Long Range Ethernet transmission over twisted-pair cable

10baseT can be used for up to 200 meters on CAT5 UTP--it may go longer on CAT6.

Beyond that, google "long reach Ethernet"

How much longer do you need?

The lmiting factor on 10MB ethernet is said to loss, so looking for the detailed specs (or measuring the gauge) and looking for the small Gauge number) will give youi the wire with the best shot of working. I *think* I've heard of 10MB ethernet working at close to 1000 ft.

You can test it on the spool if you can get to both ends.

The equipment you use at both ends needs to be settable to 10MB. This rules out cheap switches and routers. The NIC in a PC can be set if you look in driver properties.

You need to force the setting because if the wire is marginal it might handshake at 100MB at powerup but not be able to actually carry any data atr than speed.

Sure, but.....

One of the quickest is the Megabit Modem. You pick your speed, and buy the appropriate unit and it runs on a single twisted pair.

try

for the info on the 310/320 combination. IIRC it will push a signal for a couple of miles on a single pair.

Carl Navarro

What does a pair of these things cost?

10baseT will go at least 150m over cat5 cable.

There is a home networking system which is somewhat like a

1Mb/s ethernet designed to run over home phone lines.

There are special systems, such as one previously mentioned here by Tut systems. I haven't heard about that lately, so I don't know that they are still around. I believe that one has the appropriate bridge at each end.

For longer distances you can go to T1. You can connect T1 devices point to point through two twisted pair cables (as in no phone switch in between) for some number of km. That is about 1.5Mb/s, and appropriate routers aren't so hard to find.

-- glen

Check those:

Cisco LRE

VDSL technology

I installed a link using Cat 5 cable for a customer in a factory, all that was wanted was a slow speed link to a remote machine. Fibre was a rather expensive option.

My Cat 5 tester (set to 10BaseT) measured the installed cable at 187M, it was maybe within 5% of the attenuation specs for a 10baseT link. The link has worked just fine for 6 years or more.

Although, erroneously, the tester failed the link because it was 87M over length.

Stuart.

I believe at 150m there isn't much question about it. Over that, it is probably best to measure the attenuation, as you did.

Well, the tester doesn't know that someone won't try 100baseTX on it someday, or even gigabit. I would probably put enough labels on the link to be sure that everyone knows it is for 10baseT only.

-- glen

If a "tester" is doing a TDR, it's going to call any cable length over the spec as a "FAILED" based on the fact that it didn't get a pulse back in time.

Right?

Al Dykes wrote: (snip)

(snip)

I would probably say that a tester set to 10baseT shouldn't fail a cable based on length. Note, though, that 568A and 568B specify

100m so, unless the tester has separate settings for that it probably is reasonable to warn you on 100m.

As current production machines likely have 10/100 interfaces it isn't unreasonable to assume 100baseTX.

-- glen

[...]

The length limits are there because of propagation delay issues and round-trip times that come into play because Ethernet is CSMA/CD.

Of course, with more modern, switched, networks this is less of an issue, but those are the design assumptions that Ethernet was constructed on.

Regardless, the timings on 10 Mb/sec Ethernet are based on multiple 500 foot coaxial segments with an additional length of drop cable at each end. With

10 Mb/sec on UTP signal degradation will almost always be the limit, not timing. This is not the case for 100 Mb and faster, most of the copper variants of which were designed around UTP with 100 meter segments.

The cable scanner, however, is not doing a "test to see if Ethernet will barely work if everything else goes perfectly", it is generally doing a test to verify that the installation complies with EIA/TIA 568 or other applicable standards, and those standards generally specify a maximum length, so the scanner is quite right to fail a segment that is over length.

The propagation delay limit for 10 megabit ethernet is long enough that you should never reach it with 10baseT. Well, with 200m cables you might. The maximum number of segments for 10baseT is six, set by the IFG (inter frame gap) loss in a repeater. I know that six 150m cables is not too long, it might be that six 200m cables is, but a net with six 200m segments separated by five repeaters is very unlikely, especially with current switch prices.

Now, with 10baseFL and 2000m segments (fiber, not copper) it is easy to reach the length limit.

But the collision domain, in terms of propagation delay, is 10 times smaller for 100baseX.

-- glen

(snip)

If the tester is set to EIA/TIA 568 mode then yes. It it is set to 10baseT then I would say no. Most likely it doesn't offer a separate 568 mode, though.

-- glen

If "current production machines" is defined as a typical (whatever that means) PC or x86 server, they almost certainly have 10/100 and probably have 10/100/1000. If we step up into RISC et all systems I'd posit that virtually all systems currently on vendor's CPLs are

10/100/1000.

I've no idea what the embedded space looks like though.

rick jones

The DSP-4000 has 12 screens of test configurations (that's with firmware that predates release of the gigabit standard--when I update it there will be more). It includes TIA CAT 3, 4, 5, 5e, and 6 that I can recall, both link and channel. Also has 10baseT, 100baseTX, 10base2, 10base5, a bunch of types of fiber, and so on.

Quite.

The tester was a DSP100 in 10BaseT mode, and it should not have failed a cable on length since the 10BaseT ethernet spec does not impose a length limit.

Stuart.

Yep. VDSL does just that.

Just so you know, you should not pay more than $100.00US for any of these devices. Tut used to sell their equipment for $1000-$4000/unit, but that is far too much now.

The DSL testing lab at Cisco contains several huge cardboard boxes filled with DSL modems, most of them brand new. They are in boxes because... well...lets just say that their cost to Cisco is less than expensive lunch. You might be able to get someone there to ship you two or four with company permission since it was rumored that these devices are not to be sold anyway.

-Le Chaud Lapin-

Suggest you set (both ends for) full duplex.