Developments in Ground Bonding Networks
Editor’s Note: In this article, the words “ground,” “grounded” or “grounding” are used interchangeably with “earth,” “earthed,” or “earthing.”
The first part of this article introduced the first protective equipotential bonding/grounding systems, which only had requirements for human safety. It showed how – as electronics became more commonplace and more interconnected and variable-speed motor drives increased in power – these early structures developed into bonding networks (BNs) to protect electronics from damage due to insulation failures and lightning surges. Site-wide BNs are costly to create, so in those early days it was common to only provide BNs for the parts of a site where electronic equipment was installed. This led to the development of the isolated bonding network (IBN), which is where this Part 2 picks up.
Isolated Bonding Networks (IBNs)
An IBN is a BN that is isolated from the rest of the protective equipotential bonding system, except for at one single point of connection (SPC) (see Figure 1).
The idea of the IBN is that when fault or lightning currents occur in the rest of the building (or vehicle), their isolation prevents those currents from flowing through the nice low impedance created within the IBN, helping to protect the equipment it contains.
The usual guidance is that – with all of its mains power supplies isolated at the IBN’s distribution cabinet(s) and any uninterruptible power supplies (UPSs) switched off, and then its SPC temporarily disconnected – an IBN should be able to withstand a voltage of at least 10kVDC with respect to the rest of the building’s protective equipotential bonding system for at least one minute, without any current flowing in “sneak paths,” including via corona discharges, arcs or sparks, once the IBN’s stray capacitances have been charged up.
(It should go without saying that if an IBN is constructed where there could possibly be a potentially flammable or explosive atmosphere, its isolation should never be tested with high voltages as described above! Also, always remember to reconnect SPCs after successful voltage withstand tests, and do not reconnect the mains power supplies to any equipment within an IBN until after its SPC has been properly reconnected.)
Never rely on simply switching off the items of equipment within an IBN individually before testing its isolation as briefly described above. This is because all items of electronic equipment are fitted with EMI/RFI filters that “leak” milliamps of stray currents into the protective grounding conductor in their mains leads, and it does not take many such items for these leakage currents to build up to lethal levels. The EMI filters in high-power variable speed drives (VSDs) and other switching power converters can individually leak hundreds of mA, even Amps, into their protective ground.
These filters are usually fitted before the mains on/off switch, so they remain powered up and leaking current when the equipment has apparently been switched off using its own controls. This is why, before testing the voltage isolation of an IBN, all of its mains power supplies (there may be more than one) must be isolated at the IBN’s power distribution cabinet(s), and any uninterruptible power supplies (UPSs) within the IBN switched off.1
In the old days, each commercial or industrial building had a dedicated electrical manager, a skilled electrical engineer who ensured that no one compromised its protective equipotential bonding system or did anything else that might cause fires, shocks, unreliability, etc., and also supervised any/all upgrades and modifications. These knowledgeable professionals maintained the electrical drawings and knew them like the backs of their hands.
But these days it is much more common not to employ an electrical manager. Instead, suitably skilled subcontractors are hired when upgrades and modifications are done, or for annual inspections. Of course, they may not be familiar with a particular building’s electrical installation, or its history. And, if my experience is any guide, the building’s owners or operators may not have ensured that its electrical drawings have been kept up-to-date, and may not even know where they are, or which subcontractor had them last!
In such situations, it is possible for very-carefully-designed IBNs to be seriously compromised by changes and modifications made by people who are unaware of their importance (or even existence). I have seen it happen even in major national infrastructure plants. All it takes to compromise an IBN is for a person to string an Ethernet cable from their office outside an IBN to a computer inside an IBN. The consequences for equipment damage, and even for significant fire and shock hazards, especially during a thunderstorm, can be very severe indeed.
So, it is good general safety and reliability guidance to use CBNs, and not to use IBNs unless the building or site has 24/7/365 supervision by permanently-employed competent electrical engineers or technicians who understand where all the IBNs are and how (and why) to keep them isolated. These engineers or technicians should also approve any changes to any wiring (even Ethernet cables) and supervise all maintenance.
Common Bonding Networks (CBNs)
A CBN is a single BN that is “common” to an entire building (see Figure 2).
The big advantage of a CBN is that signal/data cables may be run around anywhere in the building – ideally strapped to bonding conductors/metalwork along their entire lengths to use them as PECs – without having to make any alterations to its protective equipotential bonding system. This makes adding new equipment in the future easy to do and relatively inexpensive.
The previous discussion has only concerned human safety as regards electric shock hazards, and the protection of electronics from damage by surge transients caused (indirectly) by lightning. However, all conductive items behave like “accidental antennas”.2 This fact means that for good EMC, all conductors and any pieces of metal – that are not functional conductive parts in any electrical/electronic circuits, of course – should be interconnected so as to be integral parts of any BNs, IBNs or CBNs – whether these conductors or pieces of metal have anything to do with electrical safety or not.
