Can't connect to network via switch but can directly throuch PC, possible PoE issue?

We added a new building to our facility and ran a cat5e cable underground to it (E to F). Not sure the distance, but I'm sure it's greater than 300 ft (maybe 450?).

A = Modem in Server Closet
B = Router in Server Closet
C = Switch in Server Closet (not on-site now, but I think it's a netgear prosafe PoE switch)
D = Switch in Parts Room (about 100ft from C)
E = Wall Panel in Machine Shop (about 100ft from D)
F = New External Building (maybe 350-450 ft from E)

Buildings were erected in unplanned phases, so we ran a wire from D to E to F.

If I plug in my laptop at E, it works
If i plug in a switch at E, and connect laptop to switch, it works
If I put a couple at E to connect F, and then plug only one computer at F, it works

When I move the switch to F (and use a coupler or another switch at E) , it doesn't work
When I move laptop to F (without switch at F), it doesn't work
BUT any single Desktop PC at F will work, and works perfectly

Problem is that I have 4 computers at F that need to be connected.

Why would a single desktop work, but not a laptop or switch?

Also, I changed the switch, but same result

When plugging in switch at E, still one computer will work, but not through a second switch at F. If I keep the switches at E and F, and plug one computer into F switch, it keeps trying to connect, and even says internet access, but it won't ping anything on the LAN and event viewer shows a warning for DNS timeout.

Is it possible that the Desktops have enough power to send the signal back up that far (if switch isn't in the way), whereas the switch and laptop don't?

When a single desktop computer is plugged straight in at F, I get about 40ms ping to Google, and internet seems fine.

So why would a single desktop work, but not multiple desktops through a switch.

Computers are Win 7 pro and switches are normal d-link and tplink (i think) 8 port desktop switches.
alupkamanCTOAsked:
Who is Participating?

[Product update] Infrastructure Analysis Tool is now available with Business Accounts.Learn More

x
I wear a lot of hats...

"The solutions and answers provided on Experts Exchange have been extremely helpful to me over the last few years. I wear a lot of hats - Developer, Database Administrator, Help Desk, etc., so I know a lot of things but not a lot about one thing. Experts Exchange gives me answers from people who do know a lot about one thing, in a easy to use platform." -Todd S.

JohnBusiness Consultant (Owner)Commented:
The cable overall is too long and signal strength will be BIG variable here. The Desktop NIC card is probably more powerful than a Laptop NIC card. Several machines will overload a signal dampened by strength.

You need to use higher speed cable over the main run (fiber) or else put in two good switches along the length to effectively boost signal.
alupkamanCTOAuthor Commented:
Thanks for quick response; it seems to fit what we are seeing. Would adding PoE injectors on either or both ends help? If not, what should I recommend to the Boss?
alupkamanCTOAuthor Commented:
Or, would adding a PoE switch at E and F help?
SolarWinds® Network Configuration Manager (NCM)

SolarWinds® Network Configuration Manager brings structure and peace of mind to configuration management. Bulk config deployment, automatic backups, change detection, vulnerability assessments, and config change templates reduce the time needed for repetitive tasks.

JohnBusiness Consultant (Owner)Commented:
You need a switch (I do not know if a PoE injector qualifies) in the middle, not at the ends. You need to shorten the Ethernet run.

If you cannot do anything in the middle, you would need to use fiber.
alupkamanCTOAuthor Commented:
Well E was supposed to be the middle. In fact, there is no need to have multiple devices at E because it's just where the cable terminated from D. E is basically at the back of the first building, and F is the next building over.

Cat5e range is about 300ft correct? What is fiber range?

Any other options beside fiber? Such as longer range wifi (yeah, I know I'm reaching here).
alupkamanCTOAuthor Commented:
Also, I wouldn't hesitate to just say fiber, etc. But I'm perplexed by the fact that (any) one desktop computer can connect at F with no problem. Am I wrong in assuming that it's really close to working?

...which is why I was hoping maybe a PoE injecter would be enough to resolve it.
JohnBusiness Consultant (Owner)Commented:
Fiber range is VERY long. I don't know the exact length but it will easily to this.

