Print

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

[burakkucat]

So the question: what happened at 23:45 approx and then unhappened at 10:35.

 

The revenge of the haggis? 

[Weaver]

Indeed so. That’s about all I have to suggest too. Is it possible that there is some underlying parameter somewhere that changes slowly until some threshold tipping point is reached in a highly nonlinear system, at a point where there is a very very large dy/dx and then a dependent variable’s value jumps sharply, giving us this +/-4.9dB-high upstream SNR jump. Like a transistor being used as a switch perhaps.

Would a semiconducting point-contact dodgy metal-metal semi-make-break contact work for this, a crack or metal-metal joint without sufficient inter-contact pressure, something that expands and contracts with temperature ? But I’m not sure at all that I can believe the timings in that case, as the timings aren’t right for temperature changes vs sunrise/sunset. I can’t think how an rf interference source might get gated on/off on those timings. But perhaps it’s a detector that is changing, not a source: a non-linear RFI receiver that is getting enabled/disabled somehow and is either hearing or not hearing, or relaying vs not relaying detected noise.

[burakkucat]

I've tried to think of anything that might match . . . even inverting the logic to consider that the low SNRM period is good, the high SNRM period is bad.

The fact that we see a very distinct step just keeps "screaming" at me: switched on/off or switched off/on.

[Weaver]

Im sure you’re right; Only switched on/off really makes sense to me; the rest is speculation that is stretched to breaking point. I just cannot think what it is and why the times. But worst of all, why only that line line #4 ; line #2 sometimes has similar trouble but a different pattern and lower height.

That’s why I like the copper with a bad joint/crack in it idea; it allows line #4 to be different from its brothers. But that fantasy has so many problems.

[burakkucat]

We do know that one out of the four lines, that carry your service, was provisioned differently. A pair was diverted from one cable to another . . . either in Harapul or Broadford.

However we do not know if that situation still exists subsequent to the high energy lightening zap which took out all circuits to Heasta and caused significant mischief in the Broadford <--> Harapul cabling. We have heard (in the third party sense) that some new cables had to be installed somewhere in the afore mentioned section of the route. Hence there is the possibility that your four lines now take a different route though the network to the Broadford exchange building.

[Weaver]

I see your point but it’s line 2, the last ordered, that has the different route Broadford exchange to Harapul no?

I don’t understand why it is line 3 that is the odd man out though; I could understand it if it was line 2 that was the odd man out, not line 3; well line 2 is odd but that isn’t the point. why is line 4 ok ? Line 3 was the second line ordered historically. That’s what’s worrying me.

What story can we invent that gives a different noise ingress or reception picture for differing E-sides here anyway? How much in general can E-sides differ?

[burakkucat]

Last question first. It's not so much as to how can E-sides, in general, differ but where in the cable does your pair reside. What is the effect of its neighbouring pairs? As a new section of cable has been "patched in", it certainly will not preserve the relative positioning of the pairs which existed in the old cable.

I can never remember the logical ordering of your pairs, hence why you (privately) shared the four BBEU numbers with me. Can you be absolutely sure that what you originally called line 3 (for example) is still line 3?


[OT]Did anything ever come of CarlT's offer to ask somebody, who knows somebody, about the Internet service provision situation for your remote location?[/OT]

[Weaver]

> I can never remember the logical ordering of your pairs, hence why you (privately) shared the four BBEU numbers with me. Can you be absolutely sure that what you originally called line 3 (for example) is still line 3?

Chronological ordering

• Line 1 first
• Line 3 second
• Line 4 third
• Line 2 2018

I’m not completely sure about the assignments into the individual drop cables.

I don’t understand the point about relative position in the cable. Perhaps you could elucidate?

[burakkucat]

I don’t understand the point about relative position in the cable. Perhaps you could elucidate?

Some circuits can be noisier that others. So perhaps your line was surrounded by telephony-only carrying pairs, in the original cable, but now has xDSL carrying pairs adjacent to it, in the new cable.

[Weaver]

So a neighbour crosstalker, carrying DSL, which was only switched on half the day, and only produces low-frequency interference not higher frequency - is that what we’re talking about?

Some people do turn their kit off at certain times.

For 02-03 feb, the line #3 upstream pattern would have to be :

• 00:00 - 11:30  :- crosstalker ON
• 11:30 - 21:40  :- crosstalker OFF - SNRM 4.9dB lower
• 21:40 - 00:00  :- crosstalker ON

[PhilipD]

Hi

I had a similar thing, for 2 weeks over the summer holidays on ADSL2+ it would suddenly improve, then drop again exactly 2 weeks later, happened 3 years on the trot.  Was it crosstalk or did someone have some noisy electrical equipment that went off when they went away, was it a factory taking a 2 week shutdown over the summer, no way of knowing.

If I was constantly monitoring the stats then no doubt I would be seeing little peaks and troughs constantly. People do turn their stuff off and on and noise comes and go, hence why there is a margin.  You will even see changes in noise due to certain weather conditions and how it affects radio waves.

I think you just have to accept that the way the technology is, xDSL is highly variable and it is just constantly shifting sands, especially long lines such as yours that is already optimised to the maximum.  I would just stop looking constantly, you'll drive yourself mad 

Hopefully at some point you will get FTTP or even could take a satellite service, and the issues of noise and cross-talk will become a distant memory.

Regards

Phil

[Weaver]

I ignore line 2, but line 3 has a 4.9db variation in upstream SNRM, and that is causing a loss of upstream speed of 25%

[PhilipD]

Hi

I ignore line 2, but line 3 has a 4.9db variation in upstream SNRM, and that is causing a loss of upstream speed of 25%

Still kind of within the bounds of design hence a 6db margin. Of course you could look at it another way, you sometimes get an increase of 25%, and the lower speed is the norm for that line.

Also your 25% lower upload speed on one line when taken into account over 3 lines luckily means only a ~8% overall difference in upload speed.  Given it is ADSL even with 3 lines it's never going to be a fast upload speed, so it means something large uploading might take 100 minutes sometimes and 92 minutes another, and you would not notice that in day to day use, 92 minutes is painfully slow the same as 100 minutes is painfully slow 

Any sign of fibre coming your way or options to move to 4G yet?  Satellite might be the best option for you in the coming months as they launch new services.

Regards

Phil

[Weaver]

I don’t know about FTTC/FTTP but unless they decide to put one or two cabs in the village we are just stuffed, as it’s 7300m from me to the exchange, and after the first mile there’s not a single house, tree or wall

[Alex Atkin UK]

Time to stick WiFi repeaters on sheep?