[Long but hopefully interesting...]
I hope someone with more experience than me can help with some ideas / what to do next.
I have a rural 4+ km copper connection back to the exchange (no fibre possible yet) and have had periodic problems over the past 10 years, periodically sorted out by BT finding a fault on the line. Two relevant examples were:
(i) about 6 years ago I experienced the line dropping every exactly 30 mins and 4+/-3 secs - but only when it was wet/raining. BT couldn't find any problem with the line per their normal metrics, but on further investigation found a logged fault on the pair physically next to ours (which had been taken out of service) and hypothesised that maybe the fault had propagated to our pair. They moved our pair to a completely different bundle and problem went away. The fault location was along a road through the two halves of a local quarry and my suspicion is something like: that the quarry had a sump pump which cleared collected rain water and were on a 30-minute timer, and when their motor started the resultant RF bust broke through onto one partially-broken wire of the pair, causing the drop. Maybe!
(ii) a few years later we had another series of frequently-dropping lines during wet weather which BT this time fairly easily tracked down to a water ingress issue, which they fixed satisfactorily. Chance meant that the weather was wet enough to find the problem during their visit.
I now have another dropping line issue. Line drops fairly "reliably" when the weather is wet but also, at least once a night it either drops or the SNRM falls away (more detail below), but this time BT have come out and couldn't find anything according to their normal metrics/tests. Weather was good when they came out
The interesting thing to me is the shape of the SNRM trace. I run Routerstats (ISP is Plusnet, standard Technicolor TG582n router, we are on ADSL2+) to monitor.
The behaviour is:
- when the router has re-synced the line at an SNRM at or close to the target 6 dB, then the interference is generally sufficient to drop the line. However,
- if the line was slightly more noisy at the instant of last sync such that the line has been running at a higher normal SNRM of around 8 dB, then the interference is usually
insufficient to drop the line - BUT - the shape of the SNRM trace / noise burst is essentially identical each time
See pics attached:
(i) Full_SNRM_interference_pattern.png for the full trace over the whole duration of the interference - initial burst then about 2-1/2 hrs of noise before it disappears and
(ii) Shape_of_SNRM_interference.png file which demonstrates the repetition in the initial noise burst shape from day to day.
This happens even when the weather is reasonably dry and looks to me like some sort of RF burst - not quite repetitive in the sense if REIN, but certainly daily. The noise burst generally happens at about the same time every day - between 7 and 9 in the evening, but has been as early as noon and as late as 11 pm+
When the weather is wet then you get a general worsening of the noise - see
(iii) 3_SNRM_during_overnight_storm.png
I have observed joint wet-weather and ?RF burst? behaviour.
I am not sure if the noise burst is a totally different problem from the wet weather - certainly such as characteristic noise burst shape is
not created by wet weather alone, but if for example, there was an imbalanced pair mini-break, then perhaps wet weather could make it worse.
One supposition (purely) is that the location of the wet weather problem in "old" example (ii) above was close to a very small water treatment works which has a cabinet that sports an aerial - perhaps it's sending a nightly data-dump back to the control room?
Another is that the water treatment works has aerators with motors (and probably pumps too) but I have been out next to it and when an aareator starts the line didn't see the characteristic noise burst. But that doesn't mean it's not a motor or some such at the water works.
One further point - our lines are, I believe, 100% underground, no overhead sections.
Problem is, this is now getting irritating in terms of frequency of the line dropping, but Plusnet/BT have essentially said I just have to wait until the problem gets worse so they can find the problem.
Maybe so, but it won't necessarily find the noise burst problem as that definitely looks to me like RFI of some description
Any ideas how I can progress this, and/or what the problem might be?
Thanks in advance