Future proofing and the NEC

[hbiss]

The cable is designed to have a specific impedance (Z or AC resistance), it's a transmission line. This characteristic impedance is determined by such things as the size of the copper conductors, the composition and type of insulation surrounding them, the twist, and the space between the insulated conductors. A "virgin" length of cable simply lying on the floor between point A and point B will result in the design data transmission. When you bend it sharply or kink it you alter all those parameters causing a shift in the characteristic impedance at those points which will cause standing waves and a loss of bandwidth and an increase in attenuation.

-Hal

 

[egnlsn]

Why does the bending radius effect the speed? Do some of the electrons go flying out of the wire cause the curves too tight? They are moving fast.

Packet loss. Too tight of a bend can also cause reflections (same thing with pinching (staple, zip-tie, etc.)), which can cause packet loss. I've seen it with cable TV and satellite TV as well.

 

[tallgirl]

If you work from home you maybe have higher needs. or even if you have a lot of users maybe those needs start to increase.

I don't have new enough computer to even be able to keep up with potential demands of some the high bandwith applications, it wasn't even top of the line processor available at the time when it was new, just somewhere in the middle of what was typically available I'd guess.

It's all about the very large file transfers which happen inside the house. That's usually why people have a NAS / SAN inside their home.

I forget how much total data I downloaded immediately after we were locked down here, but being able to access data faster over a network than I would have straight from the hard drive was nice. My NAS has multiple SSD drives on multiple controllers.

 

[tallgirl]

Packet loss. Too tight of a bend can also cause reflections (same thing with pinching (staple, zip-tie, etc.)), which can cause packet loss. I've seen it with cable TV and satellite TV as well.

The packet loss is caused by changes in the impedance.

It's like when an electrician puts a high impedance meter on a dead circuit that's run in the same raceway as a live circuit. Some amount of current is induced into the adjacent wiring, and the newbie electrician is left scratching their head until they get out a wiggy and check the voltage.

The tighter the bend the more the signal is distorted by itself. Distort it enough and the receiver won't be able to lock onto the signal correctly because the waveform won't be correct.

 

[tallgirl]

No expert here, but sure it has to do with the high rate speeds that aren't the same as having a high frequency sine wave. It won't exacty effect "current" from flowing in the conductor, but can create problems with losing data bits at those higher speeds. I remember some training on how to install these cables, you don't even want them pinched too tight in a clamp or other securing device in order to get the most speed out of them. Also don't pull through an LB and pull them extremely tight, better yet was to not use conduit bodies at all and make longer radius bends- in a pull box if necessary.

at the very least such pinching and tight bending distorts the rate of twist to the pairs within the cable, which supposedly is selected for optimum performance.

The "performance" is all about maintaining the correct waveform shape. They aren't going to be sine waves anyway. They won't be perfectly square, but they are most definitely not sine waves.

Impedance has two components -- capacitance and inductance. What the twists in the wire, along with the insulation, do is ensure that the signal reinforces itself, rather than degrades. This is why the pairs must be wired as pairs to the same transceiver -- you can't split a pair because then one signal is next to another, rather than the positive and negative -- differential -- signals being together.

The current and voltage waveforms must be kept together. The wrong amount of either capacitance or inductance will cause those two waveforms to get out of phase with respect to each other, the same was with any other wiring. I was working on a 400MHz circuit over the summer and it had issues until we had everything matched perfectly. It went from something that looked like a box someone had stomped on in the middle to a nice pretty box with well-rounded corners and a nice flat top.

 

[drcampbell]

Geek porn.

 

[tortuga]

Interesting thread thanks Tallgirl for all the info.

But did you ever do any kind of testing to see what max speed you could get out of it? And not just by testing via speedtest.net or similar but actual max speed ability of the cable before there is any losses or errors?

Even to get what Cat 5 or 6 cables boast they can provide you have certain installation criteria that if you don't follow can give lower speeds before developing losses/errors

No but that would be fun to do, perhaps ill do it this weekend and post back.

Anecdotal test is I hooked it up that cat3 cable back in 2017 and its been running streaming internet to a Roku off a 'average' consumer grade cable internet connection (HFC) for 3 years with no issues.
We have streamed countless movies / shows /music thru it.
The router led glows amber which means its chip-set auto negotiated 10/100 mbps, where as the cat 6 cable I ran to my desk glows green for 1000 mbps.

It would be interesting to see what these (or any) chip-sets try to auto negotiate with all 4 pairs showing and cat3 cable, 10BASE-T2, 100BASE-T2 or 100BASE-T4?
I would have to hook a scope or something up to see.
If its using all 4 pairs and running at 100mbps (IEEE 802.3u) that would be neat.

 

[hbiss]

I would have to hook a scope or something up to see.

A 'scope wouldn't show anything. There are data analyzers specifically for that.

For 10BASE-T2, 100BASE-T2 or 100BASE-T4 you would have to have NIC's , routers or switches that support the topology. I haven't looked but I don't remember seeing any. Probably because there isn't any demand. Who uses CAT3 anymore for data?

