The purpose of ground wire is just to carry current in milliseconds enough to trip the circuit breaker, it won't even get hot. So what is wrong with using very small ground wire? For example. You are using #4 AWG in the Hot and Neutral wires (70 ampere capacity) and you use #14 AWG as ground (15 ampere capacity). If the ground is connected to an enclosure and the Hot wire touches it. It would take very fast for the breaker to trip, not enough to heat the #14 AWG. What is the reasons this is not allowed?

migrated from electronics.stackexchange.com Oct 14 '18 at 1:33

This question came from our site for electronics and electrical engineering professionals, students, and enthusiasts.

  • 1
    It is a bit more complicated than that. I am not an expert (someone else who is will answer more completely). But a couple of things: 1 - With a GFCI, even a little current going to ground will result in a very fast trip as the difference between neutral & hot is detected, but not all circuits have a GFCI; 2 - With a regular breaker, it is the current on the Hot wire that trips the breaker, and depending on the breaker (some allow for more time for motor startup, etc.) it can take a while. A regular breaker only trips really fast with a very high current overload. – manassehkatz Oct 14 '18 at 1:47
  • 2
    @manassehkatz interesting notion with the GFCI ground only having to carry 8-30 milliamps. However that relies on a key assumption: that this ground Is monogamous to this circuit. Given the retrofit grounding rules (grounds are allowed and encouraged to be promiscuous), that ground could easily be grounding other stuff that may not have GFCI protection. The GFCI would never know since GFCI's do not interact with ground. – Harper Oct 14 '18 at 2:27
  • 1
    @Harper I'm not actually suggesting using a small ground wire. I'm trying to say that the ONLY time where it could possibly make sense is with a GFCI - i.e., the same place where you might end up not using a ground wire at all. – manassehkatz Oct 14 '18 at 2:30
  • 1
    @manassehkatz heh, yes, true. – Harper Oct 14 '18 at 2:30

Total heat notwithstanding, wires also have resistance. As in "ohm" resistance.

What you may not be understanding is breaker trip curves. Breakers have a thermal trip mode from 1.0x up to about 10.0x breaker rating, where the action is delayed as a bimetal strip warms up. This is to be permissive of motor starts, inrush current, or "toaster and microwave at the same time" short term overloads. The thermal trip is sized to warm at about the same speed as the wire in the walls, so it trips somewhat before the wire warms enough to be dangerous. That requires matching between breaker and wire size.

Breakers also have a magnetic trip mode that instantly trips at around 10x breaker rating.

With too-small a ground wire, the breaker cannot magnetic-trip because it cannot flow enough current to hit 10x, because of the too-high resistance of the wire. And as you can guess, that's trouble.

Now we're in thermal trip mode. We have a 70A breaker's bimetal strip, in a race with a #14 wire to see if the breaker trips before the #14 wire hits Fahrenheit 451. Obviously, the breaker is going to lose.

Now as a stupidly practical thing, one way you can solve this is do the big stuff in EMT or other fixed metal conduit. In those cases, the steel conduit and boxes are the ground path. And steel conduit has such good conductance (1/resistance) that it can ground pretty much any wire that will fit in it.

  • 1
    One point this glosses over is that ground wires are undersized relative to the circuit conductors on circuits that size -- a 70A or 80A circuit using 4AWG for the CCCs will use an 8AWG ground wire as per NEC table 250.122 (in fact, that 8AWG EGC is good all the way out to 100A). – ThreePhaseEel Oct 14 '18 at 2:49
  • Supposed my wire size is 3.5mm^2 (or about 12 AWG). And I add another 3.5mm^2.. would the equivalent be 7mm^2 (or about 9 AWG)? Given two wires or paths.. would the electrons cooperate to distribute themselves or diffuse among the two path or would they cramp into one maybe because it's closer to their turns, etc? – Samzun Oct 14 '18 at 3:14
  • 1
    Code says don't run wires in parallel unless they are huge and are cut and terminated to the exact same length. Esp. grounding conductors. When you join two wires together at both ends you make a loop. Loops in grounding conductors are a bad idea. They can resonate with RF fixtures and appliances, like fluorescent lights and microwaves. – HardScale Oct 14 '18 at 5:00
  • 2
    Well there shouldn't be any current at all on grounds, so fix your ground faults and leave the ground loops to the airplanes. If you have two separate #12 circuits wired to the same junction box, very likely those grounds will get lumped together, an example is a kitchen 2-gang with a receptacle and a garbage disposal switch in the same box, since the disposal can't be on the receptacle circuit. Or if you have a too-weak ground and use the retrofit rules to fit a proper size ground, again paralleling to your old ground. Or if you have a matrix of EMT conduit. Lotsa paralleling. – Harper Oct 15 '18 at 1:26
  • 2
    Not after one breaks and you don't detect it, no. Use the right size ground wire. If you are adding a ground to compensate for a deficient ground, use the right size wire. – Harper Oct 15 '18 at 16:03

Under CEI rules your PE, on residential installation, should be at leas as thick as phase conductor. That's because it has to discharge all the current potentially carried by the phase(s). RCD are there to increase the protection, not to replace PE. So it's true that an RCD trips with 30mA current, but it's also true that you don't have to risk a fire if RCD is gone bad and you have a big phase-to-groung failure. Anyway in some older homes you still find 1,5mm^2 PE for 16A circuits (so with 2,5mm^2 phase and neutral).


