RS485 is bidirectional does it mean it is fullduplex?

Jim Stewart <jstewart@jkmicro.com> wrote:
>
>As much as I hate jumping into this shitstorm,
>I guess I have to.
>
>You are presumably talking about the static
>drain and foil shield that wraps around all
>the pairs.  I've seen the cables and I'll
>certainly acknowledge its existance.
>
>OTOH, when I think of a telephone cable, I
>think of a pair.  The pair is *not* grounded.

However...  get out your handy dandy voltmeter, and you'll find
that the tip is at ground potential.  Which means, from the
perspective that most people see it, it's "grounded"!  And in
fact it *is* (through a 200 Ohm resistor).

There are some odd results because of that.  Try grounding your
telephone line, first tip and then ring.  Pick up the receiver
and listen to it.  Also try ringing it from another line if you
can.  There is a significant difference (with most but not all
switches) in what happens.

>In fact, FCC part 68 is quite clear on it
>being balanced and the balance having to be
>maintained.

FCC part 68 does not apply.  The FCC does not regulate telephone
company technical specifications.

However, the *telephone companies* insist that *you* as a
subscriber not do anything that will unbalance their subscriber
loop.  To that effect, FCC part 68 requires that anyone who
manufactures equipment for sale that a customer will connect to
a subscriber loop must pass type certification, part of which
requires proof that it will not impair loop balance.

There are other requirements too, all intended to protect the
Public Switched Telephone Network.  A few years back one example
became well known when PCM modems were first released and it was
discovered that the FCC regulations prohibited signal power
levels that were required for these modems to actually achieve
56 Kbps, rather than only 53 Kbps.

Prior to that, the big deal was you were not allowed to pretend
to be Captain Crunch, and putting any level of a pure 2600 Hz
tone on a telephone line was illegal.  (It's still illegal,
but 2600 Hz doesn't do anything now, so nobody cares.)

>As to the splice points, no, not all of them
>are grounded.  They are grounded where the
>cable loops into a building, but not necessarily
>on the pole.

When the cable is installed, at each point where a new section
(roll) is started, the cables are spliced and grounded.

I think you are referring to there being a ground at the network
interface on a building, but that is not grounding a shield.  It
grounds a surge protection device, not the cable.

-- 
Floyd L. Davidson           <http://web.newsguy.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska)                         floyd@barrow.com

"Steve at fivetrees" <steve@NOSPAMTAfivetrees.com> wrote:
>"Floyd L. Davidson" <floyd@barrow.com> wrote:
>>>So how do you reconcile even a 20V ground potential difference with the
>>>+/-7V common-mode maximum of RS-485?
>>
>> Any of a number of ways.
>>
>> But the first consideration would be whether RS-485 is a
>> appropriate protocol in that situation.   The answer is
>> almost certainly "No."
>
>But that's precisely what we're dealing with, routinely. RS-485 is fine in
>these situations (which, as Paul said, are far more common than you seem to
>realise) - so long as one pays attention to common-mode references.

If you are using it in inappropriate environments, don't be
complaining about how difficult it is to deal with.  "Routine"
doesn't make it good engineering practice.

I can't say that what you are doing is right or wrong, but if
you are dealing with the 80 volt common mode differences that
you and Paul claim, maybe somebody needs to review what you are
doing.

>Please give examples of the "any number of ways" you mention.

We've gone over *several* of them.  If you can't remember, try
Google.

>(<pedant>RS-485 isn't a protocol, it's a hardware signalling
>standard.</pedant>

It's a protocol.

A standard agreed to by a group, and then more or less adhered
to by others, is definitely a protocol.

Or would you prefer it to be made into a law?

