Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Greybeard33]

I have read that Christmas lights can be a potent source of REIN, and suspect that I may have been a victim this season.

My 1km-ish line (Huawei modem & DSLAM) has generally been reasonably stable, with one period of 50 days without a resync last autumn (see attached modem logfile). However, in the run-up to Christmas it suffered frequent short bursts of errors and associated resyncs. Eventually DLM reacted by increasing interleaving from 700-ish to over 2000, with INP going from 3 to 8.5 and delay from 8 to 16. This successfully stabilised the line, but at the price of a drop in DS sync speed from 40000 to 32750.

Over this period a house across the road was sporting a garish array of flashing lights all over its front wall. These were ripped down on Boxing Day, and after that the error rates and dips in attainable rate seemed to reduce. 8 days later there was another on-the-fly resync and DLM restored the line parameters to the previous values.

This does seem to be a coincidence if not cause and effect, but I am puzzled why the lights, which were flashing 24/7, should have caused only occasional bursts of errors rather than a continual high level of noise. In any event DLM, left to get on with it, seemed pretty effective in "making the best of a bad job".

On-going stats of the period in question attached, unfortunately with some gaps. Also current stats both before and after the most recent resync.

[burakkucat]

I will make the suggestion that perhaps the lights power supply was not responsible for the error bursts but was just adding to the general RF 'mush' that was present. Then when other events 'happened', they were regarded as significant due to the degraded background environment.

Now where is that analytical Eagle? 

[Ezzer]

The worst problem with locating REIN (second to the REIN disapearing on arrival) was the noise would often be transmitted back through the mains cabling and anything connected to it.

In this case it could have been; that the lighting amplified pre existing REIN to the extent where it's picked up by your DSL. If not being actualy being caused by the lighting itself.

Otherwise with exterior christmas lights, the "outdoor use" quality may be questionable. The family lights we have here have a socket at the other end and the plugs have a socket on the back end. So they can be daisy chained. There is no weather protection for these terminals yet they are deemed ok for outdoor use, so water ingress, a bit of corrosion 

[Greybeard33]

Thanks for the theories. The line has been reasonably stable since my previous post, with just two on-the-fly resyncs that slightly increased the US sync rate without change to the DS sync. It is still prone to transient error bursts (see attached stats) but not now bad enough to cause loss of sync. This does seem to confirm that the Christmas lights were a factor.

The line has always had a curious characteristic that the DS max attainable rate gradually declines from 47000-48000 immediately after a resync, to settle down at about 41000-42000 a couple of weeks later. This causes a loss of about 3dB in DS SNRM. DLM usually triggers a resync in the small hours of the morning, when the line is fairly quiet. This is good in that it maximises the sync speed, but it does mean that the DS SNRM can eventually start dipping below 3dB in the evenings.

I believe that the gradual loss of SNRM is due to bitswapping in the D1 band - bitloading is increased on some tones and reduced on others, presumably due to the modem "learning" which are more noise-prone. This can be seen by comparing the attached HiRes bitloading plots, one taken a day after a resync and the other a week later. However, similar bitswapping occurs in the U2 band, without any loss of US SNRM, which remains around 6dB.