|
|
I have been experimenting with modem cables for a few days. I got a Tandy one quite a while ago which is twisted pair and shielded. I also got another Cat 6 twisted pair through the post today. I have a couple more on order with different cables and plugs. Ideally, I would like a cable that provides the highest sync speed with an RJ11 at one end and an RJ45 at the other that is 2m long.
So far the speed tests
Tandy Cable 72/17
Cat 6 cable 76/17
Flat Billion modem cable 78½/17
So nothing beats the flat untwisted Billion cable so have ordered an RJ11 to RJ45 with flat cable hoping it is as good as the Billion modem cable.
All the experts say that twisted pair cable should be the best but my experiments prove otherwise so far and this is repeatable.
How strange.
More experiments with Billion 8800 R2 modem lead has given me another half a meg. Could be just the time of day but this cable is slightly shorter and round.
Current sync with 8800 cable 79112.
|
|
|
Even more experimenting. Tried the ZyXel VMG1312-B10A modem cable and sync is 77. Back to the Billion 8800 R2 cable and back to 79 with 80 attainable. It seems not all modem cables are equal.
|
|
|
You could also experiment with hard wiring one end onto the filtered faceplate's terminals A&B and a rj11 plug at the other end of a cat5e 2 metre cable - guess you don't like the idea of a rj11 plug wobbling about in the faceplate's rj45 socket 
|
|
Register (or login) on our website and you will not see this ad.
|
|
|
guess you don't like the idea of a rj11 plug wobbling about in the faceplate's rj45 socket
Yes that is the main reason for ordering Cat 6 cables with RJ45 and RJ11 plugs but it seems the cable isnt up to it. Tonight I ordered a standard modem cable with RJ45 and RJ11 so will test that when it arrives.
But so far the best syncing modem cable is the round Billion 8800NL R2 closeley followed by the Billion 7800 flat modem cable. More experiments to be conducted.
|
|
|
Don�t care for the flat 7800 cable myself - seems a bit long and flimsy, although having said that, for me, differing filtered faceplates seem to result in more noticeable changes in line stats than differing good quality cables. However that is with ADSL and a Mk3 faceplate seems about the best on my line.
Edited by 4M2 (Wed 16-Aug-17 01:24:28)
|
|
|
I have an MK4 5C box on my wall fitted by OR. At least I think it's a Mk4 but it is definitely a 5c as it just snaps on.
|
|
|
I have an MK4 5C box on my wall fitted by OR. At least I think it's a Mk4 but it is definitely a 5c as it just snaps on.
Yes I think the Mk4's internals are pretty much the same as a Mk3, just a different unique fitting. Certainly the Mk3 has only two contacts into the NTE5A test socket which may be advantageous.
|
|
|
|
Given the distance the signal has to travel I'm quite surprised that the modem cable makes any difference at all, however...
Twisted pair has benefits in multi-circuit systems:
Increased immunity to external interference and reduced leakage
Reduced crosstalk in multi-circuit systems but only if the twist is properly randomised
Paired wires held together give a consistent impedance
And of course there's the fact that they are clearly identifiable as pairs reducing wiring errors
On a single pair system the benefits of twisting don't really apply.
You probably don't have interference sources nearby
There's only one circuit so nothing to crosstalk to. This is especially true for "cheap" bundled leads whichfrequently have only two conductors, the outer wires are left off.
The insulation holds the wires at a fixed spacing.
There's really only one way to wire it.
It is possible that the simple flat cable could be better.
There might even be a slight benefit to a longer lead. The added loss should be trivial but all junctions introduce unwanted reflections and there may be a benefit to separating them so the reflections are less likely to add.
|
|
|
Whether the metal core is copper to just copper coated Al plus its gauge will be a factor too
|
|
The author of the above post is a thinkbroadband staff member. It may not constitute an official statement on behalf of thinkbroadband.
|
|
|
Whether the metal core is copper to just copper coated Al plus its gauge will be a factor too
Don't forget copper coated steel ... or even sliver, or silver coated copper.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
M H C
taurus excreta cerebrum vincit
|
|
|
|
You seem to have forgotten we are talking about analogue signals in a home environment where there are plenty of noise sources.
|
|
|
So got a nice thick Cat 6 RJ11 to RJ45 through the post today. Top sync - 77, can't beat the Billion 8800 R2 cable still at 79109.
|
|
|
Thick tells you nothing about the conductors, might just be an extra chunky layer of insulation or the centre plastic spacer making the difference
|
|
The author of the above post is a thinkbroadband staff member. It may not constitute an official statement on behalf of thinkbroadband.
|
|
|
It's an interesting academic exercise but particularly interesting is that Billion supply the fastest syncing modem cables.
|
|
|
it could also that the action of plugging and unplugging cables is wiping the contact pins clean of any dirt / oxidisation so thereby improving speeds? i used to do this with double enders in jackfields when I worked for the bbc
Edited by threelegs (Wed 16-Aug-17 18:22:18)
|
|
|
Possible but you re-insert a poor cable and the speed goes back to where it was on first test. I used to make jackfields.
|
|
|
|
Regarding longer lead, I have experience with "Thin" Ethernet co-ax cables.
These either had to be extremely short, a few inches, or 8 feet (2.5 Metres) long minimum.
Anything in between gave problems.
Management complained about the general "excessive" length, so I experimented with varying lengths, to confirm or otherwise my general observations.
The equipment being connected was just too far apart for the "inches" leads.
It worked perfectly with the 8 foot and longer leads.
But generally was very poor with in-between lengths.
-------
With "Thick" Ethernet co-ax, it has markings at about 8 foot intervals.