-Hal

 

[tortuga]

A 'scope wouldn't show anything. There are data analyzers specifically for that.

For 10BASE-T2, 100BASE-T2 or 100BASE-T4 you would have to have NIC's , routers or switches that support the topology. I haven't looked but I don't remember seeing any. Probably because there isn't any demand. Who uses CAT3 anymore for data?

-Hal

Hal my understanding is 10BASE-T2 and 100BASE-T2 are in any standard Ethernet chip, thats what would be in use when non-gigabit is in use.
100BASE-T4 is an open question.

 

[paulengr]

I think I also tried to say that earlier in the thread and is part of reason why the CAT 3 cable happened to work out.

Nothing magically different about the conductors themselves, just the design of the cable happens to help lessen outside influences that may not let it operate at as high of speed data transmission. Twisted pairs, varying twist rate amongst the pairs...then installation factors like too sharp of bends can lessen max data transmission abilities.

I’ve had lots of intermittent problems with CAT 3. But I’ve never seen anything that CAT 5E UTP did where something else worked better. CAT 6 is all but abandoned since it does not support anything that CAT 5E does not do. Jumping to CAT 7 or 8 is “future proofing” but terminating those is less than straight forward. The biggest argument for CAT 6 is easier pulls.

Granted there are now standards using CAT 7 that get into 10+ Gbps speeds but there are all kinds of issues to the point where fiber is a simpler solution.

 

[kwired]

A 'scope wouldn't show anything. There are data analyzers specifically for that.

For 10BASE-T2, 100BASE-T2 or 100BASE-T4 you would have to have NIC's , routers or switches that support the topology. I haven't looked but I don't remember seeing any. Probably because there isn't any demand. Who uses CAT3 anymore for data?

-Hal

If I understood correctly, it was already there, so he decided to see if it would work. It did, but likely can't deliver the same max speed the cat 5 or 6 could if everything else was able to support that speed.

 

[paulengr]

A 'scope wouldn't show anything. There are data analyzers specifically for that.

For 10BASE-T2, 100BASE-T2 or 100BASE-T4 you would have to have NIC's , routers or switches that support the topology. I haven't looked but I don't remember seeing any. Probably because there isn't any demand. Who uses CAT3 anymore for data?

-Hal

I beg to differ. A scope will definitely show ringing issues and other distortions clearly running 10BASE-T. It would be difficult to see it in 100BASE-T specifically because the modulation pattern is not easy to make out in a time series by itself. But if the scope has a clock recovery trigger you can trivially produce an eye diagram. 100BASE-T applies “whitening” to the data so it naturally shows all modulation patterns evenly. Visually on an eye diagram if you can make out a clean separation between signals then you should have no problems. But if the eyes collapse to where the pattern is not clear then the Ethernet receiver will struggle to read it, too.

Lay your EYES on this! – How to set up an... | Keysight Community

Eye diagrams are extremely helpful in testing the physical layer fidelity of clock or serial data, but many engineers don’t know: What they are

community.keysight.com

This method is somewhat “old school”. The reality is most people never get past looking at the statistical data that most Ethernet software interfaces collect “for free”. But eye diagrams can not only provide visual confirmation but can be used to calculate SINR.

The last time I saw CAT 3 was in 2007. It was in use because as you said, it was there. As long as you force the Ethernet card to 10BASE-T manually, it’s not really a problem. The problems are with 100BASE-T. CAT 3 is (was) often “free” from leftover analog PBXs.

 

[tortuga]

I beg to differ. A scope will definitely show ringing issues and other distortions clearly running 10BASE-T. It would be difficult to see it in 100BASE-T specifically because the modulation pattern is not easy to make out in a time series by itself. But if the scope has a clock recovery trigger you can trivially produce an eye diagram. 100BASE-T applies “whitening” to the data so it naturally shows all modulation patterns evenly. Visually on an eye diagram if you can make out a clean separation between signals then you should have no problems. But if the eyes collapse to where the pattern is not clear then the Ethernet receiver will struggle to read it, too.

Lay your EYES on this! – How to set up an... | Keysight Community

Eye diagrams are extremely helpful in testing the physical layer fidelity of clock or serial data, but many engineers don’t know: What they are

community.keysight.com

Wow Paul thats really cool, my home scope is a old school analogue one a (BK Precision 2120B) . I love an excuse to fire it up and learn. A retired engineer gave it to me many years ago to encourage my electronics hobbies to this day I barely know how to use it.

 

[paulengr]

Wow Paul thats really cool, my home scope is a old school analogue one a (BK Precision 2120B) . I love an excuse to fire it up and learn. A retired engineer gave it to me many years ago to encourage my electronics hobbies to this day I barely know how to use it.

It’s a standard technique for communications system engineering from almost 30 years ago.


Sent from my iPhone using Tapatalk

 

[steven765]

Dear lord the amount of misinformation and bad internet rumors in this thread is cringe worthy.

