Homemade network cables? 1Gbps possible?

[xSauronx]

Originally posted by: spidey07
xSauronx,

It is really starting to look like you guys have a physical layer problem (I'm confused on what services and over what layer1 media you provide?).

I work for an Wireless Internet Service Provider that provides wireless broadband internet service to parts of rural Kansas (that may be redundant, im pretty sure "rural" isnt necessary as an adjective here ). The sister company offers VOIP services, and we have a number of wireless internet customers using the VOIP service. (indeed, they try to sell most of the wireless customers on the phone service)

For 95% of the wireless transmission we used Motorola Canopy and Canopy Advantage for Access Points, and Canopy Subscriber Modules at each customers home. The SMs are mounted outside, usually on a roof, side of home, tv tower, or polemount in the yard.

From each SM we run a cat5 cable (that carries POE from an adapter inside) to a clients PC or router.

So basically:

AP (on a tower or other high structure) === wireless=== Subscriber Module (outside)---cat5 w/ poe------PC/Router in customers home.

The wireless signal is good when we do the install, because it simply isn't worth having the customer if the signal is going to be bad. Link tests between the SM and AP are required, by the Motorola Canopy System User Guide, to be at least 90% both ways, though is considered acceptable at 45% both ways if a 2x feature is enabled (we do not use this feature) and we dont do an install if its not 95% both ways (its not often that its under 100% at all, though on some installs done when the company was under its former ownership....well, he did a few he *Absolutely* shouldnt have done)

The first thing we check if theres *any* kind of problem, either with the internet service or VOIP service, is the RSSI/Jitter and Link Test.

But after all of this discussing, i understand some of the problems (where the wireless signal is clearly not to blame) could easily be the cables that are from from the module to the client pc/router.

So, a tool aside, how do you learn to properly make the cables anyway? Is it possible, with practice and/or proper instruction (as well as good cable), to crimp a cable that can meet category 5e requirements when tested?

Running factory cables from each SM to the clients pc/router isnt feasible, since each run has to be a custom job. At the least, as nweaver mentioned, I have to crimp the end that goes into the mod, but *may* be able to convince them that we should start doing an in-wall box instead of some ugly pigtail and using factory cables from the keystone in the wall to the next device.

Like I said, they cramp me to the minute on the installs. And since the modules run from $250 - $400, plus additional parts (a short arm or long arm, in some cases a reflector, or an antenna) plus what they pay me, and installation can cost the company as much as $550, where we only charge the customer $149 for the install, Im not sure if a few extra parts for the quality would be an easy pitch on top of everything else (figure, what, 7 or 8 bucks tops for an old-work box, a keystone jack, and a faceplate?) or if theyd cringe at the idea of a few more bucks per job (ill hope for the former) nevermind me finding a way to pull another 10 minutes out of thin air to do the box and such.

Thanks for takng the time to read all of this
/longwinded

 

[cmetz]

xSauronx, the correct way to do this would be to run a patch cable from the SM to a wall-jack box, a permanent cable run from the wall-jack box to another wall-jack box, and a patch cable from that box to the customer's equipment. This is standard structured cabling. I know it sounds silly, after all, now there would be three cables and more connections. But generally speaking, 110 punch-down connections are easier to do right. And that's also technically the right model, where there's a permanent cable (solid) from location to location, and then replacable patches to the equipment.

In your situation, though, a good first step would be to just get a cable tester. That would only cost a one-time capital cost ($1k or $2k) per installer, and it would take maybe 5-10 minutes of your time, unless you did something wrong to begin with. While these are real costs, you need to compare them against the cost of a call-back, where the customer is having problems, is now unhappy (in rural areas, word of mouth will make or break you), and you have to send someone out with nontrivial travel time to spend more time on site to fix it. It's a matter of predictable expense up front versus unpredictable expenses later.

Assuming your business is making some money, most of the time a smaller to equal expense that is predictable is greatly preferred to a bunch of larger unpredictable expenses.

 

[xSauronx]

Originally posted by: cmetz
xSauronx, the correct way to do this would be to run a patch cable from the SM to a wall-jack box, a permanent cable run from the wall-jack box to another wall-jack box, and a patch cable from that box to the customer's equipment. This is standard structured cabling. I know it sounds silly, after all, now there would be three cables and more connections. But generally speaking, 110 punch-down connections are easier to do right. And that's also technically the right model, where there's a permanent cable (solid) from location to location, and then replacable patches to the equipment.

In your situation, though, a good first step would be to just get a cable tester. That would only cost a one-time capital cost ($1k or $2k) per installer, and it would take maybe 5-10 minutes of your time, unless you did something wrong to begin with. While these are real costs, you need to compare them against the cost of a call-back, where the customer is having problems, is now unhappy (in rural areas, word of mouth will make or break you), and you have to send someone out with nontrivial travel time to spend more time on site to fix it. It's a matter of predictable expense up front versus unpredictable expenses later.

