Newsgroups: rec.radio.amateur.antenna,rec.radio.amateur.homebrew,rec.radio.amateur.misc,rec.radio.amateur.digital.misc
From: gary@ke4zv.atl.ga.us (Gary Coffman)
Subject: Re: NEED HELP! Respond : Grounding 'Experts' !
Reply-To: gary@ke4zv.atl.ga.us (Gary Coffman)
Organization: Destructive Testing Systems
Date: Thu, 8 Jun 1995 14:11:25 GMT
In article <3r4qj0$kuj@noc.tor.hookup.net> Jeff Krul <jkrul@equist.com> writes:
>I am in the process of putting together a new HAM 'shack', and I have
>come across an interesting 'snag'.
>
>I have heard, many times before, to be sure the HAM radio ground is NOT
>grounded to the house wiring ground system. Various explanations follow,
>usually to avoid having the house wiring turn into a massive antenna.
Like many folk tales, while there is a grain of truth to this idea in
some *very* special circumstances, it's generally false, not to mention
being a violation of the National Electrical Code.
>Anyways, I was SURE to separate the two and overlooked one point . . .
>packet ! My TNC uses a serial cable, connected to my computer, which in
>turn has a grounded chassis! Argh! To top it all off, the serial cable
>is NOT the right way to ground to a 'secondary' ground. Talk about a
>'high' resistance path to ground, forming an 'ideal' ground loop!!!!
>
>My question is, "How can I avoid this, and if I cannot, what effect will
>grounding to the house ground REALLY have?" "What are my OTHER options?"
Ok, this isn't a simple issue, or a single issue. There are several
related concepts and factors involved. Bear with me and I'll try to
touch on most of them.
There are three general reasons given for grounding a station.
1. Improved RF performance.
2. Elimination of stray RF in the shack.
3. Electrical safety.
Of these three reasons, only the latter has merit except under
special conditions. Lets look at them one by one and see where
the truth lies.
1. Improved RF performance.
This is usually a reason claimed for needing a good ground. It is
actually false except with antennas that must work against ground,
and even then the ground needs to be at the antenna feedpoint and
not at the transmitter (unless the transmitter *is* the feedpoint,
as can be true with some random longwire antennas). A balanced antenna,
such as a dipole, yagi, quad, etc, is ground independent. It does not
need a ground connection to work, and will in fact work in free space
with no ground connection present at all. Many verticals, longwires,
and some other asymmetric designs need to work against a ground reference.
That ground reference, however, doesn't necessarily need to be a
connection to Earth. A counterpoise, radials, or the like can serve
just as well, or better, than an Earth connection. And that connection
needs to be under the antenna, and not back at the shack.
2. Elimination of stray RF in the shack.
This is often a reason given for a RF ground connection to Earth.
However, this is like taking aspirn for a brain tumor. It may
suppress the symptoms to some extent, but it doesn't address the
real underlying problem. The real underlying problem is a station
design or layout fault. If the equipment has poor Faraday shielding,
or if the feeder currents aren't well balanced, or if station
interconnections create daisy chains or ground loops, then there
will be stray RF in the shack. Grounding one or more cabinets may
actually create more circulating currents which can *increase*
the problems of stray RF. In some cases, attachment to a good RF
ground will reduce stray RF problems at some frequency, but may
increase problems at another frequency. This is not the proper
approach to dealing with stray RF. The preferred approach is to
eliminate the *cause* of the stray RF.
3. Electrical safety.
This is the *real* reason for an effective grounding system for
the shack. There are two different hazards that an effective
grounding system will solve. One is ordinary electrical shock
hazards. In the US, the National Electrical Code is the standard
that addresses this. One of the cardinal rules of the NEC is that
*all ground connections must be bonded together*. If you fail to
do this, you can have a shock hazard between cabinets connected
to different ground references. So the myth that you need to
keep utility and RF grounds separate is not only false, it can
be a safety hazard. (There are a couple of *very* specialized
circumstances where the Code permits isolated grounds, but as
a general rule, failing to bond all grounds together is a major
Code violation.)
