Why is a "degraded" line not a faulty line?

[#1]

Valid point about the PSU, I'll come back to that in a mo.
Yes, by all means send me a PM John, that's fine.
Nice screened modem lead as well, I see. So on the phone wiring side all covered apart from the incoming BT cable. I'm guessing that before the cable was painted (and outside) it looks as though it's the modern standard CW1411 drop wire, black sheathed, round, ~5.3mm in dia. If so, we can forget about fixed phone side wiring.
Now, how long have you had the power socket and modem/router & cable arrangement setup like that? Not since around your problems started by any chance?
Rich has already commented about the PSU - they can go noisey (if it's switch mode rather than just an AC transformer), the coiled up modem lead where it is may be causing issues, along with the proximity of the Homeplug (also mentioned by Rich whilst I was compiling this!), or none of it is a problem, but best eliminate it.
Can you establish a wireless connection to a computer temporarily, or got a long ethernet lead?
What I'd suggest is, grab a copy of your ADSL stats from the modem/router (when you have plenty of  daylight left so you'll later be resyncing in daylight an hour before sunset), then do the disconnect and power down routine I've mentioned.
Then uncoil the modem lead and let it just hang down in a U shape. Turn off/unplug all your home plugs (hence the need for wireless or long ethernet lead), if it's not just an AC transformer PSU, unplug it and uncoil its lead and plug it into an extension lead run from another power socket (not the current one) and support the PSU on a table or pair of steps so that it's as far away from the modem/router as it's lead will allow.
OK, then power everything up and promptly grab a copy of the ADSL stats and post both lots.
If you see a noticeable increase in the sync speed and/or a reduction in the SNRM, then something in the current arrangement is causing a problem. So it's process of elimination.
Leave the modem/router connected and powered up and reconnect and turn on the homeplugs first and when they are all working have another grab of the stats. If you lose sync or the SNRM drops, that's the source of one problem. Whether or not there's a problem there, then try moving the modem lead about, if you can recoil it without disturbing the connectors (so that you don't drop sync) do that and grab some more stats. If there's a risk of dropping sync, then do the disconnect and power down routine and move the PSU back to it's socket first and leave it's power lead and the modem lead in a U.
Power up again and grab another copy of the ADSL stats.
Finally (doing the disconnect routine etc if needed) coil the leads back again and the grab yet another copy of the ADSL stats.
Edit: That's my suggestion of a logical way to proceed John, I'm sure you don't need me to tell about other permutations, what ever you think/need. Grabbing stats before and after each change is the key as I'm sure you realise.

[#2]

Ask PN for the delta report history for your line, it will show the historic speed of your line. I had to ask for this data and then plot it on a chart and give it back to PN before they would see and agree there is an issue.
Also, looks like I am going to go through the same issue as you on the open ticket I have.

[#3]

Quote from: JohnJ

adslchecker currently says:

The annoying thing about the BT ADSL checker is that it clearly just spits out a number based on some recent history of the current connection speed! So of course it just says the expected speed is about the speed you currently have. That is why the delta report history is so useful.

[#4]

Hi there,
It looks like things are moving along the right lines. If there is anything to do we'd be happy to step in but agree that resetting the SNR wouldn't result in a very good speed as things stand. Please let us know though.
Re speed history, it's wavered between 4500kbit/s and just above 2000 since 2010.

[#5]

@ Matt
Thanks. I recognise those figures as the profiles that I have been on over that period (I have screen captures of the BT Speedtester results).
January 2010 to April 2011 the profile was 4000 kbps.
April 2011 to May 2011 the profile was 4500kbps (with a short drop back to 4000kbps).
June 2011 the speed started to degrade.

[#6]

I have carried out the drop session - (wait a minute) - power down - (wait 1 minute) - disconnect ADSL sequence. Then rearranged the wiring, waited 10 minutes, reconnected everything and powered up again.
As I could easily remove the router form the wall, I varied the layout a little. To maximise the separation, I moved the PSU down to the floor and placed the router on a step ladder a short distance away horizontally. This meant that the power cable was now at 45 degrees up from floor to the router and the ADSL cable was at 45 degrees up from the router to the BT socket. I hope that was still a valid layout.
Attached are the screen captures from before at 15:04 and after at 15:24.
I was surprised how similar they were.
To save you looking , the only difference is that the SNR increased a little from 12.4dB to 13.7dB

[#7]

Did you reconnect between those two sets of stats? If you did I'm very surprised that the sync speed is the same.

