Grounding review and No.

100M9371 Rev NC Page 2 of 13

troubleshooting

1. Purpose
This document will describe the grounding as it applies to a Bently Nevada vibration monitoring
system defining common terms, basics of noise, basics of a plant grounding system, ground
loops and discuss a plan to located the noise/ground issue and remove it from the system.

2. Scope
Since these areas are broad and deep in scope, this document will discuss items as they apply to
a Bently Nevada monitoring system.

Disclaimer - This is a guide it is designed to give ideas and discuss known issues resolved with
our products. Always follow the National Electric Code when designing grounding and bonding
schemes.

3. Reference
IEEE Std 1143-1994, IEEE Guide on Shielding Practice for low voltage cables

Mardiguian, Michael. EMI troubleshooting Techniques, McGraw-Hill

4. Requirements
Previous knowledge and grounding experience.

5. Issue of why it may be grounding:

When the issue moves around and is not able to be consistently duplicated.

Improper wiring is a leading cause of noise/grounding issues. Each shield wire should be
insulated and only making contact to ground at a single point, the monitor.

Need to understand and diagram ALL grounds and bonding jumpers that are connected to
the rack and or cabinet. If a loop exists it can carry current and create a voltage potential
that may affect the system.

6. Grounding basics:
First and foremost grounding of electronic equipment is for personal safety by clearing any
faults and is no different than that of other equipment. Safe grounding requires fast opening
Grounding review and No. 100M9371 Rev NC Page 3 of 13
troubleshooting

of circuit breakers or fuses and minimization of voltage differences between exposed metal
surfaces on all of the involved electrical system and equipment, to levels that are safe for
people.

What makes electronic systems different is the sensitivity of their circuit components to
relatively small transient currents and voltages. It is also inherent in the nature of solid state
devices to be very fast, so they are affected by equally “fast” electrical disturbances.
Lightning is a slow transient compared to the response of almost any electronic device.
Typical threats to proper operation of electronic devices and systems include:

1. Lightning - Direct strikes, but the effects also include overhead cloud-to-
cloud, and nearby strikes causing induced voltages
2. Switching transients from power network operations and power factor
capacitor switching, lightning arrestor operation, and fault clearing activities-
especially on nearby power circuits.
3. Static electricity - Directly applied arcs to the equipment, but sometimes
arcs near the equipment will also affect the equipment.
4. Electrical fast transients - Typically as caused by arcing contacts or
collapsing magnetic fields in the coils of contactors in equipment - usually very
near the affected equipment or field wiring.

If a Bently Nevada system is not working properly due to grounding issues caused by
unwanted noise. The solution can be obtained by controlling/eliminating any one of the
three elements.

Solving transient problems is never easy. They may be random or repetitive. In general, they
have wave shapes which are not easily analyzed. Transients though are capable of being
controlled by:
1. Limiting overvoltage’s (surge voltages) on the ac power conductors with
surge protective devices (SPDs)
2. Reducing the chances of electrical noise getting on power circuits
connected to electronic equipment and the data signal circuit cables that
interconnect the units of equipment. This can often be accomplished by
observing the requirements for proper routing and grounding of field wiring,
branch circuits and their conduits along with ensuring adequate separation of
power and data signal wiring.
3. Proper grounding involving the correct installation of equipment grounding
conductors of all types, and neutral terminal grounding and bonding at the
service entrance and for separately derived ac systems.

While the above are all within the scope of the contractors’ job, it needs to be emphasized
that we, Bently Nevada, must be the most knowledgeable regarding the proper installation of
the vibration monitoring system. This vibration monitoring system is defined as from the
transducer or probe to the System 1 server. Otherwise, errors in the installation will cause
extensive effort and great expenditures in order to get this vibration monitoring system to
work in an acceptable way.
Grounding review and No. 100M9371 Rev NC Page 4 of 13
troubleshooting

6.1 Types of grounding:

Figure 1. Typical ideal installation

6.1.1 Ideal but not typical

Figure 1 shows an ideal grounding system. When the monitoring system is installed at site
and connected to the plants grounding system there is a high probability it will not be similar
to figure 1. If the monitoring system is installed at the machine there will be unintended
ground connections from clean to dirty grounds in the plant. If the monitoring system is
located in the control room the grounding system can be similar to figure 1 with some
variations providing a clean ground.

