On Tue, 8 Jul 2008 07:33:34 -0700 (PDT), Bob <S...@gmail.com>
wrote:

>On Jul 7, 12:33 pm, James Beck <j...@reallykillersystems.com> wrote:
>> In article <QoWdnZl_Ce8o9-_VnZ2dnUVZWhedn...@posted.visi>,
>> gra...@visi.com says...
>>
>>
>>
>> > On 2008-07-07, Vladimir Vassilevsky <antispam_bo...@hotmail.com> wrote:
>>
>> > > Paul Keinanen wrote:
>>
>> > >>>I am trying to interface two systems through RS422 link. There
>> > >>>are only four lines connected between them (tx+/- & Rx+/-) and
>> > >>>no ground.
>>
>> > >> This is quite OK as long as there is a termination resistor
>> > >> (typically 100..120 ohms) between the Tx+ and Tx- lines and
>> > >> also between the Rx+ and Rx- lines. The ground connection is
>> > >> not needed,
>>
>> > > Ground connection is required unless the transceivers are
>> > > floating.
>>
>> > Even when they are, I've had problems unless there's a 5th
>> > wire.  In theory they should float together and work fine, but
>> > in my experience you can't count on it.
>>
>> Yep, gotta' have a reference for the signals, even if they are
>> "differential", at least in this case.
>> You hit it on the head when you made the comparison to a current loop.
>> I'll bet you dimes to doughnuts that once a good common is run between
>> the units all the problems go away.  At the very least that is one more
>> thing to mark off the checklist.
>>
>>                             Jim
>
>I've found the same thing. Ya gotta have a common ground for reliable
>operation. It has always seemed to me that it should work with twisted
>pair shielded cable grounded at one end - but it doesn't in practice :

Why not run the signal lines into an opto? That's even the _required_
configuration for the RS-422 receivers in the NMEA 0183 standard. Data
rate could be an issue, optos tend to be slow; 0183 runs at 4800 baud so
it's pretty forgiving in this regard.

-- 
Rich Webb     Norfolk, VA

On 2008-07-08, Paul Keinanen <k...@sci.fi> wrote:
> On Mon, 07 Jul 2008 13:53:32 -0500, Grant Edwards <g...@visi.com>
> wrote:
>
>>On 2008-07-07, Paul Keinanen <k...@sci.fi> wrote:
>>> On Mon, 07 Jul 2008 12:15:17 -0500, "svyamuna" <s...@yahoo.co.in>
>>> wrote:
>>>
>>>>I am trying to interface two systems through RS422 link. There are only
>>>>four lines connected between them (tx+/- & Rx+/-) and no ground.
>>>
>>> This is quite OK as long as there is a termination resistor (typically
>>> 100..120 ohms) between the Tx+ and Tx- lines and also between the Rx+
>>> and Rx- lines. The ground connection is not needed,
>>
>>Wrong -- take a look at the receiver's spec for max common-mode
>>voltage.  It's usually only 8-12V. You've got to have a common
>>ground to make sure you don't exceed the receiver common-mode
>>voltage spec.
>
> And what would a flimsy signal ground wire help to reduce the
> common mode voltage in mains powered systems?

I'm not really sure what you're asking.  In the systems I'm
familiar with, RS-422 drivers and receivers are usually
galvanically isolated from mains power, and signal ground isn't
always connected to either case or earth ground.

> To understand why the SG is not required,

Analyze all you want.  I've seen a lots of RS-485 and RS-422
installations, and not connecting a signal ground often makes
them fail.

-- 
Grant Edwards                   grante             Yow! I feel like I am
                                  at               sharing a ``CORN-DOG''
                               visi.com            with NIKITA KHRUSCHEV ...

