Thanks Neil. I will look into the A20. I have had my problems with Time Machine/Time Capsule as well. I do periodic backups with CCC to an external drive as well as the regular ones to Time Capsule, as does my girlfriend. I do regularly turn my machine off, but I’ll check out your backup solution, which sounds possible.
The other access points are wireless: the garage is probably 75 feet from the home office router, which has to stay where it is (my girlfriend, who is often on video calls, needs it there), and the Airstream is another 25-40 feet, so I don’t want to deal with cables. Thanks for your thoughts!
Thank you for that pointer. Since I have the new Wi-Fi router, the AirPort Express is unplugged from everything. But I used to enable every privacy and security setting I could find, so I could easily have shot myself in the foot. If I get a spare hour or two, I’ll hook up the AirPort. connect to it, and poke around with AirPort Utility for settings that might be over-zealous.
It happened in Firefox on two different computers, but either more dramatically on mine or the other user is simply more tolerant. (Well, the other user is absolutely more tolerant. But maybe the issue was less significant on that computer.) Of course, if Firefox is wonky, then it could be wonky on both computers. Neither of us uses Safari regularly, but I experience this again, I will try it.
In two tests, the latency was in the high and low 30s milliseconds on average, and the max was under 50 ms. If my math is correct, 50 ms would be 1/20 of a second and so latency would not account for the delay that caused my complaint.
As I said, I now have a new router (AX50) and I haven’t noticed the significant delays with it.
Gotta say that sounds like the router is the weak point. But sub-50ms latency should be enough for most purposes. It shouldn’t impact Zoom. But I don’t know the real-time latency issues of remote ham radio operation.
Another good tip: storms and cable runs. I have a garage over 75’ from home. I had foresight to install twin conduits for comm/network and for power from the house/switching for lighting at garage-3way.
A few years back, lightning struck the woodlands behind my property. I never witnessed it (heard it!) but I suspect the EMP from it was the culprit of why the next day, the nearest garage door opener died. Other one was fine. So troubleshooting found the board was dead (it was the newest of openers in the garage but 1 month past warranty. Board was $40 and some reprogramming of remote code-pair).
My point is that having a long run of ethernet in the ground as now resulted in a possible storm issue. I seem to have some issue with a switch in the garage although everything is on good UPS/surge protection, that started after a big storm this summer. An article about a storm had me thinking maybe the storm was the issue with my hardware in the garage. Maybe I need to think about end-to-end fiber (which will still need either special hardware switch or Fiber to ethernet transceivers at each end.
I am also asking if ANYONE has experience and luck with Orbi, reach out to me. I am frustrated with their support that I have RBR40 w/ two active satellites (a spare third I am thinking of using) and I backhauled with ethernet. Well, on the Orbi app, it shows one is ethernet connected, but the other is 2.4Ghz. I tried reseting, bring to house (this is the garage one btw) close to main Orbi base station, but it doesn’t reset (amber) but goes red. I think there is too long a process to connect to ethernet and wait for it to show up. There is no diags for the unit, but it did let me update all to latest firmware (which, for me, I have to use Google Chrome as all other browsers don’t work right). The iOS app is just dumbed down (I prefer using IP login and management). Plus the garage is not climate controlled and fear even this summer heatwaves might have affected the ORBI.
I am considering replacing it, but I may only stay in this home another year, so doing fiber run AND updating hardware might be foolish. For now, I will try using the spare, but Orbi-Netgear support is just… silly.
If you want to consider this, look for Ethernet switches with one or more SFP ports in addition to the usual RJ-45 (copper) Ethernet jacks. Although you can get specialized transceivers, they will cost more than using SFP-capable switches.
Put one switch on either end of the conduit. Get appropriate SFP transceivers to plug into the switches’ SFP ports and run the fiber between the transceivers.
The important things too keep in mind when looking for equipment:
Switches with SFP ports tend to cost more than those without, but not always. An Amazon search finds several small Gigabit Ethernet switches with SFP ports for under $100, including some really inexpensive units (e.g. this 5-port switch for $10).
(I am not recommending any specific product here. I only mention that $10 switch as an example of an inexpensive switch that has an SFP port.)
You need to plug an SFP transceiver into each switch’s SFP port. They are sold separately. While they can be expensive, a basic one for Gigabit Ethernet can be quite affordable.
Note, however, that they aren’t universally plug-and-play. Although the technology is pretty generic, some switch vendors have firmware preventing you from using SFPs that were not certified by them (which can cost a lot). If you get a brand-name switch (e.g. Cisco), then you will need to ensure that the SFP transceiver you get is compatible. If you get a generic no-name switch, then it probably won’t matter.
