Wiring machine shop equipment from Delta grounded b to 120/208/Y 3 phase 4 wire

[tryinghard]

gar, So adding electrodes does not rectify the noise - correct?

To answer your question correct equipment grounding is an effective fault path to/from source (250.4(A)(3)-(5))

 

[gar]

110202-1928 EST

tryinghard:

Thanks for the definition. "Correct equipment ground" from your definition seems like a relatively definable value in contrast to what CNC suppliers say about one needing a good ground with no concrete definition of good.

Anywhere that I was able to look at ground path noise levels where there was a supplemental ground rod I saw very little reduction in noise by this addition. Again meaning the ground rod at the machine was simply added in parallel with the EGC.

With my isolation system I have experimentally applied a 1000 V RMS 60 Hz sine wave directly to one of the wires in the interconnect cable, and simultaneously sent a continuous stream of data at 115.2 kbaud with no data errors. This is a vastly larger noise signal than what a direct RS232 connection can tolerate.

Any time a direct RS232 connection is used the difference in voltage between the source and the destination must be very small. Very small means maybe 3 V peak if good drivers and receivers are used, and maybe a fraction of a volt if the receiver is simply the input of a bipolar transistor.

On measuring voltage difference between a source and destination. There is no possible way to make a voltage measurement between two points without creating a closed loop. If a changing magnetic field links this loop, then a voltage is induced in the loop. There are ways to try to minimize the loop area and separately it may be possible inject into the measurement system a means to cancel or partially cancel the induced effect.

Broadly there are two ways to introduce a ground path noise signal into an RS232 connection. One is magnetic induction of a voltage, and the other is a current injection into the ground path conductive path.

.

 

[tryinghard]

...Thanks for the definition. "Correct equipment ground" from your definition seems like a relatively definable value in contrast to what CNC suppliers say about one needing a good ground with no concrete definition of good.

These type suppliers do not address the whole grounding picture and they are misleading installers to add electrodes "good ground" = good earth connection, rather than correct bonding and effective fault path to supply source.

Anywhere that I was able to look at ground path noise levels where there was a supplemental ground rod I saw very little reduction in noise by this addition. Again meaning the ground rod at the machine was simply added in parallel with the EGC.

My point exactly the earth ground is not an effective path to source, drive as many rods as you want install big grounding electrode conductor and have no better application - still noisy.

Broadly there are two ways to introduce a ground path noise signal into an RS232 connection. One is magnetic induction of a voltage, and the other is a current injection into the ground path conductive path.

Aren't both of these only seeking source rather than data equipment - hence an effective fault path, coupled with correct neutral bonding, ensures these through a controlled route.

 

[gar]

110202-2156EST

tryinghard:

The installation of a "correct equipment ground" may not and probably does not solve an RS232 electrical noise problem if one exists. A "correct equipment ground" in no way guarantees that you will not have a CNC RS232 noise problem.

A supplementary ground rod appears to do far more, even though not much, in noise reduction than if you simply compared the EGC DC resistance with the resistive component of the earth. This probably results from a lower impedance of the earth at high frequencies compared to an EGC wire.


I do not understand this statement:

Aren't both of these only seeking source rather than data equipment - hence an effective fault path, coupled with correct neutral bonding, ensures these through a controlled route.

If I have a conductor and a current flows thru the conductor, then there is a voltage drop between any two points on that conductor. Does not matter where this conductor goes to get its source of current. Alternatively if I have a DC circuit and between any two points in this circuit there is a DC voltage drop, then there is current flowing in that circuit. The only reason to describe this as a DC circuit was to avoid problems from induced voltage (current) in a loop.

.

 

[tryinghard]

Broadly there are two ways to introduce a ground path noise signal into an RS232 connection. One is magnetic induction of a voltage, and the other is a current injection into the ground path conductive path.

gar, I'll understand your terminology "current injection into the ground path conductive path" to be objectionable current, meaning it's bonded wrong so this just needs to be corrected not accepted (unless I'm misunderstanding). But why would induced voltage travel to the data equipment?

 

[gar]

110203-0842 EST

tryinghard:

You are assuming that an EGC starts at the main bonding point (main panel where it is also earthed) and goes in a continuous path, never looping back, and nothing connects to this EGC wire that can introduce (inject) a current into the wire. This does not happen in the real world.

One example: A large cement slab is poured into the ground where a large machine is to be located. In this slab are large steel plates to which the machine chassis is attached. These are conductive paths into the earth. Thus, when the machine is attached to this slab and the EGC is attached to the machine chassis there is a closed loop path.

