• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

ethernet cards vs cable lengths

R

Raland

Junior Member
At work I sometimes need to extend the network (p2p NetBEUI) out into several large rooms that require me to use 25ft - 100ft cables connected to the closest RJ45 wall outlet in the office area. In doing this I have ran into situations were I could not get connected to the network. The first was with a sony laptop using a usb to ethernet adapter (linksys I think). I tried all the usual suspects but it refused to connect to the network till I moved it back into the office area. I chalk this up the adapter not having enough juice to drive the longer cable run. I looked around some to see if there was a shorter maximum cable length when using this type of adapter but never found anything. I only have this one laptop and adapter so I can't do much to testing. Is this typical performance??
I also had a similar problem when I need to setup a regular pc using a 100ft cable (about 150 ft total run), it was a MSI 6378 motherboard with onboard Realtek 8100 nic. The system would not connect to the network. I checked the cable at a different location and it worked fine and the system would connect to the network when using a typical 12 cable. I again assumed that the onboard nic was somehow too wimpy to drive the cable so I installed an Intel Pro 100+ nic in the system and then there was no problem connecting to the network. I know there are good and bad quality nics but find it hard to believe that the Realtek couldn't drive 150ft of cable. There are of course serveral connection points along the way with wall outlets, patchpanel, hub etc but I've never seen anything that said "subtract 20 ft from the maximum cable length for each connection point"?!?
I have made most of the cabling from a bulk roll of high quality cat5 and they all test ok. I've swapped in good retail cable when I was having these problems and it didn't help. I'd like to hear your similar experiences with both wired and wireless networking. TIA

Ralph
 
By specification, any 10BASE-T or 100BASE-TX NIC should be able to drive through a Category 5 rated cable end-to-end where 90 meters is solid-core, and ten meters (5 at each end) is stranded (total of 100 meters, end-to-end). THe specification also provides for two intermediate cross-connects (IDFs, actually).

If you terminated your own bulk cable, I'd suggest that you didn't use the correct pair-ordering (color sequence) when you crimped on your connectors: this would have a major effect on the cable's ability to carry the signal to the distances you mentioned. You also didn't mention the quality or grade of the cable you used. If it was flat & silver colored, that's the problem.

FWIW


Scott

 
If you terminated your own bulk cable, I'd suggest that you didn't use the correct pair-ordering (color sequence) when you crimped on your connectors: this would have a major effect on the cable's ability to carry the signal to the distances you mentioned. You also didn't mention the quality or grade of the cable you used. If it was flat & silver colored, that's the problem.

this might not be the case... I have a feeling your NIC is faulty. I have seen cards work over short distances and not over a slightly greater distance before... and I don't know if it coincidence but they were also Realtek cards.

Interesting problem... If the Intel card is working fine I would say that your wiring is up to spec. and that your card is just loosing the plot!

Have you checked to see if the card is transferring at a full 100meg? Or does it fluctuate?
 
If you terminated your own bulk cable, I'd suggest that you didn't use the correct pair-ordering (color sequence) when you crimped on your connectors: this would have a major effect on the cable's ability to carry the signal to the distances you mentioned. You also didn't mention the quality or grade of the cable you used. If it was flat & silver colored, that's the problem.

I used the same 1000 ft box for most of the wiring I added here. I don't remember the exact brand (Belkin maybe?), picked it up at Fry's, the usual blue jacket stuff. It is solid core. The cable says:
06096FEET E1580 12 UTP 4PR 24AWG CM/MP LAN CAT 5 75C (UL), C-(UL) VERIFIED TIA/EIA 568A & (ETL), (EC) ISO/IEC 11801

Most of which I recognize. I'm a little troubled by the "568A" as I have wired everything per 568B though I don't ever recall seeing "A" cable or "B" cable. So does using 568A vs B have any effect on how well a cable performs? I know they have different wiring layouts but have no idea when to use A vs when to use B other than to simply pick one and stay consistant.

The 100 ft cable I'm using is good quality solid 5e.

By specification, any 10BASE-T or 100BASE-TX NIC should be able to drive through a Category 5 rated cable end-to-end where 90 meters is solid-core, and ten meters (5 at each end) is stranded (total of 100 meters, end-to-end). THe specification also provides for two intermediate cross-connects (IDFs, actually).

"IDF" ??

That's great info. I know situations vary but how obtainable is the 100 meter max. I used to think that if the max spec was 300 ft then unless I had some sub-standard components I could go 299 ft no problem. Now I'm starting to think along the lines of "your performance may vary" and if I'm out past say 150 ft I should be happy it works at all.


As for the level of bandwidth, I've never tested it. If the "100" light comes at the hub I've always figured the connection was working as good as it could. Can you recommend something to test with?

Ralph

 
I'm with Scott on this one. This really screams cabling problems. What might be happening is you have the polarity on this run reversed or just have a single pin not making good contact. Double check your punch downs on both ends or re-crimp/punch.

I believe the Intel nics will automatically adjust their polarity if you have a pair miswired (say w/orange and orange swapped) which could be why it works.

I don't really know of a "cheap" cable tester that will certify cable. sorry. But if you are making cables you really need to get one. fluke and microtest come to mind. A link light coming on just means the NIC is sensing current on its receive port. If you plug a 100 meg NIC into a switch with barbed wire you'll probably get a link, just won't be movin' any data. good luck!
 