The second reason for an effective grounding system is for lightning
mitigation. This can be a difficult problem and must be approached
with care. A single mistake can be very costly. Lightning is RF,
and like any other RF, downlead inductance is a primary concern in
setting up a grounding system. Lightning surges also represent peak
currents on the order of 8,000 amperes (with ocasional so called
"super bolts" going up to 200,000 amperes). These discharges only
last for microseconds, so the average power is low, but the peak
power is intense. It only takes a slight amount of inductance in
a downlead to generate a potential difference of tens of thousands
of volts between one end and the other. If these potentials are
allowed to express themselves between equipment cabinets, devastating
damage can occur to sensitive electronic components.
There are two concepts you must apply to your station ground system
as your first lines of defense against these potentials occurring
across your equipment. The first concept is that of a single point
ground, and the second is called ground window technique.
Single point grounding is simple in concept, but often subtle in
execution. Basically, all connections to Earth must be interconnected
at a single common point, and all connections from equipment that
needs a ground termination must run to that single point and to
*no other connection with Earth*. Daisy chained connections are
strictly prohibited. Every connection to the single point must
be straight and direct from the equipment requiring the ground
connection. Ground "busses" are a serious no-no despite their
being touted in amateur literature. They form instant ground
loops for the equipment, and at the high currents present in
a lightning discharge, they can cause thousands of volts to
express themselves across the various cabinets connected to the
"ground" buss. Don't do this.
The idea here is that because everything is forced to a single
potential, that of the single point, *whatever it is*, no current
can flow through the equipment. Zero potential difference results
in zero current flow. It's those damaging currents you are
trying to control. You don't want the path *to* ground to be
*through* any of your equipment. Rather, you want the path to
ground to be through the single point, and only through the
single point. Note well that "ground" is *not* zero volts,
and will not *be* zero volts in any sense except a *relative*
one during equipment operation or during a lightning strike.
This doesn't matter if your layout topology is correct. All
"ground" is supposed to be is a *common* potential for everything
in the system to reference against. Use of a single point
connection *enforces* this single common potential, whatever it
may be at any given instant.
When you start thinking about the many "ground" potential
interconnections in your station, achieving a true single
point ground system seems a daunting task. This is where
the related idea of a ground window can help. The purpose
of the ground window is to *short out* any possible ground
loops in your system at the station entry point.
A ground window is simple in concept, and simple in execution.
Basically, it is a physically small conductive plate through
which *every* cable that enters or leaves your station must
pass. Every cable that is supposed to be at "ground" potential
is bonded directly to this plate as it passes through. Every
conductor that is not supposed to be at "ground" potential
is bonded to the plate via an appropriate suppressor device.
(A book could be written on that subject alone.) Then the
plate is connected to your single point ground by a heavy
low inductance strap conductor. Now any large potential
difference that tries to express itself across cables
connected to various cabinets in your shack will find itself
shorted out at the ground window.
Note carefully that *every* cable must pass through the ground
window plate on its way in or out of your shack. That includes
*power*, telco, CATV (if present), and any network wiring as
well as the usual antenna feeders, rotator control leads, etc.
If only *one* wire bypasses the ground window, the protection
it offers is lost. Don't run extension cords to power outlets
that don't pass through the ground window. It can cost you
everything you own in a strike, including your life.
Now since you've taken great pains to do this protective grounding
system right, you'll find that it makes an excellent general purpose
RF grounding system as well. This is not to say that you'll generally
*need* a general purpose RF ground for the shack if you've done your
homework on antenna feeds, Faraday shielding, and the like, but you'll
have one anyway. And if you've done the job right (which requires
attention to a level of detail I haven't fully addressed here), your
station will be able to work even during the worst thunderstorms,
and even during direct strikes, without damage or interruption.
Proper grounding is a specialized, but well understood, part of the
protective process that radio and television stations, commercial
and military two way, and *amateur* radio stations should understand
and apply.
Gary
--
Gary Coffman KE4ZV | You make it, | gatech!wa4mei!ke4zv!gary
Destructive Testing Systems | we break it. | emory!kd4nc!ke4zv!gary
534 Shannon Way | Guaranteed! | gary@ke4zv.atl.ga.us
Lawrenceville, GA 30244 | |
From: DB Wilhelm <w3fpr@nando.net>
Newsgroups: rec.radio.amateur.antenna,rec.radio.amateur.homebrew,rec.radio.amateur.misc,rec.radio.amateur.digital.misc
Subject: Re: NEED HELP! Respond : Grounding 'Experts' !
