5
\$\begingroup\$

Bottom-line up-front question:

I'm wiring my house for Cat6 Ethernet. Can I harm networking equipment if the equipment receives 3 V of AC voltage on any of the Ethernet pins with respect to ground?

Edit, and how to handle this question

After reading the answers, I have found that this post talks about two separate issues:

  1. What interactions (through induction, or capacitive coupling) exists between low-voltage and high-voltage lines in close proximity? How would this proximity impact devices, specifically Cat6 networking equipment, designed for low-voltage signalling? This is an electrical engineering question, and I believe belongs on the Electrical Engineering Stack Exchange site, which was the original site of this post.
  2. Is this a building code compliant setup? The answer is, not in its current form. This is because of the risk of high-voltage lines coming into direct contact with low-voltage lines. There are mitigations that can be employed, though the preferred approach is to put low-voltage and high-voltage lines in separate electrical boxes. I believe the issue is better addressed on the DIY stack exchange site, though I did not post this question on that site.

There are two directions to take, both for me, and for this site. For me the path is to work out a different setup for my Ethernet and coax ports. Although I ended up finding answers for electrical code requirements most helpful, the answer I am planning to accept (as of 9/18/2022) will be one addressing question 1, because I believe that is a type of question the Electrical Engineering site is meant to address.

Details:

I set up my first electrical box on this project, then I realized that there's the possibility for inductance within the electrical box between the 120 V wires to the outlet and the Ethernet cables. I looped each cable in the box once to provide 6" of slack to allow enough room to connect everything as suggested here.

Image shows double wide electrical cover with a 110 V, 15 A outlet on the right slot and two Ethernet ports and a coax cable, arranged vertically, on the left slot

The top outlet is connected directly to the circuit, whereas the bottom outlet is for a light and is interrupted by a switch on the door.

Because I am concerned about an inductive relationship between the power lines and the Ethernet lines, I measured the voltage between the pins of the Ethernet cable and ground and found that the maximum voltage detected was 3 VAC @ 60 Hz.

Image shows digital multi-meter testing double wide electrical cover, with negative probe inserted into ground hole on the outlet on the right side and with positive probe inserted into pin on the Ethernet port on the left side. The multi-meter shows an alternating current voltage of 3.105 volts on the display.

The light consumes 21 W of power on its max setting, so at 120 V that would be 0.175 A. To explore the relationship between current and voltage between the pins and ground, the following table is presented.

Setting Wattage (of light) Measured voltage (between Ethernet pin and ground)
switch off / light off 0 0.3 V
switch on / light off 0 2.584 V
1 7 W 2.629 V
2 15 W 2.630 V
3 21 W 2.748 V

Observe that the dominant influence on voltage between Ethernet pins and ground is dependent on the presence of voltage, and not current, in the power line. This indicates that the voltage I'm observing is dominated by electric fields and not influenced much by magnetic fields. This is an interesting result, and would indicate that my initial theory and reason for writing this post is not supported by data.

Regardless, my empirical measurements would indicate cause for concern. Can 3 V of AC at 60 Hz impact networking equipment?

Research on this topic:

This site indicates that voltage across the differential lines of Ethernet is about +/- 2.5 V. However, I do not believe that this is an answer because that indicates the voltage differential between individual pins. The voltage differential that I am measuring between pins is small, and does not exceed 0.050 V, well below the 2.5 V threshold. The voltage I am concerned about is between any of the Ethernet pins and equipment ground.

I have also looked at the electrical codes for guidance in this situation. NFPA NEC (2011): National Electrical Code, section 800.133(A)(2) states that power lines must be separated from networking lines by at least 2 inches:

Communications wires and cables shall be separated at least 50 mm (2 in.) from conductors of any electric light, power. Class 1, non-power-limited lire alarm, or medium-power network-powered broadband communications circuits.

But I have not found anything about the maximum voltage thresholds between the communication lines and ground. The EMF across the leads would also depend, not only on distance, but on the number of loops in the cable, and the expected current in the power cables. You could still achieve 2" of separation in the lines, but I have not found, in the code, how the cables should be handled once they're in an electrical box. I think this is really crucial because this is where I believe most inductive interactions would happen, though they seem to be small in this case.

I want to understand the tolerances that are built into networking equipment before attempting to attach and routers, switches, or computers.