CAT6 cable is no better than CAT5e for length. CAT7 not out yet in any quantity is no better.
alupkamanCTOAuthor Commented:
My first guess when troubleshooting was that the run was too long, but...

Also, from E to F, I can put a cable tester and it shows that all 4 pairs are working properly. The question I'm getting asked is why does the cable tester and the desktop work if the wire is too long?

What is a good answer to this?

P.S. I didn't run these cables; I'm helping troubleshoot after the fact. Also, I have been told that multiple computers did work before at F, but I haven't been able to confirm that is true.

If I say their cable is too long, they're going to say, then why did it work before? And through one computer now?
JohnBusiness Consultant (Owner)Commented:
Make sure you are using a good cable tester. Your tester may only be testing for continuity and that is not the issue here (I do not think).

Computers working or not working can easily be related to the network card.

This ALL worked before and not NOW. That seems strange. You could have damage in the cable (e.g. pulled tight and now broken). That would be the devil to prove.

What about having a professional install new cable?
alupkamanCTOAuthor Commented:
Our next step, after possibly trying PoE injectors is to run a new cable on top of the ground to test it. I just didn't want to do that if there was a good chance a PoE injector might help. Also, if the problem is the distance, then running new cable (the same distance) would be a waste of money. However, I have no problem doing it just to rule it out if necessary. In fact, we were going to do that earlier today, but didn't have enough Cat5e left.

Regarding it working before, I was told that they had installed a router at F when it was supposedly working. They thought it was the same as a switch, but they weren't able to access the corporate network, which they need. I'm skeptical about "it working before" because if everything was fine, they would have never called me out to look at it. Unless, like the desktop NIC you mentioned before, their old router was able to supply enough power (more than the new switch) before it went out.
JohnBusiness Consultant (Owner)Commented:
I put a CAT5e cable in our church - one level and spread out. Total run from base router to back office was just short of 300 feet (maybe 275 feet).  I did that 15 years ago, left lots of slack and it has worked fine for 15 years.
alupkamanCTOAuthor Commented:
oh, and yes, the tester is just a cheap continuity tester AFAIK. But if the continuity signal can reach, then why can't data?
JohnBusiness Consultant (Owner)Commented:
Ethernet is a data transmission setup with interconducter capacitance via the twisted pairs. It forms a distributed transmission line. This has nothing to do with raw continuity. You are trying to move 100 Mbit or faster signals.
alupkamanCTOAuthor Commented:
OK, thanks for all the info. I also read online that data at 100Mb will go farther than data at 1Gb, something about the faster you try to send it, the more power you need. Does anyone know if that's true?

If so, would using a 100Mb switch work better than a Gb switch?
JohnBusiness Consultant (Owner)Commented:
100 Mbit data may well go farther than 1 Gbit data because it is slower speed. The delay line in a 50 MHz analog oscilloscope is 3 or 4 feet long. The delay line in a 1 GHz  analogue scope is about 6 inches long.

the faster you try to send it, the more power you need.  <-- Possibly, but power levels in Ethernet are limited (small wire, long distances).

However, that does not change the overall length limit of 300 feet.

would using a 100Mb switch work better than a Gb switch  <-- No.

You are up against long established limits for twisted pair. CAT6 and CAT7 cables are designed for faster signals over the same length.
alupkamanCTOAuthor Commented:
If we are just beyond the distance limit, do you think adding a PoE injector (or replacing the cheap switches at each end with  PoE switches) is worth trying? The reason I ask is because I don't have any on hand to test and so I would have to purchase just to test. Likewise, I would hate to purchase if it has nothing to do with our problem.
alupkamanCTOAuthor Commented:
I know adding switches every 250 feet, for example, would help, but our problem (maybe) is that one of our runs is longer than 300 ft.

Also, is there anything other than wire distance that could be causing this issue, such as nested switches causing problems, etc.?

The distance or damaged cable are the only things I can think of, but I don't want to overlook other possibilities that may be causing our problem.
JohnBusiness Consultant (Owner)Commented:
replacing the cheap switches at each end with  PoE switches) is worth trying?  <-- That is hard to say. Unless the cheap switch is really shoddy, it should provide a proper replacement signal and a PoE switch should do the same thing.
JohnBusiness Consultant (Owner)Commented:
is there anything other than wire distance that could be causing this issue, such as nested switches causing problems

Switches should not cause an issue. Make sure everything is connected snugly.