On Tue, 8 Jul 2008 07:33:34 -0700 (PDT), Bob
<S...@gmail.com> wrote:

>On Jul 7, 12:33 pm, James Beck <j...@reallykillersystems.com> wrote:
>> In article <QoWdnZl_Ce8o9-_VnZ2dnUVZWhedn...@posted.visi>,
>> gra...@visi.com says...
>>
>>
>>
>> > On 2008-07-07, Vladimir Vassilevsky <antispam_bo...@hotmail.com> wrote:
>>
>> > > Paul Keinanen wrote:
>>
>> > >>>I am trying to interface two systems through RS422 link. There
>> > >>>are only four lines connected between them (tx+/- & Rx+/-) and
>> > >>>no ground.
>>
>> > >> This is quite OK as long as there is a termination resistor
>> > >> (typically 100..120 ohms) between the Tx+ and Tx- lines and
>> > >> also between the Rx+ and Rx- lines. The ground connection is
>> > >> not needed,
>>
>> > > Ground connection is required unless the transceivers are
>> > > floating.
>>
>> > Even when they are, I've had problems unless there's a 5th
>> > wire.  In theory they should float together and work fine, but
>> > in my experience you can't count on it.
>>
>> Yep, gotta' have a reference for the signals, even if they are
>> "differential", at least in this case.
>> You hit it on the head when you made the comparison to a current loop.
>> I'll bet you dimes to doughnuts that once a good common is run between
>> the units all the problems go away.  At the very least that is one more
>> thing to mark off the checklist.
>>
>>                             Jim
>
>I've found the same thing. Ya gotta have a common ground for reliable
>operation. It has always seemed to me that it should work with twisted
>pair shielded cable grounded at one end - but it doesn't in practice :
>(
>
>Bob

Ground it at *both* ends. 

Best regards, 
Spehro Pefhany
-- 
"it's the network..."                          "The Journey is the reward"
s...@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

On Tue, 08 Jul 2008 10:24:53 -0500, Grant Edwards <g...@visi.com>
wrote:

>On 2008-07-08, Paul Keinanen <k...@sci.fi> wrote:
>> On Mon, 07 Jul 2008 13:53:32 -0500, Grant Edwards <g...@visi.com>
>> wrote:
>>
>>>On 2008-07-07, Paul Keinanen <k...@sci.fi> wrote:
>>>> On Mon, 07 Jul 2008 12:15:17 -0500, "svyamuna" <s...@yahoo.co.in>
>>>> wrote:
>>>>
>>>>>I am trying to interface two systems through RS422 link. There are only
>>>>>four lines connected between them (tx+/- & Rx+/-) and no ground.
>>>>
>>>> This is quite OK as long as there is a termination resistor (typically
>>>> 100..120 ohms) between the Tx+ and Tx- lines and also between the Rx+
>>>> and Rx- lines. The ground connection is not needed,
>>>
>>>Wrong -- take a look at the receiver's spec for max common-mode
>>>voltage.  It's usually only 8-12V. You've got to have a common
>>>ground to make sure you don't exceed the receiver common-mode
>>>voltage spec.
>>
>> And what would a flimsy signal ground wire help to reduce the
>> common mode voltage in mains powered systems?
>
>I'm not really sure what you're asking.  

Let me rephrase, in a TN-C electric distribution system, what
difference does a (say 0,3 mm) signal ground conductor have when
connected in parallel with a (say 1 mm) diameter PEN conductor ? 

Regardless of the presence of the signal ground, the voltage
difference along the PEN wire is more or less the  same.
  
>In the systems I'm
>familiar with, RS-422 drivers and receivers are usually
>galvanically isolated from mains power, and signal ground isn't
>always connected to either case or earth ground.

As a rule of thumb, I have used the following principle: 

* RS-232 is OK when connecting equipment in the same rack (such as
various modems and current loop converters), 

* differential RS-422/485 is OK for galvanic connections in the same
equipment room, 

* galvanic isolation (500 V/2500 V) is OK within a building. 

* However, if you are connecting equipment in different buildings, an
optical fiber should be used, since during a thunderbolt and due to
the grounding resistance, the ground potential might have s several
kilovolt peak relative to the surrounding countryside.

>> To understand why the SG is not required,
>
>Analyze all you want.  I've seen a lots of RS-485 and RS-422
>installations, and not connecting a signal ground often makes
>them fail.

Using galvanic isolation between sites A and B, I have never noticed
any RS-422/486 problems due to the removal of the SG connector.
However, the loop current did cause some problems when the connection
was installed.

Paul

On 2008-07-08, Paul Keinanen <k...@sci.fi> wrote:

> Using galvanic isolation between sites A and B, I have never
> noticed any RS-422/486 problems due to the removal of the SG
> connector.

I suppose it's possible that the galvanic isolation in
equipment I've used is slightly "leaky" and both ends weren't
completely floating.

