Technology, Support, and News Forum

[sirnewf Profile]

We got a cat5 drop here that we are experiencing network problems with.

The cable is a Nordx NT-BDR + CAT5, I ran some detailed tests on it and the Propagation delay is out of spec, up in the 250ns range. Acording to spec sheet on the NT-BDR + Propagation delay should be around 20ns. The drop is about 150 feet, well within specs. All other tests pass with flying colors, and in fact the default tests of my tester pass too. Only when I run the advanced tests I get failures.

The wrkstn will run fine for hours on end, but then just drop off the network for a sec. Thus causing the user to loose documents from time to time. The issue is not with the wrkstn, since we have just replace it recently, and the issue existed before replacement. Looking at the syslogger for the cisco 4000 switch, the port in question drops off 2 or 3 times a day.

What could we do to reduce this delay?

What would cause this delay to be so high?

SirNewf


CABLE TEST

Cable ID: 28 - SJOVIC Test Summary: FAIL

Headroom: 10.6 dB (NEXT 36-45)

SITE: Date / Time: 04/29/2004

OPERATOR: City of Yellowknife Test Limit: CAT5 BELDEN

Limits Version: Cable Type: NT-BDN +

Software Version: V05.00 PentaScanner+ S/N: 38P95EA0255

NVP: 68% Injector S/N: 38T95D00633



Wire Map: PASS Expected Actual

Penta: 12345678 12345678

Remote: 12345678 12345678



|Length |Prop. |Delay |Resistance |Impedance |Attenuation |

| |Delay |Skew | | |Result Freq. Limit|



Propagation delay

Pair |(ft) Limit |ns Limit |ns Limit |ohms Limit |ohms Limit | (dB) MHz (dB)|

12 |152 0-328 |227F 45 |2F | | |13.8 99.0 23.8 |

36 |153 0-328 |228F 45 |3F | | |13.7 97.0 23.6 |

45 |151 0-328 |225F 45 |0F | | |12.9 98.0 23.7 |

78 |154 0-328 |229F 45 |4F | | |13.5 94.0 23.2 |



| Main Results | Remote Results |

|Worst Margin |Worst Value |Worst Margin |Worst Value |

|Margin Freq. Limit |Margin Freq. Limit |Margin Freq. Limit |Margin Freq. Limit |

Pair | (dB) MHz (dB) | (dB) MHz (dB) | (dB) MHz (dB) | (dB) MHz (dB) |

RL | | | |

12 | | 9.7 99.0 10.0 | | OVR 80.0 10.9 |

36 | | 3.5 97.0 10.1 | | OVR 62.0 12.0 |

45 | | 9.2 96.0 10.1 | | 5.5 49.0 13.1 |

78 | | 6.4 100.0 10.0 | | OVR 92.0 10.3 |



PSNEXT | | | |

12 | | | | |

36 | | | | |

45 | | | | |

78 | | | | |



PSACR | | | |

12 | | | | |

36 | | | | |

45 | | | | |

78 | | | | |



NEXT | | | |

12-36 | | 11.9 77.9 29.0 | | 14.1 91.2 27.8 |

12-45 | | 13.7 92.7 27.6 | | 13.8 49.2 32.4 |

12-78 | | 14.6 90.9 27.8 | | 12.4 91.0 27.8 |

36-45 | | 10.6 93.5 27.6 | | 11.6 96.8 27.3 |

36-78 | | 13.3 91.7 27.7 | | 12.0 82.0 28.6 |

45-78 | | 12.2 78.9 28.9 | | 17.7 98.0 27.3 |



ACR | | | |

12-36 | | 24.0 78.0 8.3 | | 27.9 91.0 4.9 |

12-45 | | 26.4 93.0 4.5 | | 30.3 82.0 7.3 |

12-78 | | 28.2 91.0 4.9 | | 26.0 91.0 4.9 |

36-45 | | 23.4 94.0 4.4 | | 24.6 97.0 3.7 |

36-78 | | 26.8 92.0 4.7 | | 27.7 99.0 3.4 |

45-78 | | 24.4 82.0 7.3 | | 30.6 98.0 3.6 |

[dlindahl Profile]

I had one that did a similar thing.

