BB slower on fast profile

My BT BB speeds have fallen steadily for the past few weeks, and I would be very grateful for TBB�s collective advice. I�m on BT�s basic package with 10Mb per month and checks with a few ISPs show that I can expect up to 8Mb on my line, which is 3000m from the exchange by the BT engineer�s test gear. There is FTTC to a cabinet down the street, the last 250m is underground duct to a pole and about 30m of overhead line to the house. This line was renewed in 1986 and runs to a junction box on the rafter, thence 2m down to to the hall and an old master socket (65mm square, dates from 1968, has no test socket etc). There is an 8m extension from the socket to the HH2 in our study.

We have two HP computers running W7, one ethernet to HH2, the other wireless to the HH2 about 2.5m across the room. I have tried removing the extension and connecting HH2 direct to incoming line with ethernet to computer, made no difference to speeds although I�m happy to ask BT for a new socket in the study if it would help.

TBB download speeds since ADSL2 early this year have been 8/9Mb, BT wholesale being similar. Uploads were steady 0.95Mb but have slowed to 0.94. At congested times our BB is slower but satisfactory for our purposes. The router resets itself every few days, I know not why. However, between Sept 22 and 25 the line was changed from interleaved to fast, again I know not why. From then on speeds have decreased:
graph
I replaced the microfilters which were six years old and received a 20% improvement but we seem to be stuck at 5/6Mb down, 0.94 up. Noise margins, attenuation and power figures have little changed over the year and there is no new wiring or electrical equipment etc.
This reading was recorded on Sept 22 and is similar to those earlier in the year:
22.09.2013 TBB Dn 7.55Mb http6 9.66Mb lat 50ms
Downstream 8,298 Kbps
Upstream 1,144 Kbps
Latency type Interleaved
Noise margin (Down/Up) 9.6 dB / 6.4 dB
Line attenuation (Down/Up) 33.5 dB / 16.3 dB
Output power (Down/Up) 13.1 dBm / 1.7 dBm
Loss of Framing (Local/Remote) 0 / 0
Loss of Signal (Local/Remote) 0 / 0
Loss of Power (Local/Remote) 0 / 0
FEC Errors (Down/Up) 3288 / 1
CRC Errors (Down/Up) 687 / 0
HEC Errors (Down/Up) 12809 / 0
Error Seconds (Local/Remote) 235 / 0

09/10/13 Speed 6.47Mb lat 40ms
Connection time 1 days, 16:41:51
Downstream 5,720 Kbps
Upstream 1,103 Kbps
Latency type Fast
Noise margin (Down/Up) 14.9 dB / 7.1 dB
Line attenuation (Down/Up) 33.5 dB / 16.4 dB
Output power (Down/Up) 12.7 dBm / 1.6 dBm
Loss of Framing (Local/Remote) 0 / 0
Loss of Signal (Local/Remote) 0 / 0
Loss of Power (Local/Remote) 0 / 0
FEC Errors (Down/Up) 0 / 0
CRC Errors (Down/Up) 395 / 0
HEC Errors (Down/Up) 954 / 0
Error Seconds (Local/Remote) 371 / 0

18/10/2013 BT speed test 8Mb ping 40ms
Downstream 9,060 Kbps
Upstream 1,099 Kbps
Latency type Fast
Noise margin (Down/Up) 7.5 dB / 6.9 dB
Line attenuation (Down/Up) 33.4 dB / 16.2 dB
Output power (Down/Up) 12.6 dBm / 1.6 dBm
Loss of Framing (Local/Remote) 0 / 0
Loss of Signal (Local/Remote) 0 / 0
Loss of Power (Local/Remote) 0 / 0
FEC Errors (Down/Up) 0 / 0
CRC Errors (Down/Up) 30199 / 0
HEC Errors (Down/Up) 25541 / 0
Error Seconds (Local/Remote) 12539 / 0

