Problems with shorter cables? 2

[JBruyet]

Hey all,

I'm looking at getting some 18 and 24" cables to clean up my wiring closet. However, one of my vendors said that patch cable lengths less than three feet will cause problems: CRC errors and dropped packets. Sounds kind of funny, but I thought I'd post here and see what I get back. Any experiences, links, or even gut feelings will be appreciated.

Thanks,

Joe Brouillette

 

[AvayaNovice]

Hmmm. Two experiences I can relate to this:

A) I used to have a patch panel situated directly beneath a switch, and what I did was place 6" patch cables that I made from patch panel to switch, and it worked fine.

B) Trying to test a 1 foot patch cable with a few different testers has raised major problems, it's almost as if the cable needs to be long enough to attenuate the signal to allow proper transmission.

That being said, short patch cables if from patch panel to device should be OK from my experience, as the horizontal cable from the patch panel off into la la land theoretically extends the length of the cable. If from device to device it's short, I can see problems occuring.

 

[JBruyet]

Hi AvayaNovice,

That makes sense. I think I'm going to run with the shorter cables because they're going to run from the switches to the patch panels like in your experience #A.

Thanks,

Joe Brouillette

 

[skip555]

we use them all the time 1' to 3 foot

my tester won't test the cables that short but what we do is test from the workstaion end to the patch cord

 

[franklin97355]

As a passive part of a total circuit there should be no problem. I think the one meter limit came from the coax days. (but don't quote me)

 

[daronwilson]

I am almost sure there is a minimum length in the neighborhood of 18" but I can't find it right now. If I get time I'll dig around for it, can't remember what book it was in and who's rule it was, but I'm almost positive there is a minimum length for UTP patch cables.



It is only my opinion, based on my experience and education...I am always willing to learn, educate me!
Daron J. Wilson, RCDD
daron.wilson@lhmorris.com

 

[AvayaNovice]

I'm pretty sure there is too, as I've always had problems with it. Again, if it's just part of the horizontal run to the workstation, or whatever device -- then the patch cord just becomes a part of the total run. So no big deal. Where you run into problems is patching from switch to switch for instance, for an uplink, with a 12" cord.

 

[Servamatic]

There is no minimum length, the problem is strictly tester related.
TDR functions cannot accurately test a cable under 3', but on that short of a cable you don't really need TDR. As part of a Channel test there is no problem because you have the total Channel not just patch cords and equipment cords.
We have a certain engineered application that uses an 8" patch cord and everything works just fine.
Since there is no requirement to certify just a patch cord, there is no minimum length requirement. If you were a manufacturer it would be different, but then they do have specialized test equipment for patch cords.


Richard S. Anderson, RCDD

http://www.smartbuilthomes.com

 

[AvayaNovice]

Interesting, maybe it's specific to device.

I have used 8" patchcords several times to uplink Cisco 2924 switches on the gigabit side, and had major problems -- even when they passed on a channel test. It may be the output level.

 

[skip555]

I used to have a patch panel situated directly beneath a switch, and what I did was place 6" patch cables that I made from patch panel to switch, and it worked fine.
did you try it with factory made cable ?

like you say it may be the output level but it could be your crimp rather than the cable length.

just a thought.

 

[AvayaNovice]

I did, I also tested these cables using a tester that was capable of short lengths in a particular mode (can't remember what it was). I also used the same patch cable on a panel with a longer horizontal run just to confirm, worked fine.

 

[wires]

"It may be the output level."

And you really think there would be a difference in levels when using any lenght of UTP under 10?

 

[AvayaNovice]

It seems to be the only thing that could be affecting it, overloading it perhaps. All I know is that the patch cables were good, and I could not get the switches to uplink until I used a longer length. So do you have any better ideas?

 

[skip555]

Reading this I rember the old tie onyx phone systems when you connected two cabinets you needed to use the supplied 6' 8 conductor silver satin cord (pre cat 5 days)

which was a pain becouse the cabinets were about a foot apart

I seem to recall trying a shorter Known good cable and having problms and the tie tech guys saying it had to be 6'

I'm no engineer and the dynamics are a bit beyond me but the bottom line is whatever works ...let sombody else figure out why ..

 

[daronwilson]

Yeah that is a perfect example. I did several of the Tie DS1000 (very similar to the onyx) and I tried it, I swear I made cables, I tried shorter ones, etc. Flat out, it would NOT run unless that digital bus wire was that long.

Often there IS a reason for the design even though we don't readily understand it.

It is only my opinion, based on my experience and education...I am always willing to learn, educate me!
Daron J. Wilson, RCDD
daron.wilson@lhmorris.com

 

[daronwilson]

When using 10Base2 as a medium, there is most definately a minimum distance, that being .5 meters.

This minimum difference is in there because of the variations in the characteristic impedance in the cable itself. In 10Base2, the cable has a 50 ohm impedance with a +/- 2ohm variation. That 2ohm variation can cause up to 4% reflection because of the mismatch of two cables. In this case, the active electronics (MAU) will add to this mismatch with the introduction of it's noninfinite bridging impedance.

According to 802.3 (10.7.2.1), to minimize these reflections the cable length shall be a minimum of .5 meters.

That's the scoop on 10Base2, I'm still looking through the rest of the 1,538 pages to see if there is anything similar for UTP.



It is only my opinion, based on my experience and education...I am always willing to learn, educate me!
Daron J. Wilson, RCDD
daron.wilson@lhmorris.com

 

[Servamatic]

UTP and 10Base2 are two different animals, on coax you do need a minimum distance because you are in effect using a single conductor to send signals, not so with UTP. Even 10BaseT uses 4 conductors.
There are situations when specific resistance is required to achieve proper impedance; the onyx example is a good one.
Since I am no fan of making my own patch cables, my shortest ones are typically 3', except in the case I mentioned earlier where we use an 8" patch cable. That is just what the customer required in the specs.
The other place we use many 1' cables is residential, since it would be impractical to have 3' cables in a distribution device that is not wide open and has no place for slack. Leviton has factory made 1’ patch cables, as does ChannelPlus.


Richard S. Anderson, RCDD

http://www.smartbuilthomes.com

 

[wires]

My guess is that in a short cable there is insufficient "dampening" to overcome any reflections caused by impedence mismatches at the terminals. Depending on how well the interface circuits in the related equipment are designed this could cause the link to fail.

Here is an interesting token ring article:

http://www.iol.unh.edu/knowledgeBase/training/files/tokenring/vjs_thesis/vjthes6.html

"For short cable lengths, multiple reflections can occur and hence the impedance mismatch at the transmitter and cable interface has to be taken into account."

"...as the worst-case would occur at short cable lengths (less than 2 m). The reflected signal was assumed to arrive at the point of jitter measurement on the cable without undergoing any attenuation."

Both coax and UTP will exibit some of the same return loss and refection issues when used for similar purposes. 10Base2 does not use a "single conductor to send signals". If you want a demonstartion simply remove a section of shield from the cable watch the link die.

 

[Servamatic]

Sure the signal will die, but there is a big difference between actually using the shield to send signal and using it to gain reference to ground. Coax is an unbalanced cable, which means it will not compensate for signal fluctuation on the center conductor via the shield. The shield will always remain at 0 volts, because it does go to ground. If you read what I said, I said "in effect you have a single conductor" referring to data transmission.
UTP is a balanced cable, one conductor in a pair can compensate for the other conductor, since neither conductor goes to ground.
Besides the question was not about coax, it was about UTP and I was commenting on a comment making a comparison to coax.


Richard S. Anderson, RCDD

http://www.smartbuilthomes.com