6.1.2 Plant grounds:

Any discussion on grounding invariably leads to a discussion on the different types of

grounds and the corresponding definition of each. However, it is commonly accepted
that grounds in the process industry can be broadly classified as either dirty or clean.
Please refer to Figure 1 for a comparison of the different grounds.

6.1.3 Mains Supply and Large Equipment / Dirty Grounds

Dirty grounds inside the facility are typically those 120VAC, 220VAC, 480VAC power
grounds that are associated with high current level switching such as motor control
centers (MCC), lighting, power distribution, and/or grounds corrupted by radio
Grounding review and No. 100M9371 Rev NC Page 5 of 13
troubleshooting

frequencies or electromagnetic interference. Quite often the primary AC power coming

into the plant can introduce spikes, surges or “brownouts” that further erode the
cleanliness of the AC ground.

6.1.4 Instrument / Clean Grounds

Examples of clean grounds are the DC grounds, usually 24VDC, that reference the PLC,
DCS or metering/control system in the plant. Frequently, control systems engineers from
the major SCADA (Supervisory Control and Data Acquisition) vendors recommend
isolating these grounds from power grounds. Other clean grounds are those associated
with data and communication busses that, due to the vulnerability of low level CMOS
and microprocessor circuits, must be maintained relatively free of noise interference or
risk data/communications loss.

6.1.5 Structural Grounds

These are the grounds which physically and electrically tie the facility together and, quite
importantly, complete the circuit to the 0V, ground leg, of the power distribution
transformer. Structural grounds can take many forms. In a ship, it is the hull of the ship;
on an offshore oil/gas platform, it is the structural steel of the platform. In large
petrochemical or pharmaceutical plants, a ground grid or mat is installed under the plant
or the welded structural steel of the plant itself becomes the 0V electrical power ground.
In the typical plant, the 0V ground reference is most often a heavy gauge copper wire
embedded around the base of the building and tied into ground rods at the corners as
well as into the AC ground feeds at critical junctures. Not only does this copper ground
create the 0V reference for the plant’s electrical system, it becomes part of a possible
Faraday cage lightning protection system.

Star Point / Single Point Grounding

Look at the accompanying AC distribution diagram on Figure 2 and notice that all the
subsystems in the plant, instrumentation, communication, computers and control, and AC
power, are connected to a single point ground system. This is known as “star point”
grounding. Properly done, each subsystem ground is kept as short as reasonably possible
and is connected to the star point at only one point. Multiple paths to the ground plane from
a subsystem inherently have different resistances. Different resistances to ground produce,
by Ohm’s Law, different voltage potentials impressed on the control system.

The net result of not employing star point grounding is increased vulnerability to transient
surge damage as well as less reliable control system functioning. This can affect BN
monitoring system channels by enter a NOT OK state and or not collecting waveform data.
It may also be seen a ground bounce, which will cause most or all channels in the monitor to
vary amplitude not related to machine activity.
Grounding review and No. 100M9371 Rev NC Page 6 of 13
troubleshooting

Figure 2 Star/ Single point grounding example

6.2 Items worth noting:

All grounds in a plant will be tied together at one point in the plant.

Where grounding can become an issue is when the monitoring system is mounted at the
machine and away from the control room. Typically running the grounding conductor back
to a clean reference is not cost effective and plants will connect the monitoring system
ground to the closest ground available. This is typically machinery plant ground and has the
potential for more transients and noise. The control room typically has its own grounding
grid or mat which provides a low noise 0 voltage reference point.

Most lightning damage to electronic equipment occurs when a facility employs the use of
multiple earthing references that are not intentionally, and effectively, bonded together.
Proper bonding between electrode systems can reduce the voltage drops between them
and establish an equipotential plane within the facility so as to enhance personnel safety.
For example, lack of bonding between the electrical system electrode and the
communications system electrode may result in damage to modems, telephone answering
machines, etc., during lightning and system fault conditions. A typical scenario is if the
monitoring system is in one building but is communicating with a DCS in another building.

A floating shield, not grounded at either end, can create as much noise in a system as if
shield was connected at both ends creating a ground loop. If a shield on a cable is not
attached to ground at one end, this can cause the shield to act as an antenna for switching
Grounding review and No. 100M9371 Rev NC Page 7 of 13
troubleshooting

signals. These switching signals can parasitically couple to the signal wires. This noise will
enter the monitor(s) as noise and affecting the vibration signal.

