Running from the test socket, faceplate removed, makes no difference that I can discern. I even ran it with a flexible extension lead into the master socket, just a splitter but no telephone connected; this reduced the sync speed and increased the attenuation as one would expect, so although I thought that the fixed extension lines were flat, as you supposed, they are actually performing very well, so are probably twisted.
Not sure where this leaves me. You seem to suggest that noise is being picked up, which makes me wonder from where? While I can get the ISP to raise the target SNR, I should be getting a better speed - as you correctly point out. I guess the only option now is to bite the bullet and book an engineer(!) visit.
Edited by Mygri (Tue 13-Aug-13 11:53:13)