RS-485
Ground Potential Differences
or When Good Grounds turn Bad
Thomas Kugelstadt
October 2010
1
Industrial communication via field bus systems often take place over long transmission lines.
Designers, unaware of the large Ground Potential Differences (GPD) between remote bus
locations, either rely on the local earth ground as a reliable signal return path, or directly
connect remote grounds with one another, thus creating noisy ground current loops.
In both cases the integrity of the transmission signal is compromised, which can lead to
system lock-down and worst case, destruction of the bus transceivers.
To raise awareness of these design pitfalls, this session explains where in the electrical
installation GPDs originate, how ground loops are created unintentionally, and how isolation
circumvents both conditions yielding a robust data transmission system.
Mains Supply System
2
The above slide shows a simplified diagram of a mains supply system to a commercial
building or factory.
The utility transformer, here represented through its secondary winding only, provides three
live phases (L1, L2, L3) and a neutral conductor (N) to the supply system. All three phases
are routed through an electrical meter, measuring the energy consumption of each phase, and
through a circuit breaker, allowing for the shut-off of each phase for maintenance purpose, or
during an emergency event.
The mains supply strands end at a panel board, from where each phase supply is distributed
through individual wiring to various loads.
Electrical Installation Principle
GPD1
Vs GND
Supply System
Vs GND
Load
A1
Load
An
Load
B1
Load
C1
Load
Cn
Load
Bn
L1
L2
L3
N
PE
Vs GND
Vs GND
GPD2
3
Slide 3 further simplifies the supply system through a simple box, but emphasizes on the
various loading of the individual supply phases and the possible ground potential differences
that can exist between the dc-outputs of various loads.
With regards to data transmission, these loads can represent computers, printers, remote
controlled lab-equipment, basically any type of communication nodes that might require data
exchange between one another.
The main questions now arising are:
1) How are these ground potential differences created?
2) And how do they affect the DC output ground through a, supposedly, isolated power
supply ?
DC-Output to Mains Connection
Simplified low-cost, Switched-Mode Power Supply (SMPS)
SMPS chassis connects Ground of DC output to Protective Earth (PE) conductor
4
Answering Question 2 first:
The link between the DC ground of your local electronic circuit and the protective earth (PE)
reference potential of the mains is often provided by the local power supply converting the
line voltage into the required DC output.
This slide shows a simplified block diagram of a low-cost switched-mode power supply
(SMPS), widely used in personal computers, laser printers and lab equipment.
Here the DC ground of the SMPS output is referenced to the protective earth conductor (PE)
of the mains via the SMPS chassis. This direct link therefore acts as a sense conductor,
establishing the voltage at PE as the local DC ground potential.
Theoretically the PE conductor is assumed to be current free, hence there would be nothing
wrong with referencing the output ground to this potential. In reality however, low-cost
SMPS contain low-cost EMI filter, which leak non-linear currents into the earth wire.
While the leakage current of one SMPS into the PE conductor is negligible, the fact that large
buildings often have hundreds and thousands of these, non-linear loads leaking currents into
the PE makes this wire a current carrying conductor.
GPD Composition
Linear Loads:
small neutral current due to load imbalance
Non-linear Loads:
large neutral current due to distorted phase currents
(several times larger than fundamental phase current)
5
Now starting with the answer to Question 1:
A 2nd source for large PE currents comes, via the electrical installation, from the currents in
the Neutral conductor.
Firstly we look at the generation of neutral currents in a 3-phase system and then at the
electrical installation, or wiring systems, connecting the Neutral to the PE conductor.
Large office and industrial buildings operate a vast amount of non-linear loads, such as PCs,
laser printers, solid-state heater controls, fluorescent tubes, uninterruptible power supplies
(UPS), and variable speed drives (VSD). In comparison to linear loads such as incandescent
lamps, whose phase currents maintain a sinusoidal waveform, non-linear loads distort phase
currents, thus introducing large harmonic content.
In other words:
- applying a sinusoidal voltage to a linear load yields a sinusoidal phase current,
- applying a sinusoidal voltage to a non-linear load results in a distorted, non-sinusoidal phase
current.