Date: 10 Jun 1995 02:39:22 GMT
Organization: News & Observer Public Access
NNTP-Posting-Host: grail705.nando.net
k23690@proffa.cc.tut.fi (Kein{nen Paul) wrote:
>
> I very much doubt that you can find a PC installation (computer, monitor,
> printer etc..) in which the circuitry floats respective to the chassis and
> thus also floats relative to the PE bar in the building (assuming 3 wire
> power cords).
>
It is not necessary that the DC common circuit actually float. The
correct procedure is to connect DC common to the equipment ground
at only one point within any one machine. In PC type computers
there is usually done inside the power supply. In a properly
(read low noise) designed printed circuit board, the "ground" plane
is actually DC common, and should not be confused with earth ground.
In most IBM PC compatible motherboards, there is a mechanical mounting
stud near the keyboard connector that may be mistaken for a motherboard
grounding point - it is not = This mounting stud should have insulating
washers to keep it from grounding the DC common at that point. This
one item has cured many cases of mysterious behavior in PCs.
Again, the rule of thumb for grounding digital circuits is to connect
DC common to equipment ground at only one point within a machine, and
to be connect any signal cable shields to the equipment chassis at
only one end. I am aware that some terminal devices violate this
convention. My cynical response is that it helped them pass the
tests required to get it to market, but caused many service headaches
in the field, but my practical response says that there are always
exceptions to any convention - just be aware, and if you experience a
problem maybe you can identify the root cause a bit faster.
Please re-read Gary Coffman's earlier post on grounding. That is
something that could replace most of the words I have previously
read on grounding, and I checked the ARRL Handbooks section on
station grounding before placing my first post on this subject.
While it was somewhat informative, I found it woefully incomplete
in answering the all important WHY question that is so helpful
when the current conditions don't quite match the 'normal' and I
don't ever seem to encounter a totally 'normal' situation.
73,
Don W3FPR
From: k23690@proffa.cc.tut.fi (Kein{nen Paul)
Newsgroups: rec.radio.amateur.antenna,rec.radio.amateur.homebrew,rec.radio.amateur.misc,rec.radio.amateur.digital.misc
Subject: Re: NEED HELP! Respond : Grounding 'Experts' !
Date: 12 Jun 1995 11:37:23 +0300
Organization: Tampere University of Technology
Distribution: world
NNTP-Posting-Host: proffa.cc.tut.fi
This is drifting a bit off the original topic, but here is some thoughts of
how to operate computers, computers and TNCs or computers or remote rotator
control boxes in different rooms or different buildings and how to solve
the ground potential/ground loop problems.
This is based on my experience planning and maintaining computer terminal
networks that are spread over a large area (one or more buildings).
It is also based on the 4 wire wiring practice (three phases and neutral)
used over here (230/400V) and thus is not directly aplicable to the U.S.
wiring practices, but the principles are the same.
In a three phase power distribution system, no curent is supposed to
flow in the neutral wire if all phases are equally loaded. The neutral
wire is connected to the earth at the power company and usually in every
house. In homes and offices most loads are single phase (connected between
one phase and neutral) and thus the full load current is flowing in both
the live and neutral wire in the cable between the fuse box and the
load (wall outlet, lamps etc.) The neutrals from these cables are
connected together and some cancelling takes places if loads are powered
by different phases.
The protective earth (PE) in a 3 terminal wall outlet is connected to
neutral in any convinient place, previously this was directly in the
wall outlet, but nowadays this is usually done in a distribution box
feedig the outlet.
In the ASCII drawing below, two loads are connected to the same phase
(this is the worst case, but full cancellation will not occure even
if they are connected to two different phases) and some computer
equipment is connected to the same outlets as these big loads.
Current I1 is returning from load 1 flowing thrugh a lead resistance R1.
In a similar way the return current I2 from load 2 flows through lead
resistance R2. Two equipment X and Y (say a PC and a TNC) are connected
as shown, the == symbol indicates Neutral and PE (the live connection
is not shown). The PE is connected to N at the wall outlet or some other
distribution box close to it.
Live I1 + I2
--------------------------------------
! !
Load1 Load2
! !
