Why did this repair work? 5

[wahnula]

Hello wire-heads,

I am looking for some enlightenment. I had a problem that was repaired, but when I posted this solution in another forum I was questioned as to why it worked, and since I didn't do the repair, I have no clue. The backstory:

I ran a 100' length of CAT6 between a gigabit switch and the LAN NIC of my server, and had about 10-15' of cable left over, so I coiled it neatly in a 12" loop and laid it on the floor by the switch. I started having dropped connectivity issues between the switch and the SBS2003 server. It also showed up as an error in my server eventlog that the connection was lost and then regained several times a week. The workstations all confirmed this with the requisite Windows pop-up balloon.

When I asked my MCSE tech, who was in-office on another issue, about the lost connectivity, he simply picked up the loop, gave it a half-twist, and placed a zip-tie in the middle, leaving a "figure-eight" looking coil. The problem has been gone ever since. He muttered something about "canceling it out" as he is reluctant to share "trade secrets" with me that could put him out of a job, I guess.

I would like to know if anyone is familiar with this repair and could explain it to me. It seemed so intuitive to the tech that I thought it must be a common problem, but nobody in forum602 had heard of it and some were doubting that it could have worked...but it did...so here I am seeking the science behind the repair.

Can anyone shed some light on this? Thanks.

Tony

Users helping Users...

 

[Trotski]

Farradays Law !

 

[edfair]

I'm not sure that it fixed the problem, which is one of crosstalk, but it changed the effect of it to something that isn't bothering you.

Now that you know how to resolve it, why not go back and create the problem again and see what smaller or larger diameters of the loop do.

Ed Fair
Give the wrong symptoms, get the wrong solutions.

 

[dagwoodsystems]

BICSI recommends storing service loops in a figure 8 pattern to reduce EMI coupling. That's fancy way of saying "by coiling your extra wire like this, you'll lessen the chance of your data getting contaminated by the antenna-like properties of wire".

Faraday's Law, as Trotski alluded to, refers to one of several discoveries about the interaction between wire and magnets. Current (electrons moving through wire) creates a small magnetic field around that wire. Conversely, a varying magnetic field near a wire causes current to flow through that wire. The latter effect is called induction.

Take a length of copper wire and wind it into a coil. Strip the insulation off either end of the wire an apply a 9V battery. You have made a magnet...specifically an electomagnet. Now let's see if we can recreate the opposite effect of Faraday's Law: Run an unshielded CAT6 cable over several florescent light fixtures, past a bank of printers and fax machines, then behind the microwave oven in the break room. Measure induced voltage at either end of the wire with a multimeter. You don't have one? No problem. We can observe the effect by instead driving data over this wire at gigabit speeds and frequency to see if the unwanted induction creates data contamination and connection issues.

So what would happen if you made coil of wire, but wound half of it in one direction and half of it in the opposite direction? Wouldn't the net effect of EMF be cancelled out? I suppose that's the idea behind the figure 8 coiling technique.

I feel that the cancelling effect of such a coil--if any--is negliable. The way to control EMF is by pair-twisting, shielding and grounding, differential mode signalling and sheer physical distance from noise sources. Termination techniques are critical as well. If you ask me, something else was at play in your data room. The manner in which fifteen feet of slack wire is coiled will not measurably affect the stability of that connection.

Tim Alberstein

http://www.dagwoodsystems.com

 

[edfair]

The coil of wire is the primary of a transformer if there is a secondary coil. By itself it is a choke to change the phase between the received signal's voltage and current.

All this is then tied up with the crosstalk prevention of the different pair twists so there is no telling what was really getting to the receiver. But whatever it was, it was failing intermittently.

Ed Fair
Give the wrong symptoms, get the wrong solutions.

