Long cat5 run question

Will I be able to run a 200ft distance to another computer in another building? Suggestions on what type of router to get for this endevour would be helpful. It will be running outside. Can cat5 wire be buried?

Thank you

------------------------------------------------------------------------ View this thread:

formatting link

Reply to
ldone1
Loading thread data ...

100BaseT supports up to 330 feet. That's total distance, not line of sight. So don't forget to take into account the lengths necessary to get from the router, up/down the wall, over to the building exit, down/up the wall, under the ground, etc, all the way to the other end. You'd be surprised how much distance gets chewed up.

If all you're doing is adding to the current subnet there's no router needed. You can just hang another switch off the connection. Just don't chain more than 3 switches. If you needed more than that you'd need to use a router. How much traffic or how many computers are expected to be active in the other building?

Direct-burial jacket covered wire is available. Conduit is better as it allows much better maintenance. But for inter-building connections it's often common to use fiber optic cable. There are 100mb converters available for the purpose, you'd use one on each end. Using fiber allows going longer distances and avoids risks like lightning and surges. I'd much rather have fiber in conduit that anything directly buried into the soil.

-Bill Kearney

Reply to
Bill Kearney

The spec actually states 100 meters being the max length which is 328 feet, this includes all patch cables.

Plug it into a switch or computer on each end. It doesn't have to be anything special. But you might want to let us know what your doing just in case further equipment is needed.

Yes, there are bury grade/outdoor cable available. I bought the last stuff I used to put on a tower and between some buildings from

formatting link

True, however there is nothing wrong with using cat5e cable if 10/100 is all that's required. Fiber cable, the terminations and and media converters can be pretty expensive. If i wanted gigabit speed i'd consider fiber for a run that short other wise the cat5 will be fine, just all depends on what you want to do and what you have planned in the future.

Adair

Reply to
Adair Winter

Indeed. I've had runs slightly longer that worked, and ones shorter that didn't. There are enough variables to make it reasonable to think of it as a 300' limit. My point wasn't to quibble over the total limit, but to stress that the distance needs to include ALL portions of the connection. To the wall, up and over the ceiling, etc. If the other build is really only 200' away then he's got plenty of distance to play with inside the buildings.

Two reasons I refuse to use ethernet wire buried: lightning and voltage transients. If only because I've had personal experience with both wrecking equipment.

Not as expensive as the hassle of replacing fried equipment. I'd much rather spend a few bucks on media converters instead of hundreds of dollars on blown routers, switches and PC motherboards. But hey, feel free to gamble.

Yes, gigabit media converters ARE a lot more expensive. And quite probably not worth doing in most situations. Heck, I'd probably go for bonding multiple 100mb fiber links first. More often than not the network is not the bottleneck. The host CPU, local disk drives, and processing times are usually bigger bottlenecks.

-Bill Kearney

Reply to
Bill Kearney

Nope. One of my fun demos is to take a 1000ft roll of CAT5, crimp connectors at both ends, and connect it between my laptop and a Cisco

1900 managed switch. It won't do 100mbits/sec (100baseTX) in either HDX or FDX, but will certainly do 10mbits/sec (10baseT) in either HDX or FDX. The trick is to force the connection to 10mbits/sec and don't let NWAY try to set it to 100mbits/sec. SNMP from the switch and netstat shows no transmission errors. Thruput is also at wire speed with FDX.

Yep. You're also correct about the extra cable for bends and turns. I've gone over 300ft several times by forgetting about such details.

Note that the 100meter limit is between switches. You can add as many ethernet switches as needed along the line to deal with regenerating the signal. I've done 1500ft that way with 3 switches between endpoints. I think that's about the limit as it was a bit flakey when the switches got hot. I think (not sure) that the limit is set by the end to end latency (delay), which causes ACK timing problems.

Sigh. We don't get much lightning in California, but the few lightning bolts that hit a local tower managed to blow up most of my media converters. Fiber works much better.

The equipment is fairly cheap. My time to drive up to some mountain top, with the wind blowing, rain falling, and lightning zapping, is considerably more expensive. In retrospect, I should charge hazzard or combat pay. However, if you wanna do gigabit fiber, the media converters are VERY expensive. It's often easier to settle for

100mbits/sec and save your dollars until the prices drop.
Reply to
Jeff Liebermann

My point was not to piss over the actual distance, if you've had shorter cables fail something was wrong with your termination or equipment. I install and terminate cat5/cat6 cables daily and have never had a short cable fail but we did have a customer that had us install 600ft of cat5e against our will and it would not work.

Not always possible for some applications say for traditional phone cabling of course, as with anything standard grounding and bonding techniques needs to be followed. I agree that fiber is the way to go, it's just not cheap. I just installed ST connectors on a 6 strand multimode fiber (12 total connectors) for a customer who installed the cable themselves, still cost them around $800 for my travel, connectors and test equipment. Lets add the cost of media converters (around $100 each) and cable ($1 per foot?). Lets not consider what connectors, and the termination and test equipment cost. Our fluke DTX-1800 with fiber moduals was about $16,000 and our splice kits were several hundred dollars. This is all assuming that someone has the know how to install, terminate and test the fiber correctly.

Agreed

Right, that's my point, what does he need to do with this link? If it's one computer or access point can he justify the cost of a fiber solution? Reguardless

Reply to
Adair Winter

Isn't there an issue with the potential difference between the grounds of the two buildings?