In some cases the NEC in the US does allow a reduced size for the EGC. Old Romex cable had a smaller EGC, I think it was 16 AWG on 12 AWG Romex. Since the ampacity charts are conservative for current carrying conductors, I don't think it was ever a problem.

However the NEC only allow limited reduction in size for good reasons. NEC table 250.122 spells out the minimum EGC sizes by breaker size. I would not go lower than these values.

NEC 250.122

Number one, even in the milliseconds it takes for a breaker to trip, in the event of a very high current fault, the wire can melt or even explode. This is more likely with a smaller wire. If the wire melts before the fault is cleared, you have a very bad situation.

Often breakers don't function as well as they should - I've been very surprised how long some breakers take to trip when deliberately shorted. (This isn't something I would recommend or condone, but someone I worked with thought shorting it was an acceptable way to trace a circuit.) Again, in this situation, the smaller wire is more dangerous.

The other possibility is a ground fault that does not clear the breaker, this could conduct at near or slightly over the full capacity of the breaker for a while - long enough for a very undersized wire to overheat and cause a fire. This should never occur with a GFCI, but like breakers, GFCI's are not perfect.

  • What would happen if I don't use neutral (our country doesnt because its only 240v 2-pole) and I'll tap the neutral in the service entrance and the utility pole doesn't have a grounded rod below it inserted into the soil (and neither in my service entrance). Will the neutral become live and all the metal enclosure I connected to the neutral bus via the bonded ground becomes live? Won't this be more dangerous situation? – Samzun Nov 23 '18 at 23:06
  • For my 125A main breaker. Should I use #8(100A) or #6 (200A) neutral wire to the service entrance? See related post at diy.stackexchange.com/questions/151105/… – Samzun Nov 24 '18 at 2:46
  • @Samzun - sorry you lost me there. – batsplatsterson Nov 24 '18 at 12:04
  • 1
    @samzun - I think you are referring to the NEC table 250.122 rating, that table says the breaker is not to exceed 100A for 8AWG so 8AWG no good for 125A, 6AWG ok for up to 200A, so you need 6 AWG. – batsplatsterson Nov 24 '18 at 12:25
  • 1
    @samzun, the resistance of the 75m of 6awg is certainly low enough that a short will trip any breaker in your house, by a wide margin. – batsplatsterson Nov 24 '18 at 21:42

A fuse, especially a slow-blow could allow enough current to pass for enough time to completely blow off a small ground wire, leaving the device connected to the mains through a nice fat wire, ready to cause death and mayhem.

The fuse should be specified to protect the mains wiring (AWG4 in your case) so it will also protect an AWG4 ground wire, should a ground fault occur.


My answer is a little different than most of the comments.

NEC Table 250.122 and if a service grounding conductor 220.66 gives us the correct ground wire sizes for the corresponding over-current protection and conductor size. In my state we use those sizes because if we don't we are breaking the law. This precludes any other discussion we have on selecting a smaller conductors size, because many people smarter than us have decided what we are allowed to use.

The final comment I would have on the subject would be, if you did not follow the NEC. and someone was subject to a burn or shack hazard that caused damage or loss of life. Then it was determined that an undersized ground caused the accident. Would you be willing to absorb the liability? That's why when there is already a rule we follow it, end of discussion.

Stay safe.

  • 1
    That's curious, where you live -- are you saying you can't go larger than a 250.122 size either? – ThreePhaseEel Oct 14 '18 at 21:03
  • 1
    @ThreePhaseEel - Sure you can the NEC is a minimum standard. Most engineering firms have an Engineering Standard which are more strict and is allowed by the NEC. The question specific to this question would be: Why would you increase it? – Retired Master Electrician Oct 15 '18 at 12:24
  • 1
    Good, I was thinking that you had a mighty strange local amendment roaming around :) As to your question about why such an increase would happen, it typically happens because you're using premade cable such as SER, quadplex, or mobile home feeder, which will most often have more EGC in it than the NEC strictly demands. – ThreePhaseEel Oct 15 '18 at 22:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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