You are up against distance and the things around the edges you are bringing forward are not likely to yield dividends.
alupkamanCTOAuthor Commented:
Since a single desktop PC works, presumably due to more powerful NIC or whatnot, it seems the distance is within reach with proper network devices of equal or greater power.

Is that a correct assumption? If so, what network devices would allow the transmission to work >= to the desktop NIC?

The boss says, if one computer can work, why can't the rest of them? What is the answer to this question?
JohnBusiness Consultant (Owner)Commented:
A better NIC can probably make do with a weaker signal but it will not transmit at higher power.

I think you are clutching at straws. The length is TOO long and all the discussion in the world will not shorten the distance.

Make sure everything at each end the cable is high quality and snug to ensure no flaws, but that will not compensate for the overly long distance.
alupkamanCTOAuthor Commented:
I see your point, and you may be correct...and I would close this thread if none of the computers worked at this distance. However....

If the distance is too long, then why does one computer work flawlessly with no packet loss and full network speed?

If one computer works perfectly, why can't a switch work perfectly?

If the switch is underpowered, are there not more powerful alternatives that can accomplish what a desktop NIC can accomplish?

Also, I'm not sure of the distance, it may only be about 350 feet, which many others online have reported that it should work. The buildings are adjacent and it's a very short walk between them. It's basically 2 buildings with a small parking area between them (where we buried the cable).

It may be possible to shorten the distance by going to the very corner of one building to the very corner of the other (and placing switches or some other repeaters there).

But what I"m stuck on is why a single computer (any of the 4 desktops at F) can work at that distance without any problems whatsoever. That's what is perplexing me and that is the question that keeps getting asked of me.
JohnBusiness Consultant (Owner)Commented:
why does one computer work flawlessly  <-- As I have said before a NIC may be more sensitive and deal with a lower signal level than a different network card or a given switch. The signal is low enough that any variation in sensitivity will cause issues.

I cannot do more. Greater than 300 feet is too long and will lead to the problems you have. Simple physics.
alupkamanCTOAuthor Commented:
Well thanks for all the feedback, and it's pretty much what I told them after our first round of troubleshooting.

But, I'll leave the question open a bit to see if anyone else has any other ideas/experience with this. Namely, if a computer can work at this distance without issue; what network equipment (switch) will work at the same distance?

In other words, the computer connecting flawlessly proves that it CAN be done over this distance, so now I just need to figure out how to add more than one computer

...or have a good answer for the boss as to why there are no business networking devices on the market that can perform networking as well as a very low budget PC.
JohnBusiness Consultant (Owner)Commented:
the computer connecting flawlessly proves that it CAN be done over this distance, <-- This is really bad logic. All it proves is one computer has a sensitive network card. It does not prove anything else.

At this point, you wore me out. I cannot do more, so I have now unsubscribed and I am gone.
alupkamanCTOAuthor Commented:
Thanks, but that question was for other experts and was not directed at you, so I apologize if it sounded redundant.

I don't see how testing several different PCs across the cable and getting 0 packet loss and normal internet speeds = bad logic to say that it can work at that distance.

And it's not one special computer, it's every normal computer when plugged straight in to the cable. I only can't connect when I add a switch there.

Also, my primary questions have not yet been answered, which is will PoE injectors and/or PoE switches help boost the signal...or in your words, be as sensitive as a PC NIC?

Shouldn't a more "sensitive" switch do the same thing as a more "sensitive" NIC.

Obviously, if the computer can connect, without any dropped packets, then yes, it does prove that data can be reliably sent over the existing cable.

I'm sure you all can understand that when the boss asks why any random PC can do it, I'm not seriously going to reply, "your question isn't valid, or those PCs are just more sensitive". And I'm sure you can understand that I wouldn't be doing my job if I reported back based solely on one person's opinion just a few minutes after posting the question.

If a PC can do it, surely there is a network device that can do it....right...maybe? After all, isn't the PC NIC just a network device?

I just want to get as much feedback from other people as possible because I don't want to tell him to go dig a trench and lay fiber, only to find out the next day there's a switch that can do it for a couple hundred bucks.

