Big jump in attenuation

My downstream attenuation on adsl2+ has always been around 34dB and with a 6dB noise margin I could sync at around 12Mbs. My true distance from the exchange is about 2.2 km and all the figures stack up. Earlier this year I migrated to Andrews and Arnold and my line stats were much the same and as you would expect even being moved to LLU with the talkTalk backhaul.

One day, the downstream attenuation jumped to about 50dB, with no change in the upstream attenuation but I continue to get much the same sync rate. AA support couldn't explain it but I would love to know what has caused it and why I continue to get the same sort of performance.

I am using a Billion 7800DXL router, but I get the same attenuation figures with other routers so it doesn't seem to be a function of how the attenuation is measured.

Answers on a postcard please

You can't get 12Mbps sync at 50dB attenuation, so something somewhere is distorting that reading. Given that it is occurring on multiple routers it is indeed interesting.

What are the power figures, and do you have older ones to compare with?

A bin plot might be the way to understand this (emphasis on might)

Current stats are:
Attenuation (down/up) 50/21.6 dB
Output power (down/up) 19.8/12.1 dBm
Noise Margin* (down/up) 2.3/7.8 dB
Attainable Rate (down/up) 13500/1249 kbps
Rate (down/up) 12367/1021 kbps

* I am forcing my noise margin to 3dB rather than the standard 6dB and the current figure is just the result of normal variations, and everything works fine with 3dB, even error rates are comparable.

I did take a bin plot a while ago but I don't have it now, so I'll try and get another one, if I can remember how I got it!

You implied in your OP you were on 6dB noise margin. What sync does that give? I'm wondering if it's the 34dB that was incorrect. The upstream 21.6dB is more suggestive of 50dB downstream.

Edited by RobertoS (Sat 15-Nov-14 14:28:04)

With a 6dB margin the figures are:
Attainable Rate (down/up) 11076/1234
Rate (down/up) 10586/1021 kbps

The difference again is commensurate with the difference in margin.

I take your pint about the upstream attenuation, I hadn't thought of that, yet the sync figures aren't consistent with 50 dB attanuation, at least according to my reading of the subject.

By "bin plot" do you mean SNRM/bin? I can get these plots using DSLstats now, using Ubuntu on my Mac Mini.

Bit loading per bin to examine attenuation.

Does this help?
Bitloading 1

This one covers the full range
Bitloading 2

Edited by Adrian (Sat 22-Nov-14 23:07:27)

That's the one .

I'll leave it to Andrew who requested it to analyse.