Grounded conductor with voltage...why?

[ssynoground]

Ok, here is the skinny. I have come across a situation in a commercial setting where the grounded conductor has 2V to ground at its panel and in the field. The set up is this: 208V single phase/ 100A service fed with #3AWG THHN STR COPPER. The Panel is a IG panel that is fed this way: 208V from the XFMR to an 100A breaker at another panel, then from there it goes underground(50ft) to a disconnect#1, then to an UPS, then to disconnect#2, then back underground(50ft) to the IG panel. Now this is what I found, the IG from the IG panel should be unbroken to the XFMR but, the contractor spliced it at the panel where the IG panel is being fed from. Secondly, I found at disconnect#2: the grounded conductor is bonded. My thinking is the grounded conductor may need to be larger to handle the load. The load is high on the grounded conductor at 48A. The customer is having problems with their computers and the computer guys are telling me the problem is the 2V on the neutral. I hope you can sift thru this muck and find some gold!

 

[iwire]

I have come across a situation in a commercial setting where the grounded conductor has 2V to ground at its panel and in the field.

Not unusual or problematic at all. It is caused by the voltage drop on the neutral.

Now this is what I found, the IG from the IG panel should be unbroken to the XFMR but, the contractor spliced it at the panel where the IG panel is being fed from.

There is nothing wrong with that either assuming the contractor kept it isolated.

Secondly, I found at disconnect#2: the grounded conductor is bonded.

This could be a violation or it could be required depending on the UPS system setup.

My thinking is the grounded conductor may need to be larger to handle the load. The load is high on the grounded conductor at 48A.

You should be able reduce that load by balancing the panel load better.

The customer is having problems with their computers and the computer guys are telling me the problem is the 2V on the neutral.

I think they are just grasping at straws but if the customer wants to pay you to reduce this minor difference of potential by all means, balance the panel, increase the grounded conductor size.

I hope you can sift thru this muck and find some gold!

Well I think the entire IG deal is a waste of time but I guess thats more muck than gold. :grin:

 

[mcclary's electrical]

Ok, here is the skinny. I have come across a situation in a commercial setting where the grounded conductor has 2V to ground at its panel and in the field. The set up is this: 208V single phase/ 100A service fed with #3AWG THHN STR COPPER. The Panel is a IG panel that is fed this way: 208V from the XFMR to an 100A breaker at another panel, then from there it goes underground(50ft) to a disconnect#1, then to an UPS, then to disconnect#2, then back underground(50ft) to the IG panel. Now this is what I found, the IG from the IG panel should be unbroken to the XFMR but, the contractor spliced it at the panel where the IG panel is being fed from. Secondly, I found at disconnect#2: the grounded conductor is bonded. My thinking is the grounded conductor may need to be larger to handle the load. The load is high on the grounded conductor at 48A. The customer is having problems with their computers and the computer guys are telling me the problem is the 2V on the neutral. I hope you can sift thru this muck and find some gold!

I've been told this doesn't become a problem until the number gets over 5 volts

 

[iwire]

I've been told this doesn't become a problem until the number gets over 5 volts

That is pretty much what I have heard as well.

But that said I would sure like to get an explanation why that should matter at all considering the equipment is powered through a power supply that essentially is an SDS.

 

[brian john]

The customer is having problems with their computers and the computer guys are telling me the problem is the 2V on the neutral.

They may be having problems but it is not related to the 2 volts. Test have been conducted where voltages as high as 15 volts were measured between the neutral and EGC. with no adverse effects.

We had a case where there was 115 volts between neutral and ground. The site operated this was for an undetermined amount of time. This SDS system neutral was not connected to ground and at some point a branch circuit phase conductor shorted to ground.

Tell the tech they are wrong and to check hardware and software.

 

[hardworkingstiff]

But that said I would sure like to get an explanation why that should matter at all considering the equipment is powered through a power supply that essentially is an SDS.

I've been asking "techs" the same thing for decades. I've never got an answer that I liked. Usually it comes down to a referencing a data line to ground and getting a ground loop going. Somehow that comes out in my mind that "their" equipment is connecting neutral to ground somehow and they are causing their own problems and blaming the EC.

 

[hurk27]

I've been asking "techs" the same thing for decades. I've never got an answer that I liked. Usually it comes down to a referencing a data line to ground and getting a ground loop going. Somehow that comes out in my mind that "their" equipment is connecting neutral to ground somehow and they are causing their own problems and blaming the EC.

