Maximal length of a serial (V.24/RS232) cable?

Maximal length of a serial (V.24/RS232) cable?

Post by Christian Brac » Mon, 06 Dec 1999 04:00:00



Hello,

does anybody know, how long a serial (RS232) cable between computer and
modem could maximal be to ensure no communication errors?

Thanx in advance for any answer

Regards
 Christian

 
 
 

Maximal length of a serial (V.24/RS232) cable?

Post by Hooda Ges » Mon, 06 Dec 1999 04:00:00



>Hello,

>does anybody know, how long a serial (RS232) cable between computer and
>modem could maximal be to ensure no communication errors?

It would depend on the quality of the cable but, generally, under 50 feet is
fine. Over that and you might start getting problems.

Hooda

 
 
 

Maximal length of a serial (V.24/RS232) cable?

Post by budgi » Tue, 07 Dec 1999 04:00:00





>>Hello,

>>does anybody know, how long a serial (RS232) cable between computer and
>>modem could maximal be to ensure no communication errors?

>It would depend on the quality of the cable but, generally, under 50 feet is
>fine. Over that and you might start getting problems.

with the reservation that waveform degradation introduces errors as
length increases, along with longitudinal frame voltage differences
appearing on the signal ground line.

The original standard (sic) quoted a max distance of 50 ft IIRC, but
this was to be read the other way around - devices HAD TO ACHIEVE this
distance at the prescribed baud rate under worst case drive
conditions.  As the baud rate increases the length capability
decreases, this is a fact of nature/physics.

Having said that, back in my undergraduate engineering days we sent
9600 baud (that was fast then) over standard outdoor telephone cable
some 500 metres across campus.

The basic nature of RS232 reception is for bit sampling to take place
at 8/16ths of the bit time after the leading edge.  As the edge is
usually defined as a fixed voltage and not at half-swing, this can
lead to an upper limit also.

Back to Hooda's point, 50 feet should pose no problem with unshielded
twisted pair (not necessarily Cat3/5) cable in a clean environment..
Shielding will extend the range.   Straight 25-way ribbon has shown
errors at 2400 across a room in a noisy environment.

HTH

Peter

 
 
 

Maximal length of a serial (V.24/RS232) cable?

Post by Hooda Ges » Tue, 07 Dec 1999 04:00:00






>>>Hello,

>>>does anybody know, how long a serial (RS232) cable between computer and
>>>modem could maximal be to ensure no communication errors?

>>It would depend on the quality of the cable but, generally, under 50 feet
is
>>fine. Over that and you might start getting problems.

>with the reservation that waveform degradation introduces errors as
>length increases, along with longitudinal frame voltage differences
>appearing on the signal ground line.

>The original standard (sic) quoted a max distance of 50 ft IIRC, but
>this was to be read the other way around - devices HAD TO ACHIEVE this
>distance at the prescribed baud rate under worst case drive
>conditions.  As the baud rate increases the length capability
>decreases, this is a fact of nature/physics.

>Having said that, back in my undergraduate engineering days we sent
>9600 baud (that was fast then) over standard outdoor telephone cable
>some 500 metres across campus.

>The basic nature of RS232 reception is for bit sampling to take place
>at 8/16ths of the bit time after the leading edge.  As the edge is
>usually defined as a fixed voltage and not at half-swing, this can
>lead to an upper limit also.

>Back to Hooda's point, 50 feet should pose no problem with unshielded
>twisted pair (not necessarily Cat3/5) cable in a clean environment..
>Shielding will extend the range.   Straight 25-way ribbon has shown
>errors at 2400 across a room in a noisy environment.

Good points, Peter. I was thinking only of standard modem cable and not flat
or ribbon cable with no twisting. We used to connect up to 500 feet on
serial connections (system to printer/terminals) but that was at 1200bps.
Since the null modem connection would likely be at 115200bps, I figure
dividing by a factor of 10 would be prudent.

Hooda

 
 
 

Maximal length of a serial (V.24/RS232) cable?

Post by Floyd Davids » Tue, 07 Dec 1999 04:00:00




>>It would depend on the quality of the cable but, generally,
>>under 50 feet is fine. Over that and you might start getting
>>problems.

At 9600 bps.  At higher speeds there is no guarantee that
even 50 feet is possible without errors, though it usually
does work at well beyond that.

