Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[sheddyian]

Some months back, I happened to try swapping the RJ11 lead from the modem to the phone socket with a different one.  The original was very thin, a flat 2 wire thing, marked 28AWG.  With this cable I typically had an UP S/N Margin of around 8.  The replacement, was a similar but thicker flat cable, marked 26AWG, and with this I consistently got an UP S/N Margin of around 9.5 - this seemed a significant jump, although synch speeds seemed largely unaffected.

Curious, I ordered an ADSL Nation twisted pair RJ11 modem lead.  It certainly looks good, nice thick cable, quality RJ11 plugs (with screening).

Using this extra-super cable, I get a S/N Margin of around 8, even if I earth the screened plugs rather than having them float.

Anyone have any idea why a seemingly superior twisted-pair cable would give me a lower S/N Margin than a cheap flat one?


Cable                              S/N Margin UP

Original flat RJ11 28AWG             8.1
Similar RJ11 26AWG                   9.9
With ADSL Nation T/P RJ11            8.0


I'm assuming here that a higher S/N Margin indicates a generally better line/signal quality.  Thoughts?

The ADSL Nation cable has 2 pairs, all 4 RJ11 pins connected - so one pair is probably floating.  Acting as an aerial? But then the cable is foil screened as well.  The other two flat RJ11 cables only have the centre two pins connected to one pair.  All are broadly the same length, from about 1 metre to 1.5 metres, the ADSL Nation lead being the shortest.

Ian

[roseway]

A snapshot of the noise margin on its own doesn't really tell you much. To make a proper comparison you need to look at the connection speed and also how the noise margin varies over time.

[guest]

Lower AWG = larger diameter which = more conductor area to use (skin effect and all that) which should in turn lead to lower insertion loss - all else being equal. 26AWG is 130Ω/km, 28AWG is 210Ω/km or thereabouts.

Twisted pair cabling radiates RF less than flat cable and is less prone to induced noise.

Screening isn't going to do anything unless the screen is grounded at both ends and in fact if its not then it makes for a nice antenna, especially where foil/centre earth is involved.

I wouldn't expect to see much of a difference over 1-1.5 metres. The 26AWG should be marginally better than 28AWG is about all that can be said. All much of a muchness really.

[c6em]

I was always taught that the grounding at BOTH ends of a screened cable leads to earth loops and makes things worse, and was told to only ground the foil/drainwire at one end.

[roseway]

I'm a bit rusty on such matters, but I think that earth loops are a problem at audio frequencies, and can lead to hum in a sound system. I don't think that they are a problem at RF frequencies.

On the other hand, screening can be effective without the screen being earthed at all, because it acts as a Faraday cage (although once again I'm unsure in the case of a screened cable, in which the screen is very close to the wires). I suppose it could be the case that earthing it at one end only is the worst option.

[sheddyian]

Been a bit busy, will post up a bit more info on line speed etc (though I don't think it changed much, even though the S/N margin did).

What I've noticed is that the S/N margin doesn't vary very much, but the readings I get are consistent with the specific cables.  The thicker cable always gives me a higher S/N margin than the thinner one, and the fancy screened twisted pair ADSL Nation one always gives me around 8db. 

A quick look now. my S/N Margin UP is 8.0 and DOWN is 6, which is where it was when I originally connected this particular cable.

The modem *does* refresh the readings, so I don't think they're stuck - I have very occasionally seen the DOWN S/N Margin drop to 3.8 ish for a while, before creeping back to 6.

I will see if I can get some Routerstats graphs from it in due course.

I can confirm that I have grounded only one end of the ADSL nation shielded RJ11 cable, and it's connected to the earth that runs up the unused pairs of the CAT5 telephone wire, grounded to an earth rod in the garden. 

Ian

[guest]

Grounding the screen at only one end leads to the cable looking like an antenna to RF signals.

Given the length of the cable in question I don't see what the point of screening is. I guess it may be useful if you have a noisy DECT base unit beside the cable but I wouldn't imagine it'd make any difference for a 1.5m cable.

The ADSLNation cable sounds rather like the smoke and mirrors "audiophiles" come out with when they're wittering on about gold connectors and oxygen-free cable - ie makes no difference at all

FWIW I use a cable that came with a 56k Motorola modem to connect the router. Its 2m long & I've never seen any difference in stats/sync rate when I've been swapping routers about (using the cable which they came with).

Unless its a long run of cable (10m+) and/or its got stranded wire/is mega-cheap flat cable I don't think it makes much odds what you use.

[burakkucat]

Grounding the screen at only one end leads to the cable looking like an antenna to RF signals.

Thank you for explaining the above. That is my new thing, learnt today. 

[HPsauce]

Grounding the screen at only one end leads to the cable looking like an antenna to RF signals.