[#8]

I was also quite surprised by the similarity. So, just in case my browser had cached the previous version, I forced a refresh of the page. However the SNR had gone up so there had been some difference. See also attached Routerstats screen capture showing the disconnect just after at 3pm.
I really need to use the wired network again so,even though sunset was approaching at 4pm, I repeated the disconnect-reconnect sequence again and returned everything back the way it was, as can also be seen on the Routerstats screen capture.
As a check I viewed the ADSL Status again and have attached the result. Again there was not much difference, but the sync speed increased a few kbps.
The BT wire into the master socket is actually flat black and will be at least 30 years old. I took a look along it's length to see if it had any spec embossed into the insulation, but didn't see anything. However I did spot a break in the insulation of the cable. Could that be relevant?
That possibly about 14 January 2012 when an Openreach engineer rerouted the cable. He had replaced the pole in the next door property (about 15 metres away) with a new taller one. As our existing cable would no longer reach the top of the pole, he then ran a new cable from the pole to our house and joined it to the existing cable on our gable wall. The route of the old cable had previously included a 90 degree bend up to the position of the old unused insulator, but the engineer straightened the cable at the bend and ran it horizontally to join with the new cable. The former location of the bend is where the insulation is damaged.
I've attached an image of the point with the damaged insulation and you can see a red and black wire inside. I've also attached a view of the gable of our house. The telephone pole is behind me to the left and the cable runs across the gable and enters the front of the house on the right. The old insulator show the old point where the cable came from the pole.

Another question was when I had moved the wiring into its current position. I'll try to find a clue in the Routerstats log, but I think it was sometime in early  this year or late last year.
After there was no fix via PN late last year, I moved the 2700HGV near to the master socket and used a long ethernet cable to my ethernet hub, so bypassing the hard wired extension. There was a small increase in sync speed (about 400kbps from memory). Nothing like the former speed but enough to justify making the permanent wiring change that you see now. I think I bought the Billion just before making the wiring change, so that I could tweak the SNR.

[#9]

I too was a bit surprised that the sync figures were identical, not just the DS but US as well. It is possible, obviously, and I have seen it before and experience it myself with the DS, but to have a 21CN uncapped upstream resync at the same speed as well, has to be a rare event.
No matter, the fact that the noise margin has increased on the DS means that the proximity of something in the normal layout is causing a reduction in margin which will result in lower sync speed.
On the final set of stats, both the DS & US sync speeds have improved marginally as has the DS Margin, so even some marginal difference in cable (or equipment) position is having an effect.
The BT incoming cable - why oh why did that engineer not replace the whole run of cable into the NTE5a, an unnecessary joint plus old cable, and why did he not continue to use the old insulator, is it not securely fixed to the wall still? I certainly wouldn't have allowed them to go drilling my wall unnecessarily.
Anyway, that length of old cable will be one of the major issues causing you grief as far as broadband performance is concerned, it is not twisted pair and will be continuously be picking up interference. In today's environment of ever increasing RF due to wireless devices, extra mobile masts and phones. TV and Radio transmissions, it's bad news and will be costing you some more significant reduction in sync speed (due to the loss of potential margin). What effect the damaged insulation is having is impossible to assess, any damage to the inner cores will have a significant effect - there may be none visible at the break in the outer, but moisture will track down the inside of the cable, possibly all the way to the NTE5a. That engineer did a sloppy and unprofessional job.

[#10]

My ticket is now updated, pointing out the damaged cable.

[#11]

On Anotherone's advice, I'm doing more testing of the router/PSU/cables arrangement this morning.
I setup a PC beside me with RouterStatsLite set at an interval of 5 seconds and monitored the noise margin. It started at 12.8dB.
I tried wrapping my hand around the BT cable at various points near to the master socket and outside at the broken insulation and the cable joint, but there was no noticeable effect on the noise margin.
Without disconnecting anything, I uncoiled the various cables and then started with the PSU cable..