Quote:>The basic nature of RS232 reception is for bit sampling to take
>place at 8/16ths of the bit time after the leading edge.  

That is not inherent in RS232.  It is the basic nature of
current serial IO chip sets.  (A rather pedantic point.)

Quote:>As the edge is usually defined as a fixed voltage and not at
>half-swing, this can lead to an upper limit also.

And that is significant!

Quote:>Back to Hooda's point, 50 feet should pose no problem with unshielded
>twisted pair (not necessarily Cat3/5) cable in a clean environment..
>Shielding will extend the range.   Straight 25-way ribbon has shown
>errors at 2400 across a room in a noisy environment.

Whoa.  Shielding will increase the cable capacitance, and thus
will _reduce_ the range.  It may increase the noise immunity
though, but proper implentation is critical and not commonly
understood.  Noise immunity and cable capacitance are two
distinctly separate issues even if the user might not see a
distinction in the results.  

Twisted pair is not recommended for RS-232 use unless each
active wire is not paired with another signal wire that will
experience transistions.  Hence just shorting the tip/ring wires
together and using them as a single conductor would probably
increase the range, while putting receive data and transmit data
on tip and ring of the same twisted pair would probably greatly
reduce the range.

To summarize:

  1) The distance depends on the cable and the bit rate.  

  2) Lower capacitance and lower resistance will increase
     the useful distance at a given bit rate.

  3) Shielding may enhance noise immunity, but will otherwise
     reduce the distance at a given bit rate for any given
     cable design.

  4) Twisting signal pairs will reduce the distance.

For extended distances, high reliability, noisy environments, or
high bit rates it is recommended that an interface other than
RS-232 be used.  RS-422, RS-423, and V.35 should be investigated
if indicated.

  Floyd

--

Ukpeagvik (Barrow, Alaska)

 
 
 

Maximal length of a serial (V.24/RS232) cable?

Post by budgi » Wed, 08 Dec 1999 04:00:00





(snip)
>>The basic nature of RS232 reception is for bit sampling to take
>>place at 8/16ths of the bit time after the leading edge.  

>That is not inherent in RS232.  It is the basic nature of
>current serial IO chip sets.  (A rather pedantic point.)

It's not written into the standard, but it was the practice back as
far as the pre-IC interfaces in the '60s, through the 1489 and on to
today's VLSI et al.  Repeating, it's the basic nature of RS232
reception.

(snip)

Quote:>>Back to Hooda's point, 50 feet should pose no problem with unshielded
>>twisted pair (not necessarily Cat3/5) cable in a clean environment..
>>Shielding will extend the range.   Straight 25-way ribbon has shown
>>errors at 2400 across a room in a noisy environment.

>Whoa.  Shielding will increase the cable capacitance, and thus
>will _reduce_ the range.  It may increase the noise immunity
>though, but proper implentation is critical and not commonly
>understood.  Noise immunity and cable capacitance are two
>distinctly separate issues even if the user might not see a
>distinction in the results.  

Cable capacitance will certainly set an upper speed limit.  Noise will
blow the living s##t out of a system.  I was involved a couple of
years back on a NC system in a sheet metal shop with welders and all
that nice stuff.  In this real world environment, you couldn't make
50ft at 1200 baud.  Using shielded cable (selected for low shunt cap)
we could run 19200 the length of the shop error free.  The exercise is
not only understandig how these factors are trade-offs, but also in
understanding the receiver comparator operation in a noisy
environment.  And it doesn't need to be a sheet-fab to be very noisy.
That ribbon cable case was twelve feet from a '286 to a Calcomp
plotter in an office environment.

Quote:>Twisted pair is not recommended for RS-232 use unless each
>active wire is not paired with another signal wire that will
>experience transistions.  Hence just shorting the tip/ring wires
>together and using them as a single conductor would probably
>increase the range, while putting receive data and transmit data
>on tip and ring of the same twisted pair would probably greatly
>reduce the range.

Probably

Quote:>To summarize:

>  1) The distance depends on the cable and the bit rate.  

Absolutely

Quote:>  2) Lower capacitance and lower resistance will increase
>     the useful distance at a given bit rate.

Absolutely, but the cable selection plays a BIG part in capacitance,
and one shouldn't become paranoid about capacitance.  It isn't the
biggest problem.