Exactly

Back in the day of signal reference grounds, it was a problem, but with todays Ethernet protocols it should not be a problem, unless they are using old out dated networking equipment. I used to install old MUX cabling for the state of Indiana employment offices, up to about 1999 (yea they were behind the times), and ground loops was a problem which we tracked down why we had current on the grounding, it was never a problem from the difference of potential between the grounding and the neutral, but it was a problem when we had current on the grounding in a building, and was also against code so we looked for the cause and repaired it.

Current on the grounding will share current on signal paths that are also parallel to the grounding, and 60hz on singnal paths can disrupt data flow.

But they don't get taught that in computer classes.

 

[hardworkingstiff]

Current on the grounding will share current on signal paths that are also parallel to the grounding, and 60hz on singnal paths can disrupt data flow.

But but, that's a no no. (oh yea, it's the EC's fault)

 

[iwire]

But but, that's a no no.

Well kind of yes and no.

A no for use not for them.

Even today when RS-232 or RS-485 data links are used they carry a reference ground. We need Gar in here I think he makes a product to fix this issue.

But now most data links are Ethernet or fiber with no ground reference.

 

[hurk27]

But but, that's a no no. (oh yea, it's the EC's fault)

signal wires can parallel grounding all they want, not a good idea, but it's done all the time, now paralleling the grounded (neutral) conductor, is a NO NO, but when you have neutral current on the grounding, then its a problem for all the paralleled signal grounds

 

[suemarkp]

The ground and neutral are tied together at some point (and hopefully just at one point). But, a cable has two ends, and there is no guarantee that each is is going to the same grounded power source and therefore have the grounds at the same reference level.

I noticed the most problems when RS232 or video cables were run between buildings -- the ground won't be common then. But even in the same building, issues could occur (typically two different power transformers were involved). As much as we'd like to think that a ground grid and building steel is all at the same potential, it isn't.

Using differential circuits, neither line of which is tied to a chassis ground, was the key to making interfaces reliable over a distance.

 

[gar]

091004-0847 EST

Some general comments.

Suppose I have a laptop computer, battery powered, and connected to nothing else, then I can put that computer chassis at 1,000,000 V above ground and it will work just fine. This assumes the computer is adequately insulated from ground, not much capacitance to ground, and no rapid rates of change of this voltage. In other words I do not couple any external signals into the computer.

Somewhat similar to a bird flying up to and perching on a high tension wire.

Getting down to lower voltages. If the computer chassis is connected to an EGC at some point and the neutral and hot wires supplying the computer have some substantial voltage, maybe up to 1000 V, relative to the EGC, then there should be no problem from this voltage difference other than the breakdown voltage in the power supply in the computer between the hot and neutral with the computer chassis and its connection to the EGC.

Where problems occur is with the connection of a computer to other components external to the computer chassis and there is no signal isolation between these separate components.

Note: Current Ethernet devices have built-in isolation via a transformer or use of fiber-optic cable. Devices connected in this manner should have no problem.

Keyboards, mice, RS232, RS422 or 485, printers, and other devices may not and probably do not have isolation. This means that the common reference for these signals is probably the chassis of each component. Many will call the chassis ground and this derives from the early days of radio when everything in the radio was referenced to the chassis, and the chassis was connected to earth as part of the antenna system. I much prefer to call this reference common rather than ground. And I generally assume that common is not at ground potential.

Any two different points on earth are not the same potential. Same is true of any two different points on a wire. The important question is the magnitude of the difference in potential and how this affects the circuits connected to these points.

For a moment consider a parallel port connection from a computer to a printer. In general the voltage levels on these signal lines are all referenced to the chassis. This is because the electronics common is connected to the chassis. Parallel port outputs are TTL signals and have voltage levels of 0 to 5 volts.

A standard SN7400 TTL gate has the following specifications. Any voltage from 0 to 0.8 V at the input is a logic 0. Any voltage from 2.0 to 5 is a logic 1. Indeterminate from 0.8 to 2.0 V. The outputs are at least 2.4 V for logic 1, and less than 0.4 V for a logic 0. Thus, under worst case conditions a shift of the reference voltage by 0.4 V could cause an error. Thus, if the voltage difference between the source and destination chassis is this great, then an error could occur. The voltage difference between the two receptacle EGCs supplying the two ends of the communication path are the primary concern.

Ask questions and I can add more later. Will be gone for a while.

ssynoground:

You may need to solve the problem, but I would not suggest you are at fault. Most likely the people supplying or setting up the computer system do not adequately understand their system.

.

