Hubs
Kyle Schuiteman
Here is the probelm. I put in a new 3com superstack entry hub to replace
an older 3com linkbuilder hub. The server will be seen on the old hub but
not on the new one.( I can ping the server on the old one and not on the
new one). What gets stranger is if I link the old hub to the new one and
plug the server into the old hub that works too.
I have a test machine here with same software setup etc and it connects
just fine to the new hub. Any ideas. It has to be a setup issue with the
network card.
please respond vi email
ky...@transova.com
Ben Rosenthal
autosensing. Try fixing the NIC to 10 or 100mbs. You might also try
un-pluging and re-pluging in the patch cable between the server and the HUB
after the server is booted.
HTH,
Ben Rosenthal
Optical Retail POS Solutions
bcr...@ibm.net
Kyle Schuiteman wrote:
Sorry, Post here, Read here.
Bob Willey
Did you switch from 10BaseT to 10/100BaseT??
If so, you have a cabling problem, we ran into the same thing. Newer
100BaseT requires a different pinout configuration than does 10BaseT. I
don't recall the technical spec #, but 3Com pointed us to the problem. We
upgraded to a 3Com Baseline 10/100 Switching Hub and ran into the same
problem. Switched the cabling over, and voila, everything worked like a champ.
At 05:39 PM 7/20/99 +0000, Kyle Schuiteman wrote:
>Hi I have a openserver machine here with a 3com etherlink xl card in it.
>
>Here is the probelm. I put in a new 3com superstack entry hub to replace
>an older 3com linkbuilder hub. The server will be seen on the old hub but
>not on the new one.( I can ping the server on the old one and not on the
>new one). What gets stranger is if I link the old hub to the new one and
>plug the server into the old hub that works too.
>I have a test machine here with same software setup etc and it connects
>just fine to the new hub. Any ideas. It has to be a setup issue with the
>network card.
>
>please respond vi email
>ky...@transova.com
-> Bob Willey - b...@ccs.com <-
CCS Enterprises, Inc.
PO Box 35 Linkwood MD 21835
http://www.ccs.com Come Visit Us !!
http://www.global-homebiz.com/bwilley.html
http://www.intercom.net/user/consult
-> (410) 228-9211 -- FAX: (410) 901-1105 <-
Scott Roberts
Bob Willey wrote in message <4.1.19990720135050.00968860@consult>...
>I suspect another problem.
>Did you switch from 10BaseT to 10/100BaseT??
>If so, you have a cabling problem, we ran into the same thing. Newer
>100BaseT requires a different pinout configuration than does 10BaseT.
I don't think so, can this be true??? 10BaseT may be more forgiving
and run on inferior cabling where 100BaseT will not, but I believe
that both run fine on the same cabling so long as it is installed
correctly (eg Cat5 network). Someone please correct me if I'm wrong on
this.
Scott
Ken Wolff
I also was not aware of a pin-out difference. But I have had problems
trying to use a Cat3 cable with a 100BaseT connection. Once I made (yes,
we make many of our cables, which is why I wasn't aware of the pin
difference) a new cable with Cat5 wire, it worked just fine.
------------------------------------------------------
Ken Wolff
Phone: 616-957-4949 Ext: 111
FAX: 616-957-1614
------------------------------------------------------
James Andrews
difference between 10baseT and 100BaseT was the max length of the cable
from the hub. For 10BaseT I believe the max distance was around 100
meters and for 100BaseT the max distance was around 20 meters. Then
again I could be wrong...
Bob Willey
10BaseT used a different spec which used twisted pair, 100BaseT requires
that the pairs be seperated, and is a different pinout. 10BaseT NIC cards
will work with the new pinouts, but 100BaseT NIC's will usually not work
(or at least not reliably) with the old pinouts. EIA 568B pinouts.
There was a different spec used for 10BaseT, I don't have the spec #, but
if you did not wire with the EIA 568B spec AND used Cat5 cabling, then a
change is required. I know many people who thought they needed just cat5
cabling, but it also affects the pinouts also. I learned the hard way.
(had to have connector redone).
Here are some details:
http://www.csot.com/Ethernet/100BaseT.htm
The same rules apply as for 10BaseT except as noted above.
The 100BaseT standard is made up of 3 versions:
100BASE TX is full-duplex capable in point to point unshared
applications because it uses 1 pair to receive and 1 pair to transmit.
Designed to run over 2 pairs of category 5 unshielded twisted pair cable
with RJ45 connectors and EIA/TIA 568B pinning. It can also be run on
IBM type 1 sheilded twisted pair (existing Token Ring wiring) with an
impedence matching device and DB9 connectors or regular STP and DB9
connectors. Max segment length is 100m.
100BASE T4 designed to run over 4 pairs of category 3, 4 or 5 UTP
cable with RJ45 connectors and EIA/TIA 568B pinning. It can also be run
over STP. 1 pair is used to receive while 3 pairs are used to transmit,
however full-duplex operation does NOT work because specific pairs are
not designated to transmit or receive. Max segment length is 100m.
