Print

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

[burakkucat]

. . . what do you mean by raised noise floor?

The average, flat, line of a QLN plot across the observed spectrum (ignoring any spikes or peaks originating from RF ingress).

A "good" noise floor would be -140 dBm/Hz. Estimating it by eye, (as I don't have the raw data to manipulate) and being generous, I'll say the plot we currently can see via MDWS has a noise floor of about -115 dBm/Hz. Hence my mention of a "raised noise floor".

[Edited to add the last sentence.]

[kitz]

Here in this graph you can see the crosstalker in action

Yep - thats the culprit. 
Using MDWS I checked the figures - it spiked at 10.1dB and then went down to 6.3dB, so the cost is 3.8dB of SNR.

I have worked out your crosstalker has taken 13912 Kbps or 13.9Mbps away from your line sync

Based on 3dB = 11Mbps of sync (wombats rule of thumb)

(11 Mbps / 3) * 3.8 = 13.93Mbps.   Voila we have a match!

I've no doubt that it's not what mlmclaren wanted to hear, but proof that his missing 13.9 Mbps is down to that one disturber 

[burakkucat]

burakkucat: could you attach Hlog text data? I'd like to plot Hlog graphs for ADSL2+ band (first 511 tones).

Sorry K5, I don't have access to the raw data. The graph was served up by MDWS and I just took a screen-shot of it.

(For ADSL2+ mode there are 512 sub-carriers (tones), numbered 0 through to 511.)

[S.Stephenson]

Data rate:   16994 / 65735
Maximum data rate:   20003 / 65650
Noise margin:   6.5 / 6.5
Line attenuation:   13.3 / 14.2
Signal attenuation:   13.2 / 14.2

I have a single crosstalker that takes me from ~attainable 81mbps to ~60mbps, see below (data from before G.INP on a HG612)

[kitz]

...  and sadly yet another example of why I say "Its seldom anything over 14dB manages to keep hold of the full 80Mbps once the cab starts to fill. " 

The theory is that 14dB should easily be able to manage 80Mbps...  but unfortunately once crosstalk kicks in then the speeds start to deteriorate.    Its partially one of the reasons why I haven't bothered to do a speed estimator for VDSL.   There are far too many variables.   

Not only is there crosstalk which can vary so much from line to line..  but even lines with the same attenuation may have different PSD masks which will also affect the maximum rate.  A line which is 300m from a cab which is near to the exchange is going to get different PCB and PSD masks than a line which is 300m from a cab which is 1 mile from the exchange. 

[S.Stephenson]

Well considering my area pre FTTC was 2-3mbps I'm not surprised it filled up quickly, sadly the people most effected by crosstalk are the ones who probably had bad speeds before FTTC.

If only they would roll out vectoring 

[mlmclaren]

Hi everyone, sorry for lack of responses in this thread was out on the lash last night...

Thanks for everyone's input on my issue... as some of you might have noticed my connection jumped back to normal for a minute or two at around 17:50 last night, but ropped again.

So whats the situation at present? It looks like there was some concern about the power levels on my line or potential an issue with one of my legs... it would be helpful to have something to tell BT when it (probably inevitably) goes to complaints level.

I'm also understrand that this could well be a crosstalk issue, maybe a new line has been put in next to mine in the DP, I however haven't yet plugged a phone in to check for issues, my concern at the moment remains to be the stabilty of the line and with it spikeing up and down intermittently I would say I have little faith in its reliability.

Will have a look at a few connectons on MDWS now and compare some figures and see what's different 'if anything'

[Chrysalis]

a split pair will make crosstalk way worse than it normally would be, certainly not impossible but getting an engineer to consider it as a reality I imagine will be very difficult. these wont show up on the very basic SIN test's they do with the JDSU's which only are to see if the line complies with voice line spec.

[mlmclaren]

a split pair will make crosstalk way worse than it normally would be, certainly not impossible but getting an engineer to consider it as a reality I imagine will be very difficult. these wont show up on the very basic SIN test's they do with the JDSU's which only are to see if the line complies with voice line spec.

Apologies if this seems dumb but by split pair, do you mean 'sheath cracked' exposing copper core?

If so I reckon this could be possible... I knwo somebody had stripped a load of the spare pairs from the cat5's for some reason, it may be a possibilty that when the covers been replaced one of these could be touching my pair/port.

Might go have a look at the state of the cupboard

[Dray]

Why is CAT5 involved? That's for ethernet, not phone connections.

[kitz]

s some of you might have noticed my connection jumped back to normal for a minute or two at around 17:50 last night, but ropped again.

We did,  NS pointed it out and that's what triggered the conversation last night.

I'm also understrand that this could well be a crosstalk issue,

The calculations performed last night were based on the the differentials during that brief period and is proof beyond all reasonable doubt that your speed loss is down to crosstalk from one disturber. 

[kitz]

Apologies if this seems dumb but by split pair, do you mean 'sheath cracked' exposing copper core?

Not a dumb question at all.   Its better to ask if you dont know

A split pair is a condition which results when wires from two different twisted pairs are tied together by mistake. In telecommunications, most wiring is done using the twisted-pair configuration, in which two conductors are wrapped together for the purpose of reducing line interference. It's quite common for several twisted pairs to be grouped together in one large cable which is heavily insulated, and can carry a variety of different signals.

Ref: http://blog.t3innovation.com/what-is-a-split-pair

[mlmclaren]

Why is CAT5 involved? That's for ethernet, not phone connections.

All the flats are wired with there own length of cat5e for the phonelines... anoying when you consider there could be some sort of FTTB service in place rather than this super sensitive bull.

[mlmclaren]

Thanks Kitz, will have to see what comes ack from BT... I've been in the DP and its pretty tidy, though you can see where some less trained/untidy engineers put a connection in as they've left the pair really lose and kept it tight like the first Openreach ones.... I did check that nothing was touching anything and that all pairs where correctly punched but all seemed alright... no change to SNR either so don't think anything was lose

[Dray]

Eek, sounds a bit amateurish