Quote:>  3) Shielding may enhance noise immunity, but will otherwise
>     reduce the distance at a given bit rate for any given
>     cable design.

Emphasis on "for any given cable design".  Again, pick your cable for
the task.

Quote:>  4) Twisting signal pairs will reduce the distance.

I have yet to see any cogent argument, evidence or experiance that
would support this view.  It isn't going to increase the pair cap, and
it WILL improve rejection of noise somewhat, even though it is still
an unbalanced TL.

Quote:>For extended distances, high reliability, noisy environments, or
>high bit rates it is recommended that an interface other than
>RS-232 be used.  RS-422, RS-423, and V.35 should be investigated
>if indicated.

Absolutely, but they will continue to be rare in SOHO's.

Peter

 
 
 

Maximal length of a serial (V.24/RS232) cable?

Post by Floyd Davids » Wed, 08 Dec 1999 04:00:00





>>>Back to Hooda's point, 50 feet should pose no problem with unshielded
>>>twisted pair (not necessarily Cat3/5) cable in a clean environment..
>>>Shielding will extend the range.   Straight 25-way ribbon has shown
>>>errors at 2400 across a room in a noisy environment.

>>Whoa.  Shielding will increase the cable capacitance, and thus
>>will _reduce_ the range.  It may increase the noise immunity
>>though, but proper implentation is critical and not commonly
>>understood.  Noise immunity and cable capacitance are two
>>distinctly separate issues even if the user might not see a
>>distinction in the results.  

>Cable capacitance will certainly set an upper speed limit.  Noise will
>blow the living s##t out of a system.  I was involved a couple of
>years back on a NC system in a sheet metal shop with welders and all
>that nice stuff.  In this real world environment, you couldn't make
>50ft at 1200 baud.  Using shielded cable (selected for low shunt cap)
>we could run 19200 the length of the shop error free.  The exercise is
>not only understandig how these factors are trade-offs, but also in
>understanding the receiver comparator operation in a noisy
>environment.  And it doesn't need to be a sheet-fab to be very noisy.
>That ribbon cable case was twelve feet from a '286 to a Calcomp
>plotter in an office environment.

Your ribbon cable clearly had a severe problem that is not
being mentioned and is probably unknown.  Granted that ribbon
cable is very close to the worst possible case, but the fact is
you would almost have to wrap it around a fluorescent lamp
to have a problem with a twelve foot length.  I've seen greater
than 50 foot runs of ribbon cable work at 9600 bps in a telephone
office, tied right in the cable racks along with more noise
radiating garbage than any "office environment", and it functioned
quite well for years.  Actually, I have some of that in the
room I'm in at the moment!

On the other hand, the shop environment described and the effort
needed to overcome the effects of noise is not at all uncommon.
And as you say, shielded wire does have its uses.

Quote:>>  3) Shielding may enhance noise immunity, but will otherwise
>>     reduce the distance at a given bit rate for any given
>>     cable design.

>Emphasis on "for any given cable design".  Again, pick your cable for
>the task.

But if the length is approaching critical distances, which the
subject of this thread implies , shielding is not recommended.

Quote:>>  4) Twisting signal pairs will reduce the distance.

>I have yet to see any cogent argument, evidence or experiance that
>would support this view.  It isn't going to increase the pair cap, and
>it WILL improve rejection of noise somewhat, even though it is still
>an unbalanced TL.

It will never improve noise rejection.  Instead it will degrade
it.  The two twisted wires are coupled more tightly than two
wires in a non-twisted cable (which would also tend to use
thicker insulation specifically to separate the conductors
farther, as opposed to twisted pairs where closer proximity is
made up for by noise immunity due to common mode rejection
provided by balanced circuits).  Hence, if the transmit signal
and the receive signal, for example, are place on the tip and
ring of a twisted pair, the spike from each signal transition on
each wire will be induced into the paired wire at a much higher
level than it will be into other conductors in the cable.
Basically it will suffer greatly from near end crosstalk.  

Given that the transmit and receive data lines are likely to
have the greatest number of signal transitions (compared for
example to a ground lead, the RTS or CTS leads, etc. etc.), they
are the two absolutely worst signals to put together on a
twisted pair.