SFPs come in different speeds. For a Gigabit switch, you want to use a plain SFP, which may be advertised as 1.25Gbit or Gigabit Ethernet.
You do not want to get a transceiver labeled “SFP+” (10Gbit), “SFP28” (25Gbit), “QSFP” (quad-channel, 4Gbit), “QSFP+” (quad-channel, 40Gbit) or “QSFP28” (quad-channel, 100Gbit). These will cost a lot more and probably won’t be compatible with a Gigabit switch’s SFP port. The quad-channel variants won’t physically fit in the socket either.
Within each speed SFP, they support different kinds of physical media. Common types include:
1000BASE-SX (multi-mode fiber up to 550 m distance)
1000BASE-LX (single-mode fiber up to 10 km distance)
1000BASE-EX (single-mode fiber up to 40 km distance)
1000BASE-T (your usual copper cable interface)
You will want 1000BASE-SX. Multi-mode fiber is much less expensive than single-mode fiber. At the speeds and distances you require, you have no need for more than that.
Note also that fiber types are not interchangeable. If your transceiver requires single-mode fiber, you can’t use multi-mode and vice versa. And both ends of the fiber must be the same type. If you put (for example) 1000BASE-LX on one end and 1000BASE-EX on the other, it won’t work.
Within each fiber type, there are different kinds of connectors. You will probably find your SFP using an LC connector. It really doesn’t matter what kind your SFP uses. Just make sure the fibers you buy have the same kind of connectors.
Look for fiber designed for outdoor conduit use. The connectors and glass are the same, but the insulation will be more durable. This will be easier to work with when you pull it from one building to the other.
Don’t try making your own cables. Although it is possible to buy bulk fiber and attach the terminating connectors, it is difficult and requires expensive equipment. Just buy pre-built cables for the length you require with the connectors you require.
When installing fiber, be careful not to pull on them with a lot of force or bend them sharply around corners (the bend radius) because doing so can damage the glass fiber inside the cable. If you need to tack them to a wall, don’t staple them, but use a cable clip that can’t crush the cable (e.g. this Amazon listing)
Can’t resist sharing…Due to the pandemic, I haven’t been able to my daughter and her family for over a year. My son-in-law fixed up their (unattached) garage as a movie theater, I helped him insulate it on. our last trip there.
They were having a weak Wi-Fi connection, so he and my granddaughter dug a shallow trench and ran an ethernet cable to the garage. It’s working fine, but next time I’m there, I think I’ll help them run it through a plastic conduit.
Great article, but I’d like some advice. I am a psychiatrist, I see my patients virtually from our basement, using HIPAA-compliant video conferencing software using my iPhone 8 as my device.
(Interestingly, I type my notes an Apple Bluetooth Keyboard connected to my iPad Air which is logged into the Windoze-based record system at my clinic. I use an external monitor.)
I’m often the only user on our WiFi, which is a Netgear R7900P, which uses 802.11ac; it feeds off a cable modem using our Xfinity signal. I typically get 60-80 mbps up and 5-6 down. Direct ethernet is better, like 127/6, but no way to hook up ethernet to my phone, as far as I know.
My iPhone can’t handle anything beyond 802.11ac. My signal is sometimes too slow for good connection. Would there be any point in upgrading my router?
Fellow psychiatrist here. It sounds like your speed is ok for what you are doing. I see more problem with the patient/client end of the video connections on my video appointments, and there is not much you can do about that. That said, there is a lightning to Ethernet adapter for about $20 you can try.
Curious, do you tend to use the vertical video orientation?
Thank you for your response….since we’re talking shop, I’ll mention that I use doxy.me for most appointments, doximity (which I pay for) for others, and Skype for one patient who prefers it (and knows of the HIPAA concerns).
Yes, I have a mini-tripod with an iPhone adapter, and use it in vertical orientation. My having used an adapter to have my iPad drive a display has really saved my neck, along with a sit/stand desk adapter.
I guess you’re right, most of the connection issue may be on the patients’ side of things. Doxy.me help files suggest I should use a faster connection; I may try the ethernet adapter.
Glen, I know this article is almost a fortnight old, so maybe I’ve missed the bus, but I have a naive question that I have found very hard get a definitive answer for, so here goes:
I live in a big house in France with 60cm thick stone walls. I have a very fast fibre connection which I can use via Ethernet in various parts of the house, but the house is so big that I have to also use additional routers at the ends of a couple of the Ethernet connections so as to ensure that there is wifi everywhere.
