Impedances and transmission line effects...

[Mark Hennessy]

I bought a box in 1999, and have just finished it. Now using Cat 5e. When finished, I think it'll be something like 25%/75% mix of Cat 5/Cat 5e. As having a mix of types in an installation might be a cause for concern, I looked into this carefully before starting, and decided that I might as well finish the Cat 5 box before starting on the 5e as there is no material difference between them.

This is just a home LAN, so the runs are fairly short. Using Gb Ethernet with no issues (most of my PCs have NIC diagnostics that show physical error counts, and none so far). Using Cat 6 would have been overkill, but perhaps when 10G Ethernet becomes commonplace I might regret that!

Recently I've been re-ripping my CDs as .WAV files - once a handful of discs have been done, I copy the files from the PC to my NAS box. Providing the PC isn't accessing the HDD for any other reason, the average copy speed reported by Windows is around 60MB/s, which is 480Mb/s. The limiting factor is the disc speed rather than the network. When copying backups with lots of small files, the transfer rate slows down considerably, which is obviously caused by disc seek times. It's nice to know that the network is not the bottleneck...

Cat 5 cable is certainly very flexible, and is used for many things besides Ethernet. Some people make fancy speaker cables by paralleling up runs of it! I'm tempted to use spare runs as a crude "leaky feeder" system for my radio collection...

[ppppenguin]

It's claimed that Cat5 and Cat5e are more or less hte same stuff. Just that the tolerances were tightened for 5e.
http://discountcablesusa.com/ethernet-cables100.html

Apart from yet tighter spec it looks like the main change in Cat6 is more twists per inch: http://www.howtogeek.com/70494/what-kind...uld-i-use/
The terminations are much more tightly controlled.

Ethernet cable is pretty universal stuff. There are solutions for anything from mains to light over it. I jest, but low-ish power DC, audio, video (analogue and digital) and VHF/UHF aerial distribution can all be done with standard products.

[Mike Watterson]

It's claimed that Cat5 and Cat5e are more or less hte same stuff. Just that the tolerances were tightened for 5e.
http://discountcablesusa.com/ethernet-cables100.html

Ethernet cable is pretty universal stuff. There are solutions for anything from mains to light over it. I jest, but low-ish power DC, audio, video (analogue and digital) and VHF/UHF aerial distribution can all be done with standard products.

Yes, there is just no assurance that 100m of Cat5 will perform as well as Cat5e. On shorter runs it makes no difference, and you might get more than 100m on Cat5e at 1Gbps and probably at 100Mbps, though a cheap switch as a "repeater" will reliably give 200m.

I prefer cheap "satellite" coax rather than twisted pair for UHF /VHF and even connecting a supposed 50 Ohm  Discone to a transmitter the SWR is perfectly acceptable 50MHz, 144MHz and 430MHz. The mismatch of using 75 Ohm PF125 for 50 Ohms reception isn't much and while not as cheap as regular "sat cable" it's a lot cheaper than 50 Ohm RG213. So I buy outdoor grade PF125 as well as the ordinary sat cable that's a bit better than  cable company RG6. BNC connections are a problem, but PL259 are not (drill out) and suitable N-Types and F-Type are sold cheaply for cable industry. I have the two sizes of F-Connector and crimp tools (a lot easier to make up than BNC or N), also adaptors to BNC, SMA and Belling-Lee. FME seems to be a 50 Ohm plug similar in concept (but incompatible) to F-connector. The screw on and pop-on F-connectors are rubbish.

[Mark Hennessy]

Yes, there is just no assurance that 100m of Cat5 will perform as well as Cat5e.

Honestly, for Ethernet, that's simply not true.

Gb Ethernet requires Cat 5 or better. Not Cat 5e or better.

On shorter runs it makes no difference, and you might get more than 100m on Cat5e at 1Gbps and probably at 100Mbps

On longer runs, it also makes no difference.

Maximum segment lengths are 100 metres, irrespective of cable type and network speed. Perhaps you'd get away with longer lengths if you push it, but that goes against the standards and would fail certification.

Aside from the marketing, the only difference between Cat 5 and Cat 5e is a 3dB difference in NEXT and ACR (crosstalk). That might matter in some non-standard applications, but for Gb Ethernet, there is no material difference. Properly installed Cat 5 works perfectly at Gb speeds.

Both are specified to 100MHz, and installations are certified using a test at 155MHz.

Also, don't forget that the symbol rate is the same for 100Mb and 1Gb Ethernet - the speed increase is from the use of all 4 pairs and different modulation. So it if works at 100Mb/s and all 4 pairs are present and correct, then it'll work at Gb.

Back in the late 1990s, there was a massive project to flood-wire all our sites using Cat 5, and today, most devices in use are running at Gb speeds with no issues whatsoever. Of course, newer sites will have been done using the current flavour-of-the-month (Cat 5e or Cat 6 or Cat 6A), but the remaining original Cat 5 installations are just fine, some 20 years on, and have many years ahead of them. An excellent ROI.

I repeat: Cat 5 is fine for Gb. If you find an old box of it in good condition, there is no reason not to use it unless you think you'll be using 10Gb networking one day...

OK, the argument is perhaps moot, given how little Cat 5 is out there, but let's at least get the facts right

[ppppenguin]

A whole range of devices to send professional audio and video over Cat5/5e/6:
http://www.canford.co.uk/Index/CAT5-conv...O-ADAPTERS

[Nick]

There are two standards of GbE over copper, 1000BASE-T and 1000BASE-TX.