John's answers above make perfect sense and seem to describe the problem we have. But I'm more interested in the solution than the problem and specifics instead of generalities, such as "more sensitive".

Has anyone out there successfully installed a switch across 350-450ft of Cat5e?

The reason I posted the question on EE (which is only like the 4th or 5th time I've asked here in 7 years), is because I can't find specific details about this issue and wanted to get other very specific expert opinions/explanations. I really would like a Ph.D. level explanation of why it can't be solved when a PC works fine. Otherwise I would have just asked on Yahoo and waited for someone to tell me to put my wireless dongle closer to the keyboard ;)

Thanks!
aleghartCommented:
alupkaman,

You seem very smug about wanting a PhD-level explanation.  But, you won't follow kindergarten-level instructions.  Five minutes of research should have told the first guy to never install Cat5 that far.  Another 5 minutes should have told him never to use it as an unstable electrical bridge between two buildings.

I think that you are deliberately ignoring why there are cabling standards.   I have no idea why you think using PoE will do anything to the network signal.  It's DC voltage of a few watts used to power up a small device at the other end.  If there is no handshake, the PoE switch should not inject any power into the circuit.  It's a power source for phones, cameras, small switches, wireless access points.  It's not a booster.

A 100Base-T NIC on a standard Cat5 cable with a "normal" network switch or computer at the other end should function perfectly if there are no other variables.  The twists and terminations should be within specifications.  Same for material quality, bend, even the handling of the cable and how it's laid.

Will a "normal" installation fail before 100 meters?  No.  Not unless something is outside of what's considered "normal".

Can you run a non-standard installation over 100 meters?  Yes.  But, don't complain when it fails intermittently or permanently.  Just because >100m worked one day with one computer...there is no reason to expect it will work again the next.  You were lucky.

If you are running <100m, with all variables within "normal" spec, there is 100% expectation that a network will function correctly.  Those times that <100m doesn't work?  Usually termination problem or cable handling/stress or damage.  Those, are not "normal", thus it's reasonable to expect failure.

Once you introduce problems like exceeding the 100m nominal distace for copper network cables, all bets are off.

Run unshielded power parallel to your network, you're out of the bounds of a "normal" installation.

Run through water?  Violate bend radius?  You can't expect "normal" behavior.

Span between two structures with a non-current-carrying conductor...it's not "normal".  It's code violation every place I know of.  It's hazardous to your equipment and to any personnel or innocent victims who touch your computers and network racks.

You _can_ use specialized switches or SFP modules to carry 100Base-T over UTP over 600m/2,000ft.  But, the network you described does not lead me to believe that your company would spend good money on the equipment to do it.

Extended network over twisted pair copper:
switch

SFP for existing switches

With these, you still have the potential problem with grounding/bonding.

A 500-foot multimode fiber cable is less than $300.  The rest of the cost is in the installation.  But, it sounds like you didn't have an installer pull that Cat5 cable.  It was a DIY project.

Maybe, you'll be lucky and there's a decent 3-inch conduit that you can pull through.  Even luckier if you have  a pull rope or pull string (or two, or three) waiting for you.

You'd have to buy new switches that can take a 1Gbps multi-mode SFP.  In some places, you can buy that off-the-shelf at an electronics supply store like Fry's.  You can find a Netgear ProSAFE GS110 for between $150-200.

Rough costs ~$800 parts/materials, plus the variable of labor to replace the copper with fiber.

If you don't understand the danger of the existing cable, ask/search about "non-current-carrying conductors" between buildings.  Stray current will find a path to ground through the least resistance.  If Building A has more resistance to ground than Building B ...and the buildings are connected by something metal... the current will travel that path to Building B.  And through a network switch, or a computer, or through the hand of a technician who picked up the cable and happened to be touching the rack.

I know it's not PhD-level.  You've exceeded what is stable and within spec.  There is no reasonable expectation that the network will perform normally.

Perhaps one computer is better grounded.  Most consumer laptops will not be.  Perhaps one NIC is slightly better at handling attenuation, and/or sends a fraction more power over the cable.  Just enough to keep the connection stable.