 

[hurk27]

There is a few things I would like to add, to this discussion:

While Mark brought up the problem with signal grounds running across different buildings, or we could say across different SDS power sources, the reason this is a problem, and one with no NEC acceptable fix, is because these separate services and SDS system are required by the NEC to be bonded in almost all cases to the primary grounded conductor, which will reference the MGN out on the pole, which will always have current on it, which by having a parallel signal common, in parallel with it, will cause current to flow also on the signal common. in this case the only solution is isolation, by some means or another.

The other problem which is not apparent to many is the dumb requirement of slamming ground rods in at each piece of equipment or network station, we see this requirement in many industrial installations, most times at equipment using old networking protocols, such as the RS-232/485, or others.

when we do this we have just introduced current on the signal reference common, because there will always be current flowing across Earth. there doesn't have to be very much voltage difference of potential, to have a conductor share parallel current.

any time you place a rod in the ground at two separate points on earth and connect a conductor between these two rods such as the signal common, you have just paralleled the current flowing in earth, and these Earth currents can be coming from as far away as the other side of Earth. there is no fix to this other than not putting the rod in.

so the next time you run up against, an IT guy who doesn't have a clue, draw the parallel path ways out and try to educate him/her, but do it in a nice way, because remember they are not getting taught this in there networking classes.

 

[gar]

091004-1411 EST

Continuing.

Consider a CNC machine with direct RS232 connection to a computer some distance away, maybe 15 to a 1000 ft away. Suppose there is one main panel with the ground bus and neutral bonded together and tied to a grounding system. Next put the computer 5 ft from the CNC machine and on a butcher board maple top table.

The CNC is supplied from a 240 open delta system with one transformer secondary center tapped for the 120 V circuits. The CNC is supplied with a 40 A 240 V delta and an EGC that all run directly back to the main panel. No 120 V circuit is brought to the CNC. Internally the CNC gets its 120 V from a step-down transformer from the 240 input. All exposed metal parts of the CNC are connected to the EGC, and so is the chassis and common of the computer electronics within the CNC connected to its EGC. In turn the RS232 common, pin 7 and 1, are connected to the chassis. This RS232 common is the reference for the transmit and receive signals.

At the computer it gets its power from a 120 branch circuit from the main panel and has its own EGC going back to the main panel. Between the computer and the CNC is an RS232 cable with wires that might be as large as #22, and this includes the common wire (pin 7 at the CNC).

Assume a distance of 200 ft to the main panel. The EGC conductors are probably #6 and #12. Even at a total of 200 ft their combined series resistance is about 0.32 + 0.08 = 0.40 ohms. Assume the RS232 cable is 15 ft and #22 wire, then its resistance is 0.24 ohms.

Next at the CNC short one hot leg to the EGC of the machine and there will be about 24 V RMS between the CNC chassis and the computer chassis. And about 100 A is trying to flow thru the common RS232 wire. Very shortly that common wire burns up and now the difference in voltage between chassis is close to 60 V RMS. It is an instructive exercise to figure out how I got these numbers. Components at both ends are likely to burn out as well as the RS232 cable.

If you have lesser voltages developed on the EGC lines there will only be data errors.

Note: I placed the computer 5 ft from the CNC and this was to illustrate that the close proximity did not solve the problem. I would solve this problem by having the customer use the existing stepped down supply of 120 in the CNC, or add a step-down transformer in the CNC. Then bring this to a receptacle on the side of the machine and connect its EGC pin to the machine EGC. Connect the computer to this receptacle. Now there will be negligible voltage difference between the computer and the CNC when the above said short occurs. However, when the short occurs both the CNC chassis and the computer chassis will rise to about 60 V RMS relative to the neutral and ground bus at the main panel.

There are implied assumptions above, but they do not grossly change the result.

If you do not understand my description, then ask questions.


ssynoground:

I believe you should sort of ignore the voltage of neutral to earth, and rather see how the different pieces of equipment are connected together. Then measure the voltage difference between the chassis of associated equipment that are DC directly connected. If you are dealing with USB or TTL signal levels, then the differences must be very small. Maybe 0.1 V or less. With RS232 possibly a couple volts, but maybe not. Depends upon many factors. Non-isolated 422 or 485 maybe 4 or 5 V.

If you are dealing with a Gaussian noise source, then your actual peak noise pulses will be many times an RMS reading, 3 to 5 times or more. Voltage measurements may not be of great help unless they are extremely small. Deduction and experimentation may be of greater help along with a knowledge of how the problems may occur.

.

