Print

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

[S.Stephenson]

From what I've seen I used to have 2 crosstalkers that knocked ~3db each from my line, one thankfully has disappeared and as a result my lines attainable has jumped from 65mbit to 79mbps.

While I understand that crosstalk is variable it just seems insane to me that differences that the 2 lines I have show.

Both of the lines come from the same DP.
Line 1(S.Step): ~320-350m
Attainable - 79116/26078
Attenuation - 14.3
PWR - 5.6

Line 2(S.Step2): ~320-350m
Attainable - 57090/19967
Attenuation - 15.5
PWR - 5.3

Is crosstalk truly this random?

From looking at the QLN Line 1 seems more noisy than Line 2

[Black Sheep]

It partly depends where you are within the cable-bundle, as to how much X-Talk you are subject to. If your 'pair' is slap bang in the middle of other 'DSL enabled Pairs', then you'll get a lot of X-talk. Conversely, If your 'pair' is surrounded by PSTN only circuits, then the X-talk would be severely reduced.

[Chrysalis]

might not be all down to crosstalk as line 2 has worse loop loss.

[Weaver]

Crosstalk changes: Speculation: This could really be what has happened to me. I've lost well over ~1 Mbps downstream TCP throughput ( ~7.7 Mbps down to ~6.5 Mbps) since January, and the only thing I can think of is more DSL users, as the odd house that never had an Internet connection has now got one, and, more importantly, there has been a lot of new house building. My upstream has gone from ~1.0 - 1.3 Mbps upstream TCP throughout down to around 0.8 Mbps too. The new houses are on EO lines that are ~5 mi [!] in the same bundle, and I'm the first house, that is nearest to the exchange.

It's a lottery being in the good place in the bundle. My three lines don't all perform the same, even though they are in the same one and only bundle.

[Black Sheep]

might not be all down to crosstalk as line 2 has worse loop loss.

Fair point, I missed that. Although only 1.2dB difference, it will still have its part to play.

[S.Stephenson]

I was under the impression that 1.2db was negligible, what could cause the difference just crapper copper?

I could get an engineer out if I resynced the line as the speed is below the A range, but there's not much point in my case it's just stupid how variable it is. I wonder how many people report problems to ISPs after visiting a close neighbors and seeing that they have superior speeds.

I would have thought there would be little variability between lines on the same DP as surely they are all in the same bundle?

Is crosstalk much worse on DP with many connections, I've seen a few with a Black Box and multiple white boxes with an insane number of lines on it.

[Weaver]

My three ultra-long lines which are in the same bundle have significant differences in sync rates, especially in the upstream.

[ejs]

For the people with multiple lines, isn't it fairly likely that the biggest source of crosstalk on one of their lines will be their other line(s)?

[Weaver]

Could well be. Don't know about the geometry in the bundle of course. But crosstalk is going to be symmetrical, in a smoothed-time fashion, because the lines are never operated independently. My lines are not equal in performance characteristics, in a way that has long-term stability, it's not subject to fluctuation, the fastest is the fastest long-term.

Edit: I meant that crosstalk is going to be symmetrical for me.

[ktz392837]

From what I've seen I used to have 2 crosstalkers that knocked ~3db each from my line, one thankfully has disappeared and as a result my lines attainable has jumped from 65mbit to 79mbps.

While I understand that crosstalk is variable it just seems insane to me that differences that the 2 lines I have show.

Both of the lines come from the same DP.
Line 1(S.Step): ~320-350m
Attainable - 79116/26078
Attenuation - 14.3
PWR - 5.6

Line 2(S.Step2): ~320-350m
Attainable - 57090/19967
Attenuation - 15.5
PWR - 5.3

Is crosstalk truly this random?

From looking at the QLN Line 1 seems more noisy than Line 2

I have a disturber that knocks off 6db, looks like they have just come back from 2 weeks holiday

[Weaver]

Correction. In my previous post, I meant that crosstalk is going to be symmetrical for me. Because I split all my traffic (even every individual TCP connection is split too) across all lines, so if line A affects line B, then B affects A as both will have traffic going through them at (approximately) the same instant when the lines are busy.

[ejs]

The amount of DSL signal does not vary with the amount of TCP/IP traffic that is or isn't being transferred over the link. For ADSL, there'll be a constant stream of ATM cells, with empty "idle" cells if there's nothing to put in them. I think the data also goes through some sort of "scrambler" function to ensure there's never any long sequence of zeroes in the sequence of bits. So the amount of crosstalk will be fairly constant while the other modem is switched on and connected. If the low power mode were used while the modem is idle, that would reduce the level of crosstalk because it saves energy by reducing the transmitted power level, I think BT might have done some trials, referring to it as "cool broadband", but I don't think it's in use, presumably any savings on their electricity bill weren't worth the hassle.

For a past few weeks I've had something which adds an almost constant extra 2 dB of noise, it could be crosstalk, judging from the occasional brief upward spike in the DS SNRM which would correspond to the other modem rebooting or maybe even retraining.

[Weaver]

@ejs - Agreed. I was wrong to put it in such a way that might well seem to imply that. Very fuzzy today. 

[burakkucat]

I think the data also goes through some sort of "scrambler" function to ensure there's never any long sequence of zeroes in the sequence of bits.

I, too, strongly believe that such a process occurs.

[Black Sheep]

The amount of DSL signal does not vary with the amount of TCP/IP traffic that is or isn't being transferred over the link. For ADSL, there'll be a constant stream of ATM cells, with empty "idle" cells if there's nothing to put in them. I think the data also goes through some sort of "scrambler" function to ensure there's never any long sequence of zeroes in the sequence of bits. So the amount of crosstalk will be fairly constant while the other modem is switched on and connected. If the low power mode were used while the modem is idle, that would reduce the level of crosstalk because it saves energy by reducing the transmitted power level, I think BT might have done some trials, referring to it as "cool broadband", but I don't think it's in use, presumably any savings on their electricity bill weren't worth the hassle.

For a past few weeks I've had something which adds an almost constant extra 2 dB of noise, it could be crosstalk, judging from the occasional brief upward spike in the DS SNRM which would correspond to the other modem rebooting or maybe even retraining.

You have a great memory ejs ................ 'Cool broadband' never took off. If my own failing memory serves, we had nothing but trouble with constant loss of PPP Session if it was implemented on a circuit.
I'm almost certain that it was only rolled-out on business lines, which added to the chaos ??

The idea was admirable though on two levels, lower ozone-affecting emissions and of course, a cheaper electric bill.