What I want is to be able to wander through the house with a device connected to the internet and not lose the connection. For me this doesn’t work. I told my routers to use the same SSID and password, but really there are as many networks as routers, just with the same name. An iPhone, iPad etc typically connected to one of these will not automatically switch to the most powerful signal if I move to a different part of the house: on the contrary, the signal will become progressively degraded until I have to turn wifi off on the device and turn it on again to get the local connection. As my household is addicted to this wandering behaviour, this is clearly a pain.
What I can’t find out looking at the descriptions of Mesh systems is if they avoid this problem, i.e. by providing one indivisible network throughout the house. If they do, the wandering problem would be solved, admittedly at a somewhat high price. Can you explain if I would be safe with Mesh? As it is, I don’t want to spend a lot of money to be disappointed.
In passing one might ask how public wifi systems work in large spaces like railway stations, museums, trains even? This problem doesn’t seem to exist in those places.
It’s a great question. Technically, roaming is a function of the device (called a station or STA) connecting to a Wi-Fi base station (a basic service set or BSS). You’ll sometimes see STA and BSS in more detailed descriptions of the process.
With standard Wi-Fi base stations, the only thing you need to is set the network name or SSID, security type (WPA2, etc.), and passphrase exactly the same across all base stations. It sounds like you’ve done this. At that point, your “station” or laptop, phone, etc., handles all the interactions. Its operating system or radio firmware makes all the decisions about when to hand off among identically named (or even differently name) networks.
I found a really terrific article at SmallNetBuilder that has some good advice in it. If you have routers from different makers, running different firmware, a little too old, etc., this might all explain what’s going on. It notes that mesh has no secret sauce on roaming, but does ensure that all the nodes are running a variety of identical 802.11 protocols and delivering information perfectly required for those protocols.
Commercial (infrastructure) Wi-Fi systems consist of multiple components:
A router. Typically just one for the entire site, although there may be redundancy in order to gracefully handle failures. A rack-mounted device somewhere in the building’s wiring closet.
A Wi-Fi controller. Again, typically one for the site, but maybe with redundancy. A server running the control-plane Wi-Fi protocols (e.g. authenticate connections, generate keys, etc.) Also typically a rack-mounted device, probably located near the router.
Depending on how it is configured, it may be controlling one or more Wi-Fi networks (e.g. public vs. employee access), and those networks will connect to different ports on the router (either physical ethernet ports or VLANs sharing a physical port).
Multiple access points (APs). These are what you see scattered throughout the building. They all connect to the controller, usually via Ethernet cables. They are often powered by the Ethernet, in order to simplify installation.
With these infrastructure networks, there will be one Wi-Fi network for each network configured on the controller. All the APs configured to be a part of that network are just that - multiple access points for a single network.
Consumer devices (like a typical SoHo Wi-Fi router), combine all the components (router, controller and AP) into one device. Of course, this is at a conceptual level - there is probably only one chip doing all the work.
While it should be possible to design a mesh system along these lines (e.g. the node connected to the modem acting as router, controller and AP, and the rest of the nodes only acting as AP), I don’t think that’s how it’s actually done.
According to a paragraph on Wikipedia Wi-Fi mesh networks appear like a commercial (infrastructure) network to clients, but have the router/controller functionality distributed among all the nodes, which negotiate among each other to determine how traffic should be routed through the mesh.
I suspect the primary reason for this is that with an infrastructure network, all of the APs connect to a single controller, whereas with a mesh network, the connections between the nodes are ad hoc. It may not even be possible for all of the nodes to directly communicate with each other, especially if they are using Wi-Fi for inter-node communication.
I can just add to this that a common complaint our network people here on campus have to follow up on is when people claim to have terrible wifi reception in locations where there is usually very good coverage. Turns out, at least in their experience, Macs are particularly affected by this.
Apparently, Macs tend to hold on to a specific access point longer than many PC notebooks running Windows. This can lead to a situation where your Mac will show very poor throughput to wifi because it has in the meantime been moved to another location but it’s still trying to communicate with an access point that it used to have a good connection to. Instead of quickly givng up on the old access point and reconnecting to a better AP in the new location, it will try to keep alive the connection to the old AP.
I remember experiencing this in one particular location a lot with a 2013 MBP. The threshold at which a Mac seeks to connect to a better AP is not something the user can affect. So the only fix they can offer for Mac users when it happens, is to turn off wifi and turn it back on. That will force the Mac to connect to whichever AP shows best throughput at the current location rather than to the last AP it was connected to. Not an elegant solution, but it works.
My reply was no more than pedantry. Fiber is clearly better than Ethernet, if by better you mean faster.
