cancel
Showing results for 
Search instead for 
Did you mean: 

Lumpy QLN graph = crosstalk ?

FIXED
hitchhiker43
Aspiring Pro
Posts: 201
Thanks: 30
Fixes: 4
Registered: ‎06-07-2016

Lumpy QLN graph = crosstalk ?

I have been suffering increasingly poor service over the last few months.  I have a long (3-4Kms) ADSL2 line about 52dB attenuation in the downlink. Sadly I am to far from the nearest cabinet (2.5Kms) to get VDSL but a lot of people nearer to it (and in the same cable) than me are.  I notice a lot of lumps in my QLN graph, could this be crosstalk from VDSL ? Has anybody else seen a lumpy QLN like this ? mine was smooth some months ago.

Downlink snr just shot up by 3dB (3rd file) so did a line removal reset and got a new QLN graph (2nd file) much less lumpy.....seems this crosstalk/interference comes and goes as I initially got downspeeded by a BT line restart around 9am this morning when I recorded the bad QLM (1st file).

I am mystified!

Tp-link W8960n on longgggggggg line!
66 REPLIES 66
MatthewWheeler
Plusnet Help Team
Plusnet Help Team
Posts: 8,896
Thanks: 1,506
Fixes: 480
Registered: ‎01-01-2012

Re: Lumpy QLN graph = crosstalk ?

I believe VDSL crosstalk shouldn't affect ADSL connections although I might be wrong.

If this post resolved your issue please click the 'This fixed my problem' button
 Matthew Wheeler
 Plusnet Help Team
hitchhiker43
Aspiring Pro
Posts: 201
Thanks: 30
Fixes: 4
Registered: ‎06-07-2016

Re: Lumpy QLN graph = crosstalk ?

Hello Mathew and thank you for the reply, you should read this http://www.analog.com/library/analogDialogue/archives/34-05/vdsl/index.html old but valid in particular this paragraph

"The present spectrum allocation proposals use the ADSL downstream frequency band for VDSL downstream as well. From the point of view of the crosstalk between ADSL and VDSL, there is a subtle difference between the two configurations of Figure 1. In the FTTEx case the presence of the VDSL does not affect ADSL performance, because VDSL power spectral density (PSD) is less than ADSL PSD. Conversely, the presence of ADSL in the same binder may have a serious impact on VDSL performance. In FTTCab configuration the situation is reversed. VDSL signals from the ONU may generate unacceptable noise levels for the ADSL downstream signal, as it becomes heavily attenuated along the path from CO to ONU [1]. " ONU is there terminology for a street cabinet.

Tp-link W8960n on longgggggggg line!
hitchhiker43
Aspiring Pro
Posts: 201
Thanks: 30
Fixes: 4
Registered: ‎06-07-2016

Who's been playing with my line today ?

Please see enclosed SNR plot, first of all my downlink was completely cut (about 11:10am) resulting in a re-synch at a considerably lower bit rate, this was then followed by a gradual increase in SNR.

Is this Openreach fingers again, or the DLM software deciding to re-synch because it didn't like the error rate ?

Tp-link W8960n on longgggggggg line!
ejs
Aspiring Hero
Posts: 5,442
Thanks: 631
Fixes: 25
Registered: ‎10-06-2010

Re: Lumpy QLN graph = crosstalk ?

That's the reason why the VDSL2 signals from the cab are reduced in power on the ADSL downstream frequencies. Exactly which tones and reduced by how much are set according to the distance from the exchange to the cabinet. Essentially, they thought of that before allowing FTTC.

hitchhiker43
Aspiring Pro
Posts: 201
Thanks: 30
Fixes: 4
Registered: ‎06-07-2016

Re: Lumpy QLN graph = crosstalk ?

Ahh thank you ejs, I thought I was on the right track, the problem is we have a pretty severe distance problem here, I don't remember the figures exactly off the top of my head but I think the cabinet is about 2.5Kms from the exchange and we are another 2Kms from the cabinet!!

