If the SNR is stable then I'm not sure what noise the ISP is seeing - I thought they would have seen that from a fluctuating noise margin and if that isn't happening then it sounds like noise isn't an issue?

In reply to a post by ian72:

If the SNR is stable then I'm not sure what noise the ISP is seeing

They aren't seeing any issues now but will have put that down to the elevated target margin.

Now that we can see for sure that there is no ring wire connected there is not much that can be done to improve things. "Prowling around with an AM radio" may or may not show some local noise sources but it isn't is going to produce a solution because nobody is likely to re-route the cabling or be able to fix the noise source. In any event as the noise margin is now apparently static I doubt anything will be found.

IMO the reason for the elevated margin is made pretty clear in the first post and in the clarification I requested a bit later...

In reply to a post by Mygri:

I've recently had to replace my long-serving Linksys router as its progressive failure was manifesting itself as increasingly intrusive phone line noise. I'm now getting good service from a modem and separate router (both D-Link, new), but the speeds are somewhat lower than results I was getting a year or two ago.

A dodgy router lead to instability. The margin was elevated to compensate. The router was replaced and the OP confirms that the audible line noise has gone. The obvious fix is for the ISP to reset the margin to its default 6dB. All that is needed is to find a support tech worthy of the name who can understand that simple senario.

Image of 24 hrs of line monitoring here - https://picasaweb.google.com/lh/photo/azJUZCmtEEE0bp... The results were obtained with RouterStats Lite from Netdear DGND4000 router.

Ignore the dropouts - these are due to interaction between the router and the NIC. It's long been a problem and doesn't actually affect the router's ability to maintain continuity with the line. It was originally solved with help from these forums (Kudos Vimtogirl) by putting an unmanaged switch between the two. I'm now using the same combination for these tests, but without the switch. Despite this unexpected behaviour, the line performance at that location was very good, so I've no reason the believe that this is affecting the current problem.

That said, I've ordered another switch so that I can use it in this situation, just to make absolutely sure there is no adverse effect.

As to the SNR, does that suggest anything untoward to you?

The SNR taking a dive suggests there is a major noise source close by.

It could be external to the house or even in your flat - and if that test was done at the master with no extensions connected then external is the most likely. You need to find the source - it could be something in another flat for example and with the randomness - who knows.

Daylight hours are definitely visible there but it isn't a large fluctuation. The question is whether it would become more pronounced at a lower margin but on the face of it the ISP should be able to lower the margin to 6dB.

What did I miss - the fluctuation of SNR seems to be less than 2dB over a 24 hour period?

I don't think you missed anything. OTOH MHC missed putting his glasses on.

The OP has already made clear that the noise margin is now pretty stable - that graph just proves the point however I'd be concerned about the large number of dropouts as it will still appear to be unstable to the ISP or any automated line management the may use.

In reply to a post by kwikbreaks:

They aren't seeing any issues now but will have put that down to the elevated target margin.

I've just posted a link to a line stats graph in response to ian72's post. I don't know whether you'd consider the SNR stable or not.

Now that we can see for sure that there is no ring wire connected there is not much that can be done to improve things. "Prowling around with an AM radio" may or may not show some local noise sources but it isn't is going to produce a solution because nobody is likely to re-route the cabling or be able to fix the noise source. In any event as the noise margin is now apparently static I doubt anything will be found.

IMO the reason for the elevated margin is made pretty clear in the first post and in the clarification I requested a bit later...

In reply to a post by Mygri:

I've recently had to replace my long-serving Linksys router as its progressive failure was manifesting itself as increasingly intrusive phone line noise. I'm now getting good service from a modem and separate router (both D-Link, new), but the speeds are somewhat lower than results I was getting a year or two ago.

A dodgy router lead to instability. The margin was elevated to compensate. The router was replaced and the OP confirms that the audible line noise has gone. The obvious fix is for the ISP to reset the margin to its default 6dB. All that is needed is to find a support tech worthy of the name who can understand that simple senario.

Indeed, rerouting or rewiring is very unlikely. However, the ISP person that I spoke to was knowledgeable, informed and helpful; even so, she was quite clear that there was evidence of 'interference', although perhaps I should have pressed her a bit harder about its nature. That is with the new router in place, which is in the same location as the old one.

In the light of the recent feedback from this forum, which has 'eliminated some possibilities, I'm coming to the conclusion that the location of the router close to the mains distribution board and the central heating controller could be a significant factor. To that end I'm investigating relocating it, but that will not be straightforward.

This is all good stuff - while it's not straightforward, I'm feeling cautiously optimistic that an improvement is possible. Thank you!

So, what are the vertical blue lines where the SNR drops to zero?

In reply to a post by MHC:

The SNR taking a dive suggests there is a major noise source close by.

As per my response to kwikbreaks, I'm now trying to sort out relocating the router to eliminate that 'internal' possibility. As I said in my post, ignore the dropouts, they do not propagate to the WAN side of the router, although I am taking measures to curb them anyway.

It could be external to the house or even in your flat - and if that test was done at the master with no extensions connected then external is the most likely.

Indeed, my big fear is that it is external, because that then becomes a whole different ballgame

You need to find the source - it could be something in another flat for example and with the randomness - who knows.

EMI and RFI are notoriously intractable - Ho hum...