 

[skip555]

I ran a 100' length of CAT6 between a gigabit switch and the LAN NIC of my server

is this solid cat6 with a end crimped on it ?

if so my guess is a poor crimp that made contact when he moved the loop . nothing about the loop per se

try wiggling the connection and see what you get

 

[dagwoodsystems]

Like Skip said, an iffy crimp could easily be at fault here. Moving the cable around (while forming the loop) wiggled the connector back into a semi-stable position.

Any chance this tech re-routed the cable prior to sparing up the excess? Depending on what you've got mounted down low, getting the cable up off the floor might have helped. Do you have other CAT6 cables running into this room? How is the slack managed for those runs?

The thing about intermittant problems is just that--it occurs randomly and on occasion. You might not be out of the woods yet.

Tim Alberstein

http://www.dagwoodsystems.com

 

[wahnula]

To all:

Thanks for the explanations. Thanks to dagwoodsystems and edfair for "dumbing down" Faraday's Law for me. I aced Architectural Calculus in 1979 but the lingo used in most descriptions of the Law prohibited me from getting my head around it.

Skip555: The cable has molded ends and was purchased complete.

dagwoodsystems: There was no other work done by the tech. The cable is paired with a CAT5 and, once it leaves the floor, travels through a rigid 3/4" Carlon PVC conduit with long sweeps at each end above the hallway that separates the two rooms. The office is a converted double-wide trailer so no safe crawl space or attic, the conduit is visible in the hallway. The CAT5 has its slack coiled at the other end, near the server.

This topology was established in 1998 when my office held the office "server", a Win98 machine that stored the company files & calendar and had an HP Travan tape drive. The "new" 1999 NT 4.0 server went in another office due to the noise, so all the cable runs terminated in my office with a single CAT 5 running into the copy room. When we migrated to SBS2003 in 2005, I added the second cable, and CAT6 was recommended. The only other CAT6 cable heads out in another direction from the switch. Slack is at the far end by the hub. All new clients are wireless!

This particular coil was on the floor next to the switch/SonicWall/Cable modem/UPS which were all on a small 12" table immediately adjacent. Since this "repair", which has been over six months, not a single dropped connection, including when the office was re-carpeted last week and all the equipment raised to about 36". The loop is now suspended in midair and needs re-doing, along with the rest of the cabling, but if there was a bad end it would be going nuts right now, as it is under tension. <Yes, I know> Remember this error shows in the eventlog when the connection resets.





Tony

Users helping Users...

 

[edfair]

Is this a patch cord? Of flexible wire rather than solid?

If so, you may be having some other issues as specs specify 22 feet of patch cable in a run.

Ed Fair
Give the wrong symptoms, get the wrong solutions.

 

[wahnula]

Patch Cable...Hmmm...This cable has PATCH CABLE laser printed directly on the cable. I bought it from a local tech shop:

http://www.directron.com/cat6100foot.html

Complete text from cable:

E142890D 1000X CAT6 GIGASYSTEM (UL) CM 75°C UTP 4PR 24AWG CSA CM 75°C FT4 LL43774 ETL VERIFIED TO TIA/EIA-568-B.2-1 draft 11 PATCH CABLE-LF-ROHS

Tony

Users helping Users...

 

[edfair]

FYI,

http://www.duxcw.com/faq/network/cablng.htm

33 feet limit on patch cables, although most people can get by with longer.

Ed Fair
Give the wrong symptoms, get the wrong solutions.

 

[skip555]

from eds link

3 meters (9.8 ft.) has been allowed for patch cords from the wall jack to the work area/PC. One could probably make these cables longer if the total length between Ethernet devices does not exceed 100 Meters. I would not make patch cables with stranded cable longer than 20 feet

so what your linking to is not a spec or standard but a opnion

a stranded cable will decrease the 100 meter but 100' should work just fine if the cable is properly terminated

 

[wahnula]

So, I'm getting the drift that "patch cables" are pre-made and of a different ilk than network cable terminated with crimp or clamp connectors. I had the rest of the office wired professionally with bulk wire and crimps, I had no idea that patch cables were different inside.