Linear versus Nonlinear Loads
Fundamental current
in comparison
6
Slide 6 gives a more detailed presentation of the phase and neutral currents.
In the ideal case all neutral conductors merge into one neutral conductor of large diameter
within the distribution panel running close to the utility transformer.
In the case of linear loads the neutral currents of different phase systems cancel each other to
a certain extent. Only a fraction of the total neutral current remains due to loading imbalance.
For non-linear loads however, the individual currents add to a total neutral current primarily
consisting of third harmonic content.
Because any conductor possesses a finite resistance, the large neutral currents of non-linear
loads cause significantly higher voltage drops across the line resistances of the electrical
installation than those of linear loads.
Phase Current Composition
100
90
120
80
80
Phase Current [A]
70
40
60
0
50
-40
40
-80
30
20
-120
0
90
180
Phase Angle [deg]
270
360
10
0
1
3
5
7
9
11
13
15
17
19
Order of Harmonics
7
Slide 7 shows the values and harmonic content of the phase currents measured at one of many
distribution panels within a large office building.
Note that phase L2 (blue) has higher loading then phase L1 (red) and L3 (black).
While the 3rd and 5th harmonic of the fundamental line frequency (50 Hz or 60 Hz) make up
the lion share of the harmonic content, the vector sum of all frequency components can reach
peak values that exceed the amplitude of the fundamental phase current by more than 100%.
Neutral Current Composition
100
90
120
80
80
70
40
60
0
50
-40
40
-80
30
-120
0
90
180
270
360
20
Phase Angle [deg]
10
0
1
3
5
7
9
11
13
15
17
19
Order of Harmonics
8
Slide 8 shows the total neutral current and its harmonic content measured at the same panel.
With its peak values varying between 70 A and 90 A, the neutral current’s main harmonic
content is made up by the 3rd harmonic of the fundamental mains frequency.
It is these high neutral currents that, via the line resistances of the various installation
systems, significantly contribute to the ground potential differences between the loads of
different supply phases.
Hence the characteristic frequency of a Ground Potential Difference is the 3rd harmonic
of the fundamental mains frequency.
Earthing System: TN-C
9
When investigating the root cause of ground potential differences, we cannot ignore the line
resistance of installation wires, as they convert the high leakage and neutral currents into
significantly varying voltage potentials.
The most commonly applied electrical installation systems are TN-C and TN-C-S.
TN stands for French Terre Neutral, and implies that the Neutral conductor is grounded to
Earth potential at the transformer. The letter C indicates the combined use of Protective Earth
(PE) and Neutral (N) via one conductor, designated as PEN.
The PEN runs through the entire system up to a distribution point, (i.e. a sub-panel), close to
the loads, where it splits into separate PE and Neutral conductors that directly connect to the
loads.
Although TN-C represents an old earthing system, it has regained in interest due to the
cost-savings achieved by avoiding the run of an additional PE conductor.
This method however, has a major drawback. Because the split into PE and Neutral occurs
close to a load, the voltage potential at the local PE connection includes the large voltage
drops across the line resistance, RL-N, of long neutral conductors, caused by high neutral
currents from many non-linear loads.
TN-C systems therefore, have the potential to cause large ground potential differences
between remote grounds in the tens-of-volts.
Earthing System: TN-C-S
RL-N
RL-N
RL-PE
RL-PE
L1
N
PE
L1
L2
N
PE
L2
Load A1
RL-N
Load Am
RL-N
L3
N
PE
L3
RL-PE
RLS PEN
N
RL-PE
Load B1
Load Bn
PE
10
The TN-C-S system reduces GPDs by starting an extra PE-conductor within the distribution
panel. In addition the star connection of the system’s Neutral and PE conductors receive a
second grounding to earth, reducing the equipotential at this point and counteracting the
otherwise large voltage drop at the PEN across the source line resistance RLS.