9:26	PSU Cable Handling the PSU cable showed some variations. The highest margin (13.6dB) was when I scrunched the cable in to a bundle in my hand about 10cm from the router. It didn't seem to matter if I coiled the cable or folded it tidily, as long as I closed my hand around the bundle. At 9:39 I  moved the bundled cable further away to about 40cm from the router and the margin increased slightly to 13.8dB so I blu-tacked it into that position. When I released it from from hand the noise margin dropped to about 12.9
9:49	PSU I wrapped my hands around the PSU and the noise margin immediately dropped to 8.3dB. When I let go again it only increased back to 11.9dB I tried touching the PSU on each side with my hand. Placing hand on the front face, the margin dropped to 10.5dB Placing hand on right, the margin dropped to 9.4dB Placing hand on left, the margin dropped to 10.4dB The margin returned back up to 12.3dB. I also happened to tap it sharply with my finger at that point, but further taps made no difference.
10:06	PSU Cooling I pointed a desk fan at the PSU for a minute but it made no difference. I wrapped a cool pack form the freezer around the PSU and the margin dropped to 10.0dB and gradually increased to 10.5dB. I removed the cool pack again.
10:06	I placed my hand on the right side of the PSU again and the margin dropped to 10.1dB

Margin has returned to 12.6dB.
So, wrapping my hand around the PSU reduces noise slightly but touching the PSU increases noise.
The PSU is as supplied with the Billion 7800N router.
Any suggestions?
Coffee break time...

[#12]

What that has clearly shown is the PSU is switch mode (which we really knew anyway) and it's chucking out a load of hash that is reducing your potential noise margin. I would try and arrange the cable so that it is as far away from the Landline cabling and the modem lead and have all the surplus cable as close to the PSU as possible and squash the coiled up cable into a flat bunch (so that it's just zig-zagging back and forth and tie it like that. The other thing worth a try, do you have any ferrite rings or clamp on ferrites, try one close to the PSU end of the cable. It really is a case of experimenting, as I said before, RF can sometime behave not quite as expected, and even sometimes the complete opposite of what you thought 

[#13]

Moving the PSU further from the landline cabling will require a disconnection, so I was trying as much as I could to try similar things, before I go for the disconnect.
So, without disconnecting anything, I tried moved the flat bunched cable to more positions; next to PSU, between PSU and router, next to router but there was no significant variance in margin (+/- 0.1 dB).
I ferreted around and found a clamp on ferrite from something else and tried that on the PCU cable at the PSU end but there was no difference in margin.
I also tried the ferrite in various positions between the PSU and the router, but there was no significant difference. I also repeated each ferrite position with the bundled PSU cable at different positions. No significant difference
Apart from the PSU, I've now also tried removing the homeplug.
Removed homeplug and ethernet cable - margin down 0.4dB
Fitted homeplug and ethernet cable - margin up 0.6dB
Removed ethernet cable - margin down 0.6dB
Fitting a longer ethernet cable didn't make any difference.
Fitting the ethernet cable into each port on the router made no difference.
Disconnected telephone cable plugged into ADSL filter - no change.
I've currently returned everything to where I think it was before starting and the margin is about 12.3dB.
Earlier on when I gripped the bundled PSU cable in my hand, the difference was about 1dB, but now gripping the PSU cable bundle in my hand only result in a margin increase of about 0.4dB.
The only visible differences could be in the directions that I have coiled/bundled the modem and PSU cables.
I suppose the next stage is disconnecting everything and switching routers and filters.

