SRGs look good on paper, but do they perform well in today's distributed network environment?
In today's distributed network environment, proper bonding and grounding are important factors in your search for power quality. When voltage levels vary on a ground grid, high-frequency noise currents can disrupt and degrade data transmissions between equipment. And if your using unshielded signal cabling or isolated ground topology, you're asking for trouble.
Signal reference grids (SRGs) work to provide an equipotential plane for the equipment contained within data processing rooms. Theoretically, an SRG provides a low-impedance path for high-frequency noise currents to dissipate before they cause havoc between equipment. The paths consist of short conductor connections between equipment enclosures and predetermined connection points along the grid. Of course, you must ground the grid to the building's grounding electrode system. Most installers and designers refer to an SRG as a multi-point grounding reference. But this is not an accurate description, since all grounding connections in any data processing room should be single-point.
Proper installation depends on proper design, and SRGs can take on several forms.
• Using 2-ft squares of floor-mounted copper straps. Designers often prefer this method. However, this is a costly technique requiring a high degree of maintenance after installation.
• Using a copper stranded conductor strung between the raised floor pedestals. Installers usually handle this in areas of the data processing room near the equipment. This method is not as costly as the 2-ft-square design; however, the stranded conductor doesn't have the low impedance of a copper strap under certain high-frequency events.
• Using the frame stringers and pedestals of the raised floor. This method is not always popular among designers and installers. Usually, they are concerned that floor traffic may compromise the conductivity of the raised floor structure. On the contrary, the pedestal/stringer method meets or exceeds the performance characteristics of the 2-ft-square design. In fact, tests prove measurable voltage differences in loose or improperly connected 2-ft copper straps, while no voltage difference appears with the proper installation of the pedestal/stringer method. This method also requires the least maintenance. The difference in conductivity of bolted or hung stringers is negligible. Therefore, you can use either.
For an SRG to be effective, you should run a grounding conductor between various points of the SRG and the building's grounding electrode system as often as possible. The high-frequency currents must dissipate to a grounding reference before having a chance to couple electrical noise onto signal cables or equipment frames. You should also run a conductor between the equipment enclosures and the SRG, choosing either a braided or copper-strap conductor.
What equipment should you connect to the SRG? For an SRG to be effective, you must bond the following items to it:
• Equipment enclosures;
• All metallic piping entering the room (including water, pneumatic, gas, hydraulic piping, etc.);
• Plenums for air handling units;
• Power conditioners (including power distribution units); and
• Metallic conduits.
Installers must bond all piping to the SRG at its entry point to the room. Unfortunately, this is a costly process because of the many hardware and electrical changes that take place over time. As a result, most equipment installers fail to maintain a connection to the SRG when retrofitting repairs.
How do you know if the SRG really works? Theoretically, SRGs work well; however, once the signal cabling leaves the data processing room their effectiveness shows limitations. The degree of success depends on how you answer the following questions:
Is your cabling shielded or unshielded? You can ground shielded cabling to the SRG as soon as it enters the room. This allows the SRG to dissipate the high-frequency disturbances before it affects equipment. Unshielded cabling allows electrical noise to travel unmitigated to the equipment.
Are you using an isolated ground (IG) topology? If so, don't expect the SRG to effectively equalize potentials between the IG system and the SRG. The IG conductor has higher impedance under high-frequency noise conditions. This causes voltage differences at equipment locations, affecting signal-to-noise ratios on data cabling.
How do you test SRGs? The following tests should help you determine whether the signal reference grid has underlying problems:
• You should make AC current measurements on all grounding and bonding conductors for the SRG. The AC current on any portion of the SRG should not exceed 250mA. AC current exceeding this level will cause voltages that can affect data signals.
• Measure the equipment-grounding conductor on all AC branch circuits and/or equipment enclosures to verify impedance less than 0.1 ohm. (You should make this measurement with a ground impedance tester.)
• Take two-point bonding measurements between the SRG and all building grounding electrodes. The resistance measurements between any two points should be less than 0.1 ohm. (You should make this measurement using an earth-ground tester in a two-point configuration.)
• Common-mode voltage measurements between different grounding planes within the data processing room can help pinpoint EMI/RFI on any grounding plane. (You should make this measurement using an oscilloscope and a line viewer.)
• Use an RFI locator to find any significant sources of radio frequency interference (RFI).
• A gaussmeter can help locate significant sources of electromagnetic interference (EMI).
• Tightening all connections once a year can uncover loose connections.
SRGs look good on paper. However, field testing reveals they do not perform effectively in today's distributed network environment; mostly due to poor installation practices and improper application. If installed properly, they can be effective supplements to the equipment grounding system, while helping to dissipate high-frequency noise currents.