DComTalk.com

Can a single Cat5e cable be used for both computer networking and for a
telephone line? I ask because I read somewhere that for computer
networking "only pair 2 and pair 3 are actually used, and pair 1 and
pair 4 are free". Our office network consists of three computers. Two
are currently connected to a D-Link DI-604 router. Internet access is
by DirecWay (we are in a VERY rural area), which is plugged into the
D-Link router. The third computer is in a separate building to which
unfortunately we buried only a single Cat5e line. So, to avoid digging
a new trench, we are hoping we can connect the third computer to the
D-Link router and to a phone line (it is not a fancy phone system, no
extensions, just a dial tone) using the single Cat5e cable. The network
cables are configured according to the T568B standard. I will greatly
appreciate any help or ideas. (If I am posting this question in the
wrong forum, I apologize.) Thanks. Peter

On 2005-09-03, zoofence@gmail.com <zoofence@gmail.com> wrote:

Can a single Cat5e cable be used for both computer networking and for a
telephone line? I ask because I read somewhere that for computer
networking "only pair 2 and pair 3 are actually used, and pair 1 and
pair 4 are free".

Yep, no problem. I have my cat5e running with 1 network and 2 phone
lines, utilizing all 4 pair in one cable.

--
This is my .sig

Mike Hunt wrote:

Yep, no problem. I have my cat5e running with 1 network and 2 phone
lines, utilizing all 4 pair in one cable.

What a relief! Thank you.

zoofence@gmail.com wrote:

Can a single Cat5e cable be used for both computer networking and for a
telephone line?

Yes you can, except for gigabit.

zoofence@gmail.com writes:

Can a single Cat5e cable be used for both computer networking and for a
telephone line?

In very many cases this works well but it is not a recommended practice.

I ask because I read somewhere that for computer
networking "only pair 2 and pair 3 are actually used, and pair 1 and
pair 4 are free".

MIst commonly used networkign technologies use only to pairs.
Examples of such technologies are 10Base-T and 100Base-TX Ethernet.

But keep in mind that there are also other ntworkign technologies
that can use different pairs and even need all pairs
(for example 1 Gbit/s Ethernet needs all four pairs).

Our office network consists of three computers. Two
are currently connected to a D-Link DI-604 router. Internet access is
by DirecWay (we are in a VERY rural area), which is plugged into the
D-Link router. The third computer is in a separate building to which
unfortunately we buried only a single Cat5e line. So, to avoid digging
a new trench, we are hoping we can connect the third computer to the
D-Link router and to a phone line (it is not a fancy phone system, no
extensions, just a dial tone) using the single Cat5e cable. The network
cables are configured according to the T568B standard. I will greatly
appreciate any help or ideas. (If I am posting this question in the
wrong forum, I apologize.) Thanks. Peter

I think this would work without problems if the line you have
comes form the local PBX device. Most propably things work
for most time even from directly outgoing line with the same
cable, but I would be a little worried about situation
when lightning strikes nearby which can cause overvoltage
on telephone line.. consider overvoltage protection
on the place where telephone line enters the building.

--
Tomi Engdahl (http://www.iki.fi/then/)
Take a look at my electronics web links and documents at
http://www.epanorama.net/

Tomi Holger Engdahl wrote:

Can a single Cat5e cable be used for both computer networking and for a
telephone line?

In very many cases this works well but it is not a recommended practice.

Twisted pair ethernet was designed to share existing telephone cables.

James Knott <james.knott@rogers.com> wrote:

Twisted pair ethernet was designed to share existing telephone cables.

Yes. But 10baseT, not 100baseTX (which still usually works).

-- Robert

Robert Redelmeier wrote:

James Knott <james.knott@rogers.com> wrote:
Twisted pair ethernet was designed to share existing telephone cables.

Yes. But 10baseT, not 100baseTX (which still usually works).

Given the great difference in frequencies involved. There's unlikely to be
much interference, between an analog voice line and ethernet. Most of the
voice energy is below 3 KHz. Ethernet signaling is well above that.

James Knott <james.knott@rogers.com> writes:

Robert Redelmeier wrote:

James Knott <james.knott@rogers.com> wrote:
Twisted pair ethernet was designed to share existing telephone cables.

Yes. But 10baseT, not 100baseTX (which still usually works).

Given the great difference in frequencies involved. There's unlikely to be
much interference, between an analog voice line and ethernet. Most of the
voice energy is below 3 KHz. Ethernet signaling is well above that.

You are right that most of the voice below 3 kHz and
the signal level is quite low.
And Ethernet main energy in 10Base-T is somewhere at
5-20 MHz range.
The most problematic I quess could be the ring signal,
that is typically around 90V AC 25 Hz.
Generally you can expect very good network performance with propably
very temporary problems (possible on telephone ring, on/off-hook,
pulse dialling etc. if the cable is not very good).

10Base-T is very robust and practically works well with
even where same cables are shared with Ethernet and telephone
signals. There are even products that run 10Base-T Ethernet and
telephone signals on the physically same wires inside the cable
(there are several companies that make such special adapter
products).

100Base-TX runs at much higher rate and uses more compplicated
coding system that is more sensitive to the noise errors.

I have done soma laboratory expriments related to both of those
Ethernet systems. With a sutiable filters I could successfully
run both 10Base-T Ethernet signals and telephone signals on the
same wire pair. I posted some news articles related to this
aroudn on eyear ago to one Finnish newsgroup...
Doing that with 100Base-TX did not work that well...