-- 
Floyd L. Davidson           <http://web.newsguy.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska)                         floyd@barrow.com

paul$@pcserviceselectronics.co.uk (Paul Carpenter) wrote:
>On Friday, in article <873brga5fa.fld@barrow.com>
>     floyd@barrow.com "Floyd L. Davidson" wrote:

>>>Now consider the case where the left hand ground moves up to 80V away from
>>>the right hand ground, but that the impedance between the two ground
>>>connections remains less than 10 Ohms (not unreasonable in some very large
>>>buildings).
>>
>>Actually, I believe the specification *is* 10 Ohms, and the
>>target is 5 or lower.  An 80 volt difference in ground potential
>>is...  so unusual that we can ignore it.  Lets assume it never
>
>In *your* experience, in two situations seen all over the UK, Railway Stations
>with multiple platforms can often see 10-20V from the main earthing point as
>they progress outwards across platforms. One near me saw a 60V in 100ft of
>cabling that went 60ft by line of sight. Certain Eurostar depot called North
>pole one contractor in its building phase was seeing all sorts of problems
>with ground potentials in excess of 70V.

We've already noted that yes you *can* find such situations, caused
by specific circumstances.  They require special engineering to do
some things.

That doesn't make them common, or normal, nor should they really be
part of this discussion in the way you want them to be.

One thing they might be, is a very good reason not to use
RS-485, which can only handle a very limited common-mode voltage
range.

>>>Where you do need to connect the screen wire to both ends then you had
>>>better include some impedance in the connection that limits dangerous
>>>current flow and yet provides sufficient coupling to ground for the higher
>>>frequency signals. As I have stated before, all the circumstances have to
>>>be well understood to make the right choice.
>>
>>No. The trick is to ground the cable every 3 or 6 thousand feet, so
>>there is never get anything like an 80 volt difference.
>
>The outer shield that is for protective earthing to stop the metal
>outer casing becoming live by any means, including internal cable fault.
>It has nothing to do with the signal levels, and if it is used for that
>then you are asking for problems.

That is simply untrue.

>>Can't you come up with something less boorish than repeating the same
>>thing over and over?
>
>Learn the difference between signal reference and protective earthing.

Sure sonny.  Your statement about cable shields above indicates you
have a shallow understanding of it.

>Keeping things floating is done for many reasons, knowing that in Medical
>Equipment in Europe you have to watch simple things like is the patient in
>an operating considered floating or earthed? Well it depends which country
>you are in, which effects signalling and protective earths how they are done
>and what is isolated from what.

You should be aware that medical equipment is an entirely different
subject than comm cables.

-- 
Floyd L. Davidson           <http://web.newsguy.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska)                         floyd@barrow.com

On Fri, 17 Jun 2005 13:34:14 -0800, floyd@barrow.com (Floyd L.
Davidson) wrote:

>>I would hope that it is obvious to all from this discussion that the issue
>>of connecting screens/shields is rather dependent on the circumstances that
>
>You've now mentioned ships and "MEGA-Amp equipment".  There are
>many special cases, probably none of which lead to any
>enlightenment except when they are explicitly discussed as
>extremes and flagged as unusual.

When making more or less general purpose systems to be installed into
customer premises by people with very varying level of experience, you
have to make sure that they can get the system to work on their own.
The product must be designed properly and the installation instruction
be clear and simple that works in all situation. It is not cost
effective to travel to the other side of the world to solve
"mysterious" communication failures or burnt transceivers.

It would be too much to expect a local instrumentation technician to
be able to figure out if the customer wiring system is 

- TN-S, in which case the grounding at both ends could be justified 
- TN-C, in which case only a single point grounding should be used
- TN-C-S, in which case you would have to check that all current and
possibly later added devices are in the _same_ TN-S branch now and
that this will remain in the future
- IT (floating mains), in which case there should not be too much
problems, even if grounded at all places 
- TT (N grounded at transformer only, PE separately in each building)
which might have some consequences in fault and thunderstorm
conditions

Even if the technician would be knowledgable about these issues,
getting reliable and up to date information about the wiring in an
existing building can be a problem, especially if the wiring has been
done during different periods.

In order to keep the instructions clear I prefer to recommend methods
that work well in most cases and firmly grounding the shield at both
end is certainly _not_  one of those methods, even if it does not
cause harm in certain special cases.
 
Paul

On Sat, 18 Jun 2005 01:21:29 +0100, "Steve at fivetrees"
<steve@NOSPAMTAfivetrees.com> wrote:

>"Floyd L. Davidson" <floyd@barrow.com> wrote in message 
>news:873brga5fa.fld@barrow.com...
>> An 80 volt difference in ground potential
>> is...  so unusual that we can ignore it.  Lets assume it never
>> gets higher than 20.  Or 30, if you like.  (Everything I recall
>> seeing was engineered for 20 V, max.)  No doubt there *are*
>> unusual instances were we might well see figures outside this
>> range.  And if we do, we deal with them as unusual instances...
>
>So how do you reconcile even a 20V ground potential difference with the 
>+/-7V common-mode maximum of RS-485?