100BASE FX designed to run over 2 strands of duplex multimode fiber
optic cable. It's also full-duplex capable because it uses one strand for
receive and one for transmit. Maximum cable segment varies depending on
the cabling used. Singlemode (depending on the manufacturer) can exceed
10 km when full-duplex. Multimode maximum length is 412 meters for
half-duplex and 2 km ful-duplex. Max length from station to repeater is 150
meters.
NOTE: For full-duplex operation on 100BASE TX or FX:
1) devices must support full-duplex
2) connection must be unshared end to end.
In Reply to: Wiring Stds 10/100 MPS posted by Jim Saunders on July 16, 1999
at 07:35:05:
10 Mbps Ethernet will run on a CAT 3 cabling system and 100 Mbps requires
CAT 5, but you spec'd a CAT 5 cabling infrastructure. It should be
sufficient to run
either 10 or 100 Mbps Ethernet, and in theory, should be transparent to any
brand equipment connected to it.
Did you require that the cabling job be certified to CAT 5 requirements
(including jacks & patch panels) before signing off on the job? This is a
must. There are not
specifically different testers for 10 vs 100 Mbps systems, because the
cable and terminations are being rated based on bandwidth. In the case of
CAT 5, to 100
MHz, which is more than adequate for 100 Mbps Ethernet. However, your
contractor should have used at least a Level II test set. A Level I tester
is inadequate to
properly qualify CAT 5 infrastructure.
Did your cabling contractor truly re-wire, or just re-terminate? In either
case, they may have corrected some poor termination practices from the
initial install. Did the
contractor originally use CAT 5 rated jacks, patch panels, and patch cords?
Poor termination practices or the use of less-than-CAT 5 components could
all result in
a network that runs 10 Mbps Ethernet fine, but chokes on 100 Mbps.
The main elements of the standard structured cabling schemes are:-
Fixed building cable comprising 4 pr single strand Unshielded Twisted Pair
Cable to Category 5 specifications wired from a wall outlet to a patch
panel mounted in a hub closet. Termination at either end in a standard
sequence (EIA 568A) and pin-out on 8-pin "modular" (RJ45) connectors.
Fly leads comprising 4 pr multi-strand Unshielded Twisted Pair cable
straight through connected to RJ45 plugs at either end. Used to connect
patch panel to data hub, and wall outlet to PC.
http://www.net.berkeley.edu/dcns/standards/UTP_Specs.html
Cables shall be installed in unbroken segments from central utility closets
to individual communication outlet locations. In the utility closets UTP
cables shall be
terminated on 110-style cross connect blocks from specified manufacturers
designated for voice and data communication service. At each workspace
location cables
shall be terminated in a wall mounted EIA/TIA 568A compliant communications
outlet from specified manufacturers. All installed cabling runs shall be
tested for
compliance with specified parameters, documentation provided, and both ends
of each cable run shall be labeled. Test results must be verified for
compliance and
entered into the CNS database prior to activation of cables for voice or
data use.
At 03:00 PM 7/20/99 -0500, Scott Roberts wrote:
>
>Bob Willey wrote in message <4.1.19990720135050.00968860@consult>...
>>I suspect another problem.
>>Did you switch from 10BaseT to 10/100BaseT??
>>If so, you have a cabling problem, we ran into the same thing. Newer
>>100BaseT requires a different pinout configuration than does 10BaseT.
>
>I don't think so, can this be true??? 10BaseT may be more forgiving
>and run on inferior cabling where 100BaseT will not, but I believe
>that both run fine on the same cabling so long as it is installed
>correctly (eg Cat5 network). Someone please correct me if I'm wrong on
>this.
>
>Scott
>
-> Bob Willey - b...@ccs.com <-
David L. Courtney
cables are twisted more tightly than the other two. 568A uses different
pairs than 568B and with Fast Ethernet it matters; with 10BaseT you
could use either one. HTH.
DLC
---------------
Jeff Liebermann
wrote:
>Hi I have a openserver machine here with a 3com etherlink xl card in it.
What, may I ask is the exact model number of the 3com etherlink XL
card? I don't have a marketting speak to model number conversion
chart handy. Be sure to include any "T4" or "TX" suffix.
>Here is the probelm. I put in a new 3com superstack entry hub to replace
>an older 3com linkbuilder hub.
Let see... My 3com product lineup shows about 24 models of superstack
hubs and 16 models of the linkbuilder flavour. Any particular models?
>The server will be seen on the old hub but
>not on the new one.( I can ping the server on the old one and not on the
>new one). What gets stranger is if I link the old hub to the new one and
>plug the server into the old hub that works too.
Chuckle. You have a reversed or null type cable in the system
somewhere. It's usually used to connect between two hubs and is
commonly orange colours. Check the tags, and watch the "LINK" lites
to make sure you have a properly wired connection. Be sure to inspect
the lights on the semi-specified 3com etherlink XL card.