All my years in the music/audio business, a pre-made cable was simply coax wire or speaker cable with ends molded on, if you knew the gauge of the wire it was exactly the same as homemade cables.

So...what exactly is the difference between "network cable" and "patch cables" inside? I know this thread has wandered off, but learning is all I care about. I have some crimpers and ice cubes, but have only made a few custom-cut cables, I will redo them all if there is/will be a difference.

As I mentioned at this time there are no connectivity issues, and Windows shows "connected, 1Gbps" on all GbE machines.


Tony

Users helping Users...

 

[skip555]

patch cord is made of stranded cable to be flexible

as a result higher resistance so less distance than solid

(but 100' should be no problem)

I had the rest of the office wired professionally with bulk wire and crimps, I had no idea that patch cables were different inside.

crimping solid wire is not professional my opinion

Best practice would be solid wire to a jack /patch panel then a patch cable to a device

 

[wahnula]

crimping solid wire is not professional my opinion

Best practice would be solid wire to a jack /patch panel then a patch cable to a device

Good point...who knew? Not me or the boss. Although to be fair it was run by a pro but on his off-time as a favor...and he did terminate ONE run into a jack. Thanks for the answers skip555.

Tony

Users helping Users...

 

[hawks]

My understanding is; A "patch cable" is normally made with stranded copper instead of solid. The stranded is to allow the cable movement and have a lower chance of loosing a connection to the plug than a solid cable would. But as we know stranded does not conduct as well as solid so that’s why the length of a patch cable is so short.

 

[edfair]

The stranded has multiple strands of a different gauge than the solid but overall it has more cross section of copper. It isn't to give a better connection, but to be flexible to keep it from breaking from bending.

Crimping solid wire is fine. But you must use the right kind of ice cube. The solid wire type has a split connector that goes on both sides of the wire while the one for stranded has a piercing type that goes into the stranded bundle. Put the wrong kind on either wire and you will have problems. Trust me on this.

Skip, the quote was what that blog site interpreted but the actual spec does have the limitation. I think it was in the 803 spec itself and possibly from NIST. It has been a while since I saw it. IIRC it was 110m with 33 feet available for patch cables, it didn't care where the 33 feet were used.

 

[skip555]

heres some past threads on the subject

http://www.tek-tips.com/viewthread.cfm?qid=1082060

http://www.tek-tips.com/search.cfm?action=Search&pid=575&requesttimeout=75

 

[skip555]

Crimping solid wire is fine. But you must use the right kind of ice cube. The solid wire type has a split connector that goes on both sides of the wire while the one for stranded has a piercing type that goes into the stranded bundle. Put the wrong kind on either wire and you will have problems. Trust me on this.

Skip, the quote was what that blog site interpreted but the actual spec does have the limitation. I think it was in the 803 spec itself and possibly from NIST. It has been a while since I saw it. IIRC it was 110m with 33 feet available for patch cables, it didn't care where the 33 feet were used.

crimping solid wire (even with a plug designed for it ) is always likely to be problematic and is at best a poor practice

thats what jacks and patch panels are designed for

if your crimping both ends of a solid wire with modular plugs in effect aren't you making a patch cord ?

so would you keep that to the 33 foot max ?

 

[edfair]

I've not done that many solid crimps so I have no history of quality for long term. I've seen a lot that have used solid wire for direct connection into gear rather than making a stop at a patch panel.

I prefer having the patch panel available to make troubleshooting easier, but if the wires are tied down and not flexed the solid is probably as good. IMHO. The problem comes up if the gear needs to be replaced. You either need an exact replacement or you flex the wire and risk losing the connector.

It is only a patch cord if you use the larger wire.
I've always used the example of pipes. The solid is like a small pipe and the flexible is a large pipe and it takes more capability on the sending unit to fill up a large pipe. You can see it on a long run of cable on RS232. The signal with smaller wire will be closer to ideal. Both rounded but the smaller wire less so.