Line noise

Hey there,

My line that used to be very clear and stable though became noisy a few months ago pretty-much overnight. I won't go into all the details here initially so will just ask this: What might cause a connection to be noisy during one syncronised session (with the DSLAM), then quiet(er) the next?

I've an image here that shows 3 different resync events with the DSLAM (Zen LLU in this case);
DS SNR

In the lower 'Last 24 Hours' part it shows the difference in downstream SNR between 3 router resyncs; the first being a nice, quiet connection; the second very noisy; and third relatively quiet.

There is no noise to be heard on the landline phone at all during a 17070 test, or during normal conversation; all seems perfectly fine there.

A telegraph pole drop comes into the household to an NTE5 fitted with a 'Clarity' faceplate, with no extensions attached. Plugging routers directly into the test socket (no filter) makes no difference. Different routers/filters/leads, etc, all show the same result.

Here's some other images, first being router error rates and such over the same period;
Errors

And a final image showing overall line characteristics (US/DS SNR, Attenuation, Sync, Power);

Stats

I can't fathom what's causing this, or what, if anything, I might be able to do about it (is it a 'fault' or should I accept it as normal/acceptable). I'd very like much to see, once again, the line I used to have, as since this noise appeared the line has required interleaving to maintain sync and the upstream sync brought down from 800kbps+ to 448kbps. I get the impression i'll be unlikely I'll get much more from the line now if I decide to upgrade the connection.

Any thoughts appreciated.

In reply to:

---

What might cause a connection to be noisy during one syncronised session (with the DSLAM), then quiet(er) the next?

---

Some interference arrives to depress the SNR in the chosen frequency bands, resyncing with the noise present picks a better set of frequencies or allocates the bits accordingly ?

Most noise is caused by internal equipment, forgotten phones, sky ect. However occasionally other things have influence. Length of line, quality of line. What you need to know is what DB level the line is at. There are tests that can be done that measure this. Contact your ISP and ask them what DB level the line is at. Maybe your pushing a 65DB level and your getting some feedback? Some modems can only handle really upto 60DB. You cant rule out there being a line fault also. The difficulty with getting the line tested is the tests these days are read differently and because of this it can be difficult to locate the interference. (DVC part of Line test) Basically the voltage coming down the line can pick up interference but since broadband came to the uk, this part of the test totally changed and became unreliable.. Changing every time you tested it, but to a the right eye you could actually work out interference.

Yeah, I see what you're saying there. 1 of the 3 routers I've had running, a Speedtouch, using the 'DMT' software showed occasional large spikes of noise either throughout the spectrum (wideband noise) or in random places. I couldn't determine the source of the noise though suspected it was caused by any noise source along the length of the line, but not local to me, as messing around with AM radio never revealed anything that related to the noise spikes the SNR graphs were showing.

I'm running a Cisco router at the moment and I can produce a 'DMT' (not sure if that's the exact name) graph using this page, which pretty-much seems to show the same typical unused bands in the spectrum as the DMT software for the Speedtouch. The problem is I don't know if these unused bands (and dips in BIN allocation at certain frequencies) were like this before the noise arrived on the 'line' (August 1st), or not, so can't compare them.

I do have quite a few DMT 'movies' I made a while ago though if they'd be useful (for seeing how noise affects the line), though I'm not even sure the noise they show is related either. :S

Edit: Do routers not typically reassign bits to bins dynamically to accomodate for fluctuations in noise and such? I assumed they do, as the Speedtouch with DMT showed quite well (moving bits about). I guess I could be wrong to think this as I do see what you're saying. It still leaves me with the question as to what could actually be causing the noise though.

Edited by deleted (Thu 02-Nov-06 02:25:49)

Hey,

It's not anything internal here. The line is about as simple as it gets; a drop to an NTE5, no extensions, just a Clarity faceplate with a B.T. answerphone plugged into it. Removing both phone/filter, and trying allsorts of other hardware, make no difference, so I'm sure the problem is external.

The statistics for the line are in the images (stats). It's a fairly long line for sure @ 60dB attentuation, though this same line used to be great considering that (very quiet, stable as it gets with no interleaving and 768kbps upstream rate). Since August 1st it went to hell, resyncing the connection every 'few seconds' due to the arrival of the large noise spikes. Zen supports answer to this was interleaving and capping of the upstream sync, which I've not been to happy with really, but hey the line 'is now stable'. :/

Typical line values are:
DS Attenuation 60dB
US Attenuation 31dB
DS SNR 15dB
US SNR 17dB (@ 448,000)

3 different routers show the same noise issue (Draytek 2600, Speedtouch 846, Cisco 837).

I understand it could be difficult, and possibly not worthwhile, to try to get it resolved. B.T. did do a remote test in August but basically said everything was fine to the ISP. They're right, I think, to a degree, as the line does 'work', but I can't understand how a line can go from supercool to noisy as hell overnight and no-one recognise something significant changed. I'm just interested to see what other folks think here as I'd like to find out some reasons for what might have happened more than anything, to see if it's considered pretty normal or likely fault that I could get fixed.

Here's an image of August 1st that shows: before; fault (mega-noise); and after (interleaved version of same noise).

Edited by deleted (Thu 02-Nov-06 02:44:31)