--
Tomi Engdahl (http://www.iki.fi/then/)
Take a look at my electronics web links and documents at
http://www.epanorama.net/

James Knott <james.knott@rogers.com> wrote:

Robert Redelmeier wrote:

James Knott <james.knott@rogers.com> wrote:
Twisted pair ethernet was designed to share existing telephone cables.

Yes. But 10baseT, not 100baseTX (which still usually works).

Given the great difference in frequencies involved. There's unlikely to be
much interference, between an analog voice line and ethernet. Most of the
voice energy is below 3 KHz. Ethernet signaling is well above that.

Granted. But the _design_ was for POTS & 10baseT shared-sheath.

-- Robert

The far bigger issue is.. the phone line goes outside...
and it brings in Murphy's messenger -- lightning...

You may like lightning on your LAN....I'll pass.

--
A host is a host from coast to coast.................wb8foz@nrk.com
& no one will talk to a host that's close........[v].(301) 56-LINUX
Unless the host (that isn't close).........................pob 1433
is busy, hung or dead....................................20915-1433

Tomi Holger Engdahl wrote:

The most problematic I quess could be the ring signal,
that is typically around 90V AC 25 Hz.
Generally you can expect very good network performance with propably
very temporary problems (possible on telephone ring, on/off-hook,
pulse dialling etc. if the cable is not very good).

Actually, it's usually 20 Hz. However, that's even further from ethernet
frequencies than voice. That 20 Hz signal has to couple into the data
pairs and then somehow make it through a transformer, that designed to pass
much higher frequencies. Also, it's the current, not voltage that
determines induction. So, the 20 Hz ringing current, has to couple from
one twisted pair to the next, which will result in common mode
interference, at 20 Hz. The transformer is designed to couple differential
mode signals, not common mode, to the NIC. The transformer is designed to
pass signals in the MHz range, not 20 Hz, and there's also additional
filtering in the NIC circuit. It's unlikely you'll find any interference
from ringing.

Robert Redelmeier wrote:

And how does 100baseT differ, that it would be more sensitive?
It works at higher frequencies than 10baseT.

"frequency" may be a useful concept for broadband or carrier-wave
signalling, but neither of these ethernets is.

The signal plateaux are 1/10th as long for 100baseTX, therefore
more prone to being affected by transients. Ringing current
is unlikely to be a problem because although it is high, it is
continuous and smooth AC. Going on/off hook (or pulse dialing)
might be, especially if there isn't parasitic capacitancy or
deliberate contact capacitors to round the shoulders. You also
have to cope with "bounce" from mechanical contacts.

100baseT signaling is at 125 Mbaud. You'd have to generate some incredible
harmonic levels to get interference at those frequencies. You mentioned
pulse dialing. When you have a square wave, the amplitude of the harmonics
decreases with the inverse of the harmonic, so, if you take the 10th
harmonic of the pulse dialing at 10 Hz, you'd have at most 1/10 the voltage
at 100 Hz. At 1 Khz, only 1/100th etc. Now extend that up to the MHz
range and you'll find you likely can't even measure the harmonic amplitude.
Also, power is proportional to the square of the voltage, so if the
amplitude is 1/100th, then the power is 1/10,000th. You'll soon find that
the harmonic levels are far below the desired signal level. This is even
assuming you've got a direct electrical connection. However, when you're
sharing a cable, you don't have 100% coupling, between pairs, so the
interfering signal will be considerably lower still.

In short, you'd have a tough time showing that any signal, normally present
on a phone line, is capable of causing any interference to the ethernet
signal. If an ethernet transceiver can deal with near end cross talk, from
it's own tranmitted signal, it can more than handle any possible
interference from a phone line.

Here's an experiment you can try. Connect a 100baseT ethernet through a
100M long CAT5 cable. Then place continuous ringing current on one spare
pair and continuous dial pulses on the other spare pair. Then measure your
frame errors, compared to when there's no ringing & dialing.

Robert Redelmeier wrote:

Given the great difference in frequencies involved. There's unlikely to
be
much interference, between an analog voice line and ethernet. Most of
the
voice energy is below 3 KHz. Ethernet signaling is well above that.

Granted. But the design was for POTS & 10baseT shared-sheath.

And how does 100baseT differ, that it would be more sensitive?
It works at higher frequencies than 10baseT.

James Knott <james.knott@rogers.com> wrote:

Robert Redelmeier wrote:
Granted. But the design was for POTS & 10baseT shared-sheath.

And how does 100baseT differ, that it would be more sensitive?
It works at higher frequencies than 10baseT.

"frequency" may be a useful concept for broadband or carrier-wave
signalling, but neither of these ethernets is.

The signal plateaux are 1/10th as long for 100baseTX, therefore
more prone to being affected by transients. Ringing current
is unlikely to be a problem because although it is high, it is
continuous and smooth AC. Going on/off hook (or pulse dialing)
might be, especially if there isn't parasitic capacitancy or
deliberate contact capacitors to round the shoulders. You also
have to cope with "bounce" from mechanical contacts.

Of course, ethernet is extremely robust, and higher level TCP/IP
protocols can deal with errors remarkably well. So it still
appears to work, just like the many installations with split-pairs.

-- Robert