Reply to
miso

As Jeff points out you can 'abuse' the distances but need to make sure to manually lock the port configuration. And have equipment that can actually be configured to do this, many cheapie switches will not. To say nothing of crappy drivers on various PC network cards. Know what your gear can and can't do reliably.

I was a bit suprised to see the old 3-4-5 rule has less applicability (none?) with switches. I'd still want to avoid stringing things along through too many devices, if just to avoid debugging nightmares. But a typical residential setting (or even a small office) isn't likely to run into things like arp cache overflows and the like.

At least the damage stopped there, not followed along the wire and to the computers themselves.

-Bill Kearney

Reply to
Bill Kearney

Yep, we're on the same page. It really does depend on what's really necessary over the link.

Reply to
Bill Kearney

1) I seem to remember a 1500 ft. limit for 10Mb. - not certain if that's 10BaseT or 10base2.

2) For underground, one can always run it inside a buried metal pipe. Theoretically, the pipe will isolate all static electricity (including lightening) to its outside. Note that this is for METAL pipe, so don't use ABS or PVC and think the same thing. 1/2" galvanized water pipe will do.

Reply to
D. Stussy

It's the 5-4-3 rule. It's totally non-applicable to ethernet switches and only applies to hubs (i.e. ethernet repeaters).

Hubs are useful sacrificial offerings. See below.

Not the ARP cache but the MAC address to port number table as found in all ethernet switches. The cheap switches just discard as fast as possible. The good switches have huge table space and keep the table populated as long as possible. I had to deal with an Allied Telesyn switch that had exactly 1 slot in the table as it was assumed that the backbone was connected to a router, which (usually) requires only a single MAC address. It's a rare problem, but I've seen it.

However idiots like me are certainly capeable of doing something similar. It seems that one of my coffee shop customers was complaining that *SOME* customers are connecting, but not getting DHCP assigned IP's. An important clue was that regular customers were working, while new customers and visitors were not getting IP's.

Dumping the DHCP lease database from DD-WRT v24 sp3 (old), I found that the table had filled up all 100 addresses that I allowed in the DHCP addressable space. It had saved 100 leases and said it's time to give up. After upgrading to v24 RC6.2, I found a check box to "Save DHCP table in NVRAM". Unchecking the box allowed the DHCP lease database to flush normally.

I have a sacrificial ethernet hub at each end of the link. One of the media converters was smoked, as were both hubs. Sacrifices must be made to the thunder gods. A $10 junk hub apparently placated the thunder god's anger and restored connectivity to us mere mortals.

As far as following the rules are concerned, I rarely do that. My definition of experience is knowing what rules can be broken, under what circumstances, and with what limitations.

Reply to
Jeff Liebermann

I think ethernet is guaranteed to have 1500 volt (common mode) isolation between any of the signal wires and ground. I'm too lazy to look it up.

The only way that a hot ground is gonna cause a problem is if you:

  1. Ground one or both ends of the data lines. (won't happen)
  2. If the CAT5 is shielded and you ground both ends.
  3. If the tiny isolation xformers on the ethernet interfaces decide to arc over or short in some weird way.

Even if you were running 10base2 (RG-58a/u coax) between buildings, the coax is suppose to be grounded at one end only. Now, if you grounded both ends of the coax, then you've got a real problem. At best, you'll have a ground loop which couples 60Hz crud onto the data stream. At worst, you'll have a smoking mess as the current heats up the coax and eventually melts it. That's happened to me when I ran RG-58a/u between building for a WWV and stabilized clock distribution system for the test equipment clocks. It was not suppose to be grounded at either end, but someone decided I was in error and grounded everything. It took a few minutes to start a fire and fill the building with smog as several hundred feet of coax literally melted. I think the potential difference was only about 10-20VAC, but it had full PG&E power behind it.

Reply to
Jeff Liebermann

10base5 "thick coax" has a 500 meter limitation.

The metal conduit won't help for lightning. The large difference in potential between "building A" and "building B" is what toasts stuff.

Reply to
George

one of the wiring standards for putting Cat5 into buildings puts this as 90m for the fixed wiring, and 10m for 2 patch leads (1 each end). AFAIR this was for that wiring system rather than the 100 Base-Tx standard.

You can add as many

not really. dealy thru a modern switch hardware is a few 10s of uSec.

Then you have added a store and forward delay (in each direction) - but a 1500 byte packet is 12k bits, so even at 10 Mbps that is 1.2 mSec - say 2 mSec at 10M, and 200 - 300 uSec at 100M

ACK timers at the IP and above layers are Sec or more so are not going to care, and there arent any L2 ones in Ethernet (unlike wireless).

FWIW Ethernet bridging over SDH is now fairly common in telco networks - these have a bridge each end and then adapt the bits from the Ethernet frames to run across 1 or more VC-x "pipes" inside the SDH cloud.

They work fine over 100s of Km, so a few hops over Cat5 or local fibre are not going to be a problem.

Maybe you need better switches?

Agreed. And the "fault caused by a hit can be spectactular.

We had a wall mounted Cisco router took a hit on an overhead cable carrying its ISDN line (which was "surge protected).

The smoke marks were 2 or 3 feet long, and most of the innards had very briefly been molten.....

So - a significant fire risk as well if your area is exposed to this risk - and as lightning is classed as "act of god" might be a problem with insurance...

Yes - if you have external maintenance, the cost of an engineer visit is probably more than that of a router or switch.

And - act of god again, so it may all be chargeable......

Reply to
stephen

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.