Additional Info: (edit 2)

For some of the questions in the chat on this answer, it would be helpful for me to share the contents of the electrical box. The specific box that I got, according to my HomeDepot receipt is: TWO-GANG K.O 2 GANG HANDY BOX 2-1/8" DEEP 1/2+3/4

Open Electrical box, showing

\$\endgroup\$
11
  • 2
    \$\begingroup\$ What if you measure it without those power cables right in the way, touching your probe? \$\endgroup\$
    – Passerby
    Sep 17, 2022 at 23:08
  • 4
    \$\begingroup\$ @mkeith are you sure? diy.stackexchange.com/questions/59164/…. \$\endgroup\$
    – Passerby
    Sep 17, 2022 at 23:36
  • 3
    \$\begingroup\$ @mkeith the answer in that link and nec code both say the same as you, that a physical divider needs to be in place??? \$\endgroup\$
    – Passerby
    Sep 18, 2022 at 0:54
  • 6
    \$\begingroup\$ Can you get a suitable partition/divider for the box you chose? That'd fix the electrical code issues with having low voltage and AC mains in the same box \$\endgroup\$ Sep 18, 2022 at 13:05
  • 2
    \$\begingroup\$ @user1509669 -- you should be able to find dividers that work for your metal box, although you'll probably need to look to see what size box you have and who made it \$\endgroup\$ Sep 19, 2022 at 2:38

4 Answers 4

17
\$\begingroup\$

Twisted pair Ethernet is floating in regards to the electrical ground. It is only concerned with the currents that run in the pairs. The voltage you measure is 'common mode' i.e.: it appears on each of the wires and not creating a voltage difference between the wires.

What you are measuring is capacitive coupling and by using a multimeter it is easy to get fooled as the multimeter presents a very light load (around 10 MΩ) and only needs a small amount of parasitic capacitance to mains to show a voltage. If you tried to measure the current, it would be very small - only microamps.

Depending on the actual equipment, the Ethernet port should tolerate >1500 V for short times.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ I chose this answer because it addresses the question I had for this site (electrical engineering). Although I found that the answer from TonyM and manassehkatz-Moving 2 Codidact gave me the most actionable information for my situation. \$\endgroup\$ Sep 21, 2022 at 14:24
25
\$\begingroup\$

STOP! YOU CAN'T DO THIS AT ALL!

The rules and/or recommendations regarding distance between 120V/240V AC power cables and networking cables are for where the cables pass each other in the walls/ceilings/etc. However, you are not allowed to put low voltage (Ethernet, telephone, cable TV, burglar alarm, doorbell, thermostat, etc.) wires inside the same junction boxes as regular 120V/240V AC power. There are two exceptions:

  • Non-conductive fiber. Which unfortunately doesn't help much because most people only have fiber from service entrance to router, if at all. Fiber to the desktop might never happen due to the combination of ever-increasing speeds for twisted pair ethernet and the functional advantage of Power over Ethernet.
  • Strict separation. A classic example is a doorbell (or similar transformer). It is actually usually mounted on the outside of a metal junction box. 120V AC on the inside, low voltage (typically 10V - 24V) on the outside. There are dividers available for some multi-gang boxes, but you need to use them properly (no homemade versions) and I really don't recommend them.

So as convenient as it may be, you can't put ethernet (or similar) in a 2-gang box together with AC power. You might be able to do this legally with a metal box with a metal separator between the two gangs, but that runs into sizing issues (multi-gang metal boxes built from single-gang metal boxes normally have the sides of what become the middle sections removed so that it is one big box).

If it turns out you actually have two separate boxes or one real box (for 120V AC) and one box eliminator (for ethernet) then you are OK. But if, as appears to be the case, you have a 2-gang box with everything inside the same box, you MUST REPLACE IT as this is a truly dangerous situation.

This is not about a few V of induced current. This is about LIFE SAFETY. Simple example: Hot wire comes loose from receptacle:

  • If it touches neutral, ground or grounded metal box - circuit breaker trips (which is OK)
  • If it hangs around loose, serious danger if you open up the box to work on ethernet without turning off the breaker first.
  • If it comes into contact with an ethernet jack, you could have 120V AC going into your computer, either zapping your network card, or zapping you. The ground of the computer won't automatically solve the problem, particularly with a laptop if it is not plugged in to power at the time as it then has no connection to the building ground.

After reading a bit further about this being in a masonry wall, the practical solution is:

Leviton wall plate

  • Drill a hole outside the box just big enough to get the ethernet cables through. If you are concerned about cracks near the existing box, make the new hole a few inches away.
  • Use surface mount ethernet jacks, such as these from Amazon (example for picture):

TrippLite 4-port surface mount ethernet jacks

Just realized the bottom connector is coax. There are surface mount boxes for those too, or you can do the typical industrial/basement look: Mount a standard 4" steel box on the wall and put on an appropriate cover that can hold a Decora style device (i.e., the same thing you would use for a GFCI/receptacle) and reuse the existing 2 ethernet + 1 coax plate.