The "568A" spec referenced on the cable jacket refers to the overall cabling specification. The "568A versus 568B" punch/termination is a part of this overall 568A specification.

EVERY production 10BASE-T or 100BASE-TX port should be able to do the 100 meters, or it's defective (or your cable plant is out of spec).
We have some lab test equipment that was not designed for 100 meters, but it's strictly for testing, not production.

Cable guys have two points of reference: the Main Distribution Frame ("MDF") and the Intermediate Distribution Frame ("IDF"). The MDF is the premesis entrance point. It's where all the cable from the street enters the building and is terminated (complete with "entrance protection" for the copper....proper grounding). Signals from the MDF are sent to the "vertical," where they terminate at an IDF (veritcal = MDF-->IDF). From that IDF the signals are sent through the "horizontal:" either to the endpoint (phone, computer, whatever) or another IDF. From the final IDF (you can only have two for "Category" data cabling) it has to go to the end point (Horizontal= IDF-->Endpoint OR IDF-->IDF-->Endpoint).

"Vertical" cabling (sometimes called a "Riser Cable" or "Trunk") is most often fiber, second to that, it's likely to be 25-pair (Cat5 is the highest rated 25-pair so far). If there's an elevator in your building, the shaft will have a grouping of some kind of riser, and the "wire closet" is usually adjacent to the elevator shaft...coincidence? I think not.....

Anyway...sorry to ramble...I hope this answers your questions.....


FWIW

Scott
 
Well as luck would have it I had some time to experiment this morning and it looks like Spidey gets the golden crimper award today. I finally narrowed it down to the wall outlet I was using. The wire layout was right so lacking a better idea I took the punchdown tool to the connection at the patch panel. This seemed to fix the problem, the Realtek connects and operates fine. When I originally had the problem I did try other outlets and was unable to get on with the realtek. My best guess is that the poor connection may have locked up the nic or scrambled Win98's networking so that when I plugged into a good outlet it wouldn't work, then rebooting the system after installing the intel card cleared things up. I'm also guessing that the intel card must have a more powerful output or superior signal processing that allowed it to work when the realtek couldn't.

This is the second time I've gotten bent over by a problem at the patch panel. The last time I had to move the cable to a completely different slot, I tried everything I could think of and still couldn't get a good connection through the panel. It's a 24 port Unicom and it says OWS Cat 5 Panel. Compared to the Gruber 24 port I bought recently the Unicom seem not as well made. Is it typical to have problems with punchdown connections or do I just suck with the punchdown tool?

I do have a good cable tester and check all the cables I make up before putting them in service. It's much too easy to mess up and I don't want to see what happens when a mis-wired cable gets plugged in. I was mainly interested in software for testing throughput or network performance. This seems to be the best way to find a sub-par connection like this without dropping a lot of money on testing equipment.

Thanks for the cabling descriptions. I was also drafted as the "Phone Guy" during the relocation so I have seen most of the things you mentioned but didn't know the lingo. Our phone system is actually pc based and is connected to the network. Ask me how much fun it was to get the nimda worm off it. Thanks again,

Ralph

 
I believe the statistic is something like "70% of all network problems are cable/passive infrasturcture related."

Cable is usually the best place to look for flakey/intermitent/head-scratching type problems. 70% is a pretty safe bet.

Glad you got it resolved.

Take care, Happy Holidays.

Scott
 


<< Cable guys have two points of reference: the Main Distribution Frame ("MDF") and the Intermediate Distribution Frame ("IDF"). The MDF is the premesis entrance point. It's where all the cable from the street enters the building and is terminated (complete with "entrance protection" for the copper....proper grounding). Signals from the MDF are sent to the "vertical," where they terminate at an IDF (veritcal = MDF-->IDF). From that IDF the signals are sent through the "horizontal:" either to the endpoint (phone, computer, whatever) or another IDF. From the final IDF (you can only have two for "Category" data cabling) it has to go to the end point (Horizontal= IDF-->Endpoint OR IDF-->IDF-->Endpoint).

"Vertical" cabling (sometimes called a "Riser Cable" or "Trunk") is most often fiber, second to that, it's likely to be 25-pair (Cat5 is the highest rated 25-pair so far). If there's an elevator in your building, the shaft will have a grouping of some kind of riser, and the "wire closet" is usually adjacent to the elevator shaft...coincidence? I think not..... >>



I'm not entirely sure I understand this correctly. If I got this right, then the MDF could also be referred to as the Demarc where the local loop ends? Is this right? (Take it easy on me, I'm still trying to clarify all the terminology and signal flow of network infrastructure and such.)
 
Actually, a demarc is a specific physical termination. Each entry would be some sort of demarc. For a T1, it's usually a "Smart jack," for a T3, it'll be some sort of Fiber Mux with a couple of coax outputs....So, I believe the goal is to have all your demarcs somewhere in the area of the MDF.

"MDF" is actually/physically a big cabling frame (like a wall-sized punch-down block), but has been sort of generalized into being "the main cabling (and/or equipment) center of this building's network"...and it would contain most/all of your demarcs.

FWIW

Scott
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top