>Any ideas. It has to be a setup issue with the
>network card.
Hint: That which is most obviously correct, beyond any need of
checking is usually the problem. Try different 10baseT cables and
watch out for ports labeled "MDI" which are wired for null connections
between hubs.
[x]email [x]news [ ]mailing list
Chad Lemmen
"David L. Courtney" wrote:
> A wiring expert that I use told me that two of the four pairs in the
> cables are twisted more tightly than the other two. 568A uses different
> pairs than 568B and with Fast Ethernet it matters; with 10BaseT you
Is 568A and 568B published somewhere?
Jeff Liebermann
Try:
http://www.cabletron.com/support/techtips/tk0365-9.html
for the human readable incantation. The actual spec from the EIA is
rather complex.
The cable pairs are different twist rates in CAT5 cable to reduce
crosstalk between pairs. The result is that different pairs cross
couple different amounts of garbage to each other. This is not much
of an issue with the common half duplex system but a serious problem
in full duplex which has a receiver listening at the same time as a
transmitter and cannot tolerate any level of garbage from the
transmitter. The original wiring scheme (EIA-568A) was originally
concocted for a half duplex system where the "unused" two pairs were
to carry voice (POTS) traffic as in the AT&T StarLan system. Pairs
could be selected almost at random and apparently were for EIA-568A.
However, when we get to full duplex 100baseT, it was deemed necessary
to squeeze every last bit of performance out of the system. Test were
performed to determine which pairs had the least cross talk and
EIA-568B was invented.
Drivel: Long ago, my accomplis was in the server room, crimping RJ-45
connectors wired for 568A while I was at the other (workstation) end,
wiring for 568B. After crimping about 30 cables, the problem was
discovered. Be consistant and use EIA-568B.
Ben Rosenthal
There are "really" two cable configurations for the patch cable (The home run
from the patch panel to the terminating plug is straight through). There are
cross-over cables, for connecting HUBs together and the full 8 conductor. The
full 8 conductor specifies the pin outs for each specific color. My
understanding is if you follow this (568B) setup you will be as current as
possible for 100BaseT. Any other pin out variation is _not_ standard.
A picture is worth 1k words:
http://www.cabletron.com/support/techtips/tk0231-9.html
Ben Rosenthal
bcr...@ibm.net
Jeff Liebermann
wrote:
>There are "really" two cable configurations for the patch cable (The home run
>from the patch panel to the terminating plug is straight through). There are
>cross-over cables, for connecting HUBs together and the full 8 conductor. The
>full 8 conductor specifies the pin outs for each specific color. My
>understanding is if you follow this (568B) setup you will be as current as
>possible for 100BaseT. Any other pin out variation is _not_ standard.
>
>A picture is worth 1k words:
>http://www.cabletron.com/support/techtips/tk0231-9.html
Correct. EIA-568A is *NOT* supported for 100baseT. Only 568B. For
1000baseT, there are some additional restrictions (which I'm too lazy to
find).
The bulk of the commercial cross-over cables only have 2 pairs and are
therefore totally unsuitable for full-duplex connections. To make
matters worse, some clueless manufactories have been delivering cross
wired data connections, but straight through for the "unused" pairs.
This might work with Starlan and half duplex, but not full duplex. I
worked out the exact full duplex cable wiring plan a while back, but as
usual, can't find it. Duz anyone need or want it?
Drivel: I've run a TDR (time domain reflectometer) through some CAT5
spagetti and am amazed at how poor the RJ45 connectors and 110 punch down
blocks appear. It's amazing anything gets through that bad an impedance
bump. I've also tested 25pair telco cables and found them useable for
half duplex 10baseT and a lost cause for 100baseT as long as there are no
Type 66 blocks in the system.
--
Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
(831)421-6491 pgr (831)426-1240 fax (831)336-2558 home
http://www.cruzio.com/~jeffl WB6SSY
je...@comix.santa-cruz.ca.us je...@cruzio.com
David L. Courtney
>
> The cable pairs are different twist rates in CAT5 cable to reduce
> crosstalk between pairs. The result is that different pairs cross
> couple different amounts of garbage to each other. This is not much
> of an issue with the common half duplex system but a serious problem
> in full duplex which has a receiver listening at the same time as a
> transmitter and cannot tolerate any level of garbage from the
> transmitter. The original wiring scheme (EIA-568A) was originally
> concocted for a half duplex system where the "unused" two pairs were
> to carry voice (POTS) traffic as in the AT&T StarLan system. Pairs
> could be selected almost at random and apparently were for EIA-568A.
> However, when we get to full duplex 100baseT, it was deemed necessary
> to squeeze every last bit of performance out of the system. Test were
> performed to determine which pairs had the least cross talk and
> EIA-568B was invented.
> Hi, All.
After looking at this a little closer, it appears to me that 568A and B
use the same pairs (two and three) but reverse the transmit and receive.