> However, the loop current did cause some problems when the
> connection was installed.

If they're both glavanically isolated, how can there be loop
current?

-- 
Grant Edwards                   grante             Yow! I am NOT a nut....
                                  at               
                               visi.com            

On Tue, 08 Jul 2008 14:25:34 -0500, Grant Edwards <g...@visi.com>
wrote:

>On 2008-07-08, Paul Keinanen <k...@sci.fi> wrote:
>
>> Using galvanic isolation between sites A and B, I have never
>> noticed any RS-422/486 problems due to the removal of the SG
>> connector.
>
>I suppose it's possible that the galvanic isolation in
>equipment I've used is slightly "leaky" and both ends weren't
>completely floating.
>
>> However, the loop current did cause some problems when the
>> connection was installed.
>
>If they're both glavanically isolated, how can there be loop
>current?

Unless you really understand what you are doing, using galvanic
isolation on the equipment level _might_ still cause a lot of problems
at the system level, when cable shields are connected to points, in
which they should not be connected. 

Paul

On Tue, 08 Jul 2008 14:25:34 -0500, Grant Edwards <g...@visi.com>
wrote:

>On 2008-07-08, Paul Keinanen <k...@sci.fi> wrote:
>
>> Using galvanic isolation between sites A and B, I have never
>> noticed any RS-422/486 problems due to the removal of the SG
>> connector.
>
>I suppose it's possible that the galvanic isolation in
>equipment I've used is slightly "leaky" and both ends weren't
>completely floating.

As you can see, it has to be pretty `leaky` for the signals to go
outside the CM range of +- 7VDC. 

http://komponenten.es.aau.dk/fileadmin/komponenten/Data_Sheet/Interface/MC3486.pdf


>> However, the loop current did cause some problems when the
>> connection was installed.
>
>If they're both glavanically isolated, how can there be loop
>current?

The + and - outputs are not isolated from each other. ;-) 
Best regards, 
Spehro Pefhany
-- 
"it's the network..."                          "The Journey is the reward"
s...@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

Grant Edwards wrote:
> Paul Keinanen <k...@sci.fi> wrote:
>> Grant Edwards <g...@visi.com> wrote:
>>
... snip ...
>>
>>> Wrong -- take a look at the receiver's spec for max common-mode
>>> voltage.  It's usually only 8-12V. You've got to have a common
>>> ground to make sure you don't exceed the receiver common-mode
>>> voltage spec.
>>
>> And what would a flimsy signal ground wire help to reduce the
>> common mode voltage in mains powered systems?
> 
> I'm not really sure what you're asking.  In the systems I'm
> familiar with, RS-422 drivers and receivers are usually
> galvanically isolated from mains power, and signal ground isn't
> always connected to either case or earth ground.

Without some form of common ground, there is absolutely no control
on the common mode voltage applied.  This is how we build
Van-de-Graafs.

-- 
 [mail]: Chuck F (cbfalconer at maineline dot net) 
 [page]: <http://cbfalconer.home.att.net>;
            Try the download section.