Assuming your business is making some money, most of the time a smaller to equal expense that is predictable is greatly preferred to a bunch of larger unpredictable expenses.

Personally, I like the sound of it. I dont mind spending some extra time to do the job right.

But I dont make the decisions.

However, anything I can take to the first guy above me (the VOIP sales manager, its complicated as to why hes the first guy above me, dont ask) that would improve the wisp/voip service is a good thing.

I have a meeting with the guy sunday afternoon. Ill be working on fitting all of these suggestions into a "proposal" of sorts. I really appreciate all of the input from everyone. (and Im sorry i basically threadjacked this entire thing, noreage0221)

Id like to be doing the jobs right, to within cat5e specs all around. From the ends, jacks keystones and cables, to tools, testing, and even running the wire.

I just have to convince this one guy the minor extra time and money per-install are worth it. If I can convince him, he can convince the WISP president, and, though it may take a little time, we can really start doing this the way it should be done.

 

[cmetz]

xSauronx, try to find out how much it costs to do a repair/trouble telephone call, and how much it costs to do a truck roll to fix things on site. Those will be useful in arguing your point.

 

[spidey07]

Originally posted by: cmetz
xSauronx, try to find out how much it costs to do a repair/trouble telephone call, and how much it costs to do a truck roll to fix things on site. Those will be useful in arguing your point.

Also most studies show that capex (capital expenditure) is always the smallest portion of a network. opex (operations, constant money out) is the big one...the overall running and maintaining/support.

So it really should be an easy sell. There are certain tools that are necessary for any network. A cable certifier is the top one on the list.

 

[Countryman]

Hi,

I am building a new house so the info in this thread is really interesting. I want to put SqueezeBox3s into most rooms so will be pulling a lot of cables. I take it that there is not much point going to CAT6, even for the Music Server to Switch link. I looked at the Deep Surplus link that InlineFive mentioned, should I terminate the cables at the room end with 'punch-down' or 'toolless' keystone jacks.

Cheers!

 

[noreaga0221]

I would do cat6 or fiber personally.

 

[xSauronx]

Originally posted by: spidey07

Originally posted by: cmetz
xSauronx, try to find out how much it costs to do a repair/trouble telephone call, and how much it costs to do a truck roll to fix things on site. Those will be useful in arguing your point.

Also most studies show that capex (capital expenditure) is always the smallest portion of a network. opex (operations, constant money out) is the big one...the overall running and maintaining/support.

So it really should be an easy sell. There are certain tools that are necessary for any network. A cable certifier is the top one on the list.

I talked to my boss today and suggested all of the recommendations everyone here has given. His eyes bugged slightly at the cost of the tester, but he understands the necessity of it, as well as making other changes that have been mentioned here, among some other things.

Hes going to pitch it to the WISP president and see what we can do. I really appreciate all of the input from everyone here

 

[Countryman]

Originally posted by: noreaga0221
I would do cat6 or fiber personally.

I would love to go the fiber route due to the proximity of 220v power cables but termination seems even more fraught than CAT 5e/6 for a non-professional.

Brian

 

[InlineFive]

Originally posted by: Countryman

Originally posted by: noreaga0221
I would do cat6 or fiber personally.

I would love to go the fiber route due to the proximity of 220v power cables but termination seems even more fraught than CAT 5e/6 for a non-professional.

Brian

Then don't worry about it. Fiber is more expensive all around and isn't always necessary for home users. Category 6 will give you 10Gbps to 30ft and 1Gbps to 300ft so you'll be somewhat future proof.

 

[xSauronx]

is this the kind of Fluke tester you guys were talking about? CableIQ Qualification Tester

 

[spidey07]

You need a certification tester in order to certify it - this is absolutely necessary. I'd avoid that product if I were you.

But it does look to offer some good features at a reasonable price. I wouldn't trust it to tell me "the cable is good" though.

 

[xSauronx]

Originally posted by: spidey07
You need a certification tester in order to certify it - this is absolutely necessary. I'd avoid that product if I were you.

But it does look to offer some good features at a reasonable price. I wouldn't trust it to tell me "the cable is good" though.

have a link or product name that you *would* reccomend, then?

after searching for "cat5e certification tester" i get a device at a similar price

nt950

keep in mind i clearly havent any experience with these types of devices so i may still be *way* off on what im looking at, but this *does* say

Certifies CAT5, CAT5E, and CAT6 cables to be IEEE compliant to 1 gigabit speed for all Ethernet systems, Tests to TIA 568/570 interconnect standards, Tests coax, telephone, audio, and security cables.

