This is a reply to my own post, dated 17 August 2010. It has taken this long for me to be reasonably sure I have tracked down the problem with my ADSL conection.
In my last post I stated a BT engineer attended in February 2010 but did not do anything specific to fix the problem. Actually there were 2 BT engineer visits. The first did some checks but never made a report back to the ISP, so the ISP requested a second visit. The Second BT Openreach engineer did some checks and changed the connection block in the junction box attached to the outside wall of our house. By the way the phone line goes immediately underground from this outside junction box; as far as I know it is underground all the way to the exchange (other than going into the odd green cabinet en-route).
As noted in my last post none of the Openreach visits cured the intermittent problem. The problem came back in August / September 2010.
I have been suffering from this intermittent problem for literally years and have looked into many of the popular claimed reasons for bad connections (radio interference, domestic appliances, neon lights in power connector blocks). The frustration is, when the line is ok it is as good as it gets - 8128kbps downstream.
Something that has changed since early 2010 is I retired, and have had more time to fault trace and test what was going on.
In early 2011 I decided to go back to basics and start with the router (Netgear DG834GT) connected direct to the Master Socket 'test socket', without any phone equipment connected. In setting up the router in this way, I realised something I had not noticed before. In February 2010 the Openreach engineer had changed the Mater Socket inside the house, as well as the connection block in the junction box on the outside wall - BUT he had not changed the faceplate. The old Master Socket had the BT and "piper" logo in the top left corner (from 1993, when the house was built), the new Master Socket has the Openreach logo in the top left corner. But I realised the faceplate was the original from the original (1993) Master Socket.
In the past I had checked extensively the phone wiring in the house. And the wiring to the exchange is all underground and relatively new (and all copper according to the 2nd Openreach engineer). I began to wonder whether the problem was something to do with the faceplate itself. However I have thought I had found the cause of the problem before, only for it to return at a later date. So I needed to do some testing to check out the faceplate theory.
There are two 'spur' phone lines in the house, leading from the Master Socket. By the way, over the (literally) years I have had this intermittent problem, these two spur lines have been connected:
1. both plugged into the front socket of the faceplate (using a doubler plug);
2. One connected to the IDC connectors on the back of the faceplate, the second plugged into the front socket of the faceplate.
When the Openreach engineer attended, the two spurs were connected as in 2 above; maybe that is why he did not touch it - but that is beside the point.
I did about 2 months of testing with the faceplate out of the connection path. I will not go into details about what that involved but it was all legal, I did not touch the Master Socket itself. I concluded it is somewhere in the circuit board in the faceplate leading to the 'plug' on the back of the faceplate, or a fault in the faceplate 'plug' connectors.
Having convinced myself it was something to do with an intermittent connectivity or connection issue in the faceplate, in March 2011 I went to Maplin and bought a standard NTE5A Master Socket. I took the faceplate from the Maplin Master Socket and put it on the Openreach Master Socket. This new faceplate does not have any microfilter, it is a standard faceplate. The two spur lines are connected to the new faceplate using method 2 above (one on the IDC connectors, the other plugged into the front).
From the time I started my testing in February 2011 to now (end January 2012), the intermittent connection problem has not occurred, most of that time has been with the new faceplate. Of course there has been the odd issue with power cuts causing re-syncs. As of today, since last re-sync, the ADSL connection has been running for 40 days (downstream: speed 8128kbps, attenuation 22dB, noise margin 8.3dB - RCO 86%, 9440kbps) and the total CRC errors are 77 in 77 error seconds, with no Severe Error Seconds. The CRC errors seem to occur mostly when there is an incoming call and so a ring in event. Sometimes an incoming call does not cause and CRC erros, and somtimes they seem to occur with no obvious reason. I noticed this happens even when the router alone is connected direct to the test socket. But as this level of CRC errors does not cause any problems, I have not tried to sort it.
Since changing the faceplate, the longest continuous run has been 100 days (8128kbps downstream speed) - in Autumn 2011. That ended when I realised that although the downstream speed was 8128kbps, and my ISP showed the current speed as 7150kbps (correct IP profile number for the speed of 8128), the bRAS had somehow become stuck at 2000kbps. It took over 2 weeks of ticket tennis with my ISP to get BT to reset the bRAS.
For me, the circumstantial evidence is my intermittent ADSL connection issue was a simple continuity / connectivity issue in some part of the original Master Socket faceplate. The faceplate was certainly not changed when the rest of the Master Socket was changed in early 2010. Since changing the faceplate for a new one the intermittent problem has not recurred.