Well yes and no, in that it's an earthed low-resistance antenna and so still shields the cables inside.

Unless you've very close to a transmitter the "earth-loop" induction from mains equipment is likely to be much higher in magnitude, which is why only one end is earthed.

Have a read: http://en.wikipedia.org/wiki/Ground_loop_(electricity)

(something rather odd about copying that link - click on it then follow the "did you mean"? ) 

[guest]

Grounding the screen at only one end leads to the cable looking like an antenna to RF signals.

Well yes and no, in that it's an earthed low-resistance antenna and so still shields the cables inside.

Unless you've very close to a transmitter the "earth-loop" induction from mains equipment is likely to be much higher in magnitude, which is why only one end is earthed.

Have a read: http://en.wikipedia.org/wiki/Ground_loop_(electricity)

(something rather odd about copying that link - click on it then follow the "did you mean"? ) 

I wasn't talking about antenna reception - try grounding/earthing a coax at one end and pump RF down it, see what happens It's the classical half/quarter wave dipole.

[HPsauce]

Grounding the screen at only one end leads to the cable looking like an antenna to RF signals.

I wasn't talking about antenna reception

 

[sheddyian]

Grounding the screen at only one end leads to the cable looking like an antenna to RF signals.

Given the length of the cable in question I don't see what the point of screening is. I guess it may be useful if you have a noisy DECT base unit beside the cable but I wouldn't imagine it'd make any difference for a 1.5m cable.

The ADSLNation cable sounds rather like the smoke and mirrors "audiophiles" come out with when they're wittering on about gold connectors and oxygen-free cable - ie makes no difference at all

FWIW I use a cable that came with a 56k Motorola modem to connect the router. Its 2m long & I've never seen any difference in stats/sync rate when I've been swapping routers about (using the cable which they came with).

Unless its a long run of cable (10m+) and/or its got stranded wire/is mega-cheap flat cable I don't think it makes much odds what you use.

Agree re the audiophile cables, though there can be a benefit in having a good quality cable compared to a nasty cheap thing - recently swapped a friends scart cable for a nice thick chunky one, and the picture was immediately much sharper, far less blurred.  But that's more down to the old scart being cheap and nasty, I think!  Reading some of the nonsense on expensive digital interconnects (this HDMI cable gives better colour definition than that HDMI cable...  )

Te reason I'd bought the ADSL Nation cable was curiosity - if swapping from a basic 28AWG lead to a basic 26AWG lead had gained me 1db extra noise margin, what would putting a twisted pair lead in it's place do?  Answer, strangely, drop me back by 1db ! 

As I said, this is all repeatable - if I put the basic 26AWG lead back, I will get around 9db S/N margin once more, even though the lead is only about a metre long.  It doesn't make a lot of sense, but that's the numbers I get.

Grounding, or not, the screen on the RJ11 doesn't appear to make any difference.  It's a metallic plug that's evidently intended to connect with a metal socket, but since the ADSL faceplate and the modem have plastic sockets, the shielding will usually float, which is how it'll be with most people who buy one of these leads.  I deliberately connected a stiff copper ground wire to the RJ11 outer to connect it to ground - with no perceivable difference in stats.

Ian

[guest]

Grounding the screen at only one end leads to the cable looking like an antenna to RF signals.

I wasn't talking about antenna reception

 

I knocked out our Freeview DVB-T channels and everyone else's channels within 50m before the switchover by grounding one end of CAT5e cabling and shoving gigabit ethernet down it. It took 3Mbps off the ADSL speed too.

I should probably be clearer on what I say - sorry. To me an antenna is an antenna, it has gain and maybe some directional ability and that's it - screening will make bugger all difference to reception in this case as the conducted/induced noise from the E/D-side cabling will be much greater.

You can radiate some local noise yourself as I found out, should have known better but I haven't worked on MF/HF stuff for 25 years - recent stuff is all above 6GHz, usually nearer 40GHz these days.

[guest]

As I said, this is all repeatable - if I put the basic 26AWG lead back, I will get around 9db S/N margin once more, even though the lead is only about a metre long.  It doesn't make a lot of sense, but that's the numbers I get.

Well reducing the AWG - ie increasing cable diameter and more importantly circumference - is likely to reduce insertion loss. I don't know about 1dB/metre but that seems to be what you've got.

When you increase cable circumference (within reason) then RF signals* suffer less insertion loss. That means your attenuation figure has reduced. It doesn't mean the noise on the line has reduced.

So putting in 26AWG cable and getting a better result is perfectly logical - I'm just somewhat surprised the difference over a metre is noticeable.

*google skin effect

[HPsauce]

grounding one end of CAT5e cabling and shoving gigabit ethernet down it.

Surely Cat5e is normally unshielded twisted pair...... so what did you ground?