For me, better means the best compromise between performance and price. Using that metric, I have to admit that Ethernet is better. I sure wouldn’t use Fiber in my home network — it’s far more than I need, and far more costly than I can justify.
As a guy who came up from AppleTalk/Localtalk, and Farallon Phonenet, I’m thrilled with my Cat 5e and Gigabit switch (though my Airport Time Capsule (3rd gen) is the weak link, and I plan to upgrade it shortly.
Well, fiber can definitely go faster. The theoretical maximum bandwidth is hundreds of Gbit/s on each of multiple optical wavelengths.
But for use in your home, it’s far less clear. Most people won’t be buying equipment with faster than gigabit speeds, because faster equipment costs a lot more. If everything is Gigabit Ethernet, then fiber will have the same bandwidth as copper, if everything is properly installed.
And fiber transceivers (probably SFP modules in your routers and switches) and interfaces for your computers are an additional cost.
Fiber’s main advantage, for home use is:
Immune to electrical interference. Especially important if you run it outdoors (in a conduit or overhead) - it won’t attract lightening and can’t create ground loops (e.g. if two buildings have different ground potentials)
Future proof. Properly installed fiber can go much faster than Gigabit Ethernet speeds. So you can move to higher speeds in the future by replacing your switches, interfaces and transceivers without pulling new cables through your walls and conduits.
But I suspect that it’s not worth the added cost for most people.
Yep…if one has an outbuilding needing wired access a run of buried fiber with a pair of transceivers on the ends is better for protection against lightning strikes…but beyond that for the vast majority of us the added expense of fiber isn’t worth it. That said…if I was building a house today I would run both fiber and cat 6E cables everywhere…but would just put the fiber in as a future proofing thing and not actually use it everywhere because the cost of tipping and transceivers doesn’t really buy one much beyond the lightning protection.
Not wanting to monopolize…but I’m SO lost…
As I said before: I’m often the only user on our WiFi, which is a Netgear R7900P, which uses 802.11ac; it feeds off a cable modem using our Xfinity signal. I typically get 60-80 mbps up and 5-6 down.
So all that is about the basement. Upstairs, I have used a TP-Link RE200 (v.2, with latest firmware) which was giving us pretty good signal, but often with a message on my iPhone to the effect that I need better security than WPA. Reading on a user forum, I was able to get to WPA2 through a complex hack, after a couple of months, it’s gone wonky. I often get messages that I’m using the wrong password, or that my Wi-Fi is not connected to the internet. Unplugging it, we get weaker, slower signal (from downstairs), but do connect. Rebooting and setting it up again does not fix this.
Our house is actually wired with ethernet. Can I just run a cable to a second router, such as TP-Link Archer A7, and give its network the same name as what I have downstairs, or will there be a conflict? Or should I try a different extender?
Wireless range extenders (like your TP-Link RE200) work, but they’re not ideal. They’re using the same Wi-Fi signal for communication with the primary router that they are using for communication with your devices. So they will, by necessity, reduce the total amount of usable bandwidth.
Mesh systems (e.g. the 3-band Linksys Velop devices) often work around this by operating on multiple bands (e.g. two 5 GHz radios and a 2.4 GHz radio), reserving one for communication between the nodes and making the others available for your devices. Systems that only have two radios generally don’t do this.
If you’ve already got Ethernet in the walls, then you’ve got a better option:
In the room where your main router is (the Netgear connected to your cable modem), run an Ethernet wire from one of its four LAN ports to a wall jack.
In the room where you need better connectivity, set up another router. Configure it to operate in bridge mode (so it won’t be creating a new LAN). Configure it’s Wi-Fi as you like (I do this in my home and configure the remote router to use the same SSID and password as the main router so my devices will automatically connect to either). Run an Ethernet wire from one of its ports (shouldn’t matter if it’s LAN or WAN, since it’s in bridge mode) to a wall jack.
In the location where the various wall jacks come together (typically a panel in a utility or storage room), set up a Gigabit Ethernet switch. Connect the lines from each room to the switch. If the switch doesn’t have enough ports for all the rooms, then make sure to connect the ones where you have connected equipment.
This is pretty much the same as what we’ve been discussing above, but using Ethernet instead of MoCA or powerline adapters.
You may see some issues with communication being interrupted as you roam throughout the house, as your devices switch from one Wi-Fi router to another. If this is a problem, then the solution is (as we discussed above) to replace your routers with two or more nodes of a mesh network. They cost more, but they’re designed for this purpose and tend to work better. You can (and should) continue using your Ethernet for the nodes to use when communicating among themselves.