Normally, it doesn't matter.  When the spec says "don't exceed 100m", most people won't deliberately exceed it by 30-40 meters.  A few meters...maybe.  If you've done it before, and you know you're installing it cleanly...and you know you'll absorb the cost of doing it right the second time.

If you have the footprint for it, consider a non-licensed microwave radio bridge.  I have a few at different sites.  The cheapest is ~$3,000 for a pair of dishes.  You supply the tripod or other mount.  600-700Mbps over 400m, full-duplex.  Ubiquiti airFiber.  They can go much farther (6-8 miles), but throughput decreases, and aiming becomes more difficult.

Who knows?  You might be back here looking for more PhD-level explanations when you can't get max throughput while shooting through trees, in the rain, over 10 miles, while a parade of mylar balloons is traveling down the street.

First-hand knowledge, no PhD to prove it:  standing in front of a microwave dish will negatively affect network performance.
alupkamanCTOAuthor Commented:
Hi and thanks for your suggestions, info and links.

I'm not intending to be smug and I'm not sure why asking for specific detailed scientific answers from a community of "experts" would be considered smug. And I'm not sure why, in a professional community, such an assertion even needs to be made.

But if you prefer, we can say I'm looking for a scientific (instead of Ph.D. level) explanation regarding my specific issues.

Furthermore, asking for a more scientific explanation does not equal refusing to follow kindergarten instructions, so I didn't quite understand what your point was there.

As I mentioned above, I did not run this cable, and yes it was a DIY job by company electricians. But it has already been laid and buried in the ground. And it was allegedly working fine for some time to 4 computers at F.  I was only asked to intervene when the in-house IT guy changed the router (after it supposedly failed) and could then only get one computer to work directly, but the same computer didn't work when plugged into a switch (or the new router he bought).

I do not work directly for this company and it is not my network.

I'm not ignoring standards; I'm responding to an existing problem, which showed some very interesting results and raised some interesting questions, which prompted me to want to understand the issue in more detail, instead of just lazily citing an industry standard. Especially considering there are countless cases where the standard has been exceeded.

I've asked the company for an exact measurement of the cable run but haven't got that information yet, as I'm sure that would help all of us to better understand what the options/problems are.

I never disputed any of the comments here; in fact what John said above is exactly what I first told the owner of the company. But I didn't have a good scientific answer as to why any random PC works perfectly, 100% of the time, if plugged in directly, but when going through three different switches, the connection failed.

It would have all made sense if the 1 PC kind of worked, or had packet loss or something.

It led me to think that I may be missing something, or if not, I was curious to know exactly why the computers work so well.

The answers given here thus far are just guesses, and possibly accurate guesses, but I'm looking for a definitive answer that explains which particular factor allows a computer to work but not a switch.  For example, if the issue is power, as suggested above, how much power? What is the power discrepancy between a computer nic and a switch? Is there a way to boost the power of a common switch?

When I first suspected the distance was too great, I also tested from my laptop directly. As I mentioned in my OP, the laptop also didn't work. I then plugged in the power adapter of the laptop and still couldn't connect. The only difference I could think of at the time was that the desktop PC was more powerful.

But if a computer can operate over X Cat5e cable, isn't it reasonable to think that maybe, at least in theory, there is some other network device out there that can also do the same thing over the exact same cable?

I understand everything that was said above.

I agree with everything that was said above.

I just want to get a deeper understanding of exactly why one works and the other doesn't? Not sure why people are getting upset about that.

For example, if the desktop has slightly more power as you say, then it seems like we should be able to purchase a slightly more powerful switch?

I'm perfectly fine if everyone confirms my initial assessment, which is also what the first responder said. In fact, it would just help show that my initial assessment of the problem was correct. But it still doesn't help me to understand exactly why. Could we, for example, physically modify the switch to send more power (in theory).

Just like taking a poll of 1 person isn't helpful (not because you disagree with their feedback, but because it's simply not enough information for which to base any conclusions), I'm just trying to gather enough feedback to be more certain about the answer.

Hopefully this will help people to understand what (and why) I'm asking these questions. If not, please let me know and I'll try to think of a better way to ask my questions.

Having said all that, I do appreciate your constructive suggestions. So I'll count you as +1 for fiber and I'm really interested in the air fiber as a possible solution.  Also thanks for the links and the detailed cost amounts; much of what you shared is helpful.