I1 ! ! I2
! Rs !
!==X..............Y==!
! !
R1 R2
! ! I1 + I2
--------------------------------- Neutral
Putting in some typical values to illustrate what happens.
If R1 = 0.3 ohms and R2 = 0.3 ohms and I1 = 1 A and I2 = 10 A,
then X is at 0.3 V and Y at 3 V and a 2.7 V potential difference
between X and Y. If the DC-ground is connected to the chassis in
both equipment, then this voltage difference can be measured between
the signal ground terminals (pin 7). If a serial cable is connected
between the equipment with signal ground lead resistance Rs of 2.7 ohm,
allmost 1 A is flowing through the signal ground lead and also through
the circuit boards of both equipment.
The noise margin is greatly deteriorating, since the transmitter
generates the output voltage relative to its local ground reference
and the receiver senses the input voltage relative to the local ground
at its end of the line.
If a thicker wire is used, the current between X and Y will increase
and the voltage difference will drop improving the noise margin.
If Rs could be set to 0 ohms, the voltage difference would dissapear
but 4.5 A would flow through the traces on the PCB causing lots of
problems.
However, if the protective ground (pin 1) or a separate pig tail is
connected directly to the chassis _at_both_ends_ and a better conductor
(such as the shield of the cable) are used for protective ground, the
current flow through the PCB is reduced to a tolerable level.
IMHO, this is far from a good situation, since large currents are
flowing in the shield with a potential fire risk and the current will
induce interference currents into the signal lead. Connecting and
disconnecting the serial cable will most likely burn the receivers or
transmitters if done in the wrong order.
There are several ways to improve the situation
a) Rebuild the power distribution system in the building so that the
PE is kept separate from the neutral and the PEs from the wall outlets
are brought to a single point. This point is also connecting to the
neutral and the grounding electrode of the building. In the normal
(no isolation fault) case, no current is flowing in the PE network
and consequently there are no voltage drops in the wiring, thus all
PE terminals in every wall outlet (and all chassises) are at the
same potential and no ground loop currents are flowing in the serial
cable, even if the the signal/protecting grounds are connected.
This 5 wire wiring system (3 phases, Neutral and PE) are installed
on all new office buildings. However, this solution is usually not
practical for old building, neither does it help if you are connecting
two computers in two different buildings with each own grounding
electrode.
b) If the DC ground of one of the equipment can safely be floated, such
as an external modem powered by a double isolated wallwart, this is the
simplest solution. The cable shield should only be connected at the
grounded equipment end.
c) Use balanced (such as RS-422) lines. In this case both equipment can
be grounded and no signal/protective ground lines are connected in the
serial cable. As long as the ground potential difference between X
and Y are smaller than the common mode voltage range of the receiver
(usually 6.. 30 V) this will work fine. This is currently often used in
industrial sites, where there can be quite large ground potential
differences.
d) Use completely isolated (no galvanic connection) connection between
X and Y. The simplest way is to use current loop (20 mA) signaling
and optoisolators. The cable shield should be connected at one end only.
This was previously (at relatively low data rates) the most common way
to get around the ground potential problem. My suggestion (in a previous
post) to use audio transformer between the TNC and the rig is a variation
of this principle.
Now, responding to Don W3FPR
DB Wilhelm <w3fpr@nando.net> wrote:
> k23690@proffa.cc.tut.fi (Kein{nen Paul) wrote:
>>
>> I very much doubt that you can find a PC installation (computer, monitor,
>> printer etc..) in which the circuitry floats respective to the chassis and
>> thus also floats relative to the PE bar in the building (assuming 3 wire
>> power cords).
>>
> It is not necessary that the DC common circuit actually float. The
> correct procedure is to connect DC common to the equipment ground
> at only one point within any one machine.
=======================
Yes, this is good engineering practice, but look what happens when we look
at the situation at network level. If we have one computer and some
floating device at the other end, then we can apply principle b) above.
However, if you are trying to connect two computers, then principle b)
can no longer be applied due to the grounding in _each_ PC. This was
just the situation with a cumputer and multiple VT100/VT220 terminals,
which are connected just like PCs you described. It does not matter what
you do with pin 1 (protective ground), you will have circulating ground
currents through pin 7 anyway. In this case it might be better to connect
both ends of the cable shields for improved noise margin, but I do not
recommend it after having replaced line receivers on several terminals :-)
You should newer break the signal ground (pin 7) connection on RS-232 to
break the ground current, since you will end up with burnt receivers and
transmitters very quickly during normal operation.