My wiring guy was actually telling me that the other two pairs in the
cable cannot be used with 100BaseT; people used to use the other two
pairs for a second connection sometimes with 10BaseT. He said, "Don't
do that with 100BaseT." I had not recalled accurately at first. He
insists that A or B is okay as long as everyone doing any wiring is in
agreement (see Jeff's drivel section). BTW, my Novell guide to NW4.11
says to use 568A. I am beginning to think that the main thing is to not
use pairs 1 and 4 for anything.
Regards,
David C.
-bill-
why are two pair inadequate for a patch cable to run at full duplex?
It would seem that the transmit pair and the receive pair would provide
for full duplex 100-TX
-bill-
Technical Service Systems - bi...@TechServSys.com
Bill Vermillion
-bill- <bi...@TechServSys.com> wrote:
>Jeff,
>why are two pair inadequate for a patch cable to run at full duplex?
^^^^^^^^^^^
>It would seem that the transmit pair and the receive pair would provide
>for full duplex 100-TX
The full/half duplex is the keyword - no matter whether it is
ethernet/modems/etc. (almost any communications device).
In half-duplex one pair is in the receive mode the other is
in the trasmit mode. As the transmit side is transmitting it
is using the receive side for the ack/naks/folderol.
In FDX both side can transmit and receive at the same time.
That means that if you are running a 100mbs second system in
full-duplex mode you have a maximum of 200mbs second throughput.
That's why you need two transmit pairs and two receive pairs.
Talking and listening on the same wire at the same time is
mostly found only in analog environments - eg modems.
Bill
--
Bill Vermillion bv @ wjv.com
Vania Lolham
You are right about the twist difference in all 4 pairs.
I was called out to many sites big or small from 5 user to 500 use with the
problems most of you facing now!
Example: you or your client started with co-ax cable BNC connector 4 to 10
Mbps depend on who done the cabling and
quality of the NIC's and it was fine then.
and as your business grow, your cable length grows too, till
you reach your 185meter limit and accordingly
more traffic in the cable and resulting in a lot of data
collision, packet loss and holts.
you or your client will take the plunge and change to
10base-T, using patch panels and hubs.
at this point the cable guy is the key to achieve the best
performance or a nightmare.
most cable guys didn't know anything about category 1,2,3,4,5
now 6. and run the cables from one side
of the office to the other and over the power cables and
light fittings and bend and twist and squeeze the cable.
end result was okay, the improvement over Co-Ax was 50%
better, and everyone was happy.
until the client decided to go super highway with 100Mbps and
intelligent Auto Sense hubs.
this is where someone will get fired. the bloody hub turned
in to a Christmas tree!, signals keep jumping
up and down flashing from 10 to 100 and data collision light
flashing and so on.
WHY? the answer is not knowing how to design a bullet proof
network from the start to allow growth.
A good practice is to spend a bit more from the start on the
quality of the network cabling and
accessories than finding yourself in a situation that you
have to Re-Do-It all over again.
Now this is my advise to you.
-------------------------------
Cat-5 Cabling terminology : each pair has a full colour wire twisted with
another of the same colour with stripe or just
white called "mate".(instead of saying "white" or "white with colour X
stripe).
four twisted pairs are: Green-Orange-Blue-Brown
Wiring spec for T568A: pinout:(hold the plug in front of you and clip
pointing up left to write is 1 to 8)
1-Green(Mate)
2-Green
3-Orange(Mate)
4-Blue
5-Blue(Mate)
6-Orange
7-Brown(Mate)
8-Brown
Wiring spec for T568B: pinout:
1-Orange(Mate)
2-Orange
3-Green(Mate)
4-Blue
5-Blue(Mate)
6-Green
7-Brown(Mate)
8-Brown
Important Note: To achieve 100Mbps (100Mhz) you must follow these
instruction carefully.
1- Not all Cat-5 cables are suitable for 100Mbps (choose the known
brand), but any Cat-5 will be adequate for 10Mbps.
2- Cat-5 cable for 100Mbps or 100Mhz freq have to have IEEE standard and
FDDI networks compliant and
each copper wire have to be 22AWG for Cat-5 and 24AWG for Cat-6,
read the compliance on the
package or the cable's jacket.
3- The maximum bending radius of the cable is 25.4mm without the
insulation cracking.
4- UTP Cat-5 cable shall not cross any power cable carrying more than
60volts, minimum separation 60cm
between the Data cable and Power cable of low voltage.
5- If you are making your own patch cables or terminating your patch
panel have to follow this important method:
1-When patching the cable, jacket should not exposed
the wire more than 6mm after termination.
2-When terminate each pair, do not UN-twist them
more then 4mm. major losses occurs at the
patch panel and workstation's
socket(wall-mount).
3- patch cable have to match your patch panel
sockets(read on the socket T568A or T568B)
and entire network have to have one type of
wiring implementation.
6- Never run the cable near fluoro light ceiling unit ( Electro Magnetic
Interference will cause packet loss).
7- Maximum cable length(segment) from Patch panel to workstation's NIC
is 100 meter no more(best is stay under
90 meter from the patch panel to the wall socket then allow 2 to 3
meter for the patch cable from wall socket to
the NIC.
