Trouble connecting over 10Mbps Full Duplex

[bud375]

I have a Cisco 3750 connected to a wireless radio on a tower. They connect via Cat5e and the radio is powered via power over ethernet. I have a patch cable between the power injector and the switch, a patch from the power injector to an inline lightning arrestor, and then a cable from the lightning arrestor to the top of the tower where the radio is, all of which total about 150ft of cat5e.

Now that you understand the setup, my problem is that anything over 10mbps full duplex starts having connection issues, packet loss or complete connection drop. I have set both switch and radio to Auto/Auto, hard coded 100mbps Full Duplex on both, and hard coded 10mbps full duplex on both. The only combination that works with 0 loss is 10mbps Full Duplex. When i run the cable test from the Cisco while coded to 100mbps, the test fails and says there is a problem at about 10 meters which is about the distance to the lightning arrestor, however when coded to 10mbps, the test passes.

I understand that this looks like the lightning arrestor to be the issue and that is always my first check, but my question is:

Why would the connection have a problem at 100mbps, but not 10mbps?

Thanks

 

[bill001g]

Hard to say how a lightning arrestor works with ethernet cables. I would test connecting the injector directly to the radio to prove that is what it is.

This is always some form of cable problem when devices do not negotiate the correct rates. With all the extra stuff in your "cable" it is hard to guess what is the problem. Pretty much the main difference in the speed is the method the end devices use to encode the data into the cable. 10m is much more forgiving than 100m or 1g.

If it is the lightning arrestor I would use a shielded cable and ground the shield to the tower instead. In most cases if you take a lightning strike to the tower everything will get destroyed. I saw a 40ft steel tower that got hit and it actually melted the metal of the tower itself in places. Pretty much the ethernet cable would vaporize if it took a direct hit.

 

[bud375]

Turns out the lightning arrestor had gotten water in it which is what was expected and replacing it resolved the problem. Assuming a short across one of the pairs from the water. However, my real question here is not what was causing the problem, but why would 10mbps/FD work and 100mbps/FD wouldnt. Does 100mbps/FD use a different set or more pairs than 10mbps/FD?

Hard to say how a lightning arrestor works with ethernet cables. I would test connecting the injector directly to the radio to prove that is what it is.

This is always some form of cable problem when devices do not negotiate the correct rates. With all the extra stuff in your "cable" it is hard to guess what is the problem. Pretty much the main difference in the speed is the method the end devices use to encode the data into the cable. 10m is much more forgiving than 100m or 1g.

If it is the lightning arrestor I would use a shielded cable and ground the shield to the tower instead. In most cases if you take a lightning strike to the tower everything will get destroyed. I saw a 40ft steel tower that got hit and it actually melted the metal of the tower itself in places. Pretty much the ethernet cable would vaporize if it took a direct hit.

 

[bill001g]

No it is the same pair. Been a long time since I looked at how ethernet really puts the signals on the wire. I think it even uses the same voltages so it would have to be the length of the pulses that somehow works on one but not the other.