[HN Gopher] Residential Networking over Telephone
___________________________________________________________________
Residential Networking over Telephone
Author : zdw
Score : 87 points
Date : 2025-02-09 23:54 UTC (3 days ago)
(HTM) web link (computer.rip)
(TXT) w3m dump (computer.rip)
| EvanAnderson wrote:
| It's tangential to the topic (though mentioned in the article
| towards the end) so I'll be lazy and ask the crowd here: What's
| the current state of the art in powerline networking?
|
| I have an application to network a couple of outbuildings. I
| don't have tremendous bandwidth needs (1Gbps would be glorious,
| 100Mbps would be fine). I would prefer wired connections for
| reliability and privacy. I would prefer not to trench and run
| fiber. The outbuildings already have electrical service from a
| central panel. Powerline networking might be a good fit.
|
| There's a lot of consumer stuff out there. I'd like something a
| little more geared toward an industrial, commercial, or service
| provider use. (Analogously, think of the difference between using
| a dumb Netgear switch versus a managed name-brand managed
| switch.)
|
| Edit: I appreciate the feedback I'm hearing here. This is exactly
| the kind of stuff I wanted to hear. Sorry to hijack HN for my
| personal needs like this!
| kennethrc wrote:
| FWIW, 5 years ago a Gigabit PA pair could only get ~200Mbit/sec
| between two rooms in the same house, and even that was finicky.
|
| Maybe the tech has improved since then, but I suspect not.
| tkuraku wrote:
| I know you wanted wired, but a point to point wireless link
| might be a good fit. https://store.ui.com/us/en/products/ubb
| EvanAnderson wrote:
| Ubiquiti stockholder. I want everybody to buy some. >smile<
|
| I have a variety of Ubiquiti point-to-point setups out in the
| field-- AirFiber down to Nanostations. It's good gear, for
| sure.
|
| For this particular application I'm concerned about being in
| an RF-hostile environment and unreliability, as compared to
| wired.
| rob74 wrote:
| I tried powerline networking for a while, but got so frustrated
| with the hardware (it disconnected randomly after a few days or
| weeks and failed to reconnect automatically - all the while
| hogging the WiFi although it didn't have an Internet connection
| via powerline, so all connected devices were offline) that I
| eventually gave up and now just use a WiFi repeater. Of course,
| that might also depend on the manufacturer? I would say the
| best alternative is to try it and just send the hardware back
| if it doesn't work (reliably)?
| brk wrote:
| "State of the art" and "powerline networking" have never really
| gone together well.
|
| The most common approach to your situation is point to point
| wireless. This can be off the shelf stuff from
| Ubiquiti/D-Link/Whatever, to more specialized products from
| Aruba, Cambium, etc. Ranges of 1Km and Speeds of 1Gbps are not
| out of the question at all. You do need line of sight though.
|
| Wireless (or fiber) to link buildings is also a good approach
| as it eliminates electrical conductivity paths across networks
| for lightning strikes and power surges.
| maples37 wrote:
| Since fiber is non-conductive, I've heard that it's code-
| compliant to run optical fiber in the same conduit as your
| power lines. (Normally, high-voltage and low-voltage
| conductors are required to be in separate conduits even if
| going to the same place.)
|
| Of course, this requires that you ran your electrical in
| conduit instead of direct-bury, so this might or might not
| work for you.
| EvanAnderson wrote:
| Sadly, the electrical was direct bury. I was hoping for
| conduit because I've had really good luck with prefab fiber
| assemblies in just this kind of application. No such luck
| here.
| zrail wrote:
| fs.com will custom build direct bury OS2 assemblies for
| reasonable prices. I'm a happy customer thrice over.
|
| https://www.fs.com/products/70220.html?now_cid=1148
| petee wrote:
| Depending on your lengths, budget, and your region's
| weather, might be worth giving second thought to just
| renting a little trencher and running a spare conduit or
| two (and maybe a water line too) then you're not locked
| in with one technology, and if that direct burial fails
| you can pull a new one in no time.
|
| I enjoy getting dirty though, so I had fun hand trenching
| out to my shed -- didn't run water and I'm kicking myself
| SpikedCola wrote:
| I did exactly this when I built my shed (with the same
| justification) - running fiber in the same conduit as the
| power to the shed.