 

[cmetz]

xSauronx, that's the $1000 tester I was talking about. It's not as good as a true certification tester (about $2k), but you would still be far better off using the $1k tester than using no tester if those are the choices.

The Fluke (Microtest) testers are very easy to use and Just Work. I wouldn't buy anyone else's.

 

[xSauronx]

Originally posted by: cmetz
xSauronx, that's the $1000 tester I was talking about. It's not as good as a true certification tester (about $2k), but you would still be far better off using the $1k tester than using no tester if those are the choices.

The Fluke (Microtest) testers are very easy to use and Just Work. I wouldn't buy anyone else's.

ill pass that along, then.

you guys are great

i just wish i knew more about all of this to start with

 

[spidey07]

Originally posted by: cmetz
xSauronx, that's the $1000 tester I was talking about. It's not as good as a true certification tester (about $2k), but you would still be far better off using the $1k tester than using no tester if those are the choices.

The Fluke (Microtest) testers are very easy to use and Just Work. I wouldn't buy anyone else's.

heh, by easy to use it's press a button and hear a "doo-doo-da-deee!" or a "doo-doo-de-doo"



Fluke made a good move in buying microtest.

 

[RebateMonger]

One other point about home-made patch cables:

Even if you TEST them, self-crimped cables are much more prone to damage than factory cables with strain reliefs. I can't count how many times I've seen home-made cables (crimped by "professionals") with the jackets falling out of the RJ45 connector. This means the only thing holding the connector on the end are those tiny wires.

Yank on a home-made patch cable a few times or have the cleaning people hit it with the vacuum cleaner a few times, and that "TESTED" cable may not be quite as good as it was when it was built. That's why keystone jacks are so useful. If you suspect a damaged patch cable plugged into a keystone jack, you can quickly replace the cable with another one. If you put connectors directly on the end of a long run, you have to cut the connector off, crimp a new connector, and then re-test to be sure you have a good run.

 

[cmetz]

spidey07,
>heh, by easy to use it's press a button and hear a "doo-doo-da-deee!" or a "doo-doo-de-doo"

Exactly. A legitimate concern any management type would have with buying and deploying new tools is "how much training will I need to do for my guys to use this right?" That's a hidden cost we see time and again. This is one of the very few cases in this business where it truly is as easy as pushing a button. And that's why I wouldn't buy some cheaper tester.

RebateMonger,

>with the jackets falling out of the RJ45 connector. This means the only thing holding the connector on the end are those tiny wires.

Tested or no, improper construction is improper construction. Anyone who does it wrong and calls themself a professional...

 

[yuppiejr]

Are there any applications available to allow testing or certification of cabling/infrastructure using a standard NIC installed in a PC or laptop that might present a cheaper alternative to the Fluke equipment?

 

[noreaga0221]

I got the intel pro/1000 software from the intel website. That has the ability to test the cable and many other tests. But its just for the pro adapter line I guess. I would look into the vendor of the adapter. I bet there are some 3rd party ones out there but I have never looked.

 

[cmetz]

yuppiejr, correctly testing the cabling properties necessary for compliance requires specialized hardware. The functions necessary are not normally present in network adapters.

Several vendors have "virtual" cable testers, which is vendor DSP trickery in the PHY to try to detect certain kinds of cable properties. I would describe this functionality as better than nothing. The vendor who most boldly pushes this functionality is, well, IMO rather bold in general about their product claims.

If you're doing this professionally, you should use professional tools. If you aren't using professional tools, you are doing an amateur quality job.

In any field, there are professional tools that are a lot better than the amateur tools, and the people who do it for a living and care about doing a good job almost always opt for the best tools they can get and know them inside and out. The tools you use are part of what make you a professional. Would you build a house with a hammer and a screwdriver? It's technically possible, but no pro would do that, they've got a truck box full of special-purpose tools that help them do a job that's better and faster. Same with sofware programming. Do real programmers build large enterprise software with notepad and run the compile by hand? Or svn, emacs, and make?

I realize that folks around here usually work within a limited budget and are trying to figure out how to do a "good enough" job on as little money as possible. When doing that, there are a lot of penny-wise/pound-foolish decisions you can make. That's why you hear advice like "use machine made cables," "use a real cable tester," "hire a contractor," or "bring in an expert consultant." Sometimes, you can spend money now, or you can spend more total money in bits and pieces and downtime and upset users later. There are places where you can save money using cheaper parts and skimping and it works fine, but there are more places where doing that will come back to cost you later.

At home for your personal use, the "does it work under load" test is okay, esp. if you do that at 1Gb/s FDX. You can do that with two decent PCs. It's better than nothing, and can be done with the equipment you already have.