Wire bends, amount of untwisting, etc: This is a relation to the Q in the conductor and cancelling out the magnetic field to prevent interference and ensure propagation at distance. Each pair is in opposition so the magnetic field is cancelled up to a certain frequency. A fun experiment I make undergrads do is calculate the number of bits per inch of wire at a given GHz. It's especially important in terminations in the10GHz territory, small delta in wire length can drastically effect the arrival time of bits to the transceiver causing jitter and havoc internally as it either then has to drop the packets or buffer the signal and wait. What happens when you kink a high voltage wire? Well imagine the frequency is 10,000x that, how do you think the data will arrive at the other end if at all? FYSA at 10GHz the actual wavelength is 3cm however the tolerance will depend on how data is encoded on the wave. Is it using QPSK etc etc so then you're looking at some fraction of lambda tolerance for effective signal reception.

testing: There are 3 levels of test for UTP data cables. Verification, Qualification, Certification.
This is done with a TDR, An Oscilloscope will show you frequency vs time but unless you have a known signal source it's not going to be of much use. The comments above appear to be using coax data comm. Sure it would work if you ran RG-11 and still had copies of novell networking from 1990 kicking around. And an ethernet ISA card. Man those were the days. However, today for ethernet unless your scope has 8 channels and supports 1GHz you're not easily testing ethernet that way. They're still in the 30k price range.
The levels:
Verification: is essentially just checking you wired it correctly. These are the $150 testers you find in the big box stores.
Qualification: ensures the cable will work up to the specified speed. These testers run around 1500.
Certification: actually tests and reperforms all the EIA/TIA ISO tests post installation to certify the speeds the wire will actually support. These testers run around 15k used. see https://www.flukenetworks.com/expertise/learn-about/cable-testing.

Satellite latency: That is precisely why Starlink, Telesat occupy LEO's and have space based laser comm capability. Likely using GEO and HEO for backend burst comm later if they can afford it.

To OP's original question 800.25 Abandoned Cable. To limit the spread of fire or products of combustion within a building, the accessible portion of communications cable that isn’t terminated at equipment and not identified for future use with a tag must be removed [800.2].
There's a different article on their definition of accessible, but in walls and conduit, are not considered accessible. Their biggest issue was lazy folks letting decades of cables pile up under a comm center floor. Somewhere I have a photo next to a spool/pile of cable we took out in 2008 from a 1970's classified facility. It was taller than me.. how i didn't burn to death was beyond me.

Standards: Cat 6a is the latest. Cat 7 is proprietary, Cat 8 is under EIA/TIA review.
What would I run. As others have said.. conduit! Will you likely for home use need more than Cat6 in the next 10 years, probably not. As Bill Gates famously said I can't imagine why anyone would need a PC with greater than 640k of memory. Who knows your BW requirements in 15 years. Cat5e is on it's last legs for me internally for my networks it can no longer keep up with the camera bandwidth from 4k security cameras and my torrent streams. Then again what's your time frame?
TL;DR: Run conduit either way.
<5 years.. Run cat 6.
>5 & <10 what does your budget allow
> 10 years leave them empty

 

[tortuga]

<snip>...
.A fun experiment I make undergrads do is </snip>

Hey Steven keep in mind some of us just have a HS diploma, 26 years in the field, a masters license, a welders certificate and never even got a associates degree and nothing close to an undergrad.
So 'break it down for us'
Were here to learn for each other.
But yeah back to the OP good tip on the conduit/raceway (even smurf tube) is the best future proof option, in my house with vaulted ceilings and a rough crawl underneath I would not be using that old cat3 if I could re-pull in some cat6.
I would actually do any circuit 30 amps and up in EMT if it were my own house.

 

[drcampbell]

Hey Steven keep in mind some of us just have a HS diploma, 26 years in the field, a masters license, a welders certificate and never even got a associates degree and nothing close to an undergrad. ...

You might be closer than you think. Don't forget that you become an "undergrad" the moment you set foot inside a university.
(admittedly, very few incoming freshmen enroll in electromagnetics their first semester)

 

[steven765]

Hey Steven keep in mind some of us just have a HS diploma, 26 years in the field, a masters license, a welders certificate and never even got a associates degree and nothing close to an undergrad.
So 'break it down for us'
Were here to learn for each other.
But yeah back to the OP good tip on the conduit/raceway (even smurf tube) is the best future proof option, in my house with vaulted ceilings and a rough crawl underneath I would not be using that old cat3 if I could re-pull in some cat6.
I would actually do any circuit 30 amps and up in EMT if it were my own house.

Didn't mean for that to be as snarky. 10/100 only use 2 pairs for data. 1000MBps connections use all 4 pairs in the cable as such deviation in cable pair length becomes a problem relative to the speed information is transmitted on the wire. 10GHz with a .2m wave length accounting for propagation times in wire gives you little room. What you're doing is calculating bit length see: (https://courses.cs.washington.edu/courses/cse461/99wi/issues/definitions.html)

Also with increased speed comes greater risk for interference. This is the reason for the twist ratio and shielding as the speed and standards increase.

Both combined are the reason for the specs on cable sheath peel back .51in in cat6 and max pair untwist length