|
| But, I found out during inspection that the Canadian
| Electrical Code prohibits this, unless the fiber is
| functionally related to the power lines it is running with
| (section 56-200).
|
| The explanation I received for the rule, was that someone
| like a telco installer could try and follow the fiber line,
| and find themselves in a dangerous situation (inside an
| electrical panel they aren't trained to handle).
|
| I was able to get an exception since I am the only one who
| will ever open the cabinets/work on this fiber, since it
| was not the main internet feed to the house, but just to an
| outbuilding.
| brk wrote:
| You still have to consider conduit fill ratios (though
| fiber usually does not take up much room). But more
| importantly, you have to make sure the conduit terminates
| on both ends in some way that makes it easy for you to
| separate out the fiber and power wires accordingly. If the
| conduit goes from electrical panel to electrical panel it
| really wouldn't be ideal to pull the fiber through.
| prennert wrote:
| Also, it can be ran through water pipes to retrofit areas
| that you dont want to dig up [0]
|
| [0]: https://www.bbc.co.uk/news/articles/clygd9nznj9o
| baby_souffle wrote:
| It does get better with each generation.
|
| Biggest factor is noise (thunk inductive loads like motors) on
| the circuits. Place your termination adapter as close to the
| main panel as you can and ideally on the same circuit or at
| least the same phase. Gfci and similar will nuke a good deal of
| signal.
|
| It was a pain, but I made a matrix of every outlet and had a
| simple iperf server running on a raspberry pi on a lipo battery
| bank. I went around with a laptop and a usb eth adapter and did
| basic tests to get an idea of limits.
| MisterTea wrote:
| I have no experience with power line networking but I would
| start there and experiment.
|
| If that doesn't work I would try two things: If the electrical
| cable is in a conduit or duct of some sort you could try to run
| a 600V rated Cat 6 cable (to comply with electrical codes) or
| fiber right alongside the feeders. That will not be easy as you
| likely have to pull the feeders, pair the network cable and
| then pull it back. And that is if the conduit hasn't collapsed
| or intruded by roots/water/dirt/etc. Second is an aerial line
| ran through trees or poles.
|
| A third option is something most would avoid but just run a
| thin length of black poly tube with a fiber inside along the
| ground. Years ago a friend powered his shed at his parents
| countryside house by running a length of romex in a black
| sprinkler tube along the ground hugging the treeline of his
| property. Not exactly to code and illegal for power but
| sometimes you gotta do what you gotta do and he never had a
| problem.
|
| Industrial stuff isn't far from consumer gear. Most of it uses
| the same components but in a rugged DIN rail or panel mount
| housing. Even industrial stuff is embracing wireless.
|
| And for fiber look up BiDi - bidirectional - SFP's that use a
| single fiber that is easier to route. They sell rugged outdoor
| simplex fibers for this kind of stuff.
| ikiris wrote:
| Never run copper Ethernet between buildings. If you're going
| to run something run fiber.
| MisterTea wrote:
| Never? Nah. Sure there could be issues with lightning or
| ground loops but A. use grounded surge arrestors at each
| end before termination into equipment, B. Don't use
| shielded cable. Unshielded in a conduit will be fine with
| 60Hz power but loads downstream producing harmonics (lots
| of switching supplies, VFD's, etc) can cause interference.
| This is why I said fiber a lot.
| harrall wrote:
| Really depends on your wiring and what's on the circuit.
|
| Using powerline as a single point-to-point in 2018 worked
| flawlessly for me with 0 drop outs, 0 latency problems, and 0
| throughput problems for 4+ years and then my lines changed and
| it became useless.
|
| If it works well, it can be a good way to put off the work
| required to hardwire.
| ssl-3 wrote:
| I've used powerline gear twice, both for similar applications.
|
| In the first instance, someone wanted wifi in their detached
| garage at the back of the lot. A couple of randomly-selected
| powerline adapters and an access point later, and that rig
| worked reliably for them for over a decade.
|
| In the second instance, someone else wanted wifi in their
| detached garage at the back of the lot. A couple of randomly-
| selected powerline adapters and an access point later, and I
| couldn't make it work at all.
|
| The main difference between these two was that one had power
| fed from the house's main breaker panel, and the other had
| power fed directly from the same transformer that the house was
| using but it had its own service entrance (and meter).
