sensitivity of 100MBit switches, single/double shildet cable

Hi,

for some of you a known feature, but not for me: a modern
8-port-switch 100 MBit(soho and d-link tested) doesn't
cope with a primitive "ping" when using a single shilded
twisted pair cable cat 5e!! Attaching a old 10MBit Hub
to the cable and all (ping, telnet, DSL internet) works
fine. The 5e-cable (about 20m) sometimes is in the
near (ca 10cm distance) of 220V AC current cable or
crosses it. We testes two new single shilded cables -
no success!

We tested the connection a double shilded cable an
it works with both the switch an the hub. But this
cable was put on the floor and was not parallel
to the single shilded cable. The two tests are
not with the same conditions.

Are modern 100MBit switches that sensitive, that
the need double shildet twisted pair cable?

thx
Ekkard

Ekkard Gerlach wrote:
>Hi,
>
>for some of you a known feature, but not for me: a modern
>8-port-switch 100 MBit(soho and d-link tested) doesn't
>cope with a primitive "ping" when using a single shilded
>twisted pair cable cat 5e!! Attaching a old 10MBit Hub
>to the cable and all (ping, telnet, DSL internet) works
>fine. The 5e-cable (about 20m) sometimes is in the
>near (ca 10cm distance) of 220V AC current cable or
>crosses it. We testes two new single shilded cables -
>no success!
>
>We tested the connection a double shilded cable an
>it works with both the switch an the hub. But this
>cable was put on the floor and was not parallel
>to the single shilded cable. The two tests are
>not with the same conditions.
>
>Are modern 100MBit switches that sensitive, that
>the need double shildet twisted pair cable?

Usually not. Are you sure that the wiring of the connectors is OK?
You need to connect the pairs in the right way. If you hook up the
wires in a random order (even when they are in the same order on both
ends), you're going to have troubles at 100 Mbit.

The right order is:
1 - orange/white
2 - orange
3 - green/white
4 - blue
5 - blue/white
6 - green
7 - brown/white
8 - brown
See http://www.hardwarebook.net/connecto...et10baset.html
for the numbering of the pins.

--
Maurice

On Wed, 23 Jul 2003 17:33:51 +0000 (UTC), Ekkard Gerlach
<(E-Mail Removed)> wrote:

>Hi,
>
>for some of you a known feature, but not for me: a modern
>8-port-switch 100 MBit(soho and d-link tested) doesn't
>cope with a primitive "ping" when using a single shilded
>twisted pair cable cat 5e!! Attaching a old 10MBit Hub
>to the cable and all (ping, telnet, DSL internet) works
>fine. The 5e-cable (about 20m) sometimes is in the
>near (ca 10cm distance) of 220V AC current cable or
>crosses it. We testes two new single shilded cables -
>no success!
>
>We tested the connection a double shilded cable an
>it works with both the switch an the hub. But this
>cable was put on the floor and was not parallel
>to the single shilded cable. The two tests are
>not with the same conditions.
>
>Are modern 100MBit switches that sensitive, that
>the need double shildet twisted pair cable?
>
>thx
>Ekkard

Hmm. I may be wrong but I believe CAT5 should never be shielded. At
all! I was not even aware that there was such a thing as shielded
CAT5. Much less twisted pair! They are all four pair.


Rich Piotrowski

To E-mail use: rpiotro(at)wi(dot)rr(dot)com

> Are modern 100MBit switches that sensitive, that
> the need double shildet twisted pair cable?

I've never seen one that was. There are two things to watch out for:

1. If you're dealing with Cisco equipment, make sure that bandwidth and
duplex are manually set. Cisco's autonegotiation is pathetic. To
compound the matter, after upgrading the IOS (which many people have done
recently), the ports are often reset to autonegotiation.

2. Shielded cable isn't quite as simple as "plug it in, it will work".
If things aren't done correctly, using shielded cable can cause all sorts
of problems. If you're not familiar with the nuances of shielded cable,
try an unshielded cable, and see if that makes any difference.

