The output power looks low. It is usually around the 19dBm mark. I don't really know what causes that, but I think it is something at the exchange end.

Re the stats you didn't post, the error seconds, RS Corr or FEC (same thing) counts, and the time since the last connection (approx) are useful. Loads of stuff about bin/bit loadings isn't at the moment.

When the power from there to you is low the connection speed can be low, as yours is. So I would take that up with your ISP.

Sorry I didn't make it clear when I asked about the wiring in the sockets. It is worth checking what colours are on which terminal numbers at each and letting us know.

The output power looks low. It is usually around the 19dBm mark. I don't really know what causes that

Only being able to use a smaller number of frequency bands, as the power per tone is fixed less tones = less power.

In reply to a post by yarwell:

The output power looks low. It is usually around the 19dBm mark. I don't really know what causes that

Only being able to use a smaller number of frequency bands, as the power per tone is fixed less tones = less power.

Ah thanks. I thought it was the other way round.

Though perhaps it is when the bins are available? Less power means fewer bits per bin?

it's only just managing to connect, with a low SNR and a low speed.

Ideally you want a single pair coming in and going to the socket where you connect. A tee off isn't helpful.

You didn't say what the router was, on such lines you may benefit by trying different ones like the 2 Wire BT business hub (via ebay) also set the mode to multimod eif available in case ADSL2 can do better (or force ADSL2 as a test).

less total transmitted power in this case doesn't mean less bits per bin because the Tx power per bin is the same, there's less total power 'cos of less bins in use.

Imagine 256 modems on the line but only some of them (the lower frequency ones) in action, perhaps ?

On very short lines, and in some scenarios, the power does seem to back off if the max speed is being achieved with a sensible margin. If the power backed off on a long line then the SNR per bin would drop and consequently the bits per bin. (NB SNR not margin).

Yes, that's what I was getting at. I understood what you said, and was trying to clarify my misconception .

Much appreciate all the helpful and rapid contributions about my problem!
Re the wiring: my star-wired BT master LJU2/1A socket has white/green blob on T2, green/white blob on T5, and orange/white on T3. No other wires are connected. I note that the orange/white is connected in the BT junction box to the white/orange which serves the /3A socket. I assume this is the bell wire; should I disconnect it at both LJUs? (One of my phones is elderly � BT Relate 80 � and I�m a little worried it may not ring. Guess I�ll soon find out!)
I�ve noted elsewhere on this most informative website that disconnecting the bell wire may help a lot, but I�ve ducked doing it so far, in the hope that broadband settles down if I don�t touch anything (I�m half expecting it to go down again!). But after 48 hours, it seems to be stable, if slow.
Here are all the stats! (Can�t get the tabs right -- hope you can deduce.)

Mode: G.DMT
Type: Interleave
Line Coding: Trellis On
Status: No Defect
Link Power State: L0

Downstream Upstream
SNR Margin (dB): 4.5 13.0
Attenuation (dB): 63.5 31.5
Output Power (dBm): 12.3 13.7
Attainable Rate (Kbps): 448 764
Rate (Kbps): 352 448
K (number of bytes in DMT frame): 12 15
R (number of check bytes in RS code word): 8 4
S (RS code word size in DMT frame): 4 4
D (interleaver depth): 8 8
Delay (msec): 8 8

Super Frames: 4038550 4311498
Super Frame Errors: 202 55
RS Words: 68655350 73295466
RS Correctable Errors: 801648 225
RS Uncorrectable Errors: 844 N/A

HEC Errors: 144 35
OCD Errors: 4 0
LCD Errors: 0 0
Total Cells: 60849323 0
Data Cells: 520924 0
Bit Errors: 0 0

Total ES: 120 0
Total SES: 0 0
Total UAS: 10547 0

Router is TP-Link W8950ND. (I also have access to an Novatech/AddOn GWAR3550 which I could try.) Haven�t found a multimode facility yet.
Not sure what you mean by tee off; the socket is master /1A and the microfilter�s plugged into that. Do you mean T-shaped microfilter? It was a cheap one; is that bad news?

That wiring is correct so no worries there thank goodness . However the ring wire would be better disconnected. I would do it at all three points, just to be sure - both sockets and the junction box.

The problem with it is that it acts as an antenna, picks up wireless noise, and feeds it onto the line. Probably only feeds it on at the master, but given what you described I wouldn't be certain in your case.

Your old phones will ring anyway as the ADSL filters incorporate circuitry to make them, as do most modern phones. It is only certain phones used without filters that may stop ringing. So it is usually recommended when leaving a property to reconnect ring wires, just in case the next occupier doesn't have broadband and does have phones that need them. That's why we say don't cut them.

It will be interesting to see the stats after ring wire disconnection.

it sounded like you have a T shaped wiring setup with incomer going to two sockets. One would be ideal - can you remove the other ?

check the stats are the same with no phone in the microfilter then you'll see if the filter is OK