|
| This anecdotally suggests it can work, and that security is
| inherently enhanced by not allowing the neighbors to poke at it
| at all (if I can't use it because there's a transformer on the
| loop, then the layperson next door can't use it either).
|
| However: At my own place with my own detached garage at the
| back of the lot, I just use some fairly minimum-effort wifi.
| I've got a Mikrotik access point in a central spot in the
| house, and another one in the garage that acts as both a wifi
| client and a wifi access point (and an ethernet bridge).
|
| Works fine. Speeds are in the >100 megabit range in both
| directions, which is a whole fuckton of bandwidth for anything
| I intend to do out there.
|
| If the garage were far-enough away, I'd enhance this by
| mounting the requisite mikrotik hardware on the outside walls
| instead of the inside walls. If the garage were way, way out
| there, then I'd use different mikrotiks with directional
| antennas to backhaul, and additional radios indoors for wifi
| coverage.
|
| If I had more than one outbuilding, then I might use a one-to-
| many topology with an omni at the house and directional
| antennas on the outbuildings.
|
| (I made an extra SSID that was just for the garage to keep the
| dumb devices out there from wandering onto the house's access
| point for whatever dumb reason they elect to do that, but meh:
| They'd still do dumb shit like that even if I were to string
| fiber.)
| univerio wrote:
| If you have coax, I've found MoCA to be more reliable than
| powerline.
| ikiris wrote:
| TLDR: don't
|
| More words: no really don't. Either do wireless or bury some
| fiber depending on needs.
| gertrunde wrote:
| I replaced my home plug av2 adapter with g.hn ones a year or
| two ago, and it has been much better.
|
| Lower latency, more stable, and higher bandwidth. (The av2 ones
| would just drop out for 90-ish seconds a few times a month,
| which was quite annoying).
|
| It's interesting the article talks about home plug av2 and g.hn
| coexisting, when my experience has been that g.hn powerline is
| much better, and it feels like g.hn is generally perceived as a
| successor to av2 in the UK.
|
| Edit: also, I'm in agreement with other folk in that point-to-
| point wireless is the solution I most often see used in that
| case, or sometimes fibre.
|
| Edit2: and it's also probably the cheapest one to test. If it's
| not good enough, then you know you're looking at wireless or
| fibre.
| jcrawfordor wrote:
| The current state of HomePlug is weird, the organization
| behind it dissolved in 2016 and basically dumped the spec on
| the table (they made it public domain). Since it's closely
| related to IEEE 1901 it doesn't seem likely that HomePlug
| will ever "go away," but it's also not going to get better.
| So I think it's definitely fair to say that G.hn is the
| replacement, in that it's more actively developed, but
| HomePlug devices remain plentiful on the market. IME HomePlug
| devices are more commonly sold in the US than G.hn powerline
| devices, but I get the impression that things are different
| in Europe.
| johnklos wrote:
| It's perfect for things that just need a modest amount of
| bandwidth, like a TV. Most streams are a few tens of megabits,
| and that's about all you can reasonably expect from powerline
| networking.
|
| When you try to use as much bandwidth as you can, they fail in
| sometimes spectacular ways. I have some old Radio Shack 10 Mbps
| ones that have brought a high end Juniper to its knees when the
| Radio Shack box's "collision" LED was fully lit. Newer devices
| do better, but it's best to just use powerline networking for
| things that need steady, continuous traffic.
| jcrawfordor wrote:
| the typical "industrial" version of powerline is IEEE 1901,
| which for example is used by some electrical utilities for
| SCADA and by some smart meters. The difference can be somewhat
| academic as many modern powerline protocols are members of the
| same "family," and IEEE 1901 has two different PHYs that are
| basically adoptions of HomePlug and HD-PLC (another powerline
| protocol that I didn't mention because it never really went
| anywhere with consumers). Some HomePlug devices are certified
| as interoperable with the HomePlug-derived flavor of IEEE 1901,
| so sometimes they're literally the same thing.
|
| IEEE 1901 is only specified for 500Mbps though. The more
| industrial use-cases are rarely high-bandwidth and that market
| is often pretty satisfied with even Kbps performance, they're
| just sending Modbus and that sort of thing. IEEE 1901 equipment
| is also going to be quite a bit more expensive to obtain since
| the market is small and mostly works on large contracts.