[#14]

Well that's RF for you    As I said in the PM, a bit of a black art. I should go for that 13.8dB you got earlier 

[#15]

I have now swapped several hardware items so that involved disconnections and reconnections as follows:
1. Moved 7800N onto step ladder and PSU to socket on floor (as on 7/11) but switched ADSL filter for another BT one.
2. Switched 2metre screened modem cable for 5metre screened modem cable.
3. Swapped 7800N router and PSU for 2700HGV router and PSU and switched modem cable back to original 2metre one.
4. Switched back to 7800N router and PSU, and moved everything back to original positions, with variation in position of coiled PSU cable and modem cable.
For configurations 1, 2, and 4 the sync speed always reconnected at 3008 kbps and the noise wavered around 14dB. The one noise drop during configuration 1 was when I put my hand on the PSU to confirm what would happen.
For configuration 3, the connection speed was initially 3200, but the Training History page indicated that a higher speed should be possible, so I reset the connection a few times and the speed increased to 3264kbps then 3424kbps. I then pushed my luck and it reconnected at 3296kbps then 3232kbps.
Extract of Training History page:

	Downstream 																	Upstream
Time 				Line 		Rate 	Max1 	Max2 	Max3 	Mgn1 	Mgn2 	Attn 	Pwr 	CRCs 	FECs 	INP 	DLY 		Rate 	Max 	Mgn1 	Mgn2 	Attn 	Pwr 	CRCs 	FECs 	INP 	DLY 	Mode 		Vendor 		State 		Exit Code 			Echo 	VCXO 	Rx Gain
2012/11/10 15:36:45 GMT 	1 		0 	0 	0 	0 	0.0 	0.0 	0.0 	0.0 	0 	0 	0.00 	0.00 		0 	0 	0.0 	0.0 	0.0 	0.0 	0 	0 	0.00 	0.00 	Unknown 	Unknown 	11/13 		IDLE_REQUEST 			0.0 	0.0 	0.0
2012/11/10 15:47:59 GMT 	1 		3200 	3348 	3032 	3032 	12.7 	12.0 	48.8 	19.0 	17 	21972 	1.17 	8.00 		768 	788 	11.0 	11.0 	28.5 	12.4 	0 	0 	1.19 	8.00 	G.DMT Annex A 	Infineon 	241/104 	IDLE_REQUEST 			-17.0 	-4.5 	27.5
2012/11/10 15:50:27 GMT 	1 		3264 	3396 	3176 	3176 	12.7 	11.0 	48.8 	19.0 	4 	7465 	1.15 	8.00 		832 	848 	11.0 	11.0 	28.5 	12.5 	0 	0 	0.57 	8.00 	G.DMT Annex A 	Infineon 	241/104 	IDLE_REQUEST 			-16.3 	-4.9 	27.5
2012/11/10 15:50:41 GMT 	1 		0 	0 	0 	0 	0.0 	0.0 	48.8 	0.0 	0 	0 	0.00 	0.00 		0 	0 	0.0 	0.0 	28.5 	0.0 	0 	0 	0.00 	0.00 	G.DMT Annex A 	Infineon 	114/56 		ERR_INCONSISTENT_SIGNAL 	-16.9 	-5.2 	27.5
2012/11/10 15:52:58 GMT 	1 		3424 	3508 	3328 	3328 	12.5 	11.0 	47.8 	18.4 	58 	6895 	0.71 	8.00 		800 	816 	11.0 	11.0 	28.5 	12.4 	0 	0 	1.14 	8.00 	G.DMT Annex A 	Infineon 	241/104 	IDLE_REQUEST 			-16.7 	-5.3 	26.3
2012/11/10 15:53:58 GMT 	1 		3296 	3448 	2996 	2996 	12.7 	12.0 	48.8 	19.0 	45 	2508 	0.59 	8.00 		832 	848 	11.0 	11.0 	28.5 	12.4 	0 	0 	0.57 	8.00 	G.DMT Annex A 	Infineon 	241/104 	IDLE_REQUEST 			-19.0 	-5.5 	27.5
2012/11/10 16:00:02 GMT 	1 		3232 	3400 	2984 	2984 	12.8 	11.0 	48.8 	19.0 	11 	19793 	1.16 	8.00 		832 	856 	11.0 	11.0 	28.5 	12.5 	0 	0 	0.57 	8.00 	G.DMT Annex A 	Infineon 	240/104 	N/A 				-16.3 	-5.6 	27.5

I'd have liked to try leaving the 2700HGV router in place, or trying another PSU on the 7800n but I don't have another suitable PSU and I couldn't leave the 2700HGV in situ. The location is over a doorway, the 2700HGV is not easily wall mounted and at floor level is the mains consumer unit.
I've attached images showing the noise and speed graphs and the final hardware layout.