If we want the devices to survive after a ground fault (without any
residual current breakers), it should be noted that even with a TN-S
system with separate N and PE conductors and assuming the same cross
section as the L wire, a ground fault between L and PE would cause an
identical voltage drop in both wires. With 230 V mains, the ground
fault potential would go up to 115 V before the fuse is blown. Any
equipment connected directly to the PE close to the ground fault point
would get that potential and would most likely cause some permanent
damage to the RS-485 system.

Even with the often used 100 ohm resistors between the cable shield
and frame ground would limit the current to about 1 A (or 0.5 A if all
stations use the 100 ohm resistors), which would perhaps increase the
survivability of the data communication equipment.

It should be noted that a simple short circuit (eg. at the filament of
a dying incandescent lamp) would cause in the TN-C system a high
current peek in the common PEN conductor and some equipment connected
directly to the PEN connector would jump to 115 V (at 230 V mains).

So in practice, if high reliability is a requirement, use optical
isolation or at least keep the RS-422/485 signal grounds well away
from the PE or PEN conductor potentials.

Paul
  

On Fri, 17 Jun 2005 19:34:35 -0800, floyd@barrow.com (Floyd L.
Davidson) wrote:

>I can't say that what you are doing is right or wrong, but if
>you are dealing with the 80 volt common mode differences that
>you and Paul claim, maybe somebody needs to review what you are
>doing.

While my namesake had an example about 80 volt common mode difference,
I demonstrated in an other post, how you momentarily can have a 115 V
(for 230 V mains) or 55 V (for 110 V mains) common mode difference in
various ground fault or short circuit situations.

Paul
 

On Fri, 17 Jun 2005 13:40:54 -0800, floyd@barrow.com (Floyd L.
Davidson) wrote:

>Jim Stewart <jstewart@jkmicro.com> wrote:
>>Floyd L. Davidson wrote:
>>
>>> Reality check time: telephone cables all across the country and
>>> around the world are grounded every 3000 or 6000 feet.
>>>
>>
>>Not in the usage of the term "grounded" that
>>I'm familiar with.  Perhaps you could give us
>>a little more details.
>
>At every point where sections of cable is spliced, the
>shield on both sides of the splice is bonded, and grounded
>to an earth ground.  That is generally a copper rod driven
>into the ground.

Let's analyze this a bit closer. 

Assuming these cable splices are in poles at least some distance (say
100 m) from any buildings or transformers and their grounding
electrodes. 

What do you expect that the grounding resistance from a single copper
rod driven into the ground to be ? 

1) It could be several ohms depending on the ground conditions. This
is sufficiently to bleed any static discharge accumulated on to the
cable shield due to atmospheric conditions. Even if there is a
lighting (10-100 kA) parallel to a section of cable, quite a high
current would be induced into the shield, but the total _ground_
resistance between the grounding electrodes 1-2 km from each other
would still limit the current to manageable levels.

2) Assume for a moment that a stupid electric distribution system is
used in some country with a single overhead wire at say 10 kV AC and
the ground serving as the return conductor, i.e. each building would
have a step down transformer between the overhead wire and the
grounding electrode at the house. When walking without shoes from the
house towards the power source, there would be a small voltage
difference between your feet at every step due to the ground
resistance. 

Assume that there is a telephone cable section with one grounding
electrode close to the power source and an other close to the house.
Part of the power system load current would go through the grounding
electrode close to the house up to the pole, through the telephone
cable shield with a very low resistance to the other grounding
electrode close to the power source, down the pole and into the ground
and then back to the power source. Thus, the telephone cable shield
would carry most of the power system return current.
 
What does this have to do with RS-485 grounding you might ask ?

Situation 2) closely resembles the TN-C case, in which you have a
common PEN connector and the electric system load will cause a voltage
gradient along the PEN line, thus stations A and B will have a
different frame potential. Connecting a data cable shield between A
and B will cause some of the PEN  wire current to be diverted through
your data cable shield. 

