Where do I go from here?

[#1]

Yes, the '3.00am blip' or thereabouts, occurred again last evening [below].
Thanks for the reassuring words - happy to wait and see how speeds might continue to improve; @ present we are holding 3megs d'load which compared to recent weeks, is 'heaven'.
Appreciate the b'ground to the BT corporate developments.  Bumped into a casual acquaintance who lives down the road y'day - BT b'band man [has unofficially been a good help in the past]  & he was not aware of the PN link up at all!

[#2]

Sorry if this is obvious to others, but I just discovered this after some experiments with Routerstats...
When the "noise margin" graph drops to zero with a red coloured line, it means it failed to read the data from the router, rather than being a drop in the noise margin.
Well, the default setting in routerstats seems to be a Fuchsia pink colour, rather than red (see "graphs"->"common settings"), but it looks similar.
I guess it doesn't show up so easily on the sync-speed graph, as by default colour for this is red anyway.

[#3]

WWWombat - Interesting, looking more closely, the red line on the 'capture' I showed only goes half the length of the blue trace.  Your comment refers to RouterStats I believe & not the Lite version I am using?
On RouterStats Lite, the pink/red line is to quote, 'average value of the last 70 points'. Would that be why it is a half line - there isn't an option to change the colours etc in Lite as far as I can see.  I do recall when I was using RouterStats though, there were many different options etc, including colour of trace etc.

[#4]

Others may be interested & encouraged to persevere, albeit with some patience.  My original moan was 'where do I go from here'?.
The folks in PN [DSL Logged Faults etc] are actively working with BTw on the external issues affecting my original appalling d'load speed ticket.  From 'utter frustration', I am now at 'reasonably optimistic', feelings level. 
The line seems stable, SNR is holding, albeit still high and at least I am able to d'load @ 3mbps plus.  IP profile etc. creeping up as below.  Could not have imagined this same time last week.
So, and I hope I do not regret this, praise where it is due; if Mand or others see this, thank you.

[#5]

Are you still seeing any random bursts of noise (especially the '3am' mystery one)?
I have in mind suggesting you request a training period reset (justified by a fault being fixed) but that shouldn't be done unless the line shows no significant noise spikes.

[#6]

Interesting thought.  Not fully understanding what the plus points of a 'training period' might be, please. 
Here below is the situation Sunday evening from 1.000am to 4.00am. 
Still shows stability with SNR @ present [dakness] of 12.  Just waiting now for the PN current line speed of 3,500 to catch up with the BT IP profile of 4,500!

[#7]

Just given that a nudge for you to a 4500 IP Profile on our side.
Fletch

[#8]

Now, that's what I call a proactive service.  Thank you Fletch.
Will now wait to see how long it takes to improve d'loads.

[#9]

That PN profile change should improve download speeds straight away, but you need to disconnect from Plusnet then reconnect again to pick up the new service profile.
A training period reset would immediately set your target noise margin back to 6dB, then the DLM would start it's 10-day training period again, just as if you'd just got broadband. The downside is that bursts of noise could cause disconnections with that lower noise margin as starting point. Also with the higher sync speed it is likely that error rates will increase and the DLM might react adversely to those, especially combined with 'disconnections'. The end result could put you back to where you started with fault thresholds set to poor values making it more difficult to establish faults.
That 5 hour overnight noise margin trace looks excellent. In fact with noise margin 12-13dB I wondered if the DLM has already reduced your target noise margin from 15 to 12dB. Does the noise margin rise above 15dB during the day?
At the moment taking no action seems the best course to me - leave the router connected if possible. It will re-sync automatically if the DLM reduces target noise margin.

[#10]

Spraxyt, sound advice I'm sure.  The scenario you outline has all the makings of a BB nightmare again and agree best to leave.
The SRN seems to stay pretty constant through the day @ 15/16, dropping off as dusk approaches and just now hovering on 11.5.  Be interesting to see how the PN/BTw guys fine tune things, aiming for a more permanent, stable service.
Too early to get the carpet out & close the ticket - hopefully, soon.

[#11]

Well, whatever has been worked on by PN staff & BTw, the situation is now starting to look really good.
Never quite thought that I would be back to approaching 4megs d'loads but looks that way. [below]
I tend to use 'Thinkbroadband' & 'BBC iPlayer diagnostics' [www.bbc.co.uk/iplayer/diagnostics] to check what's ha[ppening.  They seem to me anyway, to give a more realistic view on things without some of the wide swings of other tools.
So, again thank you PN for getting on top of this [OK, yes I still have my fingers crossed]

[#12]

Glad to hear it's helped, and thanks for letting us know. If it does start to slow down again please let us know and we'll see what we can do.

[#13]

Quote from: exporters

WWWombat - Interesting, looking more closely, the red line on the 'capture' I showed only goes half the length of the blue trace.  Your comment refers to RouterStats I believe & not the Lite version I am using?

My comment did refer to the full RouterStats, but I would suspect that the author would continue to use that as a way to demark the "failed" data capture.
My graphs also show the line to only be partially red (around two-thirds of its length), but I think that is more because, while the line downwards is drawn in red, the line back upwards then partially overwrites it in blue again. Your graphs earlier on the thread seem to show exactly the same features.

Quote

On RouterStats Lite, the pink/red line is to quote, 'average value of the last 70 points'. Would that be why it is a half line - there isn't an option to change the colours etc in Lite as far as I can see.  I do recall when I was using RouterStats though, there were many different options etc, including colour of trace etc.

I would have thought that a line that is an average value of the last 70 points, would be a smoothed line, that doesn't suffer from any large spikes. So no, I don't think the two are related.

[#14]

This is latest update.  PN - CSC comment earlier today Wed 24th:-
"From our logs it would appear that your fault is now resolved. I have rechecked your connection today to ensure all looks ok and I can find no issues."
Fine - certainly stability & increased download speeds have returned.  I have deliberately not carried out a router restart for a couple of days to see if the IP profile lifts etc.
I have some concerns whether the fault is resolved.  There is a noticeable improvement in SNR between15 to 12 [previous around the 26], BT IP profile still 4500.
I just do not know what the spikes are in the RouterStats images mean [todays below].  Also included shot of Thomson stats.  Does anyone have a comment, please?

[#15]

Those results look good to me. I think the red spikes on the noise margin graphs are simply where Routerstats was unable to collect samples because the router was busy at the time. I assume the red descender should go all the way to zero with a blue ascender alongside it on the way up again. However pixel resolution causes the bottom half of the blue line to overlay the red one so we can't see the former.
Only 200 CRC and HEC errors in 1½ days is good, my connection synced at 6976kbps with noise margin around 9dB shows 500 of these errors a day. If your target noise margin is reduced your sync speed would increase and I would expect your error rates to increase too, but it's impossible to say by how much.
Taking into account the CSC's comments and the noise margin trends you've posted I think asking for a training period reset (on the basis that a noisy-line fault has been fixed) is an appropriate move. There seems little risk now of the DLM reacting adversely to that. The reset will set your target noise margin to 6dB with a sync speed over 6000kbps. It is possible the DLM might increase the target noise margin to 9dB later but it shouldn't go higher than that.