As mentioned earlier, according to the National Electric Code (NEC), the PE is supposed to
be a currentless conductor under normal operation. However, most non-linear loads leak
currents in the lower milli-amps into the PE. This amount, although small for one unit, easily
reaches amps when considering that hundreds of equipments contribute into the same strand.
Although negligible in comparison with neutral currents, leakage currents do create potential
differences between remote ground locations due to the voltage drops across the line
resistance of the PE-conductors. These GPDs however, are in the lower volt range and thus
significantly lower than in TN-C systems.
Unfortunately, TN-C-S systems are mainly found in newer buildings, below a certain age.
Older buildings often receive an installation upgrade due to the latest NEC code. In most of
the cases however, only parts of the existing installation is upgraded. In some instances, new
TN-C-S installations are simply added to existing TN-C systems.
Making matters worse, additional Earth connecting rods are installed, or simply resembled by
connecting the PE to a nearby metal water pipe (above shown as dotted Earthing symbols),
thus complicating the Earth return path and worsening the GPD issue.
Ground Potential Differences
Ground Potential Difference [V]
30
20
10
0
-10
-20
-30
0
90
180
270
360
Phase Angle [O]
11
Ground potential differences are not a problem for data links whose bus nodes are powered
by the same phase strand.
Ground potential differences, as the ones above, become a concern when designing a
communication link between bus nodes powered from different parts of the electrical
installation.
This holds true for long and short distance nodes.
Because system designers don’t have control over the electrical installation supplying their
data transmission design, one should always assume that Ground Potential Differences do
exist.
Design Pitfall: Ground Loops
2
3
d) Complex Earthing
resistance network
1
12
Ground Potential Differences add as common-mode noise, Vn, to the transmitter output, and
EIA/TIA485 specifies a maximum GPD of ±7V across which a data link must operate
reliably.
Figure a: Despite this specification, relying on the local earth ground as a stable ground
reference is risky.
Even if the total superimposed signal is within the receiver’s input common mode range,
modifications to the electrical installation, (i.e. during maintenance work), are out of the
designer’s control. The modifications can increase the GPD to the extent, that the receiver’s
input common-mode range is either sporadically or permanently exceeded. Thus a data link
which perfectly works today might stop operating sometimes in the near future.
Figure b: Removing the ground potential difference by directly connecting remote grounds
through a ground wire is also not recommended.
Bear in mind that the electrical installation constitutes a highly complex resistance network,
consisting of multiple, cross-connected line- and grounding resistances, which are caused by
different phase systems, different cable lengths and various grounding electrode paths
(Figure d).
A direct connecting between remote grounds shunts this network and creates a current loop.
The initial GPD tries to compensate its collapse by driving a large loop current through the
low-impedance ground wire. The loop-current couples into the signal wires, generating a
noise voltage that is superimposed on the transmission signal. This again carries the risk of a
highly unreliable data transmission system.
Figure c: In order to allow for a direct connection of remote grounds, the TSB89 application
bulletin to EIA/TIA 485 recommends the separation of device ground and local system
ground via the insertion of 100 Ω resistors. While this approach reduces loop current, the
existence of a large ground loop keeps the data link sensitive to noise generated somewhere
else within the loop.
Single Ground Reference
Signal Rate - Mbps
100
HVD23
-20V
HVD20
10
+25V
HVD24
HVD21
1
-7V
HVD22
0.1
10
100
Cable Length - m
+12V
1000
13
In order to support GPDs of up to ± 20V, Texas Instruments has released the SN65HVD2x
family of Super RS-485 transceivers with an extended common-mode voltage range from 20 V to + 25 V.
Single Ground Reference
Isolated
Node n
ISO-GND
3
Local
Processor
Isolated
Node 1
5V
ISO-GND
3
Local
Processor
5V
Non-isolated
Ref. Node 0
REF-GND
5V
Local
Processor
Termination
Plug
Termination
Plug
14
For even higher GPDs, such as hundreds and thousands of volts, a single ground reference
approach is a must. However, this method requires the galvanic isolation of the signal and
supply lines of all but one bus node, which actually presents the system ground reference
node (see RS-485 Design Guide Lines).