Also, I posted this question hoping to get replies from people that have had similar experiences and could offer some constructive insights or from people with the scientific knowledge to help me better understand the specifics.

If other members are in neither of these groups, I still welcome and appreciate general feedback, but please try to keep it positive. Thanks!
aleghartCommented:
Patch panels, RJ45 connectors, NICs, and switches are all designed to meet the 100m standard.  Anything beyond that is gravy.  Yes, plenty of people exceed the standard.  But none of us expect 100% to work.

I'm no engineer, so I have to rely on analogy.  (I may use this in an inservice training.)

Picture a new communication standard that is based on speaking English, unamplified, 100ft distance, half-duplex, normal conversation speed.

Imagine you have 100 people standing in field facing north.   Imagine another 100 people are 1 foot in front of them facing south.  At a "normal" talking volume, everyone should be able to understand his opposing partner.

Everyone takes a step backward to 2 feet.  Then 3 feet.  Then 4 feet.

At 5 feet, people will begin to have slight problems hearing their partner.  This is attenuation/loss by distance.  They talk louder.

At 10 feet, somebody starts speaking French.  He is ejected.  Non-compliant signalling for this experiment.

At 20 feet, another person breaks out into pantomime.  Wrong protocol.  Ejected.

At 25 feet, somebody else in the park starts blasting a boom box.  A few of the test subjects start talking in a staccato rap.  While amusing, the effectiveness of communication is reduced.  This is inductance.  The boom box is ejected.

At 30 feet, another person is having problems hearing.  He can shout just fine, but suffers from attending too many rock concerts in his youth.  Ejected because he cannot no longer effectively communicate with his partner.

At 35 feet, somebody starts getting hoarse and can't talk.  He used a megaphone to boost his voice, but amplification is not part of the standard.  Ejected.

At 40 feet, one guy can't hear because of wax buildup in his ear.  His partner's voice is rattling from phlegm from a lingering cold.  Both suffer from insertion loss.  Ejected.

At 50 feet, participants are having problems focusing on their partners.  They are hearing noise from the adjacent pairs and getting confused.  This is alien crosstalk.  The pairs are then moved away from each other.  Separated by enough distance, or by barriers, so that the foreign conversations are not distracting.

At 60 feet, one pair cannot continue because they are standing next to a flowing stream.  Water?  Well, another guy is standing in a puddle from a sprinkler and isn't complaining.  Another guy is standing in a similarly sized stream, but there are no rocks, and no animals making noise.  So, there must a certain amount of noise that interferes.  Best to avoid exposure to this potential source of noise, or limit it.  Somebody add that to the guidelines: avoid babbling brooks.

At 70 feet, some of the older guys are having problems keeping up.  They're wheezing and talking slower.  Those young guys are still talking at normal speed.  Bit-rate issues.  Some of the old guys try to re-negotiate to a slower talking speed.  Some have partners that reject the negotiation, and stop talking.  A few old guys successfully negotiate the slower speed.  But, they're still ejected from the experiment.  The standard says you have to work normally all the way out to 100 feet.

At 80 feet, one pair has a tree in between.  Lots of leaves.  The medium is not up to spec (open air).  Nobody is expected to talk through trees in this new standard, so they are ejected.

At 90 feet, one pair are in a bit of fog.  This muffles some of the tones of their voice.  Frequency attenuation.  The medium must be clear from X to Y Hertz signals.  Out of spec.  Ejected.

At 95 feet, one guy is having problems hearing.  His left ear is bigger than his right, so the sound is arriving at different times.  He's getting strange echoes, and has to ask his partner to say again.  In Category cable, that's twist rate multiplied by distance.  One reason why you can't sent analog video over Cat6...to many twists skew the different color signals.  Visibly, there are color halos around everything.  With Ethernet signals, increasing the distance increases the delay/skew problem.  The higher the frequency/bandwidth target, the more attention you have to pay to twist.  Cat6 for 10Gbps, you have to downgrade from 100m to 55m limit.  Anyway...the guy with echoes in his head: Ejected.

At 100 feet, there are still a lot of players in the field.  They're successfully communicating, and keeping up with each other.  They have all met the spec.