Meshed BNs, IBNs and CBNs
Computer electronics initially used circuits operating from 5VDC power rails, and with such low-voltage signals/data the “equipotential” voltages considered acceptable between “touchable” points during faults and thunderstorms in protective equipotential bonding systems were much too high. But the cost of fitting suitably rated insulation/isolation to every data cable regardless of how short it was would have been totally ridiculous.
So, when computer rooms and digital telephone exchanges (called Central Offices in the U.S.) started to be built in the 1970s, they invented much cheaper solutions: MESH‑BNs, ‑IBNs and ‑CBNs. The word MESH in the acronym refers to the fact that multiple cross-bonds are needed to reduce the inductances in the protective equipotential bonding systems by enough to reduce the exposure of digital electronics to lightning surge damage, and (in the 1990s, when the European Union’s EMC Directive loomed) to help achieve EMC for systems and installations.
Generally, these structures take the physical form of regular “grids” or “meshes” of bonding conductors – hence their name (see Figures 3 and 4).
Initially, these meshed conductive structures were called SRPPs (for system reference potential planes), BMs (for bonding mats) or a wide variety of jargon or proprietary terms that can be found in computer and telecom system installation guidance documents from the 1970s, 80s and 90s.3
Figure 5 shows the sort of SRPP design that was often used. The conductors used for the mesh were usually 6mm diameter copper, soldered at their joints, but some preferred to use wide copper “lightning tape” because of its lower inductance and ease of jointing using the clamps used for that purpose when constructing LPSs. Some computer/telco system installers used “natural metalwork” instead of installing a copper mesh, either by using the metal framework that supported the computer false-floor tiles as the mesh, or interconnecting the metal backs of the computer floor tiles. Figure 6 shows a modern proprietary development of the latter approach.
As time went on, these computer systems grew to occupy more than one room, so the rooms’ individual MESH‑BNs or MESH‑IBNs had to be mesh-bonded together to reduce the “surge impedance” of the new combined BNs or IBNs being created.
Remember that when the Z = √[R2 + (2 L)2] expression was introduced in Part 1 of this article, I mentioned that this was only relevant for conductors well-below their first quarter-wave resonance. We now need to correlate this with mesh dimensions.
Most lightning energy is contained in the spectrum below 1MHz, but it is still considered to have significant amount of energy up to 10MHz. The wavelength in air of 10MHz is 30 metres, making its first quarter-wavelength resonance 7.5m. So, a mesh size of 5m or less on a side (in air) is considered effective against all lightning frequencies, and the smaller the mesh size the lower its inductance between any two points and the lower the surge transient voltages that can arise due to induced lightning currents.
For good EMC, we may want our meshes to be smaller, either to control higher frequencies than 10MHz due to speedier computer data, or to provide lower impedances below 10MHz due to high-power VSDs. For example, 30MHz was a common goal in early computer systems and required mesh dimensions of around 600mm on a side, as shown in Figure 8. Modern computer systems may require meshes to control 100MHz or more.
The VSD technology that was new in the early 1990s could excite structural resonances in installations up to a few MHz, and this frequency has been steadily rising as power switching devices have developed. These frequencies are lower than those used by computer data, but on the other hand, their levels are much higher, so the sizing of a mesh size could depend more on the VSDs used on the site than on its computers. This issue will become much more important as the next generation of power switching devices replaces IGBTs and silicon powerFETS during the 2020s.4
Clearly, to be able to easily and quickly install new electronic systems or VSDs these days, it helps if you don’t have to first modify a building’s protective equipotential bonding structure (whether it is grounded to rods in the soil, or not) to create MESH‑BNs, IBNs or CBNs. Modifying existing installations to create meshed bonding networks for new equipment can easily cost more than the new equipment itself! After all, you often have to cut into floors or walls to get at the conductors that need to be meshed together.
Also, in industrial applications it has long been a simple matter to use existing metal cable support structures and/or cable armor as PECs. But this clever cost-saving measure is very vulnerable to changes and modifications being carried out by people who are not aware that these metal structures have any functionality other than mechanical. Creating a well-meshed CBN helps avoid problems of unreliability and/or EMC arising for such reasons.
So, since the mid-1990s, the general recommendation for all systems or installations is that “new-builds” should install MESH‑CBNs right from the start. It is also generally recommended that legacy buildings convert to MESH‑CBNs as soon as practical, usually a gradual process as new equipment is installed.
These recommendations are set to become much more important during the next few years, as the new generation of power switching converters and variable-speed motor drives based on HEMTs and SiC powerFETS discussed in Part 1 of this article become readily available in high power ratings.