|
| I agree with others that point-to-point wireless is probably
| your best bet. It can be similarly priced to HomePlug AV2
| transceivers and, sort of ironically, will probably turn out to
| be more reliable. HomePlug tends to have trouble when there are
| multiple neutral bonds because that adds "side legs" to the bus
| that harbor reflections. That's why people tell you that, as a
| rule of thumb, your transceivers need to be on the same circuit
| for good performance.
|
| You might also consider direct-burial Ethernet, which is not
| very expensive. However, you'll get into some complexities with
| bonding and lightning protection and you might need to study
| the code book a bit.
| rob74 wrote:
| If you look at how incredibly thin standard telephone wires are,
| it seems incredible that with the right technology you can
| transmit up to 100 Mbit/s over a pair of them...
| epcoa wrote:
| In addition to conductor thickness not being a sole direct
| factor in bandwidth (the interconnects carrying 32 Gb/s PCI-e
| or Thunderbolt are much thinner), standard telephone wire is
| the same or greater than Cat 5e.
|
| Most telephone interconnects will be at least 24 AWG, some
| short run cables may be 26 AWG (same as Cat 5). Really old
| interconnects will often be 22 and even as high as 18. And the
| thing is, these are usually worse from a high frequency
| transmission line standpoint as they weren't designed for
| anything above 4khz voice.
|
| Poor noise rejection, cross talk, poor loss characteristics,
| and inconsistent termination are the more challenging issues in
| pressing high speed out of the phone network.
| giantrobot wrote:
| I first used PhoneNet in a small school computer lab. For the
| time it was awesome. A bunch of Macs could pretty easily share
| files and print to the shared printers.
|
| A point to remember is the machines weren't sharing terabytes,
| gigabytes, or even necessarily megabytes of files. Text or Word
| docs, spreadsheets, and occasionally networked games were a
| majority of the traffic. Rarely was the speed ever an issue. You
| just had to learn some behaviors like don't open and edit a file
| directly from a network share and everything would slow down when
| sending a job to the printer.
|
| An aspect I thought was cool was it worked with cheap adapters on
| all the Macs in the lab from the old Performas to the latest
| PowerMacs. Ethernet cards were still fairly expensive as was
| cabling. A cheap RJ-11 extension cord from RadioShack could add a
| new drop to the PhoneNet network.
|
| Once Ethernet got cheap it was far and away superior to
| LocalTalk/PhoneNet networking. There was no going back once you
| tried Ethernet but for a time LocalTalk was an amazing Mac
| feature.
| autoexec wrote:
| It's a shame that a lot of telecoms have either ripped out a
| bunch of the copper wire they had installed themselves or have
| just not replaced stolen copper after they successfully lobbied
| the FTC to let them abandon the lines. It'd have been a great way
| for communities or small groups to set up their own tiny
| networks.
| icedchai wrote:
| At DSL or T1 speeds? You're better off with wireless. We'll
| also ignore the problem of all copper going back to the central
| office.
| axiolite wrote:
| That copper requires upkeep, which is expensive. The maximum
| speeds are terrible by modern standards. Using the existing
| right-of-way (established for the copper lines) to install new
| fiber that will replace the copper, is the right move.
| api wrote:
| Speeds aren't great. Directional WiFi antennas and a local mesh
| are a much better option for communities. There's also a slower
| 900mhz WiFi standard out there that can cover a lot of area and
| may still be faster than crappy abandoned copper in
| questionable shape.
| castratikron wrote:
| EV charging also uses HomePlug. When Ethernet needed to be added
| to an existing signal line on the EV connector, HomePlug was
| chosen as the way to do it. Every EV consequentially will need a
| HomePlug modem to communicate to a charger, so, the world will
| have HomePlug around for quite some time.
| gwbas1c wrote:
| According to https://en.wikipedia.org/wiki/SAE_J1772, HomePlug
| is used in Europe. The article is otherwise rather ambiguous
| about how it's used in common chargers.
|
| At least for the _typical_ level 2 charger, the signaling is an
| analog square wave.
| api wrote:
| I still use HomePlug to connect some cameras and TVs in my house.
| It works, though real world performance tends to be far lower
| than advertised.
|
| You only get close to advertised performance if all plugs are
| more or less on the same circuit branch and the distance isn't
| too large.
___________________________________________________________________
(page generated 2025-02-13 23:01 UTC)