Just for fun, a few of them take another step back.  No big deal, right?  101 feet is just 1% over the standard, and you can't even notice the difference.  Exceeded spec.

Same for 105 feet.  But at 110 feet, some start shouting louder and slower.  People are cupping their ears and shouting "Say again!".  Dropped packet.  The sender re-sends until he gets a thumbs-up acknowledging the receipt.  Unless this was UDP.  Then, he just keep talking, hoping that 90% of the conversation was enough.  

At 120 feet, a lot of people are just quitting and walking away.  Why bother?  They can only advertise to the 100 foot standard, unless they spend millions of dollars to cover another standard out to 120 feet.  But, the consensus is that the 100 foot distance is big enough for most needs.  Anything longer, and you'll be using up a lot of resources and disqualifying many other competitors.  At some point, it's easier to move to a different technology that's made for distance.

Invariably, there is a guy at 135 feet or 150 feet, shouting and hopping up and down that he can exceed the spec, and you all are wimps.  Good for him.  It's working.  But, you can't sell that to a customer who wants 100% reliability.  Who would pay for 50/50 chance of success?  Or 100% success on installation, but no guarantee that it will continue to work the minute the installer walks out the door?

You can use powered megaphones.  Same medium: voice communication, in free air.  But, now you have to create standards for amplification, noise, battery life, etc.

You have walkie-talkies, which work for miles.  Same half-duplex.  But you have channel congestion problems.  And, it's a different technology, different medium (emitted radiation).

You have Morse code with directional light.  It's open air media.  But, not voice.  You need to train in a different language.

Let's go back to your specific situation:  Cat5 cable.  Talking about fiber doesn't help.  That's a new medium, and a replacement.  So, we have to look at the standard for what is considered acceptable.  Let's assume all other installation requirements are met, and there is only the problem of distance.  You need a stronger sender and a stronger receiver.

How many companies will advertise that their FastEthernet products can exceed the standards?  None.  That's the point of having the standard.  Everyone has the same target spec.  Everyone can design and manufacture for the most reliability and lowest cost based on that specification.

There is gear that might be able to go 5% farther.  (I have an installation that is 350 feet.  It usually works, unless it rains heavily.)

There is gear that might go 10% or 25% past the spec.  But not reliably.  Not every time.  And nothing that the OEM will advertise.

You have to rely on trial-and-error experimentation to see what works.  If the infrastructure was within spec, _any_ FastEthernet gear would work.  If you're in the field of experimentation, you might like that.  If your job is to keep the network running, you have to match the proper technology to the task.

Your media can be re-used by a 2-wire technology that is designed for longer distances.  I remember a printing client back in the 1990s who started advertising to the telco industry about a single copper pair long-haul modem that could be run out to 12,000 feet at speeds far higher than 34Kbps of an analog modem.  That became DSL.

I gave you a link to a company that does something similar.  FastEthernet is converted onto a single pair of twisted copper (UTP or STP) or onto coax cable.  At the other end, a partner device will decode the signal and give FastEthernet handoff to the user.

That matches the right technology to your infrastructure.  It sounds like the "electrician" may have used a direct-burial cable.  There may or may not be foot markings on the sheath.  Subtract the smaller number from the larger, and you have your distance.  Or, connect a cable tester/qualifer that will read the distance.

In reality, no electrician should have been running a cable like that between two buildings.  It violates code.  It will never meet spec for networking.

Other suggestions for rescuing the connection while you wait for a replacement:
1. trim back both ends and terminate into proper patch panel.  Could be a loose RJ45 connector.  Patch panel is physically more stable, and stops the cable from moving (if you strap it down on the back, as designed).
2. install FastEthernet switch at both ends, not Gigabit.  This requires only two pairs, not all 4.
3. put a cable tester on to determine the length and if there are any breaks, shorts, or mis-punches.
alupkamanCTOAuthor Commented:
Wow, I get your meaning but I think you could have just said signal degrades over distance, which we all know.

"How many companies will advertise that their FastEthernet products can exceed the standards?  None.  That's the point of having the standard.  Everyone has the same target spec.  Everyone can design and manufacture for the most reliability and lowest cost based on that specification."

That's a good point.