If no other cable is available, pairing signals that have the
fewest transitions with the data leads is recommended if
necessary, but the best solution is to tie the tip and ring
together on each pair and make a single, lower resistance,
conductor.   Obviously that only works if enough conductors
are available.

Quote:>>For extended distances, high reliability, noisy environments, or
>>high bit rates it is recommended that an interface other than
>>RS-232 be used.  RS-422, RS-423, and V.35 should be investigated
>>if indicated.

>Absolutely, but they will continue to be rare in SOHO's.

They are not so rare in places where competent designs are used.

  Floyd

--

Ukpeagvik (Barrow, Alaska)

 
 
 

Maximal length of a serial (V.24/RS232) cable?

Post by budgi » Thu, 09 Dec 1999 04:00:00




Quote:>>Cable capacitance will certainly set an upper speed limit.  Noise will
>>blow the living s##t out of a system.  I was involved a couple of
>>years back on a NC system in a sheet metal shop with welders and all
>>that nice stuff.  In this real world environment, you couldn't make
>>50ft at 1200 baud.  Using shielded cable (selected for low shunt cap)
>>we could run 19200 the length of the shop error free.  The exercise is
>>not only understandig how these factors are trade-offs, but also in
>>understanding the receiver comparator operation in a noisy
>>environment.  And it doesn't need to be a sheet-fab to be very noisy.
>>That ribbon cable case was twelve feet from a '286 to a Calcomp
>>plotter in an office environment.

>Your ribbon cable clearly had a severe problem that is not
>being mentioned and is probably unknown.

Sorry to break the news, but we fabricated several more because the
client wanted under-carpet and we needed nore than TD/RD/SG.  All gave
the same problem, new or downtrodden.

Quote:> Granted that ribbon
>cable is very close to the worst possible case, but the fact is
>you would almost have to wrap it around a fluorescent lamp
>to have a problem with a twelve foot length.

I can only disagree with that surmise.

Quote:> I've seen greater
>than 50 foot runs of ribbon cable work at 9600 bps in a telephone
>office, tied right in the cable racks along with more noise
>radiating garbage than any "office environment", and it functioned
>quite well for years.  Actually, I have some of that in the
>room I'm in at the moment!

Hey, I have seen that too.  One lucky shot doesn't make a marksman,
and one lousy shot doesn't prove anything either.

Quote:>>>  3) Shielding may enhance noise immunity, but will otherwise
>>>     reduce the distance at a given bit rate for any given
>>>     cable design.

>>Emphasis on "for any given cable design".  Again, pick your cable for
>>the task.

>But if the length is approaching critical distances, which the
>subject of this thread implies , shielding is not recommended.

No, the original post did not state he was approaching any threshold
distance - rather enquiring as to what the ground rules and
expectaions were.

Quote:>>>  4) Twisting signal pairs will reduce the distance.

>>I have yet to see any cogent argument, evidence or experiance that
>>would support this view.  It isn't going to increase the pair cap, and
>>it WILL improve rejection of noise somewhat, even though it is still
>>an unbalanced TL.

>It will never improve noise rejection.  Instead it will degrade
>it.  The two twisted wires are coupled more tightly than two
>wires in a non-twisted cable (which would also tend to use
>thicker insulation specifically to separate the conductors
>farther, as opposed to twisted pairs where closer proximity is
>made up for by noise immunity due to common mode rejection
>provided by balanced circuits).

there you go again, comparing apples with polar bears.  Now watch my
lips and say slowly after me - For a given cable construction.  Got it
now?

Quote:>>>For extended distances, high reliability, noisy environments, or
>>>high bit rates it is recommended that an interface other than
>>>RS-232 be used.  RS-422, RS-423, and V.35 should be investigated
>>>if indicated.

>>Absolutely, but they will continue to be rare in SOHO's.

>They are not so rare in places where competent designs are used.

They will only be used in SOHO's when there has been evidenced a need.
It is not a case of competent design.  If they are used in the absence
of need, is this competence or over-design?

Peter

 
 
 

Maximal length of a serial (V.24/RS232) cable?