On Jul 8, 4:44 am, Paul Keinanen <keina...@sci.fi> wrote:
> On Mon, 07 Jul 2008 13:53:32 -0500, Grant Edwards <gra...@visi.com>
> wrote:
>
> >On 2008-07-07, Paul Keinanen <keina...@sci.fi> wrote:
> >> On Mon, 07 Jul 2008 12:15:17 -0500, "svyamuna" <svyam...@yahoo.co.in>
> >> wrote:
>
> >>>I am trying to interface two systems through RS422 link. There are only
> >>>four lines connected between them (tx+/- & Rx+/-) and no ground.
>
> >> This is quite OK as long as there is a termination resistor (typically
> >> 100..120 ohms) between the Tx+ and Tx- lines and also between the Rx+
> >> and Rx- lines. The ground connection is not needed,
>
> >Wrong -- take a look at the receiver's spec for max common-mode
> >voltage.  It's usually only 8-12V. You've got to have a common
> >ground to make sure you don't exceed the receiver common-mode
> >voltage spec.
>
> And what would a flimsy signal ground wire help to reduce the common
> mode voltage in mains powered systems ?
>
> To understand the situation you have to look how different earthing
> systems work e.g.http://en.wikipedia.org/wiki/Earthing_system
>
> In a TN-S system with separate protective earth (PE) and neutral (N)
> no current is flowing in normal situations in the PE conductor and
> hence all device cases (and the RS-422 transceivers are usually
> connected to the case) are at the same potential. In this situation
> using a separate signal ground does not make a difference.
>
> For TN-C-S systems, the situation is the same, provided that the two
> communicating devices A and B are after  the same split point where
> the PEN conductor is split into PE and N conductors.
>
> Looking at the TN-C system, in which there is only a single PEN
> conductor carrying the neutral current as well as connecting the
> equipment case (and usually the RS-422 ground reference), the
> situation is more complicated. The current flowing in the PEN
> conductor (typically several amps in a non-balanced three phase
> system) will cause a voltage drop between mains sockets A and B and
> hence also a reference potential difference in the RS-422 transceivers
> for equipment A and B, which can be several volts.
>
> If we connect a signal ground (SG) between equipment A and B, this
> will create an additional current path parallel to the PEN conductor
> and the neutral wire current is shared inversely proportional to the
> resistance of the PEN conductor and the SG conductor.
>
> In order to significantly reduce the ground potential difference
> between equipments A and B, the SG conductor would have to carry a
> significant part of the PEN current (say 1/2) thus the SG conductor
> cross section area should be similar to the PEN conductor.
>
> If you make the SG conductor that thick there may flow several amps of
> 50/60 Hz neutral current and all their harmonics and other noise
> currents through the SG wire as well as flow within the equipment
> (often through narrow PCB tracks) into the device case, which can have
> harmful effects or even burn a narrow PCB track.
>
> In a TN-C system, the separate SG wire does not solve any problems,
> but creates some new problems. If the ground potential difference is
> larger than the RS-422 transceiver common mode voltage range, the only
> alternative is to use a floating (galvanic isolation) system. In a
> floating system, the "fail-safe" pull -up/down resistors will pull the
> internal signal ground into the middle of the transceiver common mode
> voltage range, so there is no need for a SG wire.
>
> >> as the RS-422 works as a balanced current loop.
>
> >Wrong -- it's a differential-mode voltage signal.  It's not a
> >current loop signal.
>
> You can analyze the RS-422 system as a voltage driven system, as a
> bipolar current loop or as a transmission line with impedance matching
> to the transmission line.
>
> To understand why the SG is not required, it is easiest to look at as
> a current loop. The standard requires that the receiver voltage is at
> least +/-200 mV and assuming there is a 100 resistor at the receiver,
> at least +/- 2 mA will flow in this current loop from Tx+, through the
> termination resistor (at the receiver end) and back to Tx-. With a 5 V
> transceiver, the maximum current is below +/-50 mA.
>
> Does the unipolar 0/20 mA current loop need a separate signal ground ?
> Why would the bipolar RS-422 current loop need a signal ground ?
>
> Paul

I don't want to be argumentative, but you have make a lot of
assumptions in your discussion.  For example, you assume that the
chassis of both units "are at the same potential".  If that were true,
the OPs circuit should be working.  If they are not at the same
potential, then the reason for that needs to be identified.  Your
discussion makes assumptions for why that is and makes the assumption
that the chassis and/or the electronics are grounded to the power
ground.  Many systems are isolated just for this reason, to prevent
power supply current from changing the ground level of the circuit.

In your discussion of RS-422 as a current loop without a ground, you
don't mention what voltage is present on the conductors relative to
the associated ground, only the current through them. If you ignore
this voltage, you run the risk of blowing out the chips or at least
making them work improperly.  A common ground must be established
between an RS-422 driver and receiver.  How to best do this depends on
the equipment and the grounding method used.  But to say the common
ground conductor is not required is a significant
oversimplification.

Rick

On Jul 8, 12:40 am, "svyamuna" <svyam...@yahoo.co.in> wrote:
>I forgot to mention one thing, one unit is deriving power from the other
>
> unit i.e. the slave unit is powered up from the master. in such a case is
> it required to connect a gnd line between them? But i have even tried this
> out by connecting one more gnd line but that doesnt solve my problem.
> yamuna


You need to explain better how the power connection is made between
them and how the RS-422 chips are grounded inside the separate
systems.  Is there already a common ground between the two systems and
what current is flowing through this ground?  Is there enough current
to produce a significant voltage difference in grounds?

Rick