I'm a EE working on a scientific instrument that is large in scale, like a building. The structure is all metal and will have many server cabinets and electronics equipment, motors, sensors spread all over it. It is not yet built.

Power is supplied from a local utility feed to the instrument as 3-phase AC power where both 3-phase and single phase can be utilized. The 3-phase includes L1, L2, L3, N, and PE. PE is tied to earth ground and Neutral at the utility feed.

There is a strange requirement to keep the PE from the power source isolated from the structure itself. Equipment with a metal chassis must either have its PE-connected chassis isolated from the structure or the equipment's chassis can be bolted to the structure if PE is disconnected inside. They say this is to reduce noise and ground loops.

For example, if I installed a rack computer, I would need to either isolate the chassis from the structure or remove any PE connection from its power plug to its chassis to isolate it. Then ensure the chassis is bolted to the rack which is bolted to the instrument chassis for safety ground.

I believe that they're confusing PE with an actual current carrying conductor. They keep calling it Ground. I can't make any sense of it. Perhaps I'm misunderstanding the requirement. Have any of you ever heard of something like this?

  • \$\begingroup\$ I know of a lot of equipment connected to PE by an uninsulated PE conductor that is wired from the equipment to PE at the electric service entrance. However the equipment chassis is usually bolted to a cabinet that may be similarly grounded but also connected to metal building structural elements. I suspect that some equipment has a metal chassis that has an insulated PE conductor to service entrance PE with the chassis otherwise insulated from ground. \$\endgroup\$ – Charles Cowie May 22 '19 at 18:46
  • \$\begingroup\$ So they basically want you to defeat safety ground on everything? If you didn't remove the PE on a device, it failed, and it shorted line current through the chassis to the mounting hardware and enclosure, would this somehow ruin the measurements/function of the device at large, even if the current flow was only momentarily present? Was this designed by scientists or EEs? \$\endgroup\$ – schadjo May 22 '19 at 19:01
  • \$\begingroup\$ I haven't heard of this. In my opinion, it would create a hazard for personnel maintaining the instrument if the metal cabinets are not connected to PE. So I would be looking at isolating the cabinets from the building rather than cutting PE inside cabinets. Is the instrument somehow isolated from the actual earth? Like sitting on giant dielectric blocks? Or is it resting on soil or?? Just asking out of curiosity. \$\endgroup\$ – mkeith May 22 '19 at 22:44
  • \$\begingroup\$ whenever there are switching power supplies, the intense dV/dT will throw Efield interference all around, unless the outside world is TOTALLY hidden from view and all holes are carefully filtered. Without this mindset, the "ground loops" will exist everywhere. What is the noise floor of this scientific instrument? Does a similar "machine" exist elsewhere on Earth? \$\endgroup\$ – analogsystemsrf May 23 '19 at 3:49
  • \$\begingroup\$ Sounds like confusion between mesh grounding and the old school 'technical earth' that used to be so common on broadcast sites. Assuming the supply PE is connected to the structure somewhere, I would personally mesh the whole lot (Every earth, Chassis, structural member connected to everything else by as many routes as possible), on the theory that this creates lots of tiny low impedance loops and small currents flowing in low Z loops develop little voltage. Certainly I would be very nervous around racks of gear where the rack and the equipment cases were not firmly tied together. \$\endgroup\$ – Dan Mills May 23 '19 at 12:23

I worked in a Radio and TV studio building during equipment installation.

All broadcast equipment, and outlets for such equipment, was isolated from Building Ground (Protective Earth), but was connected to "Technical Ground". Technical Ground and Building Ground were connected only at one point, in the TV master control room. During installation, we had an alarm device connected between Tech Gnd and Building Ground to alert us if there was any connection between the two grounds. This arrangement was intended to prevent ground loops from causing interference in the broadcast equipment.