8- use a path for all the cables going from patch panel and group them
when they have to change direction.
follow the nature simple examples: Tree and Heart.
9- Always use solid wire for cable run from the patch panel to the
workstation and stranded wire for the patch cable
form patch panel to the hub and from wall socket to the computer.
--------------
In most cases for higher velocity and frequency stability use T568A
topology.
But the most important of all after you did the above is matching your
Hub with the NICs, try to use one manufacturer
or brand name for your Hubs and NICs for the right speed. Never mix and
match then.
I hope this short course will solve most of your common problems.
Regards
Vania
Kyle Schuiteman <ky...@transova.com> wrote in message
news:01bed2cc$1ed21b80$0101a8c0@transova...
> Hi I have a openserver machine here with a 3com etherlink xl card in it.
>
> Here is the probelm. I put in a new 3com superstack entry hub to replace
> an older 3com linkbuilder hub. The server will be seen on the old hub but
> not on the new one.( I can ping the server on the old one and not on the
> new one). What gets stranger is if I link the old hub to the new one and
> plug the server into the old hub that works too.
> I have a test machine here with same software setup etc and it connects
> just fine to the new hub. Any ideas. It has to be a setup issue with the
> network card.
>
Jeff Liebermann
>why are two pair inadequate for a patch cable to run at full duplex?
>It would seem that the transmit pair and the receive pair would provide
>for full duplex 100-TX
Actually, it can, and does.
There are several different flavours of 100baseT.
Flavour Wiring Symbol Rate Amplitude Mod
100base-TX 2 pairs 100mbits/sec 1 bit/baud
100base-T4 4 pairs 25mbits/sec 2 bits/baud
100base-T2 2 pairs 25mbits/sec 4 bits/baud
100base-TX will run on 2 pairs of CAT5 but is horribly sensitive to
crummy wiring, connections, moisture incursion, bend radius, and due to
the high frequencies involved. This is the most common form of 100baseT.
http://www.ora.com/reference/dictionary/terms/1/100BASE-TX.htm
100base-T4 runs on 4 pair of CAT5. However, it will not do full duplex
because 3 pairs are used for transmit/receive and one pair for collision
detection.
http://www.ora.com/reference/dictionary/terms/1/100BASE-T4.htm
100base-T2 will run on 2 pairs of CAT5 but is rare because of the
expensive processing needed to remove the inevitable crosstalk noise.
See:
http://www.trynci.com/cat/100baset.htm
for more coherent explanations.
However, 100baseT is easy. I just finished a gigabit link. See:
http://www.gigabit-ethernet.org
For info about 1000baseT (Gigabit Ethernet) technology, see:
http://www.gigabit-ethernet.org/technology/whitepapers/gige_11.97/how.html
Note that 1000baseT uses 4ea 250mbit/sec full duplex links. The actual
symbol rate is 125mbits/sec with a 5 amplitude bits (4 bits data plus one
bit of forward error correcting code) to provide the additional
bandwidth. Just to make it interesting, 1000baseT "negotiates" the
connection (NWAY) and falls back to 100baseT is 1000baseT fails. As I
recall, the order and sequence is:
1000base-T Full Duplex
100base-TX Full Duplex
100base-T4 Half Duplex
100base-TX Half Duplex
10base-T Full Duplex
10base-T Half Duplex
If you get rotten performance on 1000baseT or 100base-TX links, look for
NWAY negotation problems. There are many devices out there that will not
negotiate and need to be told what to look for.
Some of the other gigabit ethernet "white papers" are well worth reading.
Hint: Fiber rules mostly because there is not crosstalk.
Jeff Liebermann
wrote:
>There are "really" two cable configurations for the patch cable (The home run
>from the patch panel to the terminating plug is straight through). There are
>cross-over cables, for connecting HUBs together and the full 8 conductor. The
>full 8 conductor specifies the pin outs for each specific color. My
>understanding is if you follow this (568B) setup you will be as current as
>possible for 100BaseT. Any other pin out variation is _not_ standard.
Foundit. The right way to make a crossover cable with all 4 pairs.
See:
http://wwwhost.ots.utexas.edu/ethernet/100quickref/ch12qr_6.html
MrSmiley
|Hi,
[snip excellent tutorial]
| 2- Cat-5 cable for 100Mbps or 100Mhz freq have to have IEEE standard and
|FDDI networks compliant and each copper wire have to be 22AWG for Cat-5
|and 24AWG for Cat-6, read the compliance on the package or the cable's jacket.
At my outstanding local electronics store, they have never sold me 22 gauge
solid copper CAT5 in either shielded or utp. Now I am a bit concerned, and
I intend to find some specs on this. If you have a url to cat 5 specs, it'd
be neat for you to share. Jeff mentioned a few I'll check out, too.
| 3- The maximum bending radius of the cable is 25.4mm without the
|insulation cracking.