Post by Floyd Davids » Thu, 09 Dec 1999 04:00:00




>>>>  4) Twisting signal pairs will reduce the distance.
...
>>It will never improve noise rejection.  Instead it will
>>degrade it.  The two twisted wires are coupled more tightly
>>than two wires in a non-twisted cable (which would also tend
>>to use thicker insulation specifically to separate the
>>conductors farther, as opposed to twisted pairs where closer
>>proximity is made up for by noise immunity due to common mode
>>rejection provided by balanced circuits).

>there you go again, comparing apples with polar bears.  Now
>watch my lips and say slowly after me - For a given cable
>construction.  Got it now?

Here's an apple for ya, pard,

It will *never* improve noise rejection.  Instead it _will_
degrade it.

If you don't understand that, it will be a bear until you do.

Could you substantiate your earlier statement that "it WILL
improve rejection of noise somewhat, even though it is still an
unbalanced TL."  What noise rejection improvement will there be?
Please describe the mechanism.

You also stated, "It isn't going to increase the pair cap".  Can
you enlighten us about how two wires can be twisted together and
not increase the capacitance between them?

Of course, "for any given cable construction"...  but it is
reasonable to assume cables that are actually available for use.
For example, to avoid induction between circuits, cables
intended to be unbalanced transmission lines are constructed to
maintain a relatively large physical spacing between any two
conductors.  On the other hand, to accomplish the same results,
construction of a twisted pair cable will provide different
twists for different pairs to reduce induction.  Twisted pair
cable, to counter the physical bulk increase caused by the
twist, will use thinner insulation, which is effective because
common mode rejection is the primary noise reduction mechanism
for balanced circuits and the capacitance between conductors
does not cause the significant increase in noise which would be
seen with a non-balanced circuit.

Of course, attempting to match twisted pair to non-balanced
circuits or non-twisted pair to balanced circuits negates all of
the optimizations in cable construction.  It is not recommended.

  Floyd

--

Ukpeagvik (Barrow, Alaska)

 
 
 

Maximal length of a serial (V.24/RS232) cable?

Post by budgi » Fri, 10 Dec 1999 04:00:00




Quote:>You also stated, "It isn't going to increase the pair cap".  Can
>you enlighten us about how two wires can be twisted together and
>not increase the capacitance between them?

You first - you explain how twisting two wires will increase the
coupling between them.  Any transmission line model you use will
reject your theory.

Quote:>Of course, "for any given cable construction"...

Yes, that's what I was talking about - you recall?

Quote:> but it is
>reasonable to assume cables that are actually available for use.

It is.

Quote:>For example, to avoid induction between circuits, cables
>intended to be unbalanced transmission lines are constructed to
>maintain a relatively large physical spacing between any two
>conductors.

Correctamente.

Quote:>On the other hand, to accomplish the same results,
>construction of a twisted pair cable will provide different
>twists for different pairs to reduce induction.

Right again.

Quote:>Twisted pair
>cable, to counter the physical bulk increase caused by the
>twist, will use thinner insulation.

You may find that "bulk" is not the real driving force behind multiple
twisted pair cables using thinner dielectric.

Quote:> which is effective because
>common mode rejection is the primary noise reduction mechanism
>for balanced circuits
right
> and the capacitance between conductors
>does not cause the significant increase in noise which would be
>seen with a non-balanced circuit.

Exposed to an external noise source (where else?) why would the pair
cap cause ANY increase in noise?

Quote:>Of course, attempting to match twisted pair to non-balanced
>circuits or non-twisted pair to balanced circuits negates all of
>the optimizations in cable construction.  It is not recommended.

That optimisation was for the intended purpose indeed, but that
doesn't preclude cable selection for a non-designed purpose, nor does
it automatically render that selection/use either invalid, unworkable
or unwise.  You must sometime in your career think outside the
box/envelope.
 
 
 

Maximal length of a serial (V.24/RS232) cable?

Post by Floyd Davids » Fri, 10 Dec 1999 04:00:00


You seem to have clipped out the first question, which was
posed in relation to using single ended RS-232 circuits on
twisted pair cable:

   "Could you substantiate your earlier statement that "it
    WILL improve rejection of noise somewhat, even though
    it is still an unbalanced TL."  What noise rejection
    improvement will there be?  Please describe the
    mechanism."

Please do describe the mechanism, or admit there is none.

Quote:>>You also stated, "It isn't going to increase the pair cap".  Can
>>you enlighten us about how two wires can be twisted together and
>>not increase the capacitance between them?