My recommendation is to ground equipment properly for electric safety
and use isolation (optoisolators, transformers etc.) to transfer the
signals.
Sorry for this long ranting, but hope that someone might find something
useful, next time when someone is connecting together equipment in
different rooms or different buildings.
Paul OH3LWR
--
Phone : +358-31-213 3657 Mail: Hameenpuisto 42 A 26
Internet: Paul.Keinanen@cc.tut.fi FIN-33200 TAMPERE
Telex : 58-100 1825 (ATTN: Keinanen Paul) FINLAND
X.400 : G=Paul S=Keinanen O=Kotiposti A=ELISA C=FI
Newsgroups: ,,,
From: gary@ke4zv.atl.ga.us (Gary Coffman)
Subject: Re: NEED HELP! Respond : Grounding 'Experts' !
Reply-To: gary@ke4zv.atl.ga.us (Gary Coffman)
Organization: Destructive Testing Systems
Date: Tue, 13 Jun 1995 18:27:26 GMT
In article <D9wont.F4E@mv.mv.com> rapp@lmr.mv.com (L. M. Rappaport) writes:
>gary@ke4zv.atl.ga.us (Gary Coffman) wrote:
>... edited
>>Single point grounding is simple in concept, but often subtle in
>>execution.
>
> Excellent article. One question, however: Assuming you bond
>everything properly like a reverse daisy chain to a single point gound
>at the meter entrance, is it ok to extend that gound by bonding
>multiple ground rods together? Something like this
>
> Ground Rod --------+ +--------Stuff in shack
> | |
> Ground Rod ---------+----------+--------Entrance Panel
> | |
> Ground Rod ---------+ +--------Tower, etc
Daisy chaining of any sort is generally to be avoided. However,
this is not a daisy chain if I read the diagram correctly. What
you have is a star connected group of grounds (I'm assuming you
left out a couple of |'s in the drawing) that are connected to
a single entrance point. That's a perfectly acceptable way to
do it. IE
G
|
G---x-------------------->house
|
G
That's fine. Now, if you were to do this;
G------G------G-----G------G--------->house
That would be Ok too, but not as desirable because of the accumulated
voltage drops in the daisy chain. What you want to avoid is this;
A----G----G----B-----G-----C----G->house
| | |--------->
| |--------------------->
|------------------------------------>
with A, B, and C representing antennas or other external
equipment. Now you have a nasty ground loop through the
chain of G's and that's going to lead to trouble. If you
do this;
A-G B-G C-G G-house
| | |------------------->
| |-------------------------->
|--------------------------------->
This is *real* trouble. Now you're depending on the (poor) conductivity
of Earth to tie the grounds together, and they'll be at wildly different
potentials during a strike. Those potentials will be reflected on the
cables from A, B, and C, and will do big time damage to the station if
not shorted by a ground window.
Now the better way is to do this;
A---------------------->
G
|
BG===x====================>house
|
G
C--------------------->
B's cable closely parallels the star ground run in the drawing. Since
A, B, and C's grounds are tied in star, they will be at nearly the
same potential. Best, however, would be individual ground runs from
A, B, and C's grounds back to a single common point at the house.
That avoids any ground loops.
Any of the situations that create external ground loops can be
shorted by a ground window at the station entrance. That will
protect the equipment *inside*, but equipment in the external
circuit, preamps, rotators, etc, can still be damaged by the
circulating ground currents. So it's best to avoid ground loops
in the external circuit and always use a star topology.
Remember, for lightning, just having good DC conductivity isn't
enough. In the first place, at 8,000+ amps of surge, even good
conductivity is rarely enough to prevent substantial voltage
drops. More important, however, is that lightning is RF, and
low *impedances*, including reactance, are important to keep
the potential differences at bay. It's always the *difference*
in potential that matters. It doesn't matter if everything
becomes elevated in potential, as long as everything is elevated
to the *same* potential. Star topology is important to make
this happen.
