SNR/Connection speed wobble.

[#1]

@Gandalf Understood - but at a connection speed it has no chance of maintaining!

Not shouting at you, but there is something wrong with the system if it forces an unrealistic setting, surely?

[#2]

Ah fair. DLM works in odd ways...

I'd be happy to increase the target SNR to a higher level like 9dB or 12dB and ask BT to see if we can keep it at that level?

[edit]
We could also modify the DLM threshold on the line so it favours stability over speed

[#3]

That would be great - 12dB should - hopefully - give me back the ~9900 speed?

Edit I'd still like the speed, if possible.

[#4]

I've nudged the SNR target up to 12dB, your router will be going for a resync soon.

I've also emailed BT and I'm in the process of modifying the threshold.

[#5]

Cheers, Anoush - that was quick!

I'll leave the laptop running overnight to see what happens.

[#6]

To your edit I'm afraid the higher the SNR generally the lower the speed. There's not much else we can do to adjust the speed bar moving the SNR target up and down. I've placed a modify order to set the DLM threshold at "Stable"

This should go through on Monday. We can always reverse the change if needed.

 that was quick! 

Yup we're quite quiet tonight!

[#7]

That's OK - the speed is around what I had for the 100+ days and is perfectly OK for me.

Thanks again.

[#8]

Cool. Fingers crossed the changes improve things for you

[#9]

Hi John,

Looking at those plots I very much suspect you have something causing transient REIN around 9dB of it. It has a switch on spike and fairly quickly recovers and then gradually fades. Setting the target SNRM to 12dB will simply mask the problem.

I suggest that the following is happening. The line encounters REIN and drops the connection. The line resets at a speed which delivers a 3dB TARGET MARGIN. A while later when the noise has gone the OBSERVED MARGIN becomes 12dB.

If in the absence of noise the line resyncs again (manual action, power interruption) it will deliver a higher speed at a 3dB margin ... which when noise comes along breaks the link.

You say that nothing has changed in the home - what you cannot assert is that nothing has degraded. Switched Current power supplies are notorious for blasting out RFI whilst still appearing to work without issue.

Keep a watch on the plots (use the alarm setting for low SNRM and see if you can find a time correlation with something being switched on OR put under load.

I’ve seen similar issues myself - nothing being apparently switched on ... finally tracked the issue down to what it could not possibly be after trying everything else - power supply to the CCTV cameras ... as dusk fell and the IR Illumination woke up, the power drain pushed the PSU over the edge and RFI was generated. As side from what I saw on router stats, nothing seemed untoward.

Happy hunting?

[#10]

Thanks for your observations/comments, Kevin.

I understand your comment re: setting the target margin, but at the moment I am happy, if that is the word, to just maintain a stable(ish) speed.

As to REIN, again, you are probably correct - I'm no electronics engineer - but as regards the target margin at 3dB, it will never hold, as I found out prior to my last experience - every time it came back at that, my speed was stratascopic, but only for a very short while, and all three of the OR guys that attempted to solve the problem said 'too low'.

I will try and borrow a portable MW radio at some point and see if anything is emitting identifiable noise, although the last SFI did run a check for me and couldn't find anything of significance.

As to anything switching on/off causing the errors - I can't see any correlation - the only thing in my property that does that is the fridge/freezer, and that certainly is not an obvious culprit - it has about a 40 minute cycle time and does not correspond with any disruptions.

I am tempted to let things stagger along until after my holiday in just over three weeks and then get a highly recommended local telecoms engineer to have a look for me. I would rather pay an independent than BT.

Hope this makes sense, but if you have any other suggestions - I'm all ears.

/EDIT to add - I had at least three drops between 2300 when I went to bed and 0530 when I got up, but unfortunately, RSL decided to throw a strop around 2330, so I have no plot - I only spotted them when I had a look at the router logs for some inexplicable reason.

[#11]

Hi John,

You might have difficulties finding the RFI. By the looks of those plots it is short lived. If you are content with the speed at 12dB I would not bother with trying to fix the target .... unless the minute or so line drop is having a marked impact on your internet use.

The guilty party could be any power supply brick ... including that of the router or laptop. You are noting going to find it unless you see it happen and can go around the house switching stuff off at the plug and then see an instant jump in the SNRM plot.

That’s how I eventually found mine - that was when I was in the pub (lots of potential sources) and finally it was a matter of “What have I missed? Surely it cannot be the CCTV ... it sits there on all day.”. Having found it to be suspect I left it turned off for a while, then after turning it back on the same gradual decay was seen.

Hmmm thinking about that experience - having got rid of that source of REIN disclosed a second which had a similar shape to yours ... that was caused by certain types of passing trains on the nearby West Coast Mainline ... only the super pendelinos caused a massive SNRM spike not dissimilar yours!

If you are near an electric railway line I’ll give you a resource which will give you the time of every passing train.

[#12]

 Morning Kevin

As you say the interference is extremely short lived, so short that I don't notice it even if sat in front of the laptop.

With the location of most of my sockets, trying to switch off at the plug is a bit of an obstacle course, so not a good plan.

I do have a railway line fairly close by, but not electrified, so I doubt that is a source - especially given that over the 100+ days since the engineers last visit where I had a fairly steady SNR and speed plot - in fact I would call it rock-steady.

I'll leave it alone for now, but leave this thread in my subscriptions, and just monitor. Still thinking of my local telecoms engineer though - he is highly thought of.

[#13]

Hey John

Just received a response from BT advising they've hard set the SNR target to 12dB. Coupled with the modify order I've placed, hopefully it stays at that level. How have things been over/since the weekend?

[#14]

Thanks Anoush. I'll try and keep this brief.

From your intervention to me going to bed Friday (28th) at about 2245 all OK. Up at ~0530 Saturday to discover my logging software had had a hiccup and stopped operating, so checked router logs to discover I had had three 'xDSL down' events during that period, fortunately without any detrimental effects.

Since then, no major disasters, and reported d/l SNR seems steady around 13dB, with a 9458Kbps reported connection speed, actual hourly reported speed tests around 7860.

I'll keep my eyes peeled for the next 3 weeks or so before I go away for fortnight and shout up if anything suspicious grabs my attention.

Thanks for your help, as always.

[#15]

Cheers for the update John, that's good to see.