 

[SG-1]

This article by Fluke does a good job explaining the voltage reading between the grounded conductor and the EGC.

http://support.fluke.com/find-sales/Download/Asset/1989076_6003_ENG_C_W.PDF

The graphic for the Hot & neutral reversed has two Hs between the measuring points. The top H should be an N.

Does the problem start after a certain piece of equipmen is brought on-line? Like while a printer is running ?

 

[gar]

091004-1414 EST

hurk27:

As you said supplemental ground rods are probably not a solution, and may worsen the problem.

And even worse is when a manufacturer or their people or IT people suggest disconnecting the EGC at one end or the other, and maybe using only a local ground rod. This is an NEC violation and is very unsafe.

.

 

[hurk27]

091004-1414 EST

hurk27:

As you said supplemental ground rods are probably not a solution, and may worsen the problem.

And even worse is when a manufacturer or their people or IT people suggest disconnecting the EGC at one end or the other, and maybe using only a local ground rod. This is an NEC violation and is very unsafe.

.

So very true, and I have seen so many other so called quick fixes over time, like installing a 1 to 1 power transformer with no NEC required bond across the neutrals, or the neutral not bonded to an EGC at the transformer, or lifting the grounding conductor, and running it to a ground rod.

this is where they need the proper training, and education, at the IT school, as to what the real problem is, and what the solution is, whether a EC is required to fix it, or they need isolation of signal common on there end.

 

[ssynoground]

This may be the ticket!

This may be the ticket!

.[/QUOTE]

A standard SN7400 TTL gate has the following specifications. Any voltage from 0 to 0.8 V at the input is a logic 0. Any voltage from 2.0 to 5 is a logic 1. Indeterminate from 0.8 to 2.0 V. The outputs are at least 2.4 V for logic 1, and less than 0.4 V for a logic 0. Thus, under worst case conditions a shift of the reference voltage by 0.4 V could cause an error. Thus, if the voltage difference between the source and destination chassis is this great, then an error could occur. The voltage difference between the two receptacle EGCs supplying the two ends of the communication path are the primary concern.

Ask questions and I can add more later. Will be gone for a while.

ssynoground:

You may need to solve the problem, but I would not suggest you are at fault. Most likely the people supplying or setting up the computer system do not adequately understand their system.

.[/QUOTE]

You hit the mark! What I took from this was, the problem could be from other equipment that is directly connecting to the computers without some form of isolation. The customer has multiple computers that are set up the same way as the one computer that is experiencing the problems. I'll keep the MHF posted of any new problems related to this incident. Ok, one quick question: Does the fact that I can read 2V across the Neutral and the Isolated Ground cause a problem with the computer and/or produce an error?

 

[gar]

091008-1344 EST

ssynoground:

The answer to your question about the 2 V between neutral and the EGC at the computer under normal conditions should have no relationship to the problem.

Make sure there is not a wiring error somewhere. Check the voltage between the computer chassis, probably a DB25 connector shell is a good connection to the chassis, and the wall plate of the outlet socket, assuming metallic, should be zero volts within 10 MV AC. I read 3 MV on one computer and 6.8 MV on another. If there is substantial voltage, then check to the EGC pin of the receptacle. If still substantial voltage, then find some other EGC and check to it. From a copper water pipe to my 6.8 MV computer the reading is 35 MV. In this case I am assuming no voltage drop on the water line and that it is a test lead back to the main panel. There is about 60 ft of EGC from my main panel to the 6.8 MV computer.

The voltages I am measuring are probably from the computer power supply input filter which I can expect has capacitors connected to the computer chassis and thus to the computer EGC wire.

Relative to the first question in your paragraph. Interconnected equipment, the ground connections, and current flowing in the ground circuits is where many problems exist. Not the voltage neutral to ground.

.

 

[winnie]

The set up is this: 208V single phase/ 100A service fed with #3AWG THHN STR COPPER. [....] The load is high on the grounded conductor at 48A. The customer is having problems with their computers and the computer guys are telling me the problem is the 2V on the neutral.

I agree with the points that others have brought up about how a computer system _should_ be able ignore the voltage between the grounded conductor and the EGC. As a side note:

The description above (208V and _single_ phase) strongly suggests that you have two legs of a three phase system rather than a true single phase system. If this is the case, then no matter how you balance your 120V loads, you will have lots of neutral current. The only way to eliminate the neutral current is to have only 208V loads.

I am curious: how many output phases (bridges) does the ups have, and does it produce the correct phase angles to match single phase 208?

-Jon