There may be other paths. Such as black pipe air lines, lube lines, and maybe some direct attachments to building steel. Also this machine may be part of a transfer line with many machines connected together.

So there are all sorts of paths.

There is capacitive coupling and maybe some leakage resistance from the hot wires in each of the machines connected to the EGC.

You can classify these currents as "objectionable currents" and a code violation, but you won't and probably can not eliminate them.


Another loop that exists in a direct RS232 connection is the common wire in the RS232 cable. RS232 is an unbalanced signalling system. This means that the data lines are referenced to a common, and this common is usually connected to the equipment chassis at both ends. It would not work unless connected at both ends.

.

 

[tryinghard]

You are assuming that an EGC starts at the main bonding point (main panel where it is also earthed) and goes in a continuous path, never looping back

If not it is extremely unsafe in fact often its not an assumed path its actually a known intentional conductor! Why would looping matter?

and nothing connects to this EGC wire that can introduce (inject) a current into the wire. This does not happen in the real world.

Sure it happens in the real world all the time, it?s just as important as the supply voltage. Incorrect neutral bonding causes a parallel current path on all associated conductive items. This can mean dangerous touch voltage and ?Current flowing on metal parts of electrical equipment and conductive building parts causes elevated electromagnetic fields in the building or structure. These low frequency electromagnetic fields can negatively impact sensitive electronic devices, particularly video monitors and sensitive medical equipment?, from Mike Holt for EC&M magazine on Objectionable Current

There is capacitive coupling and maybe some leakage resistance from the hot wires in each of the machines connected to the EGC. You can classify these currents as "objectionable currents" and a code violation, but you won't and probably can not eliminate them.

These are not intentionally bonding the neutral to frame or intentionally using the frame as a neutral path, if these do happen as you say I?m still wondering why they would route to the data equipment?

 

[gar]

110203-2121 EST

tryinghard:

At no point in my previous discussions did I make any comment about connecting the EGC to the neutral at any point except at the required bonding point.

But now to one single point at a time.

In my statement that you quoted I said "You are assuming that an EGC starts at the main bonding point (main panel where it is also earthed) and goes in a continuous path, never looping back"

Your response:
If not it is extremely unsafe in fact often its not an assumed path its actually a known intentional conductor! Why would looping matter?

I do not understand to what "If not" applies.

What my statement said was that I concluded that you assumed the EGC was a wire that did not form a loop.

If I take a wire X ft long and connect the two ends together, then I have formed a loop. If the two ends are not connected together, then it is not a loop. I can create a loop by driving two rods in the earth and running a wire between the two rods. As well as many other ways to form loops.

If I have a loop and a changing magnetic field linked to that loop, then a current is induced in the loop. A non-looped wire in a changing magnetic field does not have current induced in it like in the looped case.

A wire with its two ends unconnected in a changing magnetic field has an induced voltage in it, but I can not prove this experimentally without forming a loop. The path forming the closed loop might have a resistance of 100,000,000,000 megohms, but it is still part of a loop.

.

 

[tryinghard]

At no point in my previous discussions did I make any comment about connecting the EGC to the neutral at any point except at the required bonding point.

When you say "inject a current into a wire" I'm not familiar with this type statement but incorrect neutral bonding is the most common mistake and it injects current where it does not belong onto an EGC and all conductive paths.

I do not understand to what "If not" applies. What my statement said was that I concluded that you assumed the EGC was a wire that did not form a loop.

You are assuming that an EGC starts at the main bonding point (main panel where it is also earthed) and goes in a continuous path, never looping back

The "if not" notes how serious this would be as you state above, in other words if the equipment grounding does not start at the main bonding point (and ground) and continue throughout it would be extremely hazardous. And in most cases this is not assumed its actually known as a EGC.

With equipment grounding I've never been concerned about looping what-so-ever only effective fault path to/from source. I would think some amount of looping would commonly occur but if so why would this current travel to data equipment?

 

[hurk27]

Not to take this post any farther sideways, I just want to point out that there are many instances of code legal installs that will place a 60hz noise on the signal shield between a CNC computer and the main computer in a office some where, every time you cross an SDS you have regrounded the neutral and put this signal shield in parallel with the neutral, every time you have a different service feeding each end you have placed this signal shield in parallel with the neutral conductor, so there are times Gars isolation is the only way, or the method of data transmission will have to change to an isolated one.

We have large steel mills around here that have SDS all over the plants, just about any large industrial plant will have this.

