Help requested with Network

[Fujiflyer]

I'm having a problem connecting my laptop to the network at work (100baseT). Up to recently I have used a short (3m) utp cable to connect it to a local hub which is a Netgear FS108.

Due to the need to work further away from the same hub I require a much longer cable - about 35m. I duly purchased the materials necessary to prepare it, from RS. Initially I have made up a 2m sample in order to check that it would work, which it did. I then went on to prepare a 60m length in order to check the longer run would be okay, so as to avoid the embarrassment of my extension not working once our electrician had put the required cable run in place. Unfortunately it doesn't. Two lamps on the hub port flash together. I have checked the cable for continuity and ensured that no short circuits exist between lines.

Cable Type: UTP, Cat 5, rated to TSB36 for speeds up to 100Mps - RS part number: 196-311

Hub: Netgear FS108, IEEE 802.3-LAN, IEEE 802.3U-LAN; Ethernet, Fast Ethernet

I would have thought that 60m was not particularly huge for a network link although I do not have the actual spec' for the max cable length allowed. From an electronics hardware point of view it seems like the data must be being excessively attenuated by the cable length or maybe some sort of impedance mismatch is setting up standing waves, however I thought that by using cat 5 utp cable on the said hub, these worries would have been taken care of.


:confused:

Please help me :(


Fujiflyer

[FL310]

There are several factors which may cause a problem.

Look here (http://www.netgear.com/product_view.asp?xrp=3&yrp=32&zrp=52) for the Netgear specs. They offer a 24h free phone helpline. Unfortunately they do not state recommended maximum length but, it would surprise me if it less than 100 metres.

Interference with the cable is always possible, don't bundle the cable to a roll when testing.

One more thing is to check the LAN adapter of your Notebook, is that capable of 60 metres???? Some are not and this is quiet frequently a problem.

[Fujiflyer]

FL310

Thanks for your reply. As you suspected I did place the test length in a coil so I'll look into this.

I believed that we'd tried our 60m cable length with a desktop PC, to no avail however I am not certain.

I'll do some more tests on Monday.


Best Regards

Fujiflyer

[PPRuNeUser0171]

The maximum length of CAT 5 is around 120meters
I personally wouldn't push is past 100 metres.


When you plug the cable in to the PC and hub do you get any lights on the hub? If you do not then the cable is faulty.

Are you running IP? If so do a PING w.x.y.z where w.x.y.z is the IP address of another PC on your LAN

Is is wired as per the IEEE (I think that's the correct body) specs?

Have you considered using 802.11? It's a wireles protocol that can easily do 35 meteres - There are some security aspects to it but it works well.

HTH,

--
Gary Williams

[Fujiflyer]

Still no success. :(

Have shortened the cable to 35m, straightened it out however the network is still not accessible.

williamsg, I get a slowly flashing led on the hub port where the "long" cable connects into it. The cable is to the IEEE specs and has been checked for dc continuity.

I'm not sure what to try next. Apparently the hub we're using is a "switch" type which should act as a repeater as far as cable lengths are concerned. However I'm still sure the problem relates to signal attenuation.

Out of interest a fairly short cable length made up using the cat 5 cable I purchased (see earlier post) - about 4m seemed to result in a longer network initial connection delay than the cat 5 cables we had been using between PC's & hub previously.

Fujiflyer

[--o-CK-o--]

Check that all works OK with a known good lead, just in case its not the lead that's at fault!
What other kit is connected to the hub? is that all working OK?
Check you are not connecting to port 8 with the uplink button in the wrong position (obviously!!!):)
I always forget the colour sequence when making up patch leads, as I don't do it every day, so I always check an existing one and copy it - in case that helps. Certainly with a correctly connected cable, you should have no problems with 60m unless you are running the cable near power cables - then you should maybe use shielded twisted pair (STP) cable.
Good luck!

[FL310]

about 4m seemed to result in a longer network initial connection delay
There is something wrong with the type of cable you are using!!!!

[SLB]

did you check for crossed pairs? DC continuity check will not show this problem. If you have not properly kept the pairs on the correct pins, you completley lose the noise canceling property of the twisted pair, in fact, you generate noise. While the noise may not be an issue on a real short run, it will be on longer runs, causing bad packets. The 4m cable that acts "slower" would be an indicaiton of this problem. Good luck:eek:

[Fujiflyer]

I just want to let you all know I have managed to get my problem solved. :)

SLB, you were spot on - the Rx pair (which are on pins 3 & 6) had not been wired into the same twisted pair. I had tried to copy the colour sequence from an existing cable however had misread a couple of the bicolour lines in this cable because they were embedded in the connectors. I guess, also that I was happy to assume that the twisted pairs would run side by side, in turn into pins 1 to 8 of the RJ45 connector. Hence the DC charactoristics were fine but the noise immunity and transmission line charactoristics were totally wrong.

Once I read the messages today I went back and checked a pre-made working cable carefully (+ internet info' check). Hence the problem was found.

FL310, williamsg, --o-CK-o-- and of course SLB - thanks very much for helping me with this problem. It is a credit to yourselves and the spirit of the PPRUNE community that I was able to get this quality of help so readily. Our company IT dept didn't have much of a clue. :rolleyes:

Fujiflyer :) :) :)

[PPRuNeUser0171]

Gary,

when you mention 35m - are you basically saying that it will work in the scenario as given by FujiF or is that the maximum it can achieve that you have seen?

Am asking since I'm about to set a wireless network for a customer but Belkin are quoting a distance of 1800feet, or 548m. Is that wishful thinking?

I'm saying that it will work in the scenario given by FujiF.

I've seen wireless working a total distance of about 3m in a test enviroment - Where I work and have my test network it's not practical to test wireless over much bigger disatances than that :(
However, I see no reason why it would not work for much longer distances!

--
Gary Williams

[PPRuNeUser0171]

FL310, williamsg, --o-CK-o-- and of course SLB - thanks very much for helping me with this problem. It is a credit to yourselves and the spirit of the PPRUNE community that I was able to get this quality of help so readily. Our company IT dept didn't have much of a clue.

You are welcome and I'm glad the problem has been fixed.

I cannot comment on other IT Dept's but I do have to deal with some companies IT Depts on a day to say basis and some are not the most helpful :rolleyes:

If you want a laugh at IT's expense take a look at http://www.rinkworks.co.uk/stupid
:)

--
Gary Williams.

[FL310]

you are most welcome....on second thought, even Linux would not have solved that problem...but http://www.edonkey-town.com/forum/black/smilies/drink.gif

[--o-CK-o--]

I love it when a plan comes together....

(love the icon FL310 too)

[SLB]

Glad to hear the problem is sorted. :D I ran across a warning for this issue when doing research on the topic before CAT 5 wiring my house in '96. As for the IT guys, they rarely install or make cables, they typically use pre-made cables for the short runs and hire installers for the stuff in the walls, etc. Professional network cable installers have a fancy (and expensive, about 300 BP) gadget that catches crossed pairs by sending tests signals and evaluating the result, so the IT guys have likely never seen this issue.:rolleyes:

[pied piper]

The gadget you refer to is either a "NEXT" (near end crosstalk) tester or an TDR (time domain reflectometer). but there is a cheap DC test set (about GB20) that will catch reverse and crossed pairs as well disconections. The cheap device is a D200 LAN cable tester from Darkstar Technologies.:cool: