Yes, that's about right, the gap at 32 is the gap between upstream and downstream, and yes tone 511 is the top 2.2 MHz frequency.
The first big dip, at about tone 46, that goes down to 6 bits, will be due to BBC Radio 4 on Long Wave 198 kHz.
It does look a bit of a mess after about 110, but some of the gaps down to zero might be deliberate gaps set by the DSLAM, and another normal dip is the pilot tone, which usually has 2 bits allocated to it on ADSL2/2+, and may be in a different place each time you connect.

OK, so yesterday at 3:41pm was stood outside the cable that seemed to be making noise. It was RF quiet. 3:41pm came and went, and it stayed quiet. So don't have a smoking gun yet.
The bits per tone files, I have a file to analyse. Quick eyeball, and over 10 hours yesterday, the bips per tone only varied by a few bits; so expect to look like graph plotted the other day. I'll check.
My script for doing bits pet tone, doesn't grab the Thomson adsl stats - so I can't see if the the router saw noise in the afternoon. I'll add that, if I can work out command to get useful info from the Thomson router.
ANyway, from the TP-Link router, here is 1 weeks with of down steam noise margin.

And here is the bits pet tone from yesterday. 2pm is when the line has been clear in last week. 10pm is when its been noisy.
Seems almost no change in the bits per tone.
So yes looks like I'll have to add the adsl wider parameteres to the log file, as I can see last night, if there was noise on the line.

Hi there, just came across this. Obviously a REIN problem and cedlor has passed on a few of the useful tips. As you've indicated, there is a distinct time pattern to this which appears to be consistent.
You've suggested the possibility of street lighting which would be the obvious choice I would have thought of - but does your local street lighting fit this time pattern? - If not, it may be some other form of lighting or associated equipment that operates on a similar time slot. It will not be storage heating, that does not come on and go off at those times. Try and spot what comes on and goes off at exactly those times.
As far as tracking with a radio, if around 1.6Mhz is the strongest source of the interference noise - be certain that it comes on and goes off exactly with the interference otherwise you have the wrong noise. As ejs mentioned make sure that you aren't listening on a frequency with a radio station and the source may change by a couple of kHz each time it comes on. The primary trick is to identify the pattern of noise you need to listen for. It doesn't matter if the precise frequency you listen on has bit loading, it's merely a means of tracking the interference source.
Use the directional properties of the radio ferrite rod antenna to try and triangulate the potential source. I wouldn't get too concerned with that cable, the interference is probably being radiated from it as well as elsewhere. The problem with this type of interference is it can have multiple harmonics and can be picked up and re-radiated by various cables, as well as transmitted along them. Because of the multiple harmonics it can appear to have a fairly broad spectrum and can cause a general depression in bit loading across multiple tones. This won't be easy to spot on a graph because of the general AM/MW interference that appears from dusk to dawn, but won't really prevent tracking with the radio.
I know all that makes it sound easier than it will be in practise, but I hope it'll be of some help. Getting a list together of potential sources and then eliminating as many as possible is a good way to start.

Yes, at the moment still collecting Thomson modem stats. Just eyebally the results - and looks like noise has gone. Now I won't for sure until I plot the full results - and I'll wait for a few days before doing that. But if so very strange, possible reasons:
. Maybe it my network printer, plugged into the TP-Link router, and in a very similar place. The Thomson router is futher away. Easy to check.
. Maybe the owner of the house with the cable, noticed me leaning against the wall with a radio, last weekend; he came out of the house before 3:41. Anyway if he has recieved complaints in the past, maybe he just switched the noise off.
Anyway a few more days of collecting stats, then I'll plot some graphs, and then home in on the noise. Hassle for me is only real time I can dig into the noise out in the field, is at weekends - just don't get time in weekday evenings ...

Alas simple mistake in logging, so lost a few days of data (well may be able to recover the files - but don't have time).
Anyway last few days of logging with the Thomson - Noise margin has been very constant. 15.5-16dB in day time, decreasing to 14.5-15dB at night. Attenuation has dropped to 39.5dB.
So Thomson isn't seeing the noise, and has less attenuation, than the TP-Link. This doesn't make much sense. I'll try switching back to the TP-Link over the weekend - and see if the noise comes back. That will help identify the source ...

Different routers often produce different figures for attenuation - mine varies form 38 to 42  wouldnt worry about that.
Maybe the noise has gone? 
I know nothing about linux but there is program IPcop?  with an addon  "Adslmonitor for IPCop v.0.2 "  which looks abit like routerstats for linux
have you looked at those.

IPcop as far as I know is a firewall implimentation. Now most firewalling is done in the adsl modem - e.g. you need to explicitly enable outgoing ports. If you have to firewall on the linux machinhe, that stanard firewalling included in linux is *very* powerful - but a bit cryptic. Anyway I had to get on top of it once for work, so can set it up if I needed it - but the router does most of it (and its linux as well OpenWRT - not that you can tell from the box).
Anyway now flipped back to the TP-Link to see if the noise has gone there as well ...

Finally had time this evening to plot the results from the Thomson and TPLink modem.
So 28/1/16 is from the Thomson
All in febuary are from the TP-Link.
So clearly on both the noise has gone. Notice how the noise margin is shown smaller on the TP-Link
So all I can do is keep the TP-Link and NAS watching.
But also should raise a plusnet ticket to reset noise margin, as clear now I can live with less than the 12-14 db it is set to ....
... hmmm how does one raise tickets with plusnet, don't seem to be able to find the link ...

They don't really use question-type tickets any more, except perhaps for a few specific subjects.
Live chat might be the quickest way.

As long as your line appears stable (no drops of late - don't go rebooting, they'll think they are drops unless you tell them it was you and can account for every one) then there's no reason they shouldn't do an SNR Reset for you.

To the first approximation, I very rarely reboot the rooter. It stays up, along with the NAS and printer to be home set up that is always up.
Eventually found out how to post a ticket again - forgot how hidden it was on the plusnet site. Ticket seemed easiest - as then I post the graphs, that give the whole story. Ticket is still wiating a staff reply, so currently have also 15dB of noise margin! In case any PN staff are reading - its ticket  119451935; should just be a simple reset ...

I've changed the SNR target on your line to 6dB. You may need to disconnect/reconnect your router

Thanks harry. Yes SNR now at 6.3db; and speed is 50% higher. Thats great. I check it stays up overnight (should do given recenet daily SNR), then close the ticket. Thanks.