4

I am at the end of a renovation, and only now  I discovered that it is not recommended to run ethernet cable in parallel and closely to power cables, even if they are in separate conduits, but I may have the option to improve the safety and the data integrity: As you can see in the photo there are 3 pvc conduits runs parallel to each other. Each with a diameter of 16mm. In one is threaded a telephone cable, in the other a cat 7 ethernet cable and in the third a 240V power cable.( I dnot know in what order) On the left side there is an additional 50mm in diameter flexible PVC conduit also parallel and closely to the three conduits, but it is empty.I estimate that there is  2-3 cm between the cables. (not between the conduits) I have the option to cancel the cat7 that is threaded inside the 16mm conduit, and thread a new cat7 cable inside a flexible metal conduit (that protects low voltage cables from electromagnetic) and thread the metal conduit into the flexible 50mm PVC conduit (conduit within a conduit) Is it worth making this change? (I am thinking of a future option that maybe I will increase the network speed, and even if there is a lot of amperage in the power cable which increases the strength of the magnetic field)     And another question: can there be an interference to the telephone cable as well because it is run parallel near to the power conduit? It is not shielded and not a twisted pairs, but I can replace it with a more high quality cable, and also transfer it in the 50mm conduit. Thanks in advance for any answersenter image description here

3
  • (I understand that you ask about interference, not lightning. I also understand that a primary lightning strike is rare. I'm not paranoid about parallel power and network cabling.) But when I do avoid that, it's because I don't want lightning induced events on the power supply to needlessly induce an additional surge on the Ethernet which could kill some additional electronics. It's like buying a surge suppressor: not strictly needed but sometimes worth it. – Jirka Hanika Dec 21 '20 at 17:16
  • 10
    As you can see in the photo... erm, did you take a moment to have a good look at the photo you have provided us? – J... Dec 21 '20 at 17:59
  • 3
    All cat7 (and most cat6, even cat5e) is already shielded. Is the extra layer of shielding offered by the metal conduit "better"? Yes. Are you going to notice? Not a chance. – dberm22 Dec 21 '20 at 22:47
11

You discovered wrong. Ethernet runs at 100MHz and up, your line voltage at 60Hz is virtually DC at that frequently difference.

With that said, you should still run your data cables in conduit. Not for interference concerns, but so that in 2040 you can easily replace the cat5 with something new for our 20K ultra VR direct brain video or whatever is the deal then. Plastic Smurf tube is fine, it doesn't need to be metal.

15
  • 11
    @RanShalom: I believe that you read something like that, but the claims in that you read in that passage are simply incorrect. This stuff matters if you are building an actual billion dollar data center with a gazillion cables and the difference in 0.01% of performance translates to thousands of dollars, but in a residential or office commercial environment, you can pretty much haphazardly space and align Ethernet and power cables as far as interference goes. – whatsisname Dec 20 '20 at 22:55
  • 2
    This answer needs some serious sources, as it flies in the face of conventional wisdom. When I was studying for CCNA certification, we were taught that interference from power cables can cause speed drops. I've also encountered, multiple times, performance problems fixed by moving power cables away from networking cables. – BlueRaja - Danny Pflughoeft Dec 21 '20 at 16:40
  • 5
    Agreed that this answer is lacking sources. Power cables can carry a lot more than just 60Hz - plenty of equipment generates harmonic noise on the mains lines and that noise can radiate to signal conductors. If you've never had to fight with interference, consider yourself lucky. It's not a bogeyman, it does exist in the wild, and not following best practices is exactly how you discover that it's a problem. Industrial environments tend to be the worst due to having a high density of electrically noisy, high power equipment. – J... Dec 21 '20 at 17:58
  • 2
    As per @J..., the problem is not the 60Hz (or 50Hz in Europe), but transients carried on it, particularly as a result of switching inductive loads. The conductors in the mains cable efficiently radiate RFI. As to what to do, remember the inverse square law - even a few inches away from the mains cable is better than right next to it. Metal conduit should act like a faraday cage (and shield you), but a better option might be to use STP (shielded twisted pair) if you are concerned. Perhaps a few lost packets when your boiler pump (or whatever) switches on or off would not be disastrous. – abligh Dec 21 '20 at 18:04
  • 3
    @whatsisname: Yes, conventional wisdom is often wrong, but it's also usually right. Especially when it's what's taught to professionals in the field, and is also backed by the experience of (probably) most professionals. Since your claim goes against all that, the burden of proof is on you. You still have not provided any sources. – BlueRaja - Danny Pflughoeft Dec 21 '20 at 18:29
8

Speaking as somebody who's spent much more time messing around with networks and power cables than with bricks and beams: keep them moderately separated but don't go full paranoid.

