Design of the test for 100baseT Ethernet.

Archived from groups: comp.dcom.lans.ethernet (More info?)

Stuart Robinson <Stuartr@nospam.please> wrote:
> Is the CAT5 test (as in TSB67) good enough to tell us in
> that a link should be functional and reliable on 100baseT ?

AFAIK, yes!

> There are circumstances where it may be convenient to set
> up a link between two 100baseT machines, where that link
> does not conform to Category 5 specification.

Then it may not work.

> So whilst a 30M patch cable is clearly NOT Category
> 5 compliant, what is the probability that it will be
> functional as a 100baseT link, 50%, 99%, 99.99% ?

99.99+% if correctly wired and crimped.

Cat5 is a _functional_ spec. If you can make the electrical
parms, you can make it of barbed wire. The "cat5" label on
cable & components isn't "offical" or ultimately meaningful.
It is mostly a fervent hope (promise) by the mfr that if you
match and install these bits correctly, your link will meet Cat5.

A 100m patch-cord may well pass Cat5, the higher attenuation
of stranded being offset by few impedenced discontinuities
and jacks.

A 30m patch-cord might offend against TIA/EIA-568 Structured
Wiring Standard. But it isn't permanent.

-- Robert

Archived from groups: comp.dcom.lans.ethernet (More info?)

William P.N. Smith wrote:

> Stuartr@nospam.please (Stuart Robinson) wrote:
>>So whilst a 30M patch cable is clearly NOT Category 5 compliant
>
> Why is a 30M patch cord not CAT5 compliant?

If one is being picky about the EIA/TIA spec that defines CAT5, the longest
allowable patch cables are IIRC 5 meters. A 30 meter patch cable doesn't
meet the letter of the spec, however if one does a channel test on it it
will pass.

--
--John
Reply to jclarke at ae tee tee global dot net
(was jclarke at eye bee em dot net)

Archived from groups: comp.dcom.lans.ethernet (More info?)

"Stuart Robinson" <Stuartr@nospam.please> wrote in message
news:memo.20050119093420.1308A@srnet.compulink.co.uk...

> So whilst a 30M patch cable is clearly NOT Category 5 compliant, what is
> the probability that it will be functional as a 100baseT link, 50%, 99%,
> 99.99% ?
> Stuart.

You are not designing a life support system around that cord, are you? If
not, then what difference does probability make?

If you are designing a (cabling) system and made a conscious decision to
disregard the known standard, then you are dismissing the factual/empirical
data collected during development of the standard. In this case, if you want
to make sure the system you design will work, you have to do tests yourself
and, after you've done statistically significant number of them, decide if
the design is viable, and what's the probability of failure (not zero!).

If you don't want to go through the trouble of proving the new design, and
use it anyways, you'd have to live with (at best) 50% probability that it
will be functioning.

In many situations in life it is sufficient! So, go for it as long as
someone's life does not depend on it. In this case you better pick a
standard solution thus reducing your probability of failure to xx% (no data
known to me in cabling, but pick any low number ;-))

Good LUCK!

--
Dmitri Abaimov, RCDD
http://www.cabling-design.com
Cabling Forum, color codes, pinouts and other useful online resources for
premises wiring users and professionals
http://www.cabling-design.com/homewiring
Downloadable Residential Cabling Guide

Archived from groups: comp.dcom.lans.ethernet (More info?)

> In this case, if you want
> to make sure the system you design will work, you have to do tests
> yourself and, after you've done statistically significant number of
> them, decide if the design is viable, and what's the probability of
> failure (not zero!).

That was rather the reason behind the question, if a link of whatever
makeup passes the Cat5 test, can we be sure to > 99.99% or so that it will
work ?

For instance has anyone ever seen a link pass a Cat5 test but fail in
operation ?


Stuart.

Archived from groups: comp.dcom.lans.ethernet (More info?)

William P.N. Smith wrote:

> "J. Clarke" <jclarke@nospam.invalid> wrote:
>>If one is being picky about the EIA/TIA spec that defines CAT5, the
>>longest
>>allowable patch cables are IIRC 5 meters. A 30 meter patch cable doesn't
>>meet the letter of the spec, however if one does a channel test on it it
>>will pass.
>
> And will work just fine for 100BaseT, which brings us full circle to
> the OP's question. The answer, AFAICT is that if it passes
> qualification by a CAT5 tester (even if it's barbed wire) it'll work
> just fine with 100BaseT.
>
> I suppose the inverse question might be: Is there a transmission line
> that will _pass_ a CAT5 certification that will _not_ work with
> 100BaseT, and if so what are it's characteristics?

As far as the EIA/TIA spec goes, I believe the channel test is the final
arbiter. But the channel test can't be conducted until the cable is in
place wherever it is going to end up. If a cable passes the channel test
but doesn't carry signal, then something pathological is going on.

--
--John
Reply to jclarke at ae tee tee global dot net
(was jclarke at eye bee em dot net)

Archived from groups: comp.dcom.lans.ethernet (More info?)

J. Clarke wrote:

(snip)

> If there is no 90 meters of solid cable and there are no jacks or plugs and
> no cross-connect then 30 meters of patch cable will, barring some really
> bizarre circumstance, work fine. And such a bizarre circumstance, if it
> would selectively prevent 30 meters of patch cable from working but not
> prevent ten meters of patch cable, two plugs, two jacks, 20 meters of solid
> cable, and a cross-connect from working, be something that I would really
> like to see.

Yes.

30m of stranded cable should work fine, even 90m. The attenuation
is slightly higher, so you might not want to go to 100m.

There are two standards. The ethernet standard, which Cat 5
satisfies, and tells you what works.

TIA/568 which tells you what works assuming the rest of the
system follows TIA/568. This allows 90m of solid wire, and
two 5m patch cables. You don't have to go measure the 90m
to be sure that a 5m patch cable will work.

Many things that won't satisfy TIA/568 will satisfy ethernet.
Two 100baseTX repeaters more than 5m apart. Even three
100baseTX repeaters if the rest of the cabling is short
enough! Look at the model 2 rules for 100baseTX.

-- glen

Archived from groups: comp.dcom.lans.ethernet (More info?)

Stuartr@nospam.please (Stuart Robinson) wrote:
>TSB67 only details tests for length, attenuation, NEXT and wire map.

In terms of a transmission line that you are moving 100BaseT over,
what else matters?

Archived from groups: comp.dcom.lans.ethernet (More info?)

> Did you include the connectors in the 99.99%?

I suppose I meant 99.99%, excluding mechanical failures caused by bad
crimps.

I agree with your point on the crimps though, I have seen many
commercially produced cables that are poorly crimped, but they seem to
work.

On reflection I cant ever recall a 100baseT link failing, but since the
critical point would be at maximum length (greatest attenuation) and most
cables will be well under max length, perhaps thats not surprising.


Stuart.