A shield wire is to be treated in the same manner as the signal wire of a transducer. It
needs to be insulated, isolated from other conductors and only connected to ground at a
single point at the monitor side. Experience has shown that improper shielding is largest
contributor to grounds loops in a monitoring system.

The only time a grounding conductor (green wire) should carry significant current is when a
short circuit or lightning strike has happened.

Newer BN cabinets may no longer connect the shield of the field wiring to the monitor inputs.
The shields can be tied to a ground bar at the entry to the cabinet. This allows any noise to
be drained off to the ground system without potentially affecting the monitoring system.

Field installed cables should be separated from power cables and conduits to the greatest
practical distance. This reduces unwanted coupling between the two circuits. To avoid noise
coupling problems where one circuit crosses over or under the other, try to make the
crossover at right angles.
Grounding review and No. 100M9371 Rev NC Page 8 of 13
troubleshooting

7. Typical installation of transducer and field wiring.

Figure 3 Bently Nevada Single point

Figure 3 shows a typical transducer installation with the system grounded at a single point at
the monitoring system.
Grounding review and No. 100M9371 Rev NC Page 9 of 13
troubleshooting

7.1 Ground loops:

Figure 4 Bently Nevada Single point with ground loop.

Figure 4 shows multiple ground loops are formed when two or more points in the monitoring
system. The voltage potential causes the current to flow in the ground loop creating noise in
the system. This change in reference voltage affects the vibration levels being measured.
Flowing current in the loop will have both an AC and DC component in the loop. Since
these are ground loops and any loop acts as an inductor or antenna for magnetic fields.
The length of the loop will determine the frequency ranges that can affect the system.

Check for ground loops:

1. Walk down the installation from the monitor to the transducer inspecting each
junction box or wire termination.
2. Lift field wiring from the monitoring system for a 3500 rack simply remove the
phoenix / euro connector from the I/O module will work. Using a Digital multi-meter
measure the resistance between each field wire and monitor chassis ground. * note
relays will have energized voltages applied to them. DO NOT perform a resistance
check on wires connected to the relays as there is a potential for shock hazard.
Grounding review and No. 100M9371 Rev NC Page 10 of 13
troubleshooting

7.2 Grounding checks:

This test uses a 3500 rack as an example.

1. Download the 3500 rack information before testing.

2. Walk down the field wiring identifying the routing of field wiring. Make note of
junction boxes, terminations of wires specifically shields. Take pictures of terminals
and connections.
3. Identify the grounding connections of the 3500 rack. Locate where the cabinet
ground is connected to in the plant. If possible without disturbing the 3500 system
confirm the position of the high pot switch on the 3500 PIM most application the
switch should be closed. Exceptions are for intrinsically safe applications. Draw a
diagram of the grounding scheme and take pictures of the connections. Does the
power feeding the 3500 rack PIM have a ground wire?
4. Unit should be running. Using a battery powered digital O-scope with the negative
probe connected to the battery powered common. Connect the positive probe lead
to several different ground points in the plant around the area of the 3500 system
and sensors. Looking for any spikes or ground bounce, leave the probe connected
for 10 – 30 minutes. If noise spikes are found record and move to another location.
In order to sweep frequencies from 50 – 200 K Hz adjust the sampling rate of the O-
scope during the test.
5. Grounding resistance checks. Using 10 AWG or equivalent wiring measure the
difference in resistance between the 3500 ground point and to other ground points,
found in step 4. Measure the resistance of the wiring and subtract that from the final
measurements. This will measure the resistance between two different plant ground
points. If the resistance value is higher than 2-4 ohms the plant ground is most likely
not adequate to provide a clean ground reference. Also suggesting the plant ground
has an issue that needs to be looked into. Measure 3500 ground point to at least 2
– 4 different ground points using the 10 AWG wire.
6. Measure voltage differential between ground points in step 5 using O-Scope. Set
both channel A and B to differential mode. Differential channel A positive probe is
connected to ground at the 3500 rack. Differential channel B positive is connected
to the end of the test ground 10 AWG wire in step 5. Invert channel B on the scope
and then sum/add channel A to Channel B. The sum output will measure and
display any differences in signal on the display. If a difference is substantial in
amplitude this may indicate ground noise or bounce. This will also test if common
mode exists. Do not hook up either negative probe from each channel.
7. Measure for ground bounce or spikes from the rack ground and shields, which
should be the same if the high pot switch is closed on the PIM. Using battery
powered O-scope attach the positive probe to shield screw on the 3500 PIM with the
negative probe connected to the battery power O-scope common. Then set the
Grounding review and No. 100M9371 Rev NC Page 11 of 13
troubleshooting

trigger level above noise floor to capture spikes on ground to rack and shields. – can
be an overnight test.
8. When testing is complete again download the 3500 information that was tested.
Grounding review and No. 100M9371 Rev NC Page 12 of 13
troubleshooting

8. Case studies

8.1 Spurious noise on GE Gas Turbine power generation system

Issue – customer was experiencing random and spurious vibration signals in the 3500
monitoring system being used for machine protecting and management.