>You first - you explain how twisting two wires will increase the
>coupling between them.  Any transmission line model you use will
>reject your theory.

Simple fact: it happens.

If you twist the wires it takes more wire to go the same
distance, with a higher twist rate requiring more original
conductor length to obtain the same twisted pair length.  That
alone will increase the coupling for any given distance of cable
when comparing two conductor non-twisted pair cable only.  But
we are talking about multi-pair cable, and in that case there is
even more reason.  If the two conductors are not twisted, they
should be constructed such that proximity to all pairs in the
cable becomes roughly equal over some minimum distance.  With
twisted pair cable, each pair is treated the same, which is why
the bundle is swirled and each pair has a different twist.  It
basically means two conductors should not lay next to each other
for the entire distance of the cable.  But with twisted pair
that is exactly what happens for those two conductors.  When
they are used with a balanced circuit that is beneficial, but
with a different single ended circuit on each conductor it is
clearly detrimental.

Hence the coupling between pairs of a twisted pair are higher
than between either of those conductors and any other conductor
in the cable, as originally stated.

If you would like to explain how you can twist two pairs
together and not cause that effect, be my guest.

Quote:>>Twisted pair
>>cable, to counter the physical bulk increase caused by the
>>twist, will use thinner insulation.

>You may find that "bulk" is not the real driving force behind multiple
>twisted pair cables using thinner dielectric.

But you aren't going to grace us with your concept of what that
might be???  In fact, physical size is in fact the primary benefit
of thinner insulation.  Every other effect can be accomplished in
other ways, but not a reduction in size.

Quote:>> which is effective because
>>common mode rejection is the primary noise reduction mechanism
>>for balanced circuits

>right

>> and the capacitance between conductors
>>does not cause the significant increase in noise which would be
>>seen with a non-balanced circuit.

>Exposed to an external noise source (where else?) why would the pair
>cap cause ANY increase in noise?

If you put the receive data signal on the tip and the transmit
data signal on the ring of a twisted pair, each signal is
nothing but noise to the other.  If those signals are placed on
conductors with less coupling, there will be less "noise" for
each signal.  That is true because with the exception of clock
signal lines, which are usually not used with asynchronous
RS-232 anyway, the two data lines are the most active as far as
the rate of signal transitions.

Quote:>>Of course, attempting to match twisted pair to non-balanced
>>circuits or non-twisted pair to balanced circuits negates all of
>>the optimizations in cable construction.  It is not recommended.

>That optimisation was for the intended purpose indeed, but that
>doesn't preclude cable selection for a non-designed purpose, nor does
>it automatically render that selection/use either invalid, unworkable
>or unwise.

It does if one doesn't have an extremely broad based understanding
of cables.  Lacking that, using cable designed for the specific
purpose will avoid the type of errors your understanding of cable
use and abuse will cause.

Quote:>  You must sometime in your career think outside the
>box/envelope.

A lot of my career has been spent tracking down the cause of and
correcting unfortunate mistakes made by people who didn't know
where the box/envelope was, much less have any idea when their
thinking was either inside or outside of it.  Unfortunately,
your perception of cable functionality fits that description,
and tends to leave those kinds of problems for trouble shooters
to correct at some later date when small problems have grown to
significant proportions.

  Floyd

--

Ukpeagvik (Barrow, Alaska)

 
 
 

1. T1 -> 24 x v.32 -> RS232

A company called Primary Access has a product that will take a T1 (24
VOICE channels) and interpret the DS0 channels as modem connections
(v.32, v.42bis, etc..) and output standard RS232 to hook to a system.

What other companies have something like this?  Comments on their
products?  Contacts to get more information?

PLEASE EMAIL me. I don't have Usenet at home right now; only email.

Thank you!


2. Palm Vx Hotsync after V

3. dropouts on QuickComm Spir

4. proxy slows down?

5. Is V.24 exactely the same as RS232?

6. Where is Atari PD

7. rs232 max cable length?

8. The maximal length of 256 Kbps V.35 interface

9. Smurf amplifier attacks involving 24.6.230.169, 24.8.22.81, 24.1.13.39, 24.13.156.131, 24.8.18.47

10. Maximal length of 100Base-FX

11. "Processed an NCP with a bad subfunction length, station XX (NCP 23 24)"