And that, for my friends in the nickle seats, is 1 inch.
| 4- UTP Cat-5 cable shall not cross any power cable carrying more than
|60volts, minimum separation 60cm between the Data cable and Power cable
|of low voltage.
Unless you are smart and run shielded twisted pair.
Why allow any interference?
| 7- Maximum cable length(segment) from Patch panel to workstation's NIC
|is 100 meter no more(best is stay under 90 meter from the patch panel
|to the wall socket then allow 2 to 3 meter for the patch cable from wall
|socket to the NIC.
Minimum cable length is 10 feet, or so I read.
|9- Always use solid wire for cable run from the patch panel to the
|workstation and stranded wire for the patch cable form patch panel
|to the hub and from wall socket to the computer.
Why the difference in solid versus stranded?
How about the patch cable from hub to nic?
Or from hub to wall socket?
Or from patch panel to wall socket?
What did you mean 22AWG earlier, in light of the fact that AWG means stranded?
Solid sizes don't match stranded as you may well know, so which one are you
quoting as 22 gauge?
| --------------
|In most cases for higher velocity and frequency stability use T568A
|topology.
Why?
Also, what do you think are the ramifications of taking T568A cable stock and
wiring it as T568B?
| Regards
| Vania
Thanks for taking the time to post that extensive analysis into history.
It'll be a nice hit on deja news in the future.
Mr:)
Jeff Liebermann
wrote:
>At my outstanding local electronics store, they have never sold me 22 gauge
>solid copper CAT5 in either shielded or utp. Now I am a bit concerned, and
>I intend to find some specs on this. If you have a url to cat 5 specs, it'd
>be neat for you to share. Jeff mentioned a few I'll check out, too.
See:
http://www.belden.com/products/catalog/help/techdataguide.htm
http://www.belden.com/products/catalog/help/techdualguide.htm
http://www.belden.com/products/catalog/help/techtechnical.htm
Please note that all CAT5 and DataTwist 350 (CAT6) are 24awg solid
wire. The flexible stranded cable, commonly found in patch cables is
also 24awg but has slightly more insulation. This is because the wire
guage is a measure of circular mils or electrical cross sectional
area. Therefore, for a given wire guage, the DC resistance is the
same whether stranded or solid. Since stranded has some air gaps
between the stands, the bundle diameter is larger. However, RF (radio
freak) travels on the outside of the wire. To maintain a consistant
100 ohms impedance, and maintain the same outside to outside wire
spaceing, the insulation thickness must be slightly larger for
stranded than solid.
The measure of the ability of a cable to pass data is the signal to
noise ratio at a given frequency. As the frequency goes up, so does
the attenuation and so does the crosstalk (garbage from other data
lines as in full duplex systems). Eventually, the crosstalk and
desired signal attenuation are equal and nothing passes through the
cable. Actually, 10db (3:1) signal to noise ratio is required for the
typical ethernet transceiver to operate and is specified in the TIA
CAT5 specification, so the cutoff frequency will be somewhat less.
Ethernet repeaters will work a lower S/N ratios. See:
http://www.belden.com/products/catalog/help/techcompetitor.htm
for a more coherent explanation.
Check out some of the other wire techy references at:
http://www.belden.com/faq/QUERY1.HTM
and search for your favorite buzzword.
>Unless you are smart and run shielded twisted pair.
Don't you dare. Shielded network cable has much higher line to ground
capacitance, requires grounded RJ45 connectors, creates miserable
ground loop problems, mangles timing because of changes in velocity
factor, and generally is a lousy idea. It also does absolutely
NOTHING for magnetic induction from flourescent ballasts and nearby
power lines. The only justification for shielded CAT5 is strong RF
fields. I use the shielded stuff at radio sites. Metricom uses a
shielded CAT5 mutation for all their Ricochet wireless installations.
However, the garden variety network does not need shielding, would not
benifit from shielding, and will certainly have ground loop problems
if used. (Remeber the "protective ground", ground loop problems with
RS-233 wiring?).
>Minimum cable length is 10 feet, or so I read.
3 ft or so I vaguely recall.
>Why the difference in solid versus stranded?
Solid wire tends to break if moved too often. I've never broken one
but that's the official pitch line. Stranded also survives abuse
(tripping over the patch cables on the floor) better than solid.
>How about the patch cable from hub to nic?
>Or from hub to wall socket?
>Or from patch panel to wall socket?
Same as above. If it moves, use stranded. However, I can never seem
to find 1000ft rolls of the bulk stranded cable, so most of my home
made patch cables are solid wire. I've had no connection problems
when I use a ratchetting crimper and test the results with a
(borrowed) cable certifier.
>What did you mean 22AWG earlier, in light of the fact that AWG means stranded?
>Solid sizes don't match stranded as you may well know, so which one are you
>quoting as 22 gauge?
AWG means "American Wire Guage".
>|In most cases for higher velocity and frequency stability use T568A
>|topology.
>Why?
No clue. I use 568B wire standard for everything. It works.