Gary
--
Gary Coffman KE4ZV | You make it, | gatech!wa4mei!ke4zv!gary
Destructive Testing Systems | we break it. | emory!kd4nc!ke4zv!gary
534 Shannon Way | Guaranteed! | gary@ke4zv.atl.ga.us
Lawrenceville, GA 30244 | |
From: jeff@wa1hco.MV.COM (Jeff Millar)
Newsgroups: rec.radio.amateur.digital.misc
Subject: Re: Ham-Digital Digest V95 #183
Date: 15 Jun 95 17:56:14 GMT
Organization: ucsd usenet gateway
NNTP-Posting-Host: ucsd.edu
Originator: daemon@ucsd.edu
>Date: 15 Jun 1995 12:46:40 GMT
>From: n7tcf@primenet.com
>Subject: NEED HELP! Respond : Grounding 'Experts' !
<Stuff Deleted>
> First, don't get bent out of shape until you find there is a problem.
> Second, build your own serial cable with only one end of the shield
ground
>connected.
> 73 Jim n7tcf@primenet.com
>
A few points about cable shields seem appropriate. A number of posts
referred to grounding a shield at only one end. This has some advantages
and disadvantages. Clearly, having the grounded shield creates electrostatic shield to prevent coupling of E-fields in the vicinity from
coupling into the wires of the cable. By not connecting it at both ends, no
currents can flow in the shield...which at first glance prevents
electromagnetic coupling into the wires of the cable. This approach works
best with low level audio signals in an environment with significant ground
current. Examples include aircraft audio distribution, building PA systems,
and so forth.
However, this has a big disadvantage in the presense of RF fields. A shield
grounded at one end forms a quarter wave resonant antenna at a frequency
dependant on its length. This antenna in effect focuses hi-Q resonanting
currents onto the shieldat one end and high voltage RF electrostatic
coupling at the other end. So, although low impedance, low frequency
grounds get prevented, RF currents become encouraged.
In practice, hams should ground shields at both ends for any practical
applications within a ham shack. Consider the other approach only if you
have to run repeater audio from the basement of a high rise to a repeater on
the top floor.
jeff, wa1hco
Newsgroups: rec.radio.amateur.antenna,rec.radio.amateur.homebrew,rec.radio.amateur.misc,rec.radio.amateur.digital.misc
From: gary@ke4zv.atl.ga.us (Gary Coffman)
Subject: Re: NEED HELP! Respond : Grounding 'Experts' !
Reply-To: gary@ke4zv.atl.ga.us (Gary Coffman)
Organization: Destructive Testing Systems
Date: Sat, 17 Jun 1995 20:56:37 GMT
In article <DABq8o.4MK@mv.mv.com> rapp@lmr.mv.com (L. M. Rappaport) writes:
>gary@ke4zv.atl.ga.us (Gary Coffman) wrote:
>>In article <DA5y6I.64A@mv.mv.com> rapp@lmr.mv.com (L. M. Rappaport) writes:
>
>>>Where my scheme breaks down at the moment is that the tower has two
>>>individual ground rods and there is one on each of three guy wires.
>>>I'd guess that could cause a loop, eh?
>
>>Connect them in a star at the tower base, and then connect that
>>point back to the star at the house. A star of stars is an acceptable
>>topology.
>
>The guys are steel and not insulated from the tower. In addition,
>the tower base is also grounded. I can connect all those grounds as a
>star which then connects to the star at the house, but doesn't the
>tower + grounds constitute 3 loops in the tower star?
>
>IE, if the tower is grounded, a guy wire is grounded, and connecting
>each of those to another, common, star create a loop?
It's a circle, but it's not a ground loop. A bussbar represents a
ground loop, but it's not a circle. Confusing isn't it? The terminology
could stand improvement, as with the case of Earth and ground, they
aren't the same thing at all.
A ground loop occurs when two current meshes share a common
element. The current of one mesh modulates the potential
seen as "ground" by the other mesh, and hence indirectly
the current flowing in the second mesh. So the "loops" being
spoken of are current loops, not physical loops. And it
takes at least two current loops interacting in a certain
way across a shared impedance to create a ground loop. A
bussbar "ground" is a classic example of a ground loop.