Maybe there power reductions are inadequate in such situations, but of course as we know they only cater for 90% of customers at all!!

I often have huge holes in my upper tones such as the 1st chart below compared with a more normal situation in the 2nd

The only thing that bemuses me is if it was VDSL xtalk I would expect it to be steady as people leave there routers on but sometimes it almost follows business hours!!

Tp-link W8960n on longgggggggg line!
ejs
Aspiring Hero
Posts: 5,442
Thanks: 631
Fixes: 25
Registered: ‎10-06-2010

Re: Lumpy QLN graph = crosstalk ?

So it's probably not crosstalk then, sounds more likely to be some other source of interference (REIN).

hitchhiker43
Aspiring Pro
Posts: 201
Thanks: 30
Fixes: 4
Registered: ‎06-07-2016

Re: Lumpy QLN graph = crosstalk ?

Well actually I was thinking of grain driers!! We are right in the middle of the barley harvest here and its just possible the grain drier motors have big nasty switching motor controllers and of course we have overhead power to the houses and farms AND I have a really old figure of 8 phone wire (overhead too) that nobody wants to replace lols All my neighbors with go ahead ISP's (nameless) have nice new 2 pair twisted drop wires!! (hint hint pnet!).

Tp-link W8960n on longgggggggg line!
Anotherone
Champion
Posts: 19,107
Thanks: 457
Fixes: 21
Registered: ‎31-08-2007

Re: Lumpy QLN graph = crosstalk ?

The pattern on your Bits/Tone strongly suggests that this is a case of REIN and from all the data it suggests that additionally have a noisy phone line which is also subject to periodic interference because part of it is not twisted pair and the combined effect is periodic drops in connection, DLM seeing issues and raising your Target SNRM to try and stabilise things, with resultant slower speed.
This is NOT as you seem to think OR messing with your line. I'm afraid the engineers are too hard pressed to be fiddling with other lines when they aren't dealing with a known fault.

Can you hear any audible noise on the line, have you heard any crackling or other noises on the line when using the phone? Have you had any problems with incoming or outgoing calls? Try the Quiet Line Test 17070 option 2 if need be. Who do you pay line rental to?

What modem/router are you using? Some xDSL stats could be useful.

hitchhiker43
Aspiring Pro
Posts: 201
Thanks: 30
Fixes: 4
Registered: ‎06-07-2016

Re: Lumpy QLN graph = crosstalk ?

Thank you for your reply, I have tried so hard after many months to isolate this, actually I am fairly certain the disconnects I had taken to be an OR engineer are in fact the DLM. It seems it reacts to certain conditions in realtime hence disconnects can occur throughout the day not just in the early hours of the morning but I cannot find a list of what those triggers might be. I have tried 17070 many times but never heard anything like impulse noise, I have tried the radio and there is a fairly continuous hash from the power lines at 510-550meters+ and as my horid figure 8 drop cable even after all my efforts still runs in parallel with them for some 5 meters this could easily couple.  Here are some DSL stats

adsl info --stats
adsl: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 1
Max: Upstream rate = 864 Kbps, Downstream rate = 5856 Kbps
Path: 0, Upstream rate = 864 Kbps, Downstream rate = 4982 Kbps

Link Power State: L0
Mode: ADSL2+
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 4.0 6.6
Attn(dB): 53.5 31.1
Pwr(dBm): 0.0 12.8
ADSL2 framing
MSGc: 59 64
B: 155 13
M: 1 8
T: 1 2
R: 12 16
S: 0.9955 4.0000
L: 1350 256
😧 32 8
Counters
SF: 4019742 839734
SFErr: 6693 0
RS: 261283258 602493
RSCorr: 10531380 0
RSUnCorr: 15467 0

HEC: 4755 0
OCD: 0 0
LCD: 0 0
Total Cells: 764111347 944327
Data Cells: 23349310 21380
Drop Cells: 0
Bit Errors: 0 0