Cable velocity factor is strictly a function of the wire insulation
type and not connector wiring scheme. Frequency stability is a
function of the NIC cards and is crystal controlled. There is nothing
the cable can do to change the clock frequency. I think I know what
he mean't but I don't wanna guess.
>Also, what do you think are the ramifications of taking T568A cable stock and
>wiring it as T568B?
The cable stock (off the roll) for 568A and 568B are identical. There
is no such thing as 568A or 568B cable. The difference is the
connector wiring plan. However, the pre-wired modular connectors
commonly found in the wall plate assemblies are 568A and 568B
specific. Careful about mixing these.
Bill Vermillion
MrSmiley <mrsm...@hotbot.com> wrote:
>| 4- UTP Cat-5 cable shall not cross any power cable carrying more than
>|60volts, minimum separation 60cm between the Data cable and Power cable
>|of low voltage.
>Unless you are smart and run shielded twisted pair. Why allow any
>interference?
Shielding and twisting offer two different types of rejection.
Shielding is use to supress electrostatic interference - high
frequency interference such a radio/tv/cell - any high end signal.
The shield must be ground to be effective, and really should
be grounded at one end only.
Twisting offers protection againt electromagnetic interference,
that being low frequency such as that emitted by transformers,
motors, flourescent lights, etc.
As the wire is twisted first one pair of the wire and then the
other pair of the wire will be the first to the the interference
from the problem device. The induced signal will be flowing
in the same direction in both pairs. If the signal induced
on the first pair traveled left to right, so would the signal
induced on the second pair. Since these are a pair to complete the
circuit these then become voltages heading the opposite direction
when viewed from the end of the and in the circuit cancel each
other out.
>|9- Always use solid wire for cable run from the patch panel to the
>|workstation and stranded wire for the patch cable form patch panel
>|to the hub and from wall socket to the computer.
>Why the difference in solid versus stranded?
Solid is best for long runs where you start with punch-down blocks
on each end of the run. Using solid cable for the final connection
will cause the cable to fail if flexed too often.
>How about the patch cable from hub to nic?
Flex
>Or from hub to wall socket?
Flex
>Or from patch panel to wall socket?
If you mean a patch panel in one room that connects to a patch
wall socket in another room it should be solid.
Tony Blackwell
(..the good stuff deleted for space reasons..)
See:
> http://www.belden.com/products/catalog/help/techcompetitor.htm
> for a more coherent explanation.
>
Not at all! Your own answer to the post was elegant and coherent.
Always good to hear someone who knows a subject.
Tony Blackwell
Seth R Arnold
>>Why the difference in solid versus stranded?
>
>but that's the official pitch line. Stranded also survives abuse
>(tripping over the patch cables on the floor) better than solid.
I have also heard that the electrons travel along the outside of the wire;
stranded wire has more surface area than solid wire, therefore (the thinking
goes :) less resistance.
my $0.02.
--
Seth Arnold | ICQ 3172483 | http://cswww.willamette.edu/~sarnold/
I prosecute unsolicited bulk emails, using the RealTime BlackHole
List. You should too. Ask me how, or visit http://maps.vix.com/rbl/
Dan Star
James Andrews wrote:
>
> I'm no network expert, but if I remember correctly the only cabling
> difference between 10baseT and 100BaseT was the max length of the cable
> from the hub. For 10BaseT I believe the max distance was around 100
> meters and for 100BaseT the max distance was around 20 meters. Then
> again I could be wrong...
Max length for 100BaseT from repeater is 100M. See:
http://wwwhost.ots.utexas.edu/ethernet/100quickref/ch10qr_4.html
--Dan
MrSmiley
|On Sat, 24 Jul 1999 Jeff Liebermann <je...@comix.santa-cruz.ca.us> wrote:
|>>Why the difference in solid versus stranded?
|>
|>Solid wire tends to break if moved too often. I've never broken one
|>but that's the official pitch line. Stranded also survives abuse
|>(tripping over the patch cables on the floor) better than solid.
|
|I have also heard that the electrons travel along the outside of the wire;
|stranded wire has more surface area than solid wire, therefore (the thinking
|goes :) less resistance.
|
|my $0.02.
Sorry, back to electronics lab for you (and me it seems).
Resistance can be expressed as
(l)
R = ---------- ;
(sigma)(A)
where
l = Length of the wire
A = Cross-sectional Area of the wire
sigma = Conductivity of the metal wire
Resistance is a function of the length of the wire and of the cross sectional area,
not of the surface area.
Also Ohm's Law for a Linear Conductor,
Given:
J = Current Density Vector
sigma = Conductance of the metal
E = Electric Field Vector.
We have:
--> -->
J = (sigma) (E) ;
Now the E field for a dc setup has a potential at one end and
ground at the other end of a straight copper wire. The E field points
along the wire and thus the current denisty points in the same direction.
This can show that for the electrons to move away from the center of the wire,
there would need to be a potential. As there is no potential source in
the center of the wire, the electrons distrubute themselves in a uniform
density. Now electrons are free to travel in metal, and if all the
electrons were somehow on the surface, they would be repel each
other into the interior of the metal.