Power cables are substantially more robust than network cables, have different failure modes, and are tested differently by people with different skills and equipment.

Network cables are invariably twisted pair and sometimes screened, which will reject what is known as "common mode" interference. Untwisted cable, as shown for example in Best way to stick a very long Ethernet cable semi-permanently to walls? is not intended for high-speed networks and should be avoided (this photo https://upload.wikimedia.org/wikipedia/commons/f/ff/Ams-ix.k.root-servers.net.jpg shows that type of untwisted cable in its intended use as a low-speed console connection).

Network interface cards have isolation circuitry, but this can be overwhelmed (inductors driven into saturation, insulation damaged) by noise of sufficient amplitude.

Power cables can often have substantial amounts of switching noise on them (e.g. from an induction hob), which can easily be sufficiently high in frequency to result in a measurable network degradation. Note that I'm saying measurable here, not necessarily detectable "by eye".

An extreme case of this is a nearby lightning strike which can result in fried network equipment, although the ingress for that is usually a computer's power supply (hence the recommendation that stuff be disconnected during storms).

There is a slim possibility that network traffic could be "sniffed" by putting monitoring equipment on the power lines. This is a VERY slight risk compared with other ways your personal data could be exfiltrated.

Broadly speaking, high-frequency noise and cross-coupled data will travel between several houses etc., until stopped or at least attenuated by the inductance of a power distribution transformer (normally described as "at the end of the street").

The "keep them separate and cross at right angles" rule applies to 'phone lines and low-speed data cables, i.e. the sort of thing that was run for substantial distances before about 1990. These days, as I've said, keep them moderately separated but there's much less need to go full paranoid.

2
  • 3
    Re "Power cables can often have substantial amounts of switching noise on them..." +1 – Arvo Dec 21 '20 at 15:30
  • 3
    Apropos switching noise for anybody unfamiliar with the art. These days most power supplies take very short sips of current from the mains, repeating this tens or hundreds of thousands of times a second. The sharp on-off transitions result in harmonics which can easily get into the MHz (million cycles-per-second) or 10s of MHz range, and that's high enough to be difficult to distinguish from the 10 to 100 MHz data on a network connection. – Mark Morgan Lloyd Dec 21 '20 at 15:44
4

It's not the problem it's made out to be.

Ethernet is running at millions of frames per second. Power is 60 frames per second.

Imagine we stretch that time frame out by a factor of 86400*60. Now, the AC frequency is akin to the rising and setting of the sun, and the Ethernet is akin to ships passing light-gun signals. Does the sun interfere with that? Of course not, it is massively, glacially too slow to affect that.

One conduit for telecomm is fine.

In those conduits, one conduit should get all telecomm - phone, ethernet and coax/cable TV if that is present.

The large conduit should be reserved for big power, or for water lines or whatever the original layer envisioned.

Consider fiber-optic.

If you're really that worried about it, maybe fiber-optic is for you. It has the merit of being immune to EM interference, but also, it can be placed in AC power conduits if the optic line is entirely non-conductive.