Figure 7 shows a MESH‑CBN covering an entire floor of a building, but of course, we may need to extend them in three dimensions to other floors too, and Figures 8 – 10 are copies of relevant slides from my training course on EMC for Systems and Installations.5
Speculation on 3G, 4G, 5G, etc., and Fibre-optics
What if low-cost high-rate digital wireless comms had been available back in the 1970s? Even 3G cellular systems would have made data cables unnecessary back then, making BNs, IBNs and CBNs unnecessary. As the complexity of the electronic systems grew, wireless datacomms would have kept pace, first with 4G and then 5G.
Perhaps when 5G is mature and proven to be robust in industrial applications (despite the high levels of interference often associated with industrial processes), we will simply be plugging 5G modems into USB 3 sockets to carry industrial Ethernets, with no longer any need for data cables, hence no need for costly MESH‑BNs, ‑IBNs, or ‑CBNs. Protective equipotential bonding/grounding networks would still be required for human safety, but nothing more complex than the original types sketched in Figure 1 of Part 1 of this article – a big reduction in the use of costly copper.
A similar speculation concerns low-cost fibre-optics. If we had had modern low-cost fibre-optics running at 25Mb/s in the 1970s, they would have been preferable to copper cables (with all the EMC problems created by their unavoidable “accidental antenna” behaviours).
These days, when people ask me for help in fixing data interference problems with cables between items of equipment in scientific/industrial systems/installations, I am increasingly recommending that they replace their copper data cables with fibre-optic “modems” connected by (metal-free) fibre-optical cables. The cost of fibre-optic systems is steadily falling, and their data rates are steadily increasing, and using them instead of copper cables avoids the need to create MESH‑BNs, ‑IBNs, or ‑CBNs.
Even though the cost of a fibre-optic solution may be a few hundred or thousand U.S. dollars, very little time is required for installation. Although creating a MESH‑BN, MESH-IBN, or MESH‑CBN might appear at first to cost less, it will almost certainly cost a lot more overall when labour costs are taken into account, never mind the costs of the lost production while these intrusive modifications are being undertaken.
Also, while the fibre-optic solution is almost guaranteed to work first time (no one with any real experience ever guarantees anything where EMI is concerned!), converting a legacy installation into a MESH‑BN, ‑IBN, or ‑CBN can be a bit of a gamble. Installing meshed bonding in legacy buildings is very labor intensive and time consuming, but going for a least-cost option might well only result in having to do it all over again! For example, the mesh size depends upon how low the overall impedance needs to be, and the highest frequency it needs to control, and these are often not understood as well as they might be.
Also, will the resulting meshed structure be future-proof, or will it need to be modified again when the existing equipment is upgraded or replaced, or when new equipment is installed nearby in a few years’ time? Even replacing failed equipment with new versions of the exact same product from the same manufacturer inevitably causes ever-increasing noise problems at ever-higher frequencies.
This problem arises because the newer versions inevitably use newer power switching devices and newer microprocessors that switch more quickly – whether we want or need them to, or not! The original, slower semiconductors are simply no longer available to manufacturers, whose products therefore tend to become ever noisier at ever-increasing frequencies – even when they remain fully compliant with the relevant emissions standards.
Generally speaking, for the best EMC with the lowest overall costs, now and in the future, copper cables should only be used for (well-filtered!) AC or DC power. And all signals, data, and controls should use either (metal-free) fibre-optic cables or proven-industrially-robust and reliable wireless datalinks.
- It is always a problem in a brief article like this for an author to know how far to go into the detail, especially where safety issues are concerned. I have to assume that my readers understand that testing an IBN by isolating it and charging it up to 10kVDC has the potential to injure people due to electric shock – therefore such tests should only be carried out by people independently certified as being competent to perform them, and who regularly perform such tests. The high-voltage test generators must be current-limited to help prevent dangerous shocks, and the area of the IBN and near to it kept reliably off-limits to all personnel not directly involved.
- All conductors (including any metalwork) are “accidental antennas,” whether we want them to be or not.
- For example, I have seen such a guide from the 1970s that said the SRPP for a computer room had to maintain an ‘equipotential voltage’ from one corner to another that should not exceed 0.7V at frequencies up to 30MHz.
- This article is not the place to discuss mesh sizing in detail. But, for more information about EMC, see EMC for Systems and Installations, Tim Williams and Keith Armstrong. Also see section 5 of “Good EMC Engineering Practices for Fixed Installation” at https://www.emcstandards.co.uk/good-emc-engineering-practices-for-fixed-instal2 for information on using rebar meshes and the like to help protect installations from the powerful electromagnetic pulses (EMP) that can be created by lightning and nuclear explosions (e.g.: LEMP, HEMP, NEMP, etc.).
- EMC for Systems and Installations training course.