Vampire taps had to positioned at that pitch between them, not necessarily on the marks themselves; but that minimum distance apart - otherwise problems.
|
|
|
|
|
|
|
|
That link shows the "Thick" Ethernet with Vampire Tap etc, so "Yes" wrt to the last part of my comment.
The first part, Thin Ethernet, used coax cable about 6 mm diameter, similar to domestic TV aerial coax; but slightly different impedance, 75 ohm if I remember correctly, rather than 72 ohn for TV.
----------
The Vampire Taps had to be carefully attached to the Thick Ethernet, including drilling in to it from its far side, often teetering at the top of a 25 foot, 3-section, ladder, with the bits and pieces tucked inside my shirt top.
On one occasion, the company insisted that a "networking" company deal with a new extension, 5 taps, at about 7 feet up.
I had to re-do 4 of them, they were so badly installed.
|
|
|
|
Just noticed that the Thick impedance is given as "50 ohm", so the same for the Thin Ethernet, as it was possible to connect them end to end , using a special but simple adaptor.
I did this on a few occasions for demonstrations to customers.
|
|
|
|
I don't see the relevance to modem cable
|
|
|
Just noticed that the Thick impedance is given as "50 ohm", so the same for the Thin Ethernet, as it was possible to connect them end to end , using a special but simple adaptor.
I did this on a few occasions for demonstrations to customers.
10 BASE T was 50 Ohms and most used a BNC connector connected to a BNC T-Piece and each end having a 50 Ohms BNC Terminator.
But yeah I recall seeing each segment being a set fixed length, we had to coil up the cables so that they didn't get tangled up behind the desk.
UHF TV and Sat Cable are 75 Ohms.
Paul
|
|
|
|
The OP had mentioned the length of cable.
When dealing with RF signals, such as the BB ADSL and VDSL signals, these are in the RF spectrum 2 MHz to 17 MHz, so reflections, standing waves etc, can be coming in to play.
I have particularly encountered this with Ethernet cabling, 10 MHz, so used that to illustrate what can happen, especially with the spread of RF frequencies used by xDSL, and the general reduction of wavelengths in cables compared to "free space".
----------------
You may have noticed the posting by threeleg, (learned) Wed 16-Aug-17 18:21:15, regarding incidental cleaning of "dirt/corrosion" may be having an effect.
That repeated insertion/withdrawal may also be clearing dendrites that form in gold, plated version particularly, as encountered in plugs and sockets.
But no obvious difference to the casual user - "The fault just disappeared and all I did was unplug and plug back in".
|
|
|
|
What is the wavelength of vdsl rf carried by a modem cable?
|
|
|
|
Taking 10 MHz in "free space", it is 30 Metres.
In a cable of any sort, it will be distinctly shorter than that, depending on the cable characteristic, let's say 20 Metres.
(It may be even less - in my training, it was mentioned that the transmission speed in cables was generally around 1/3rd of the free space figure, which would give a result of 10 Metres rather than 20 M)
You then have to particularly consider the "Quarter-Wavelength" and multiples, so now down to 5 Metres @ 10 MHz.
Given that Ethernet of the "10 Base x" is 10 MHz; xDSL, and all the channels involved, imply a "Broadband" mixture of wavelengths, ranging from in free space, about 60 Metres @ 5 MHz to 20 Metres @ 15 MHz.
Hence the relatively higher Attenuation figures seen for the generally much shorter FTTC via PCP to Home, seen when changing from ADSL to VDS.
ADSL - 24 db for 1,286 Metres
VDSL - 13 db for 300 Metres on a 40/10 contract. (Expect higher on 80/20 Contract)
Leading on to the recommendation for some that VDSL may not lead to a BB Speed increase; and may actually cause a decrease.
|
|
|
|
So the length of a modem cable is irrelevant?
|
|
|
Given the distance the signal has to travel I'm quite surprised that the modem cable makes any difference at all,
indeed, however that last 2 metres is in the noisiest (electrically) enviornments the line goes through. typically having computers screens printers and various psu devices nearby.
|
|
|
|
It clearly is Relevant, as it adds in Length; and by implication, Attenuation, Noise etc.
Only experimentation as the OP has been doing, might establish "The Perfect Modem Cable" for the specific Installation.
|
|
|
BT cable with two RJ11 plugs, sync 65m which is way down.
|
|
|
Your first "guess" was about right for coax. Speed of propagation in coax is about 2/3 that of free space and does vary. Using 0.6, 0.65, 0.66, or 0.7 is normally reasonable enough.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
M H C
taurus excreta cerebrum vincit
|
|
|
|
Important data, such as CRC errors and error seconds, are unfortunately missing from your tests. An interesting post however for purely the speed comparison.
|
|
|
DS errors are zero and upstream is about 20 a day. I have only tested for 30 minutes max at a time so difficult to tell if upstream errors have changed but I suspect there is little in it as since I changed the modem errors have greatly reduced.
|
|
|
|
They should be available on mydslwebstats
|
|
|
I think I have come to the end of my experiment. Going to give it a rest before DLM gets upset with my line. So far the Billion 8800 R2 cable combined with ZyXEL VMG1312-B10A gives me the highest sync and lowest errors.
One more cable to try but not expecting much from it as it is a flat 2m cable with RJ11 and RJ45 plugs delivered for the princely sum of £1.19.
|
|
|
£1.19 flat cable sync 77meg so that is the end of testing. Billion 8800NL R2 cable is the winner.
|
|
|
10 BASE T was 50 Ohms and most used a BNC connector connected to a BNC T-Piece and each end having a 50 Ohms BNC Terminator.
That'll be 10base2 then...10baseT is ethernet over UTP...
 Comms is hard
|
Pages in this thread: 
Print Thread
Banger