What we found was that the connector was bad and would go intermitent as the day got warmer. This appeared to be because the wire insulation would expand causing the bite pin to loose contact. Bad plastic holding the pin.

Don't know if yours could be the same issue or not.

Dave

[sirnewf Profile]

Before I ran my tests, I replace the patch cables with STP Cat5 to elimate the patch cables from the tests.

Now this does not mean that the connector jacks themselves are toast.

However I am not totally sure what can cause such a high Propagation delay. I think i may phone nordx or fluke networks to get a better understanding.

Sirnewf

[Weaver Profile Website]

Before I ran my tests, I replace the patch cables with STP Cat5 to elimate the patch cables from the tests.

Now this does not mean that the connector jacks themselves are toast.

However I am not totally sure what can cause such a high Propagation delay. I think i may phone nordx or fluke networks to get a better understanding.

Sirnewf

Are you going through a patch panel at all? Or is the node plugged directly into the switch?

If you are going through a patch panel, try a different port on the patch panel or don't use it altogether.

-Weaver

[glexp Profile YIM]

Prop. delays in switching circuits are dependent partly on the capacitance of the circuit in question. If there is added capacitance in the circuit, it will take a longer time for the signal to switch from 1 to 0 or from 0 to 1 since it will take extra time to discharge (1 to 0) or charge (0 to 1) the circuit. Since the workstation has been changed, have you checked the hardware at the other end? How many mechanical connections (jacks) are there between the workstation & switch? Every mechanical connection can add to the prop. delay. A bad contact between a plug & jack can be thought of as "capacitance"

Hope this doesn't add to the confusion

[sirnewf Profile]

This is a single run utp cat5 cable run. from the jack at the usr wrkstn to a nordx bix patch panel in the computer room. Total lenght is 150 feet. No line breaks one cable.

sirnewf

[Weaver Profile Website]

This is a single run utp cat5 cable run. from the jack at the usr wrkstn to a nordx bix patch panel in the computer room. Total lenght is 150 feet. No line breaks one cable.

sirnewf

Have you tried using a different port on the patch panel like I suggested? Or tried skipping the patch panel altogether?

-Weaver

[sirnewf Profile]

Currently, I have no advailable ports on my patch panel to try your suggestion. However I do have some on order, but will be a week before it arrives.. As for running the cable directly to the switch, (which does not conform to 568 standards) the cable will not reach.

in addition, we only use IBDN cable/jacks/ports

Sirnewf

[Weaver Profile Website]

Currently, I have no advailable ports on my patch panel to try your suggestion. However I do have some on order, but will be a week before it arrives.. As for running the cable directly to the switch, (which does not conform to 568 standards) the cable will not reach.

in addition, we only use IBDN cable/jacks/ports

Sirnewf

If you have no available patch panel ports and are unwilling to temporarily bend ANSI/EIA/TIA 568 and circumnavigate the panel itself, then you have ruled out the checking of a proven source of problems, the patch panel.

You seem like a bright guy, so let's talk like bright men. While I do not have an outline of your complete network, I do know that the Ethernet propagation delay relative to one of your nodes is less than optimal. We do not know what is causing this.

This case is an Ethernet issue, so start at the bottom. The first step is to rule out layer 1 issues, physical layer issues, hardware and cabling. This is going to include all cabling and intermittent physical (not logical) segmenting devices.

You know how to sequence according to T568A and T568B, that I am assuming. However, make sure that there isn't an excessive amount of untwisting at any wire termination point. Cat5 UTP can be "untwisted" on endpoints to a point where signal crosstalk becomes an issue.

After your cabling has passed an in depth test (crosstalk, impedance, strength, etc) by something more than a $50 tester off of Newegg, you can move to physical (not logical) segmenting devices. These include patch panels (not repeaters or switches, while repeaters are not considered layer 2 devices, they are layer 2 aware in some senses). Change the port on the patch panel, check the patch panel cable for the same deficiencies that you would check a normal patch cable for, excessive untwisting, excessive bending, etc. Skip the patch panel altogether if the problem still exists. Put ANSI/EIA/TIA 568 on hold while you fix your problems.

Then move to layer 2, bypass or physically switch layer 2 aware and capable devices, try different ports. The idea is to move up the chain changing one variable at a time, systematally isolating the issue, until the problem kicks you in the face.

-Weaver