 

[tryinghard]

Not to take this post any farther sideways, I just want to point out that there are many instances of code legal installs that will place a 60hz noise on the signal shield between a CNC computer and the main computer in a office some where, every time you cross an SDS you have regrounded the neutral and put this signal shield in parallel with the neutral, every time you have a different service feeding each end you have placed this signal shield in parallel with the neutral conductor, so there are times Gars isolation is the only way, or the method of data transmission will have to change to an isolated one.

We have large steel mills around here that have SDS all over the plants, just about any large industrial plant will have this.

hurk, I am not questioning gar's isolation I am questioning the use of ground electrodes and large GEC's to eliminate the noise. I am claiming correct bonding and equipment grounding often rectify noise issues to the sensitive equipment.

I can see as you point out with a network the server or main computer is usually plugged into a different supply then the CNC?s supply creating a parallel neutral path (bad juju?s).

 

[tryinghard]

...I just want to point out that there are many instances of code legal installs that will place a 60hz noise on the signal shield between a CNC computer and the main computer in a office some where, every time you cross an SDS you have regrounded the neutral and put this signal shield in parallel with the neutral, every time you have a different service feeding each end you have placed this signal shield in parallel with the neutral conductor...

Fiber's the way to go in cases like this.

 

[gar]

110204-2305 EST

tryinghard:

I was having Internet Explorer crash problems this morning. Actually crash of XP as well.


I do not believe in any of my posts that I suggested the neutral be connected to the EGC anywhere except at the one single required location. I used "main panel" to refer to this location. This is a little loose in some situations, but close to the location.

I did discuss that some CNC suppliers have suggested the EGC be disconnected from the machine chassis and this grounding connection be replaced with a local ground rod at the machine. I also said this was a violation and unsafe.

I also indicated that some just added a supplemental ground rod at the machine.


If I take a 100 ft piece of #12 copper wire and make connections at the 20 and 80 ft points to earth, then the existing current in the earth will inject some current into the 60 foot section of the wire.


A normally non-load current wire that is not connected to the required bonding point for an EGC wire is not an EGC. A logical conclusion from the definition of an EGC wire.

With equipment grounding I've never been concerned about looping what-so-ever only effective fault path to/from source. I would think some amount of looping would commonly occur but if so why would this current travel to data equipment?

Looping of the EGC certainly allows current and voltage to be induced in the loop from changing magnetic fields. Further current can be injected into this loop by something like a 12 V battery and 12 ohms resistor, about 1 A. Or many other source types are possible.

Why would current travel to a computer via an EGC?

Assume a common main panel for both the CNC and the computer.

An EGC runs from the main panel to the CNC. There are some noise filter capacitors in the CNC from hot lines to the machine chassis. There is capacitance in the main power transformer and other similar devices from the hot lines to the machine chassis. Maybe there is a little leakage current as well. The low voltage common of all the machine electronics is connected to the machine chassis. Some of these connections to the machine chassis cause noise current to flow into the EGC back to the main panel. Thus, a noise voltage drop on the machine EGC.

Assume no noise is generated on the EGC to the computer resulting from the computer. In reality there likely will be noise generated from the computer thru its EGC also. Next connect the RS232 cable between the computer and CNC. Now a loop is closed, and noise current flows from the CNC thru the RS232 cable to the computer, and back thru the computer EGC wire, and as a result of all of this there is a noise voltage difference between the computer and the CNC.

Fault conditions and lightning cause much worse noise levels.

,

 

[tryinghard]

gar, I was having Internet Explorer crash problems this morning. Actually crash of XP as well.

There should be a direct mail slot to Bill Gates like a prompt for an email that goes directly to him “Internet explorer has encounter a problem, would you like to send some expletives to Bill Gates – please click here and enjoy?”

...I do not believe in any of my posts that I suggested the neutral be connected to the EGC anywhere except at the one single required location. I used "main panel" to refer to this location…The low voltage common of all the machine electronics is connected to the machine chassis…Next connect the RS232 cable between the computer and CNC. Now a loop is closed, and noise current flows from the CNC thru the RS232 cable to the computer, and back thru the computer EGC wire, and as a result of all of this there is a noise voltage difference between the computer and the CNC...

I don’t think you’re intentionally saying to bond the neutral incorrectly but when you say “inject current” this statement to me means incorrect neutral bonding and this problem really does exist on many sites usually from non-qualified installers. Too bad the LV common of the electronics uses the chassis and there are so many other factors to contribute noise as you have pointed out. I believe critical locations (or any) with noise problems should use fiber instead of conductive cabling.