R1 R2
x----/\/\/\-----x-----/\/\/\------x------x
| | | _|_
/ / | ///
\ R3 \ R4 |
/ / |
\ \ |
/ / ^ |
| ^ | | I2 |
| | I1 x------(S2)--RL2--x
| |
x-----(S1)----------RL1-----------x
Ok, R1 and R2 are the distributed impedance of the bussbar. R3
and S1 represent one piece of equipment, and R4 and S2 represent
another piece of equipment. RL1 and RL2 are the working loads
for S1 and S2 respectively. There are two current meshes. Current
I2 flows through S2, R4, R2, and RL2. Current I1 flows through S1,
R3, R1, R2 and RL1. R2 is the common impedance. Any change in I2
will be reflected as a different potential at the R1 side of R2,
and will be seen as an elevation of "ground" potential by S1 which
will alter the value of I1 flowing in that mesh. Any change in I1
will similarly show an elevated "ground" potential at S2. This is
a ground loop. The potential of ground is seen as different values
depending on the current flow in the *other* mesh. If lightning
were to strike RL2, the value seen as ground at S1 would change
dramatically, and likely cause damaging currents to flow through
S1 and RL1, or between S1 and S2 if they share any interconnecting
cabling.
If we eliminate the shared impedance by collapsing the buss to
a star, the ground loop goes away.
The guy wires don't represent a ground loop because there aren't
two current meshes.
x-------------(SL1)
/ | \ |
/ | \ |
/ | \ |
/ | \ |
/ | \ |
x-/\/-x-/\/--x--------x
|Rg1 | Rg2 |
|_____*______|
star
Where Rg1 and Rg2 are the impedances of the individual strips
of Earth connecting the guy anchors with the tower base. These
are shorted by the star connection and can be ignored. The
current source is SL1 which represents a lightning bolt flowing
current to Earth. The current will divide at the tower top, with
some flowing down the tower, some flowing down one guy, and some
flowing down another. The guys are in *parallel* with the tower
between Earth and sky. There's only one current mesh, and no
ground loop. The tower looks like the common element of our
ground loop example above, but it isn't because all the currents
are flowing in one mesh.
Now if the star were missing, and some piece of equipment's
current mesh bridged a guy anchor and the tower base across
Rg1 or Rg2, *then* you'd have a ground loop. Currents flowing
across Rg1 would induce a potential that would be seen in
the equipment current mesh, and alter the current flowing
through that mesh. By starring the grounds, we short circuit
that possibility.
Gary
--
Gary Coffman KE4ZV | You make it, | gatech!wa4mei!ke4zv!gary
Destructive Testing Systems | we break it. | emory!kd4nc!ke4zv!gary
534 Shannon Way | Guaranteed! | gary@ke4zv.atl.ga.us
Lawrenceville, GA 30244 | |
Newsgroups: rec.radio.amateur.antenna,rec.radio.amateur.homebrew,rec.radio.amateur.misc,rec.radio.amateur.digital.misc
From: gary@ke4zv.atl.ga.us (Gary Coffman)
Subject: Re: NEED HELP! Respond : Grounding 'Experts' !
Reply-To: gary@ke4zv.atl.ga.us (Gary Coffman)
Organization: Destructive Testing Systems
Date: Sun, 18 Jun 1995 17:40:38 GMT
In article <DADIID.3AF@mv.mv.com> rapp@lmr.mv.com (L. M. Rappaport) writes:
>gary@ke4zv.atl.ga.us (Gary Coffman) wrote:
>
>...edited
>
>>It's a circle, but it's not a ground loop.
>
>>A ground loop occurs when two current meshes share a common
>>element.
>
>>If we eliminate the shared impedance by collapsing the buss to
>>a star, the ground loop goes away.
>
>>The guy wires don't represent a ground loop because there aren't
>>two current meshes.
>
>> x-------------(SL1)
>> / | \ |
>> / | \ |
>> / | \ |
>> / | \ |
>> / | \ |
>> x-/\/-x-/\/--x--------x
>> |Rg1 | Rg2 |
>> |_____*______|
>> star
>
>>Where Rg1 and Rg2 are the impedances of the individual strips
>>of Earth connecting the guy anchors with the tower base. These
>>are shorted by the star connection and can be ignored. The
>>current source is SL1 which represents a lightning bolt flowing
>>current to Earth. The current will divide at the tower top, with
>>some flowing down the tower, some flowing down one guy, and some
>>flowing down another. The guys are in *parallel* with the tower
>>between Earth and sky. There's only one current mesh, and no
>>ground loop. The tower looks like the common element of our
>>ground loop example above, but it isn't because all the currents
>>are flowing in one mesh.