In situation 1), the two telephone line grounding electrodes are
nearly (if not exactly) the same potential and no current would flow,
thus this corresponds to a shielded cable between two nodes in a TN-S
system with completely separate N and PE wirings. 

Even if there would be a small voltage difference between the ground
potential around stations A and B, the current would still be limited
by the grounding resistance at each electrode. This resembles the
common grounding practice which used in quite few RS-422/485 systems,
in which the shield is not connected directly to the frame ground, but
instead a 100 ohm resistor is used to limit the "ground loop" current.

While your claims about grounding the telephone cable at every 1-2 km
is believable, it does not justify your claim that RS-485 cables can
be directly grounded (without current limiting resistors) at both ends
in all cases.

Paul

Floyd L. Davidson wrote:
> "Paul E. Bennett" <peb@amleth.demon.co.uk> wrote:
> 
>>I am sensing that you have all along been talking of the cable armouring
>>and not the cable shields. When you refer to things by their proper names
>>it gets a whole lot less confusing.
> 
> 
> You are *grossly* confused, and don't know what the names are.
> 

I don't know anything about telephone cables (I *do* know something 
about RS-485, but I'm keeping quiet), but I've noticed one thing about 
this argument between Floyd and Paul - you both claim to have examined 
telephone cables, you both claim to know the regulations, and you both 
claim the other hasn't a clue and has never seen a real telephone cable. 
  Since one of you (Floyd) appears to be in the USA, and the other 
(Paul) appears to be in the UK, is it possible that there are 
differences in the standards, regulations and common practice on the two 
sides?  If this thread is to remain of interest to others, it might be 
worth establishing this.

mvh.,

David

On Sat, 18 Jun 2005 14:31:40 +0300, Paul Keinanen <keinanen@sci.fi>
wrote:

[Snipped]
>
>It would be too much to expect a local instrumentation technician to
>be able to figure out if the customer wiring system is 
>
>- TN-S, in which case the grounding at both ends could be justified 
>- TN-C, in which case only a single point grounding should be used
>- TN-C-S, in which case you would have to check that all current and
>possibly later added devices are in the _same_ TN-S branch now and
>that this will remain in the future
>- IT (floating mains), in which case there should not be too much
>problems, even if grounded at all places 
>- TT (N grounded at transformer only, PE separately in each building)
>which might have some consequences in fault and thunderstorm
>conditions
>
[Snipped]

Do you perhaps have a URL that describes the above wiring systems in
more detail ?

Regards
  Anton Erasmus

David Brown <david@westcontrol.removethisbit.com> wrote:
>Floyd L. Davidson wrote:
>> "Paul E. Bennett" <peb@amleth.demon.co.uk> wrote:
>>
>>>I am sensing that you have all along been talking of the cable armouring
>>>and not the cable shields. When you refer to things by their proper names
>>>it gets a whole lot less confusing.
>> You are *grossly* confused, and don't know what the names are.
>>
>
>I don't know anything about telephone cables (I *do* know
>something about RS-485, but I'm keeping quiet), but I've noticed
>one thing about this argument between Floyd and Paul - you both
>claim to have examined telephone cables, you both claim to know

Paul looked at a cable 20 years ago.

I merely worked in the industry for 34 years before retiring.
I've seen more at least twice as many cables! :-)

>the regulations, and you both claim the other hasn't a clue and
>has never seen a real telephone cable. Since one of you (Floyd)
>appears to be in the USA, and the other (Paul) appears to be in
>the UK, is it possible that there are differences in the
>standards, regulations and common practice on the two sides?  If
>this thread is to remain of interest to others, it might be
>worth establishing this.

There isn't any significant difference in cables as used around
the world.  Other regulations do differ significantly, though
generally there are two basic patterns an most countries follow
one or the other.  Such things as loop current specifications,
layout of the digital heirarchy (E1's vs. T1's etc), and the
type of codecs used (Mu-Law vs. A-Law), are the most significant
differences.

-- 
Floyd L. Davidson           <http://web.newsguy.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska)                         floyd@barrow.com