1000BASE-T uses all 4 pairs each running bidirectionally on CAT5 or better.

1000BASE-TX uses 2 pairs, one for TX, the other for RX, however it requires CAT6 or better.

1000BASE-TX was a commercial failure as CAT6 is expensive and most folk already have CAT5, also 1000BASE-T kit is cheap, so 1000BASE-TX never caught on. Pretty much every GbE installation is thus 1000BASE-T.

PoE (Power over Ethernet) is extremely important in structured cabling environments - with all 4 pairs in 1000BASE-T being used for data, PoE is delivered by exploiting the fact that the pairs are differentially driven to reject common mode noise, so a DC level can be imposed and extracted with no effect on the data. All Ethernet terminations use transformers for galvanic isolation, however PoE terminations use center-tapped transformers to extract the imposed DC.

One slight difference between CAT5 and CAT6 is that CAT5 is 24 AWG and CAT6 is 23 AWG, so whilst PoE on CAT5 is tecnically limited to bundles of 100 cables due to heating (assuming everyone is drawing 30W or so), CAT6 bundles can be bigger...

Finally, 1000BASE-T devices use auto negotiation (auto MDI-X), so there is never a need to use different cables (crossover vs.  straight through) for endpoints and uplinks, even when talking to older 100BASE -T devices which have distinct TX and RX pairs like 1000BASE-TX, thus reducing installation errors and simplifying switches (no longer any need for a distinct uplink port in low-end kit).

[pwdrive]

On the subject of transmission lines and loads here's a link to a series of videos for a quick intro to using the Smith chart, not fully comprehensive but easy to follow, plenty of other explanations on the web:

https://www.youtube.com/watch?v=hmqM8PnUkmo

Another by the same bloke:

https://www.youtube.com/watch?v=Cd1H5ElNBo8

Lawrence.

[Mark Hennessy]

There are two standards of GbE over copper, 1000BASE-T and 1000BASE-TX.

1000BASE-T uses all 4 pairs each running bidirectionally on CAT5 or better.

1000BASE-TX uses 2 pairs, one for TX, the other for RX, however it requires CAT6 or better.

1000BASE-TX was a commercial failure as CAT6 is expensive and most folk already have CAT5, also 1000BASE-T kit is cheap, so 1000BASE-TX never caught on. Pretty much every GbE installation is thus 1000BASE-T.

That's right - I said the same in my original post (#19). Today "Gb Ethernet" is synonymous with "BASE-T", but it wouldn't hurt to add "BASE-T" to my more recent posts for clarity.

PoE (Power over Ethernet) is extremely important in structured cabling environments - with all 4 pairs in 1000BASE-T being used for data, PoE is delivered by exploiting the fact that the pairs are differentially driven to reject common mode noise, so a DC level can be imposed and extracted with no effect on the data. All Ethernet terminations use transformers for galvanic isolation, however PoE terminations use center-tapped transformers to extract the imposed DC.

One slight difference between CAT5 and CAT6 is that CAT5 is 24 AWG and CAT6 is 23 AWG, so whilst PoE on CAT5 is tecnically limited to bundles of 100 cables due to heating (assuming everyone is drawing 30W or so), CAT6 bundles can be bigger...

The PoE point is a good one, and I remember a colleague trying that on me before I started flood-wiring my house. That was just about the only "concrete" reason I had to consider Cat 6 cable, but there were plenty of other reasons to stick with Cat 5, including cost (I already had the box!), ease of installation, etc. Ultimately, the bundle leaving the patch panel will only contain 40 or so cables when finished, and the runs are all quite short - it's only a 3-bed house. But when I win the lottery and move to that mansion in the country, it'll be Cat-8 all the way

Finally, 1000BASE-T devices use auto negotiation (auto MDI-X), so there is never a need to use different cables (crossover vs.  straight through) for endpoints and uplinks, even when talking to older 100BASE -T devices which have distinct TX and RX pairs like 1000BASE-TX, thus reducing installation errors and simplifying switches (no longer any need for a distinct uplink port in low-end kit).

Yes, auto MDI-X made life so much easier.

Auto-negotiation was a bit problematic for a while. One example I remember is the Compaq/DEC Alphaservers that were used for audio playout: they would try to run at half-duplex while the switch would try to work at full-duplex. Manually setting the switch to full-duplex (100Mb/s) worked fine. Other switches had the same behaviour; others did seem to negotiate correctly - we didn't identify any patterns between brand or anything else IIRC. Happily, things have moved on quite a bit since then.

[Terry]

As AF has low ish reactance I guess it doesn't matter that my test cable collection all inherited from various places is a mix of 50, 75 and unknown. 

Be careful of long runs though, Andy!

Many years ago I got involved with a quiz night a local club was running. However, this wasn't the common 'everybody joins in' type quiz found today but a quiz between competing teams. 

One night, for some reason I can't recall, the quiz was moved to the other end of a long, narrow bar. Microphone cables? No problem, we've got extensions - and out came three cable reels for the three microphones - one for each team and one for the QM.

Two microphones worked beautifully but the third was badly muffled with no HF response at all. The microphone was eliminated, so I gave the cables the once over.

Two were twin screened 'microphone' cable but the third was 75Ω TV downlead!

If you look at the specification of any such cable and note the capacitance per metre, you'll soon realise why 75Ω coax (or 50Ω or whatever) is useless for audio ...

 

[pwdrive]

http://uk.rs-online.com/web/p/multicore-...e/7492576/

http://www.jyebao.com.tw/files%20download/RG%20TYPE.pdf

Lawrence.