3. Close-proximity AC cabling shouldn't be affecting your ethernet lines.

steve

On Wed, 23 Jul 2003 13:01:36 -0600, "Steve Wolfe" <(E-Mail Removed)>
wrote:

>
>> Hmm. I may be wrong but I believe CAT5 should never be shielded. At
>> all! I was not even aware that there was such a thing as shielded
>> CAT5. Much less twisted pair! They are all four pair.
>
> There's UTP cat5 ("unshielded twisted pair"), and STP cat5 ("Shielded
>twisted pair"). STP uses a thin, metallic film around the wires to help
>prevent interference.
>
> STP isn't quite as simple as UTP. One of the gotchas is that the film
>should be connected to ground at one end of the cable - but *ONLY* at one
>end of the cable! If it's connected at both ends, you're connecting the
>grounds of two seperate devices through a very thin metallic film, which
>has the potential to cause very bad problems.
>
>steve
>
>

As in "ground loops"? I am familiar with those!!


Rich Piotrowski

To E-mail use: rpiotro(at)wi(dot)rr(dot)com

Ekkard Gerlach wrote:

> Are modern 100MBit switches that sensitive, that
> the need double shildet twisted pair cable?
>

No. It sounds like you miswired the cables. Look up EIA/TIA 568A & 568B,
for wiring configurations. 100 Mb ethernet should work fine on unshielded
cable.

--

Fundamentalism is fundamentally wrong.

To reply to this message, replace everything to the left of "@" with
james.knott.

Rich Piotrowski wrote:

> Hmm. I may be wrong but I believe CAT5 should never be shielded. At
> all! I was not even aware that there was such a thing as shielded
> CAT5. Much less twisted pair! They are all four pair.
>

There is shielded cable, but it's less common than unshielded.

--

Fundamentalism is fundamentally wrong.

To reply to this message, replace everything to the left of "@" with
james.knott.

Ekkard Gerlach wrote:

> this must be the fault ...

I would personally guess so - it usually behaves weird. I had two cables
(25m each), both with the same (read: with the same _wrong_) combination of
wires - one worked okay, the other did not. The right order solved it ...

It seems, that the `random` order works for shorter cables, but I would not
dare try it

--
Robert Jirik
[mailto:robert(at)aristoteles(dot)xhaven(dot)net]
public PGP key: http://xhaven.net/robert/pgp_key.asc
-
"Whenever you find yourself on the side of the majority,
it's time to pause and reflect"
-- Mark Twain

Maurice Janssen <(E-Mail Removed)> wrote in news:bfmklg$vag$1
@linux.z74.net:

[...]

> Usually not. Are you sure that the wiring of the connectors is OK?
> You need to connect the pairs in the right way. If you hook up the
> wires in a random order (even when they are in the same order on both
> ends), you're going to have troubles at 100 Mbit.
>
> The right order is:
> 1 - orange/white
> 2 - orange
> 3 - green/white
> 4 - blue
> 5 - blue/white
> 6 - green
> 7 - brown/white
> 8 - brown
> See http://www.hardwarebook.net/connecto...et10baset.html
> for the numbering of the pins.
>

Oh, Maurice, for all my life i have been wiring patch cables like this:
1 - green/white
2 - green
3 - orange/white
4 - blue
5 - blue/white
6 - orange
7 - brown/white
8 - brown
(Organge instead of Green)

What do you say - does orange instead of green affect the tx/rx speed
(power? i dont know correct word)?
- Personnaly, if testing never had any problems, so it seems that this
case does not affect the performance of the cable.


P.Krumins

Peteris Krumins wrote:
>Oh, Maurice, for all my life i have been wiring patch cables like this:
>1 - green/white
>2 - green
>3 - orange/white
>4 - blue
>5 - blue/white
>6 - orange
>7 - brown/white
>8 - brown
>(Organge instead of Green)
>
>What do you say - does orange instead of green affect the tx/rx speed
>(power? i dont know correct word)?

No, the speed is not affected by this. As long as you keep the (color
coded) pairs at pin 1/2 and 3/6, you should be fine (4/5 and 7/8 are not
used for 100baseTx).
However, if somebody else wants to put a new connector on one of your
cables, you might end up with a cross-cable instead of a straight one.

>- Personnaly, if testing never had any problems, so it seems that this
>case does not affect the performance of the cable.

That's right. http://www.siriuscomputers.com/rj45.htm has some more
information about this. Your wiring scheme is 568A, while normally
568B is used for patch cables. But as long as you consequently use
568A, it'll work fine.

--
Maurice