Fixed link.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
"Isolated Ground" for DCS server room - NEC 250.54
- Thread starter trallio
- Start date
- Status
- Not open for further replies.
zbang
Senior Member
- Location
- Roughly 5346 miles from Earls Court
Publication 1770-4.1 – February 1998
Industrial Automation Wiring and Grounding Guidelines
As it says-
Connect the ground bus to the grounding-electrode system through a grounding-electrode conductor. The grounding-electrode system is at earth-ground potential and is the central ground for all electrical equipment and ac power within any facility.
(color added)When ac power is supplied as a separately derived system through an isolation/step-down transformer, you can connect it as a grounded ac system or an ungrounded ac system. For a grounded ac system, connect one side of the transformer secondary to the ground bus as in Figure 7. For an ungrounded ac system, connect one side of each test switch for the ground-fault-detector lights to the ground bus as in Figure 8. We do not recommend an ungrounded system.
smalltime
Member
- Location
- Roanoke, VA
First, my apologies for being late to this party. I come from the perspective of an audiovisual system design and instructor.
An Isolated Ground (IG) system as permitted in 250.146(D) will, at some point, be terminated at a regular (non-isolated) equipment grounding terminal. As we know, equipment ground is bonded to the neutral so that any fault current has a pathway back to the source. The pathway back, the Effective Ground Fault Current Path, is what trips the breaker or blows the fuse.
As noted, the IG system could have its own ground electrode but… The IG and the regular equipment ground connection is still required and that “separate” ground connection will need to be bonded to the building’s GES.
The common misconception is that ground is some sort of magical noise sump and that if I can get a “separate” or “clean” ground, I can drain all my noise into the ground. Circuit theory is still circuit theory. Comes from the source, wants to return to the source. Closed loop system.
Also as noted, the shield doesn’t always carry signal. This would be true in the case of a balanced interface. “Balanced” means a balanced impedance on the two conductors and a balanced interface carries a differential signal. (I have traced the origin of the balanced interface – differential signal – all the way back to 1886 during the transition from telegraph to telephone. They needed a noise defense mechanism against noise on the telephone line. The level of the noise on the line that was acceptable for dots and dashes was not acceptable for voice transmission.) This is just like the differential voltage on the two line conductors of a 240 V single phase. Balanced interfaces are used in audio, Ethernet and HDMI, just to name a few.
“Unbalanced” interfaces use the shield as the return conductor of a circuit. Think RCA connectors on your old home stereo receiver.
Lastly, you should connect the shield at both ends. That Industrial Automation Wiring and Grounding Guidelines is an old document… 1998. The root issue is how the manufacturer terminates the shield connection on the equipment. The shield connection should be made to the chassis of the equipment (at point of entry) and not to the signal reference trace inside the equipment.
In audio, we refer to this as the “Pin 1 Problem” as Pin 1 of our common XLR connector is the shield connection. The June 1995 AES (Audio Engineering Society) Journal revealed this issue and the AES 48-2005 standard addressed specifically what the equipment manufacturers should be doing with the shield terminal on their equipment.
Think about it this way – if I have noise currents flowing on the shield, where do I want them to go? If I terminate the shield directly to the equipment chassis, I know the equipment chassis is bonded to the EGC. Noise that originated with the power source can now go back to the power course (again, circuit theory.) If however, I don’t connect the shield to the chassis but to the signal reference trace on the printed circuit board inside the equipment, noise currents could be found running around on the PC board and getting induced into my good signals. Yes, the signal trace will eventually be connected to the EGC so the noise can return to the source but that’s after I’ve let the fox into the henhouse.
An Isolated Ground (IG) system as permitted in 250.146(D) will, at some point, be terminated at a regular (non-isolated) equipment grounding terminal. As we know, equipment ground is bonded to the neutral so that any fault current has a pathway back to the source. The pathway back, the Effective Ground Fault Current Path, is what trips the breaker or blows the fuse.
As noted, the IG system could have its own ground electrode but… The IG and the regular equipment ground connection is still required and that “separate” ground connection will need to be bonded to the building’s GES.
