receiver, am not in big need of another radio or tuner. Of course, the Marantz doesn't do 10 kHz - 380 MHz, or 404 MHz - 2 GHz, but it's not a portable device either. $480 is a bit pricey to confirm the theory that the issue is from inside rather than outside. Does anybody rent these kinds of tools ?

?

#19840

On Wed, Apr 2, 2025 at 06:04 AM, Mike N2MS wrote:

Is there interference around the ISM frequencies:

That's what I was hoping a tool like the tinySA could help me determine.

For Wi-fi frequencies, an RF scan from my 2 wired Unifi U6-LR APs shows negligible interference on any channel, on either the 2.4 GHz or 5 GHz bands. Unifi is not capable of running an RF scan on the 7 meshed APs, unfortunately. I can partly understand why they don't do so on the 5 GHz frequencies - that would temporarily break the mesh, since the 5 GHz radio is used for backhaul. However, they still could perform the RF scan on 2. 4 GHz channels only, but don't. Having this 2.4 GHz channel scan on all 9 APs would be really useful, since nearly all the IoT devices use 2.4 GHz.

My Toshiba microwave oven interferes with the 2.4 GHz wifi. My manual mentions radio interference when in operation.

In my previous townhome, there was a GE combo microwave / wall oven from the mid-1980s. In the late 1990s, I bought the house, and had a satellite dish mounted on the outside wall immediately behind the the appliance.

?

When I used the microwave, the satellite receiver reception was adversely affected. No TV dinners for me !

I'm sure my microwave oven could still interfere with Wifi, but it's not being used very often, probably 5 - 10 minutes per day, and the Wifi issues do not coincide with microwave usage. I later remodeled the kitchen, changed the combo microwave/oven to a double wall oven, and installed a separate over the counter microwave, and the satellite interference disappeared.

?

If the microwave interference with Wifi is anything similar, it probably doesn't affect a very large area.

?

My microwave oven model is a Panasonic NN-SN77HS. It was top rated in Consumer Reports 3 years ago when I bought it from Best buy. I have no idea how to get information about shielding. It's certainly not in the specs. That said, I have tried placing my smartphone in the microwave oven (not running, of course) and pinging the phone from another device over Wifi, and the ping still returned. I don't recall the latency or packet loss. I could repeat the test, but I don't think it would teach me much, due to the microwave not being the root cause of the problems. But the previous result tells me the microwave oven shielding isn't that good. A perfect Faraday cage doesn't exist, though.

?

#19841

On Tue, Apr 1, 2025 at 05:00 AM, Jim Shorney wrote:

Wow. Just wow. You have created what we call RF Hell. You will inevitably have a high rate of signal collisions. Also the combined signal level of all your transmitters and EMI generators across all bands will be very high which will create strong signal overload of the receivers, both in-band, harmonic, and fundamental overload. Receiver overload causes distortion and even more interfering mixing products are generated. With fundamental overload a strong signal out of band blows through the front end filtering of a receiver so yes, devices in different bands can interfere with each other. You are effectively trying to put a swimming pool into a 5 gallon bucket. If you do buy an SA you will also need some attenuators so you don't overload it as well.

Your only cure is to shut everything down, clean up the EMI/RFI sources, and rebuild the system with careful planning possibly with the aid of an RF consultant. Even then you will still have issues. My former employer sold coaches wireless systems to the NFL some years back and throughout the life of the system we sent engineers to the Super Bowl every year to babysit the radios and manage interference. It was a job that I would not have wanted to do.

Thanks. I did not realize the radios could interact on different frequencies. I may be able to slightly reduce interactions by introducing some delays in certain automations that use different frequencies, such as the ones that use both Z-Wave and Wi-fi. I'm not sure if that would be of any help. Z-wave is generally low bandwidth and low traffic, but still gets overloaded if I flip a bunch of switches at once. I also did not know about the need for attenuators. Which ones would you recommend in my case ?

?

An RF consultant is probably not in the cards due to my limited income right now. I'm living off savings/investments only without collecting disability benefits or pension. I have the time to learn about these issues and hopefully do some of it myself, though. I can certainly flip some or all breakers during the day and test things out, if I know how to use the tool.

?

I'm sure there are many sources of issues, though. For instance just moving a vase on a table near the Simplisafe base station is the difference between some distant devices being in range or not. I can't believe they don't have at least a wall-mount for it. There are third party 3D printed ones I'm looking into, but they are not exactly pretty.

?

Placing metal objects such as pots on the kitchen countertop in the center of the property likely has an effect on various RF signals traversing as well, though I have made no specific observation, and of course don't have the tools to measure that. The induction cooktop is another surface with potential for interference.

?