Now I have to think about those other posts and that mag field stuff.
M
Barry Smith
<MPG.1206f201d...@news.earthlink.net>...
<snip>>
sectional area,
> not of the surface area.
That's kind of like saying the area of a circle is a function of the
radius, not the diameter. :>)
If you know length and cross-sectional area, can you calculate the surface
area? Yes. So, if resistance is a function of the length of the wire and
of the cross sectional area, it is *also* a function of the surface area.
-Barry
Bill Vermillion
Seth R Arnold <sar...@redwood.willamette.edu> wrote:
>On Sat, 24 Jul 1999 Jeff Liebermann <je...@comix.santa-cruz.ca.us> wrote:
>>>Why the difference in solid versus stranded?
>>Solid wire tends to break if moved too often. I've never broken one
>>but that's the official pitch line. Stranded also survives abuse
>>(tripping over the patch cables on the floor) better than solid.
>I have also heard that the electrons travel along the outside of
>the wire; stranded wire has more surface area than solid wire,
>therefore (the thinking goes :) less resistance.
Electrons traveling on the surface is true in the higher frequency
range - typically in RF - that's why transmission lines are long
big copper pipes often with nitrogen in them.
I suspect this hold true in the ethernet world, but resitance
isn't a problem in the 100 meter maximum run - it's the length that
gets you - as the pulses become more deformed because of
capacitance which rolls off the high end.
One reason for using the pipe concept in RF is that if it travels
only on the surface 90% of the copper content is not being used.
And that stuff is not cheap.
Jeff Liebermann
Arnold) wrote:
>I have also heard that the electrons travel along the outside of the wire;
>stranded wire has more surface area than solid wire, therefore (the thinking
>goes :) less resistance.
Think impedance instead of resistance. When you're working with
100-250MHz carrier rates and 50Mhz symbol rates, you're playing with RF
not DC resistance. Any difference in DC resistance of stranded vs solid
over the maximum 100meter segment length is negligible when compared to
the 100ohms impedance of the twisted pair.
Geometry will tell you that stranded wire has a greater surface area and
therefore less loss at RF frequencies. However, life is not so simple
above 50Mhz or so. Bends in the wire, irregularities, proximity to
metallic objects, and variations in twist rate, will all contribute to
losses and reflections far in exess of any differences caused by a
surface (skin effect) losses. Stranded wire tends to be worse in most of
the aformentioned. It bends easier, therefor edge reflections are worse.
It's more irregular, therefore dispersion is worse. Because it's more
flexible, the twist rate tends to move around.
However, all the above are very minor compared to losses from crappy
cable assembly techniques and the disgusting RF characteristics of the
common RJ-45 connector. If I had to put a number on the contribution of
the differences between solid and stranded CAT5, I would say that it the
surface area contributed to perhaps 1/1,000 of the total loss picture.
Belden has a nice article with mangled formatting on the topic. Note how
the impedance varies radically along the cable length, and is only close
to 100ohms in the graphs. This is the real difference between CAT5 and
the newer cables that are rated to much higher frequencies. The newer
cables have much less manufacturing and therefore impedance changes.
http://www.belden.com/products/ftvimptp.htm
The impedance becomes critical only in 1000baseT. 1000baseT uses hybrids
to seperate the direction of data flow on each pair of wires. Each end
has 4ea hybrids for each of the pairs running at 125Mhz symbol rate. The
directivity (how well it seperates data going in opposite directions) of
these hybrids is heavily dependent upon the termination impedance which
is primarily determined by the cable impedance. If the cable impedance
is all over the graph, or if variations in impedance along its length
cause waveform distortion, 1000baseT is going to have problems.
Kenneth McCormick
|Recentky, sar...@redwood.willamette.edu said...
||On Sat, 24 Jul 1999 Jeff Liebermann <je...@comix.santa-cruz.ca.us> wrote:
||>>Why the difference in solid versus stranded?
||>
||>Solid wire tends to break if moved too often. I've never broken one
||>but that's the official pitch line. Stranded also survives abuse
||>(tripping over the patch cables on the floor) better than solid.
||
||stranded wire has more surface area than solid wire, therefore (the thinking
||goes :) less resistance.
||
|
| --> -->
| J = (sigma) (E) ;
|
|
Ok, that bunch of hooey doesn't apply to ethernet, which transmits
as pulses of high frequency waves.
Ken
Vania Lolham
you haven't told us if your new HUB is autosense 10/100 and what is your old
hub's speed 10 or 100 or both?
are your network card's speed set at auto ?
you can mix 10 and 100 NIC's within your network providing your HUB is
capable of auto negotiation or autosense.
you have to remember entire network can operate at 10 or 100 if all the
cards are set at auto and no matter what type of HUB you are using.
first analyse your connection on the hub is it indicating 10 or 100 ?
this should solve your problem.
Good Luck