Puzzling redundant wiring

[Poverty]

Can any of you boffins explain this?

My telephone wiring is fed from the master socket to three other sockets via wiring that runs along outside walls and into three other rooms via the window frames.  The total distance to the furthest socket which feeds my computer is about 20 metres.  My sync speed is around 7.2Mb/s.

The cable is quite old and I recently noticed that the outside sheath was peeling off  revealing the inner wires.  I have recently replaced my telephones with cordless ones with the master connected to the master socket, so I thought that I would replace the old cable, now only used to feed the computer, with a couple of homeplugs.  I did this and, before I disconnected the old cable, I tested this arrangement.  It worked well, still giving a sync speed of around 7.2Mb/s with the router now plugged straight into the master socket.

So I disconnected the old wiring, expecting, if anything, to see a slight increase in sync speed.  However I now had a sync speed of only 4.9Mb/s, the only difference being the disconnection of the now unterminated 20 metres of old wiring.  I reconnected it, still with no termination and no filters on it, and again got a sync speed of 7.2Mb/s.

I repeated this three times and in each case obtained the much higher sync speed with the length  of old wiring connected, but feeding nothing.

At the moment , I have left it connected, but as further deterioration might cause it to short the line, I would like to remove it soon, but without reducing the sync speed.  Any clues, anyone?

[Rik]

Other than an interaction between router and phone, I can't think of a sensible reason.

[Steve]

What do you get with the router plugged into the test socket?

[Poverty]

This morning I did the following tests, with about five minute gaps in the hope that the DSLAM didn't interpret the disconnections as a bad line.

1   Router into Test Socket - no filter         5.9Mb/s
2   Router into Test Socket - via microfilter - no telephone      5.9Nb/s
3   Router into Test Socket - via microfilter - with telephone      5.9Mb/s
4   Router into Faceplate - no wiring - no telephone      4.9Mb/s
5   Router into Faceplate - no wiring - with telephone      4.9Mb/s
6   Router into Faceplate - with wiring - no telephone      7.2Mb/s
7   Router into Faceplate - with wiring - with telephone      7.2Mb/s

The redundant wiring is connected to terminals A and B on the back of the faceplate.
Perhaps I could simulate it with a 600 ohm resistor?

[Steve]

Doesn't make any sense to me, a lead that's connected but goes nowhere. I did initially wonder whether the presence of the lead behind the faceplate for some reason allowed a 'better' fit of the faceplate to the test socket.

[Lance]

I have to admit to being clueless on this one as well. As far as I can see there is no logical reason as to why having a piece of redundant cable aids a higher sync.

Does you attenuation change each time, or noise margin?

[Rik]

It's got me beat too, especially as it's on the AB terminals.

[MisterW]

The redundant wiring is connected to terminals A and B on the back of the faceplate.

That's odd, the normal faceplate doesn't have A and B terminals, there should be an IDC connection with terminals numbered 1-6.
A and B terminals are normally only on the backplate. What type of faceplate do you have ? is it a filtered faceplate ?

[nowster]

The only thing I can think of is that it allows a resonance on the line which is enhancing the higher part of the ADSL frequency band, or it's forming a notch filter which is removing something nasty.

[Rik]

We need to descend on this line with some test gear.

[Poverty]

The faceplate has a 6 terminal idc connector on the back numbered 2 3 4 5 A B.
A and B are for ADSL extensions and 2 3 5 are for telephone extensions which are filtered on the faceplate.
Yes, I had come to the conclusion that the wiring must be enhancing the speed by presenting a favourable impedance to the line.  Hence my suggestion that it could be simulated by a 600 ohm resistor, but perhaps some other component might do the trick, like a capacitor.
I gather that there are some speed enhancing faceplates available.  How do these work?
Unfortunately I have no test equipment these days.

[Poverty]

I forgot to add that the DS margin is the same for all results at 6dB which I suppose is to be expected after each sync.
The DS attenuation is around 38 dB with the wiring disconnected and 37dB with it connected, only a marginal difference, but still queer.

[nowster]

In case some have never heard about this one: http://catb.org/jargon/html/magic-story.html

[Rik]

Now that is weird.

[Poverty]

Here are some more interesting statistics, each over approx 8hrs.

Modem into faceplate - redundant wiring connected:
Sync speed = 7.1Mb/s
Total bytests received = 13298547      Bit errors = 0
Total bytes transmitted = 14157214    Bit errors = 0

Modem into test socket - no filter or telephone:
Sync speed = 5.8Mb/s
Total bytes received = 1473596     Bit errors = 849
Total bytes transmitted = 15479346    Bit errors = 2079

I'm now waiting to see whether my SNR is reduced due to my day of errors.

Does anyone know a circuit that will simulate a lenghth of line?  If so, they could make their fortune by building it into a faceplate design!!

[Rik]

I find a length of cable connected does it very well. Have you tried disconnecting the sockets at the other end of the cable?

[Polchraine]

It is almost certainly modifying the line impedance at the higher frequencies and potentially tunes the line and modem.

Try a resistor across - probably 4k7 as 600 will really kill the signal and also a small capacitor start at 100pF and then go either side, maybe up to around 10 nF.     It will be a case of play around and see what happens.