#19842

The content of this topic does no longer fit in the purpose of this group so I'm closing it.

--

Designer of the tinySA
For more info go to

#19843

开云体育

Hi.

The situation you have is - honestly - a nightmare.

With so many clients in a mesh network you mess up a lot of theoretically available bandwidth, so I am not really surprised about your problems.

As you already found out, your home theater(?) worked lots better since you changed the smart relay mode and it got worse when many clients in small time window are turned on.? That is the consequence of using WiFi.

It is not only the number of MAC addresses. On every power-on every device initiates a long sequence: It enables WiFi, then it searches for available networks, then it starts to connect to this network, then it starts requesting an IP address (if you are running on DHCP). Imagine the amount of network traffic for that!

After the device is successfully registered and connected, you still have the situation that lots of traffic moves forward and backwards through the mesh and that all APs have to talk permanently to each other (exchanging mAC addresses and signal quality of every MAC to control handover).
If your central "smart controller" is located on the other end, you increase the problem, as every packet has to be forwarded again and again...

So I do not really wonder about your issues, but I am very, very sure this is not caused by something you can solve with a tinySA*. This is mostly the internal function of WiFi, with and without mesh (doesn't really matter with not moving clients).

The only way you can get this nightmare under control is installing lots(!) of LAN cables?and install some more APs (using this LAN cables as backbone to the central switch and smarthome controller. I know, you don't want to hear that, but there is no other way. This LAN cables drastically reduce the WiFi traffic, and a LAN cable is also LOTS faster (GBit or even higher) and has a lots lower overhead and latency, this is a huge improvement. But don't do it wrong and install a chain of LAN network switches, you need a star architecture.

It is also good advise to enforce smaller wifi-meshes, so every mesh has less clients/MAC addresses inside. Configure different WiFi SSIDs, so you gain control over where a client can connect at all. So the mesh traffic will be reduced to inside every mesh, only packets that belong to outside the mesh will leve it, but using the LAN cable.?? -> This is the correct way. If and how you tell this your APs you have to tind out by yourself. They must NOT use radio as backbone to outside their mesh(es), only LAN cables.

You may configure two virtual WiFi networks on your APs, one for the individual "smart stuff" mesh (use different names to enforce small areas to enforce where?a client can connect at all), and another one for "human use" (one huge SSID and mesh across entire site, so you can walk around with your mobile phone, laptop, ...) without disconnects. This virtual meshes must not be interconnected inside any AP with the "human mesh", they have to run completely isolated. And try to keep the number of "smart clients" below 80 (because of MACs). If this doesn't work with your APs, you need two APs one for every mesh.

This I see as absolutely mandatory. What you try right now has no realistic chance to work properly. You will notice a drastic improvement when doing it this way. Without splitting all this into smaller sections and using high effeicient backbones (Copper) I see no real chance that you can solve your issues.

Am Mittwoch, 2. April 2025, 22:17:24 CEST schrieben Sie:

> Thanks. I'm unfortunately well aware of the MAC address limits. The Ubiquiti

> APs I use advertise being able to support 200 to 350 clients each. I tested

> this, but this is not reality. When I disabled all but one AP, 120 clients

> on the 2.4 GHz band is the maximum the lone AP could handle. Subsequent

> connections were rejected, with a smartphone located near the AP getting an

> authentication error when trying to connect. I filed a case with Ubiquiti

> and they claim they do this because airtime is likely to become the

> bottleneck with this many clients. It may be true, but their specs are

> still false advertising. I don't have nearly enough 5 GHz clients to see if

> the same MAC address limit applies to that band. I could easily test

> whether the airtime is the bottleneck by placing 2 APs next to each other

> on the same channel, and seeing if I can get more than 120 clients to

> connect, between the two of them. I haven't taken the time, but I strongly

> suspect I would get >120.

>

> While the clients are numerous, most of them are lightbulbs, which are very

> low bandwidth, and support only 2.4 GHz channels. For example, my porch

> light Wifi bulb, which is connected to Wifi 24/7, averages 135 bps for the

> last 24 hours in the Unifi controller. A good chunk of that traffic is

> likely from smokeping that I have been running to track packet loss and

> latency. I disable/enable smokeping periodically for testing purposes. It

> significantly increases traffic as it will be pinging all devices, but is

> not the root cause of the Wifi problems.

>

> I am not using repeaters, but I'm using a mesh network, because I only have

> Ethernet plugs in 2 indoor locations on the same bottom floor, when there

> are 17 indoor locations to be covered, and many more outdoors (I haven't

> kept a count).

Rainer

开云体育

Locked Help ! Calling Ghostbusters.

开云体育

Help ! Calling Ghostbusters.