8
  • Actually, my one real experience with fiber several years ago was on communicating with electric meters using fiber to link to a router. Distance was an issue and it was one instance (high voltage equipment in the basement of a huge building) where interference with serial and/or Ethernet cabling was a potential issue. But someone else handled the details - my job was just to configure the meters and test communications. – manassehkatz-Moving 2 Codidact Dec 21 '20 at 2:52
  • 2
    In enterprise networks, Fiber IS used for short distances in RF prone areas. I've set up fiber media converters for going from Fiber to UTP Cat 5 cabling while working for an internationally known aerospace company in one of their manufacturing buildings. – Davidw Dec 21 '20 at 7:24
  • 2
    @Steve around here, the "killer app" for fiber is you can put it in AC mains conduit and boxes. It basically ends all discussions of "I wanna put Ethernet with my power wires" :) – Harper - Reinstate Monica Dec 21 '20 at 8:02
  • 4
    @SteveSether -- the other killer app for fiber is "going between buildings" -- even if it's not a long distance, the need for a Category rated and UL497 listed primary protector at each end of the outdoor link is a major cost/installation hassle adder for running copper inter-building – ThreePhaseEel Dec 21 '20 at 12:38
  • RBR (reduced bend radius) fiber makes "special rules for bends" pretty much "don't kink it" and actually will take tighter bends than Cat5e/Cat6/Cat6A/Cat7 assuming you actually want those to work to spec. – Ecnerwal Dec 21 '20 at 13:21
1

Let's talk interference for a second. Some other answers have dismissed EMF as a concern, but it's worth noting something on the subject.

Most networks today are 1GB or less. That equipment is ubiquitous, as is the cable. Very few people are running 10GB networks because

  1. You need far better cable than Cat5e (or even Cat6)
  2. The equipment is much more expensive
  3. Very few people need bandwidth on that scale. I mean, most WiFi still struggles to top 300MB (0.3GB) and virtually nobody complains about that

If you're just running a 1GB network (which is probably 99% of DIYers), you're probably fine as-is. Metal is slightly better to limit EMF interference, and 1GB networks are fault-tolerant enough to where you probably wouldn't notice it. In fact, most devices aren't making a lot of EMF interference on electrical wires to begin with (which is why "powerline" Ethernet devices are a thing).

If you think you'll be running a 10GB network anytime soon, you'll want to rethink that and place the cables elsewhere, since you'll be paying a LOT for that extra shielding and even the minor EMF coming off electrical wires could cause interference.

The better solution here would be fiber, which doesn't use EMF. That equipment isn't that much more expensive than straight Ethernet, and the fiber itself has come down in price to reasonable levels. For comparison (late 2020 pricing), 100ft Cat5e is about $14, and 100ft two-channel fiber is $35 currently on Amazon (prices subject to change)

4
  • Can you amend your post to include the kind of equipment a DIY'er would need to add to go fiber? I appreciate the Cat5e vs 2-ch fiber, but what else would I need? – Bryan Boettcher Dec 21 '20 at 18:23
  • 1
    @BryanBoettcher That would be off topic but Super User discusses what you'd need here. Basically, you need something that can accept both Ethernet and what's called SFP. They make devices that do nothing but that bridge (one SFP and Ethernet), but if you've got a home network you should be able to find a switch that does SFP pretty easily. – Machavity Dec 21 '20 at 18:36
  • I run 500Mb on the cat7 , but it runs adjacent and parallel to the power cable that conductive a current of 14A. Is it safe, and will not interference to the integrity of the data? (From the EMF produced by the amperage in the power cable) – Ran Shalom Dec 24 '20 at 17:48
  • Cat7 is heavily shielded to reduce EMF interference. The likelihood you notice on a gigabit network is low. – Machavity Dec 24 '20 at 17:50
0

One thing rarely mentioned is the interference from the Ethernet cables to other devices.

In theory, this should be prevented by conformity testing, but in practice, it's not too uncommon to find devices with appropriate conformity labels being sold, which, when combined with particular cables, tend to radiate like crazy. The situation with not-very clear standards for Ethernet cables, and optimistic cable labeling adds to the issue.

As a result of the what I mentioned above, it might happen that you, for example, have interference to FM radio, or low VHF over the air TV. To make issue even more insidious, the "regular" users of network equipment in general aren't going to be capable of easily detecting interfering devices, so you usually won't find any comments about that in equipment reviews.

Putting the cables into metal conduits is one of the things which might help mitigate such issues, if they pop up in the future. Of course, if you don't watch terrestrial TV, and don't listen to terrestrial radio, then these potential issues won't be very interesting to you.

1
  • 1
    A link I added to my answer earlier showed somebody using what looked like Cisco console cable for Ethernet. Untwisted and not a good idea at all for a long run- even if it did happen to have the right connector at each end. – Mark Morgan Lloyd Dec 21 '20 at 15:46

Not the answer you're looking for? Browse other questions tagged or ask your own question.