And a few comments based on the latest picture:

  • I don't know if a divider is available for that type of box or not (I suspect not, but I could be wrong). However, the cables would have to all go in through the correct locations - e.g., all low voltage on left and the 12 AWG NM on the right. It doesn't look like that right now.
  • For the power (12 AWG NM) cable, you need to have a proper clamp (impossible to tell from the picture) and the sheath should be removed except for a small amount (~ 1/4"). I see a big yellow loop.
  • It looks like the NM ground is wire nutted to the receptacle ground. That is normal for a plastic box. But with a metal box the NM ground needs to go first to a ground screw on the box. In addition, if you use a better quality receptacle (e.g., Leviton 5325-S) then it is self-grounding in a metal box and you don't need the receptacle ground wire. (Your receptacle may have this feature, check the instructions.)

Recommendation does not change. Actually, based on the look of that box I urge you to pull out the low voltage cables and route them through a separate hole to some sort of surface mount box. It will not be as nice looking, but it will be a lot safer.

\$\endgroup\$
11
  • 4
    \$\begingroup\$ Thanks for posting this. \$\endgroup\$
    – user57037
    Sep 19, 2022 at 1:05
  • 1
    \$\begingroup\$ I can tell you put a lot of detail into this answer, and I appreciate that. The chief concern I've gathered from your answer is that high voltage wires could come lose and touch low voltage wires in the box. That is something I have not considered, and although it is unlikely, it is certainly possible and should be protected against. Therefore, at a minimum a electrical box with a separator that is designed for the box is required. Although this falls short of your recommendation, I consider this to be the minimum. However I agree that it is ideal to handle low and high volt wires separate. \$\endgroup\$ Sep 19, 2022 at 1:55
  • 3
    \$\begingroup\$ Yes, a minimum is an appropriate divider (UL or ETL listed or similar, ideally made for/approved specifically by the manufacturer of the 2-gang box). But really, pull the low voltage stuff out, surface mount it and call it done. The risks are not insignificant and if there were ever a fire or other serious problem, this would be found pretty quickly and insurance would not be too happy with that. Low voltage wiring has almost no rules except "keep away from 120V+ voltage". \$\endgroup\$ Sep 19, 2022 at 1:59
  • 1
    \$\begingroup\$ That faceplate configuration would be just fine is combined with a box like this: lowes.com/pd/… which specifically provides the feature "Low voltage wall dividers allow for mixed voltages in the same box". I agree it definitely should never be done with the box he has. This one would also be ok, and really shows how "no requirements" the LV side has: lowes.com/pd/CANTEX-2-Gang-Dual-Voltage-Box-BRKT/5001724919 \$\endgroup\$
    – Ben Voigt
    Sep 19, 2022 at 16:28
  • 2
    \$\begingroup\$ Alternatively, if you don't like that half-blank panel, you could always install a second 120VAC outlet pair into the box. Doesn't solve the "where to put the ethernet jack" question, but more outlets might be useful provided they're all wired up correctly. \$\endgroup\$ Sep 19, 2022 at 21:01
11
\$\begingroup\$

Ethernet is a transformer isolated interface.

Any pair of data can have 1500 VAC in respect to ground.

And likely the 3 VAC is capacitive coupling, not inductive, as CAT6 is twisted pair, which rejects magnetic fields.

3 VAC within a pair itself might cause problems, but you don't have 3 VAC b within a pair.

And besides you only measure 3 VAC because the wiring is floating, it's not connected anywhere, and you are measuring it with a multimeter that has megaohms of impedance.

When connected to a device, the pair will get AC terminated with 75 ohms to device chassis, making the voltage barely measurable.

\$\endgroup\$
3
\$\begingroup\$

There are mitigations that can be employed

They MUST be employed. They are requirements of the Electrical Code.

though the preferred approach is to put low-voltage and high-voltage lines in separate electrical boxes.

Ideally, but dividers are allowed.

In the plastic world, they have half wish boxes. One half is a complete electrical box, the other half is "you wish it was a box" LOL. Those are fine, except for the plastic doing absolutely nothing to suppress magnetic fields.

enter image description here

By the way, in metal boxes on the AC power side, it is essential the Romex cable grounds go to the box first. Don't route them to the receptacle and ignore the box, as would be done on a plastic box. Grounding the box helps further suppress crosstalk. As a bonus, switches and some receptacles automagically pick up ground via the mounting screws, so that's one less AC wire to wrestle. It's quite a good system.

Also, make sure your AC power grounds are good all the way back to the panel and not "islanded", as it creates weird dangerous problems.

Therefore, at a minimum a electrical box with a separator that is designed for the box is required.

That will satisfy Code, yes. And a metal divider in that metal box should tamp down crosstalk, for obvious reasons.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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