System – 8 - GE frame 7FA Gas Turbines were at this site. GE Mark VI control systems
were are controlling the turbines. All vibration transducers were powered from and
connected to the Mark VI TVIB card. TVIB cards provided buffered output signals for each
transducer that was then connected to the 3500 monitoring system using a standard DB 25
cable for external terminations. This site used one 3500 monitoring system for two GT’s. A
3500 monitoring system was located in the PECC of each odd numbered GT PECC, or units
1, 3, 5 and 7. Each PECC was 60 meter, or 180 feet apart on center and grounded to the
plant ground system.

Discovery – with each GT and PECC tied to the overall plant ground system this created a
ground loop. Unit 1’s 3500 monitoring system and Mark VI’s grounds were connected at
the same point on the plant ground. Unit 2’s Mark VI was grounded 180 feet from unit
1. When the transducers common from the buffered output of the TVIB card was connected
to the 3500 monitoring system in Unit 1, a ground loop was created. The 180 feet between
Unit 1 and 2’s ground points created a voltage potential large enough to affect the vibration
channels output.

Solution – galvanic isolators were installed to isolate the commons for each transducer
signal for the even numbered units 2, 4, 6 and 8.

8.2 650 MW Steam turbine generator experiencing high vibration

alarms.

Issue – customer is seeing Velomitor channels on multiple bearings entering alarm states.
Proximity probes on the bearings do not indicate a change in vibration levels. It has been
determined these alarms occur when relays close for activating the lube oil system. Relays
are located above and adjacent to the 3500 racks.

Tests performed:

 Start of testing; closed relay and verified in the 3500 system that 5 – 7 mil spikes are
present. Spikes are visible, continue testing,
 Disconnected the installed Velomitor wiring from the terminals in the JBox.
Connected spare Velomitor wiring to JBox terminals. Spare Velomitor not mounted
to the turbine and is located in the JBox. Test of closing relay show 5-7 mils of noise.
Continue testing
Grounding review and No. 100M9371 Rev NC Page 13 of 13
troubleshooting

 Installed spare Velomitor on Channel 5Y (one of the worst channels for spiking) on
the external terminal block in the cabinet at the 3500 rack, this eliminated the field
wiring.
o Changed relay state and no spike on that channel. This indicates the relays close
proximity to the 3500 is not an issue, but the field wiring is a factor with the noise
problem.

Here is an additional test that could be performed to help isolate the issue. At the JBox
remove the transducer field wiring and replace with a battery powered function generator
with a DC bias. Simulate the transducer using an AC signal with DC bias (refer to
verification section in the manual for frequency and amplitude).

Solution:

After consulting with the relay design engineer a properly sized diode was connected
'backwards' across the relay coil. This diode is not to be sized or installed by Bently Nevada
it’s the customer’s responsibility.
Current flowing through a relay coil creates a magnetic field which collapses
suddenly when the current is switched off. The sudden collapse of the magnetic field
induces a brief high voltage across the relay coil which is very likely to be
transmitted to other signal wires. The protection diode allows the induced voltage to
drive a brief current through the coil (and diode) so the magnetic field dies away
quickly rather than instantly. This prevents the induced voltage becoming high
enough to transmit spikes to other systems.

9. Suggested Tool Kit:

Battery Powered O-Scope 100 MHz (storage if possible to grab transients)
Battery Powered Digital Multimeter
Battery Powered Fluke process calibrator – to simulate AC signal with a DC bias
Fluke current clamp similar to i1000s AC Current Probe, suggest being able measure
frequencies between 5 Hz – 10-50 kHz signals.
50 ft. 1 inch braded copper ground wire with clamps
Clamp on current transformer CT able to measure 20 Hz to 5000 Hz and voltages from 5
VAC to 500 VAC

Leaded resistors (with axial leads) such as 10 K Ohm, 1 K Ohm, and 100 Ohm