Print

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

[WWWombat]

Does QLN help?
We've all seen a variety of QLN graphs, and the values shown there. I wonder if these actually help us quantify the scale of crosstalk? If we assume that a quiet phone line should largely have a QLN of "-140 dBm/Hz", does that mean lines showing -110 to -120 have lost the 20-30dBm/Hz because of crosstalk? In which case, what proportion comes from the one worst disturber? Can we figure that out?

Even if we can figure some of that out, how does the "dBm/Hz" figure in the QLN translate towards the SNR values we see change when a disturber modem is switched off?

Individual Disturbers
There are quite a few crosstalk "heatmaps" available out there, usually within articles on vectoring. They show a matrix of the impact between any two pairs in a cable; some colour-code against an unlabelled scale, while some label the scale as "dBm/Hz" ... which matches the QLN units. Do these give us a clue about how big a disturber can be?

I've attached two graphs below:
- One from TDC vectoring trials, with labelled units
- One from Ikanos with unlabelled units, but for 100-pair cable like our tie pairs, showing the impact of the internal sub-structure of 4 25-pair bundles.

Interpreting TDC
If I'm interpreting that TDC heatmap correctly, then a "victim line" that is showing mostly blue colours will come up with a QLN graph around the -140 figure, while one with yellow colours will come out with a QLN around -107.

What I'm not sure about is how to figure the cumulative impact in the TDC heatmap. Does the worst disturber largely define the total crosstalk, because it tends to mask all the lesser disturbers? Or is it thoroughly accumulative - where 20 "green" disturbers can accumulate as much impact as one "orange" disturber?

See some lines where a disturber can be as high as -90dBm/Hz is ... disturbing!

Interpreting Ikanos
My understanding of crosstalk is that the biggest impact happens where the signal is strong - right back at the FTTC cabinet. That means our tie pairs are our biggest breeding ground for crosstalk, because you are pretty much guaranteed to have a neighbouring pair that is also carrying VDSL2.

I wonder if there is a correlation between those who suffer the largest crosstalk being on cabinets where the FTTC cab is somewhat removed from the PCP.

[Chrysalis]

nice data my line is in the yellowy orange area i lost 40mbit attainable and not even in the red i guess red lines get mercy from openreach via pair swaps

[Bald_Eagle1]

Does QLN help?
We've all seen a variety of QLN graphs, and the values shown there. I wonder if these actually help us quantify the scale of crosstalk? If we assume that a quiet phone line should largely have a QLN of "-140 dBm/Hz", does that mean lines showing -110 to -120 have lost the 20-30dBm/Hz because of crosstalk? In which case, what proportion comes from the one worst disturber? Can we figure that out?

Even if we can figure some of that out, how does the "dBm/Hz" figure in the QLN translate towards the SNR values we see change when a disturber modem is switched off?

I was one of the first FTTC users connected to my cabinet back in 2011.

I didn't really see much in the way of crosstalk etc. until January 2013, when it gradually worsened in small steps for around 18 months.

See the attached animated gif for my connection.
QLN hasn't really worsened much since September 2014.

Whenever we have a brief power cut, my connection can sync at up to 5 Mbps DS higher rate than before the power cut, but SNR/SNRM drops rapidly over the next few minutes, presumably as other connections sync up.



EDIT:

See the attached graph showing the effect on SNRM of presumably a single disturber being connected.

SNRM was restored to around the target 6dB when the connection resynced (not manually initiated), but sync speed was approx. 5 Mbps lower.

[Bald_Eagle1]

The QLN graphs from before the SNRM reduction & after the connection resynced as mentioned in the above message are attached.

Those graphs perhaps demonstrate the effect a single disturber can have on QLN.

[S.Stephenson]

The distance between the PCP and the ECI cabinet is quite large compared to what I've seen elsewhere, I'd say it's around 50m.

I'm pretty sure I'm near the worst case for crosstalk as I've not seen anyone with my attenuation have such relatively low speeds.

But it was bound to be the case as on ADSL everyone would get 1-3mbit.

[roseway]

The problem with QLN is that it's never refreshed until there's a resync. We had a couple of brief power cuts earlier today, and my connection was one of the first to reconnect, so it connected at the maximum speeds (about 10 Mbps downstream more than before). The QLN which resulted at the time looks crazy (see below) but I'm quite sure that it doesn't reflect the present situation. I've little doubt that if I forced a resync now my stats would look very similar to how they were before the power cuts.

[Chrysalis]

ouch thats pretty nasty crosstalk. s.stephenson.

[S.Stephenson]

I just hope I'm in range of g.fast if it is rolled out to the cabinets first, but in the mean time some g.inp and Vectoring would be appreciated.

[NewtronStar]

Find it very hard to separate crosstalk from the normal evening background RFI on the QLN graph though the spikes of RFI from 5938 - 6210 Khz and 7140 - 7500 KHz has been with me since day one on FTTC the only difference as of to-day is the amplitude has increased in those two RFI bands
 

[Starman]

It may have been aardvark or starman who about a year or so ago suddenly got a massive crosstalker?

Within the last 18 months I've had two major HR faults - but since the last one in Nov my line and SNRM has been fairly stable. <3 DS ES, and approx 250 US ES each day hopefully G.INP will get enabled on my upload one day. Looking at my QLN I mainly see effects of AM broadcasts, and my Hlog looks clean.

[NewtronStar]

Notice those two RFI bands with blue arrows

[Starman]

Yeah I did - no idea what the cause is could be cross talk I suppose.

[NewtronStar]

Yeah I did - no idea what the cause is could be cross talk I suppose.

Nope it is short wave radio interference from china and romania 

[Starman]

How rude of them

[NewtronStar]

How rude of them

It is but it also says you line is still showing the effects of HR fault those RFI signal should not be penetrating the D-side unless the joint is impaired.

Soon i'll be setting up a the NooElec NESDR Mini 2 USB RTL-SDR & ADS-B Receiver Set to look a the radio spectral graph but have done most of my testing using the handheld radio receiver and the QLN graphs and tones confirms the location from where the RFI comes from

http://www.amazon.co.uk/gp/product/B00P2UOU72?redirect=true&ref_=pe_1980281_98137581_em_1p_1_ti