21/10/2013 TBB test 5.8Mb
Connection time 2 days, 19:39:54
Downstream 6,227 Kbps
Upstream 1,115 Kbps
Latency type Fast
Noise margin (Down/Up) 13.4 dB / 6.2 dB
Line attenuation (Down/Up) 33.4 dB / 16.2 dB
Output power (Down/Up) 12.7 dBm / 1.7 dBm
Loss of Framing (Local/Remote) 0 / 0
Loss of Signal (Local/Remote) 0 / 0
Loss of Power (Local/Remote) 0 / 0
FEC Errors (Down/Up) 0 / 0
CRC Errors (Down/Up) 3117 / 0
HEC Errors (Down/Up) 4742 / 0
Error Seconds (Local/Remote) 2318 / 0

Today BTW test 5.59Mb down, 0.92 up ping 46.25
Connection time 3 days, 17:16:04
Downstream 6,227 Kbps
Upstream 1,115 Kbps
Latency type Fast
Noise margin (Down/Up) 13.8 dB / 6.3 dB
Line attenuation (Down/Up) 33.4 dB / 16.2 dB
Output power (Down/Up) 13.0 dBm / 1.7 dBm
Loss of Framing (Local/Remote) 0 / 0
Loss of Signal (Local/Remote) 0 / 0
Loss of Power (Local/Remote) 0 / 0
FEC Errors (Down/Up) 0 / 0
CRC Errors (Down/Up) 3732 / 0
HEC Errors (Down/Up) 5959 / 0
Error Seconds (Local/Remote) 2804 / 0

Big change in downstream noise margin explains drop in speed and even when you thought you had good speeds they were below what you would expect for attenuation

Http://www.coolwebhome.co.uk/calc will illustrate difference that varying noise margins makes

This line was renewed in 1986 and runs to a junction box on the rafter, thence 2m down to to the hall and an old master socket (65mm square, dates from 1968, has no test socket etc). There is an 8m extension from the socket to the HH2 in our study.

Is that the only extension wire ie only 23 sockets in house ? Is it hard wired or plugged in ? Have you disconnected the ring wire at the master socket if it's hard wired ?

There are two extensions, one hard wired to master phone in hall (BT Freelance, serves two slave phones by radio) and the plug-in extension to the HH2 in our study. The incoming line goes into push-on connectors and the hall extension is pushed on on top of it. There is no ring wire. In Sept I disconnected both extensions and tried the HH2 alone into the master socket but it made little difference to speeds, TBB average 9.88Mb via extension, 20 mins later 9.77Mb plugged direct into socket.

The phone side is quiet and has always worked perfectly. My tech very limited so all advice welcome but given our 45-year-old socket which has been hacked about over the years, would it help to have a new socket installed in the study, dispensing with all the old four-core cabling and the extensions? And should I ask to have Fast returned to interleaving, as the slowdown became apparent since then?

It is less the difference to speed test results, but more the subtle variations in noise margin and connection speed information from the router that are important.

It is less the change from fast to interleave mode, but more the change to the target noise margin you want to get corrected. Reducing the telephone wiring to a single socket might or might not help you, until you try you cannot be sure.

Modem stats from the test socket are the best guide as to whether that work is worth the effort or not.

Post deleted by RobertoS

Thanks, I did not realise the importance of noise margin. Unfortunately there is no test socket in this old fitting, I removed the cover and ensured the only cable is the incoming line, then plugged in the HH2 as the only output. A little better today, the line speed is up slightly, I presume this is DLM?
Connection time 0 days, 22:10:34
Downstream 6,907 Kbps
Upstream 1,112 Kbps
Latency type Fast
Noise margin (Down/Up) 11.5 dB / 6.8 dB
Line attenuation (Down/Up) 33.4 dB / 16.2 dB
Output power (Down/Up) 12.3 dBm / 1.6 dBm

Probably not DLM at all, just the way the physics of RF signals varies over time.

If that was the only socket working in the property, then looks like line may be capped on the downstream.

Getting ISP to lower the target noise margin would boost speeds
http://www.coolwebhome.co.uk/calc/index.php?param=RG...

But as you can see still no where close to the expected speeds. Suspect DLM may have done something odd to the line, maybe in response to bursts of noise picked up by the wiring in the property.

I shall ask BT to replace incoming line and 40-yr-old junction boxes, then we'll know where we stand. Many thanks for your advice.