ECG's from different circuits

[OpenCircuit]

I am being told not to bother separating EGC's according to circuit in j-boxes, since "it's all ground", but NEC says this - 300.3 (b) "All conductors of the same circuit, and, where used, the grounded conductor, and all equipment grounding conductors and bonding conductors, shall be contained within the same raceway, auxiliary gutter, cable tray, trench, cable or cord, unless otherwise permitted..."

If EGC's from different circuits are spliced together in j-boxes then they wouldn't be staying in the same run. So what's the right thing here?

 

[ptonsparky]

Splice them all together in the jboxes, gutters, etc. What they want is the EG to be in the same cable or raceway as the CCC, not current carrying conductors in one and the EG in a totally separate one.

 

[texie]

I am being told not to bother separating EGC's according to circuit in j-boxes, since "it's all ground", but NEC says this - 300.3 (b) "All conductors of the same circuit, and, where used, the grounded conductor, and all equipment grounding conductors and bonding conductors, shall be contained within the same raceway, auxiliary gutter, cable tray, trench, cable or cord, unless otherwise permitted..."

If EGC's from different circuits are spliced together in j-boxes then they wouldn't be staying in the same run. So what's the right thing here?

Take a look at 250.148.

 

[hbiss]

I think somebody is confusing EGC (equipment grounding conductor) with the grounded conductors (neutrals). EGCs should certainly be all spliced together where possible.

-Hal

 

[480sparky]

300.3(b) doesn't say to not splice them all together.

If you're not supposed to tie them together, I guess all those commercial bathrooms I've wired over the years with a 277v light switch and a 120v receptacle in a 2-gang box were illegal.

 

[winnie]

IMHO OpenCircuit is pointing out a potential problem that is generally not considered and usually not an issue.

When we join all EGCs in a junction box (as we are required to) then we are creating parallel paths; a fault will flow on all of the various EGCs back to the source supplying that fault. So in a very real sense these mingled EGCs are now no longer completely contained in the cable along with the circuit conductors that they are protecting.

Code requires all of these EGCs to be joined, and since the EGCs are not supposed to carry current in normal circumstances, this doesn't cause a problem.

In some situations where EGCs are subject to external magnetic fields, and are used as signal as well as safety ground, these parallel paths can be an issue, in which case you need to look at system design to avoid these interconnections.

-Jon

 

[OpenCircuit]

Take a look at 250.148.

Well, that makes it clear they need to be spliced all together + box.

 

[OpenCircuit]

IMHO OpenCircuit is pointing out a potential problem that is generally not considered and usually not an issue.

When we join all EGCs in a junction box (as we are required to) then we are creating parallel paths; a fault will flow on all of the various EGCs back to the source supplying that fault. So in a very real sense these mingled EGCs are now no longer completely contained in the cable along with the circuit conductors that they are protecting.

Code requires all of these EGCs to be joined, and since the EGCs are not supposed to carry current in normal circumstances, this doesn't cause a problem.

In some situations where EGCs are subject to external magnetic fields, and are used as signal as well as safety ground, these parallel paths can be an issue, in which case you need to look at system design to avoid these interconnections.

-Jon

Yeah, this is what I was looking for/ concerned about. This code article refers to both neturals and EGC's. As for neutrals it's a real issue for them to be paralleled. EGC's do not normally carry current - if they do it's short term (alternatively, low level faults that are not enough to trip the overcurrent device) and even less likely for ground fault occurring on several circuits simultaneously.

I wondered if it would make a "cleaner" ground if they were separated according to circuit (even if someone were to use them in the future for signals) but as someone pointed out, code now requires EGC's from ALL circuits to be bonded to the boxes... So the only option left would be isolated ground to the electronic equipment. Isolated ground is in some cases able to reduce/eliminate ground loops.

Btw if you have some good references to signal grounding design, I'd be interested in looking more into it.

 

[ptonsparky]

300.3(b) doesn't say to not splice them all together.

If you're not supposed to tie them together, I guess all those commercial bathrooms I've wired over the years with a 277v light switch and a 120v receptacle in a 2-gang box were illegal.

Seems like we find out we've been doing something wrong every three years and the world will end if we continue.

 

[winnie]

I wondered if it would make a "cleaner" ground if they were separated according to circuit (even if someone were to use them in the future for signals) but as someone pointed out, code now requires EGC's from ALL circuits to be bonded to the boxes... So the only option left would be isolated ground to the electronic equipment. Isolated ground is in some cases able to reduce/eliminate ground loops.

IMHO it is better to tie all grounds together in a particular junction box, and if you are in a situation where this will cause a problem then you should design things from the start to only have one ground in that junction box. During a fault transient you don't want the metal chassis of something plugged into one circuit to be at a different voltage than the metal chassis of something plugged into a different circuit right next to it.

If 'ground loops' are a problem (for whatever reason) then the answer is to design the grounding topology to both meet code _and_ not introduce problems.

Btw if you have some good references to signal grounding design, I'd be interested in looking more into it.

If possible, I recommend using signaling that does not rely on ground to be part of the signal path. Eg. 'differential' signals or even better yet transformer coupled differential signals. Compare RS-232 serial to RS-485 serial. Same serial protocol, but RS-232 depends on a good ground.
-Jon

 

[OpenCircuit]

Seems like we find out we've been doing something wrong every three years and the world will end if we continue.

Just wait and it will go back to being right.

 

[roger]

On a metallic raceway job all the EGC's are tied together, no stopping it.

Roger

 

[al hildenbrand]

Just wait and it will go back to being right.

You're going to wait a long time. The rule in 250.148 is old, settled Code.