It sounds like you are expected to do something similar with grounding in your scientific system.

| improve this answer | |
  • \$\begingroup\$ This can be easily achieved by wiring the PE wires to a single point and then connecting then to so called technical ground. No need to cut PE wires inside the device and then to isolate them. Even if you have a separate (technical) ground and you daisy chain the PE wire like in household wiring, you'll get ground loops. \$\endgroup\$ – Marko Buršič May 22 '19 at 20:17
  • \$\begingroup\$ This is interesting and sounds similar. Can you provide an example of how a ground loop happens through the PE? It shouldn't be a current carrying conductor except in a fault. What currents are on it? I'm talking to some people at work and trying to get more info. I'll report back. Thanks \$\endgroup\$ – DD5000 May 23 '19 at 20:51
  • \$\begingroup\$ From an industrial instrumentation/controls point of view, a similar thing is done with two separate grounding systems used - an "Instrument Earth" (analogous to your "Technical Ground") and a "Power Earth" (analogous to your "Building Ground"). This is done for the same reason - to reduce interference on the sensitive instrumentation circuits. \$\endgroup\$ – Li-aung Yip May 24 '19 at 0:45
  • \$\begingroup\$ Interesting. So there certainly seems to be precedent for this. Do you have any links to documentation for such a system? Thanks \$\endgroup\$ – DD5000 May 29 '19 at 21:18

I'm not sure it will have any effect, but I can see what they're trying to do here. Basically they want there to be no loops in the system. The structure probably has it's own earth ground connection of some sort, so they don't want the power supply PE connected to that.

I don't see any reason it wouldn't be safe to do this, as long as all devices with metal chassis have a PE connection one way or another. Whether this will actually help anything, or just cause problems I'm not sure. I think we would have to know more about the system.

The PE connection may not be intended to carry current, but it certainly can. So the architecture does matter. Furthermore, I don't think you can assume that all the equipment is internally isolated from PE, in such a way that imbalances don't matter.

| improve this answer | |
  • \$\begingroup\$ Under normal (non-fault) operation, what currents are carried on PE? If they are charged, wouldn't they be immediately discharged to ground via the low impedance connection? \$\endgroup\$ – DD5000 May 23 '19 at 20:53
  • \$\begingroup\$ One example: One of your devices has it's power supply ground connected to PE; it tries to push a large current into the ground. This causes the local ground voltage to rise. Now local ground is higher than PE and current flows. \$\endgroup\$ – Drew May 24 '19 at 18:18

Each metallic enclosure has to be connected to PE. All sockets have to have a PE connected, this is usually inspected from an electrician that has a certificate to measure the ground resistance with attested measuring instrument (regularly calibrated and attested). All original cables come with PE wire and it connects to metal shielding, chassis,...

If you intentionally hack the sockets or original equipment, then you will face consequences in case of injury or electrocution. If you are an electrician then you will be guilty, even if you have been told to do so, because they may don't have an EE knowledge and it is your responsibility to prevent such wrong decision.

EDIT: In addition to the answer of user Peter, the correct wiring from my point of view is as follows: All critical devices have to have a PE wire connected in one single point, that can be protruded to the building ground. If you wan't to eliminate the ground loop, the PE wire shall not be daisy chained into electric sockets as it is used in home installations.

| improve this answer | |
  • \$\begingroup\$ You're writing this as if it's a universal truth with no exceptions. This is a scientific instrument, not a workplace, not a building, so building codes do not necessarily apply. \$\endgroup\$ – pipe May 22 '19 at 19:35
  • \$\begingroup\$ @pipe There are no exceptions. If they were, the equipment would be already made with those exceptions. \$\endgroup\$ – Marko Buršič May 22 '19 at 19:43
  • \$\begingroup\$ It is possible that the market for large rack servers to be mounted in an all-metal scientific instrument is not large enough that you can buy them off-the-shelf. \$\endgroup\$ – pipe May 22 '19 at 19:46
  • \$\begingroup\$ Look into mesh grounding, it may or may not be appropriate (Bigger picture issues), but the approach of LOTS of small loops each of minimal area and carrying small currents has much to recommend it (It is also generally less fragile then old school 'technical earth' based approaches and better behaved with switch mode supplies (Lower effective inductance). \$\endgroup\$ – Dan Mills May 23 '19 at 12:28

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.