>
>>Now if the star were missing, and some piece of equipment's
>>current mesh bridged a guy anchor and the tower base across
>>Rg1 or Rg2, *then* you'd have a ground loop. Currents flowing
>>across Rg1 would induce a potential that would be seen in
>>the equipment current mesh, and alter the current flowing
>>through that mesh. By starring the grounds, we short circuit
>>that possibility.
>
>Ok, now I understand. It's the common point obtained by tying the
>bottom of the guys and tower together in a star which changes the
>configuration from a loop to a "mesh".
No. It's not a ground loop because there's only *one* current mesh.
To have a ground loop requires two (or more) current meshes that share
a common impedance. We'd have a ground loop if there were a *second*
mesh, that included some equipment we are trying to protect, that
shared a common impedance with the first current mesh. The common
impedance of the two meshes in this case would be either Rg1 or Rg2
depending on the topology. By connecting the guy grounds in star back
at the tower base, we drastically lower the value of the common impedance
and reduce any ground loop effects. If the star connections are low
enough in impedance, we can say we have effectively short circuited
the ground loop. The ground loop would still be *there*, it just
wouldn't have much of an effect because the common impedance would
be so small.
"Mesh" is just a way of saying "circuit" in analytic terms for
certain current paths. There's nothing more to it than that.
When we analyze a complex circuit, we break it into simpler
closed current paths and call each of those a mesh. We can
then analyze each mesh, noting any common elements with other
meshes, and then combine them all back together, using those
common elements as the interaction points, to get a complete
picture of overall circuit behavior.
A mesh with no shared common impedances with another mesh is
a closed circle, but it's not a ground loop. Earth is a giant
sheet resistance, however, and when we make a connection to
Earth anywhere, there will be a common impedance with every
other connection to Earth anywhere else. Mostly those common
impedances will be fairly high, and we can short them out by
doing a low impedance star connection between the points.
>In turn by tying the mesh into
>a the star at the house entrance panel eliminates the loop which might
>be formed between the two stars were they not connected.
Yeah. A star connection shorts out the common impedance needed to
have a ground loop. Note that we don't much care about absolute
potentials in any of this. We simply *define* one of our star
points as reference ground. All we really care about is potential
values *relative* to that reference point. If we arrange our
wiring topology right, we can keep those relative potentials to
a minimum.
In point of fact, *all* potentials are relative. We just aren't
very used to thinking of them that way. We tend to take "ground"
as zero volts, but of course it is only if we define it to be.
And if it has any impedance at all, we can only define it to
be zero at *one* point. That's our single point in our grounding
system.
Let me take a moment to explain where the term "ground" came from.
When we say "ground", we really mean *ground potential*. All
potentials are referenced to this ground potential. This idea is
modeled on the idea of a weight suspended above "the ground".
The higher the weight is above ground, the more gravitational
potential energy it has. The closer it is to ground, the less
gravitational potential energy it has. When it is sitting on
the ground, we say it has zero gravitational potential energy.
We could have the weight suspended above a table, and the table
would become "ground". Ground is the potential reference, the
defined zero state. This concept was well known when electrical
physics started, and the term "ground" just carried over. We
shouldn't confuse it with Earth. Earth is a non-homogeneous
sheet resistance. No two points on Earth have the same potential
when currents are flowing between them. Earth is not ground,
though we can pick a single point on Earth and *define* it as
ground for some particular circuit. But we can also pick some
point not on Earth and define it as ground instead, and we usually
do because we can arrange for that point to be tied to other
points via simple low impedance paths that can be easily analyzed
and controlled.
Gary
--
Gary Coffman KE4ZV | You make it, | gatech!wa4mei!ke4zv!gary
Destructive Testing Systems | we break it. | emory!kd4nc!ke4zv!gary
534 Shannon Way | Guaranteed! | gary@ke4zv.atl.ga.us
Lawrenceville, GA 30244 | |