The common misconception is that ground is some sort of magical noise sump and that if I can get a “separate” or “clean” ground, I can drain all my noise into the ground. Circuit theory is still circuit theory. Comes from the source, wants to return to the source. Closed loop system.
Also as noted, the shield doesn’t always carry signal. This would be true in the case of a balanced interface. “Balanced” means a balanced impedance on the two conductors and a balanced interface carries a differential signal. (I have traced the origin of the balanced interface – differential signal – all the way back to 1886 during the transition from telegraph to telephone. They needed a noise defense mechanism against noise on the telephone line. The level of the noise on the line that was acceptable for dots and dashes was not acceptable for voice transmission.) This is just like the differential voltage on the two line conductors of a 240 V single phase. Balanced interfaces are used in audio, Ethernet and HDMI, just to name a few.
“Unbalanced” interfaces use the shield as the return conductor of a circuit. Think RCA connectors on your old home stereo receiver.
Lastly, you should connect the shield at both ends. That Industrial Automation Wiring and Grounding Guidelines is an old document… 1998. The root issue is how the manufacturer terminates the shield connection on the equipment. The shield connection should be made to the chassis of the equipment (at point of entry) and not to the signal reference trace inside the equipment.
In audio, we refer to this as the “Pin 1 Problem” as Pin 1 of our common XLR connector is the shield connection. The June 1995 AES (Audio Engineering Society) Journal revealed this issue and the AES 48-2005 standard addressed specifically what the equipment manufacturers should be doing with the shield terminal on their equipment.
Think about it this way – if I have noise currents flowing on the shield, where do I want them to go? If I terminate the shield directly to the equipment chassis, I know the equipment chassis is bonded to the EGC. Noise that originated with the power source can now go back to the power course (again, circuit theory.) If however, I don’t connect the shield to the chassis but to the signal reference trace on the printed circuit board inside the equipment, noise currents could be found running around on the PC board and getting induced into my good signals. Yes, the signal trace will eventually be connected to the EGC so the noise can return to the source but that’s after I’ve let the fox into the henhouse.
An experiment that didn't work
An experiment that didn't work
I engineered a job where the sound system people insisted on an isolated ground system. It was joined to the main ground at one point. I insisted that this would cause problems and not solve them. I explained that if you think of the grounding system as a pool of water, a small rock falling into a large pond doesn't do much splashing. However, a small rock in a small pond creates a much larger disturbance. The equipment which they wish to isolate almost always is a big splasher. Much better to put it in the bigger pond. Each ground connection represents a quantity of pond water. The more ground connections, the bigger the pond.
I do imagine that if there was an enormous amount of VFD loads on a system, particularly if many were single phase and not balanced, that an isolated ground system might do some good. Maybe.
The end of the story above, there was noise on the system. So they tried unisolating the sound system and the noise disappeared permanently.
It is ironic that the vendors pushing isolated grounds are the vendors of equipment with the wonkiest sine waves that do the most polluting to the voltage. They need all the help they can get diluting their bad waveforms.
An experiment that didn't work
I engineered a job where the sound system people insisted on an isolated ground system. It was joined to the main ground at one point. I insisted that this would cause problems and not solve them. I explained that if you think of the grounding system as a pool of water, a small rock falling into a large pond doesn't do much splashing. However, a small rock in a small pond creates a much larger disturbance. The equipment which they wish to isolate almost always is a big splasher. Much better to put it in the bigger pond. Each ground connection represents a quantity of pond water. The more ground connections, the bigger the pond.
I do imagine that if there was an enormous amount of VFD loads on a system, particularly if many were single phase and not balanced, that an isolated ground system might do some good. Maybe.
The end of the story above, there was noise on the system. So they tried unisolating the sound system and the noise disappeared permanently.
It is ironic that the vendors pushing isolated grounds are the vendors of equipment with the wonkiest sine waves that do the most polluting to the voltage. They need all the help they can get diluting their bad waveforms.
- Status
- Not open for further replies.