However the analogy still doesn't answer the questions, which is what is the specific techincal difference or threshhold that allows a PC to work but not a laptop.

"Let's assume all other installation requirements are met, and there is only the problem of distance."

I understand what you meant by this, but to clarify, the whole reason for initially posting the questions was that I wasn't making this assumption. That is, before I go down that road of declaring it can't work at that distance, are there any other possible factors that may be the underlying cause of our issues?

When I saw how well the PCs performed, I was just worried that I may have been jumping to the conclusion of distance and missing something obvious.

Nevertheless, if in fact the problem is solely the distance, any random PC working flawlessly seems to suggest that the distance is very close to the threshold, and if so, while not ideal, it is reasonable to think other devices besides a PC NIC would work.

The information and suggestions you provided are helpful. I'm seriously thinking about buying some air fiber dishes for my house just to play with. I know it sounds ridiculous, but my house is actually pretty close to my office and I'm wondering if I could use this to replace my VPN connection. Then again, I have no way to physically secure it from theft.
aleghartCommented:
Signal does not degrade just over distance.  You have skew and reflection and the variable of temperature.

As the temp rises, the insertion loss increases.  So, what would have been fine at 100m may show more error when the temp is 10C higher.  Expect 4% greater insertion loss.  Is the desktop NIC cooler than the laptop?  Maybe.

Calculate insertion loss varied by temperature here .

There is also skin effect, which varies based on material quality, wire gauge, and signal frequency.  If your signal stays the same (100Base-TX), then you should stay below the 100MHz frequency.  I think you're at below 35MHz, so you don't need to do the loss calculations here.

With Cat5e cable attenuation at 2.2dB / 10m, you can overcome with a bit of signal boost at send/receive time.  But only to a certain point.

Skew is another factor.  At 25 ns / 100m, you shouldn't have a problem with 10Base-T (half-duplex) timing.  NLP bursts will be every 100 nanoseconds.   With FastEthernet, the auto-negotiation will be  several bursts at 125 picosecond intervals.  Increasing skew will interfere with timing and receiving the bursts within tolerance.

As the signal loses strength, the on-off pulses are not as easy to decipher.  Much like how you would yell with sharper breaks in staccato to make yourself audible over distance, the signal must be interpreted.  At what point is the voltage actually moving +1 or -1?  Did it make that move completely or intelligibly withing the timing window?  A pulse is not a pure square curve, flipping +1/-1 instantly.  Picture from here might help.

output pulse
Return loss comes into play with bad terminations and bi-directional traffic.  1000Base-TX will suffer from a partially reflected echo bouncing off the recipient's bad cable termination.  As it arrives back at the sender, it is noisy interference that must be filtered out from a real signal being sent from the far end.  At your distances, you should not be plugging in Gigabit NICs.  Each wire goes both ways.  That's why I suggested putting FastEthernet switches at both ends.  Then, it wouldn't matter if your laptop has a Gigabit NIC.  It will auto-negotiate down to FastEthernet, and only FastEthernet will be carried across the wire between the two switches.

Return loss also comes from bends or kinks in the cable, or a cable end that has been folded behind a desk, rack, computer, etc.  Picture from here .

return loss
Does your laptop have a gigabit NIC?

Experts Exchange Solution brought to you by

Your issues matter to us.

Facing a tech roadblock? Get the help and guidance you need from experienced professionals who care. Ask your question anytime, anywhere, with no hassle.

Start your 7-day free trial
alupkamanCTOAuthor Commented:
Thank you, this is exactly the kind of information I was looking for; very informative.

Yes, my laptop is newer and has Gb Nic, and all three switches we tried were Gb switches. But the desktop computers are older and quite possibly not Gb NICs.

I'm going to try a FastEthernet switch just to see, and also send someone back out to measure the length if they don't provide me the info I requested. But beyond that, I'll consider the distance is too great and look at appropriate solutions to address that.
It's more than this solution.Get answers and train to solve all your tech problems - anytime, anywhere.Try it for free Edge Out The Competitionfor your dream job with proven skills and certifications.Get started today Stand Outas the employee with proven skills.Start learning today for free Move Your Career Forwardwith certification training in the latest technologies.Start your trial today
Switches / Hubs

From novice to tech pro — start learning today.