What about all the wires on the left side? Should I disconnect to make it neater/tidier? I think that's what put me off. Wasn't as concerned about the right side.

Left side is BT's so don't touch it.

Okay so I had a bash today. I need some better tools. My slipnose pliers had started to rust! Consequently, I managed to cut one of the wires rather than just slip off the plastic coating.

Before (well last week):



After:



I need to go round and sort out the extension downstairs. The extension upstairs is a cordless phone that has its base downstairs so unconnected.

Just wanted to verify that in the junction box I posted above,

1 = orange (?/white)
2 = unconnected
3 = unconnected
4 = white
5 = orange/white ->now changed to white/blue
6 = unconnected
7 = unconnected
8 = blue/white

I know I may have gotten things the wrong way round so please advise.

That's fine. Good to see a slight improvement. You may find it's more reliable too.

If the socket upstairs isn't being used, then disconnect the cable to it at both ends. Provided you can identify it of course!

I really appreciate the help from JJW009 and Rusty.

If memory serves correct, there were two lots of wires in the "master" socket, each lot coming from one cord. One cord goes to the junction box and the other goes to the extension. Is this how they're set up? Daisy-chained? From what I could see in the garage, the extension from the master socket goes to the front room (and not upstairs). Does this mean the front room socket has an extension to upstairs? Is this why there's a REN rating on products? Can't exceed a max number without distorting the signal/being unpowered or something?

Without looking at your wiring myself, I'd hesitate to give an answer. There's any number of different ways to wire up a phone circuit.


Bingo. If the total REN on your phone line exceeds 4, things like Caller ID and broadband can start to be adversely affected. If you are particularly stupid about it, the exchange may cut you off.

However, it's a totally made up number and you can't really just add it up; impedance and other such things just don't work like that.

The Wiki entry is quite detailed: http://en.wikipedia.org/wiki/Ringer_equivalence_number

Although the impedance of all electronic devices is complex, comprising of real and imaginary components at right-angles to each other, the important thing is that the total REN of a parallel circuit will never exceed the sum of the individual REN ratings of the component parts. This is the same as saying that the length of the hypotenuse of any triangle will never exceed the sum of the two shorter sides. Since the REN value is the magnitude of a vector, we will never know the true value of the sum of two or more.

For example

If we have complex conductances a and b such that:
||a||=1
||b||=1

then all we know about the combined conductance in a parallel circuit is that:
0 =< ||a + b|| =< 2

A conductance of 0 is an infinite resistance, and would occur at a specific resonant frequency if one component was purely conductive and the other purely capacitive (with zero resistance). That's unlikely to happen with phone wiring, but it is how you tune a radio (if it's superconducting lol)

Another electrical example of "inadequate information" is the rating of a UPS and it's complex load. Typically given in VA, this is once again the magnitude of a vector, so we will never know the true value the sum of two or more. In some real life circumstances, 100VA + 100VA might actually add up to less than 100VA, which is counter-intuitive. Note that the power consumption in Watts adds up normally, since power is a scaler value. 100W + 100W = 200W always. Many cheap power-meters incorrectly label their display as Watts, when actually they are measuring VA. It requires real time integration to accurately calculates Watts.

Yes. And that's a good thing.

Your picture of the master socket indicates one cable (6 wires, 4 unused) going to the junction in the garage. The other pair of wires go to one of the extensions. We then deduce that he third socket must be daisy-chained off the second, or an additional cable would show in your picture.

The alternative would be to "star-wire" each extension back to the master. This would be a disaster for broadband, although it is often still done by engineers with out-dated education or too little time to do the job properly.

The reason star-wiring is a big fat NO for broadband is that the impedance of two lengths of cable in series remains constant (daisy-chain), where as putting them in parallel (star-wired) reduces it by as much as half. When a high frequency signal (broadband, but not speech) hits the junction, a significant amount is reflected by the impedance mismatch.

For the same reason, the untwisted length of telephone or data cable inevitable in a junction must be kept to a minimum. There are very specific standards set out in CAT5 for example, and a failure to adhere to the regulations will result in sockets failing a test and dropping packets in real life. Unfortunately a lot of cable guys don't know the regulations exist, let alone why...

Strangely, I seem to understand what you're saying about star wiring and impedance. Maybe I'm doing something wrong.

Next question. I'm trying to minimise data loss. Sticking the router next to the "master" socket is a no-no since the wi-fi signal would be too poor and also mean gaming lag for my PS3.

Instead, I have a long telephone extension wire. Recalling your post in a previous thread, this is where there's a lot of data loss (particularly as a portion of the wire goes outside).

I now plan to trim down the telephone extension to meet my needs so there's a little slack, but only a little. Is there any other way I can try and improve this signal? Eg some kind of shielding?

Okay, scratch that. Just went to check and I have an RJ11 cable that's just the perfect length fitted.

EDIT: It's this wire = http://www.amazon.co.uk/gp/product/B001 ... 00_s00_i02 - is it worth getting a different/"better" one?

It's impossible to say if that's any good from the info they give. Basically:

Flat cable, not twisted, not balanced = crap
Round cable, twisted pair, balanced = good

Easiest thing is to just try the router directly at the master socket and see what speed you get. If it's about the same as with the long cable, then the cable is fine.

It's a flat cable. I was looking at Cat5 cables but it was past midnight and I got confused over whether a Cat5 cable with RJ45 would work. Then I started looking at pin outs and how 5 connects to 2 on the telephone line and my head hurt.

Will have a look at moving the router downstairs and re-running the ping and speedtest tests.

Right, did the tests. Used the 3m wire as above when upstairs. Laptop connected via LAN rather than wireless LAN. Downstairs meant router was connected with the supplied 1m cable.

Upstairs



Downstairs



Back upstairs again

Well, I've been switched over to plusnet and am a little underwhelmed:



Ping speed is awesome though:



If I go to the Member Centre, I get this:



How long do you reckon it'd take for things to settle, or is this the most I'll ever expect to get?

What is your router actually synced at?

Given the history, I doubt it will improve much. In reality, ADSL 2+ pretty much gives you the best speed when you first plug it in, and then gets slower if the line isn't stable.

If you're up for it, you could connect a socket at the BT junction box and test with all the house wiring disconnected. That would prove once and for all if the house cabling is dragging it down.