ES: 36625 0
SES: 5193 0
UAS: 734 4294966742
AS: 65028

INP: 1.13 2.00
PER: 16.17 17.50
delay: 7.96 8.00
OR: 32.14 32.00

Bitswap: 12129 104

Total time = 1 days 19 hours 3 min 27 sec
FEC: 157999077 0
CRC: 338924 0
ES: 36625 0
SES: 5193 0
UAS: 734 4294966742
LOS: 18 4294967271
LOF: 70 4294967271
Latest 15 minutes time = 3 min 27 sec
FEC: 32702 0
CRC: 12 0
ES: 12 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Previous 15 minutes time = 15 min 0 sec
FEC: 172159 0
CRC: 132 0
ES: 49 0
SES: 2 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Latest 1 day time = 19 hours 3 min 27 sec
FEC: 10557026 0
CRC: 7153 0
ES: 5926 0
SES: 17 0
UAS: 26 4294967272
LOS: 1 4294967295
LOF: 9 4294967295
Previous 1 day time = 24 hours 0 sec
FEC: 2958929 0
CRC: 13736 0
ES: 376 0
SES: 223 0
UAS: 178 4294967190
LOS: 3 4294967289
LOF: 9 4294967289
Since Link time = 18 hours 3 min 47 sec
FEC: 10531380 0
CRC: 6693 0
ES: 5913 0
SES: 8 0
UAS: 0 4294967246
LOS: 0 4294967295
LOF: 0 4294967295
>

All comments welcome of course Smiley

 

Tp-link W8960n on longgggggggg line!
hitchhiker43
Aspiring Pro
Posts: 201
Thanks: 30
Fixes: 4
Registered: ‎06-07-2016

Re: Lumpy QLN graph = crosstalk ?

Sorry forgot modem, its a tp-link W8960n running ADSL2+ with Annex L, SRA & Bitswap on. I am presently spoofing snr (Ahem) -2dB and the DLM most of the time runs 6dB.  I don't have any problem with the error rate in the downlink (about 10Kfecs/sec) but the DLM obviously did, fortunately recently I found someone who knew what I was talking about and got the DLM to ignore errors.

Tp-link W8960n on longgggggggg line!
Anotherone
Champion
Posts: 19,107
Thanks: 457
Fixes: 21
Registered: ‎31-08-2007

Re: Lumpy QLN graph = crosstalk ?

Right, quite a lot to go through here. Where to start, perhaps the things we don't need to mention again.
SRA will have no effect on a BTw provided circuit, and I was going to ask why have you specified Annex L as you would severely limit the maximum speed you might be able to achieve (if the line were good and if specifying it actually did anything) but it certainly seems to be having no effect - and BTw circuits do not support Annex L, only Annex A and Annex M, and you certainly have Tones well above 128 in use. Annex M has to be ordered by the ISP and in any event Plusnet do not supply it on residential connections.

Lets deal with errors briefly, if you have the option to reset the error count/statistics on that TP-Link without causing a reboot, that may be useful providing you note the date & time you do so and post it with any further stats. This would help give a clearer picture of which errors are increasing when etc.
I have no doubt that when you have a burst of interference or line problem, the counts will go rather high in any event.
I'm not sure that DLM is ignoring errors as such. There are very few people in faults (first line support don't have a clue) that know how to set the Line with Custom Stability (which can effectively turn off DLM we are led to believe), otherwise the normal settings are Standard, Stable, Super Stable - the default IIRC is Stable - someone may have set stability to Standard which will prioritise Speed over stability, or managed to set a permanent (pinned) Target SNRM of 6dB.

DLM uses MTBE and MTBR to determine what it does/may do. As far as MTBE is concerned ES/SES errors are the primary factor. Often if the FEC, HEC or CRC counts go excessively high they are frequently accompanied by high levels of ES or even SES errors.
We've seen many connections that seem to be unaffected by high FEC counts, but certainly high CRC counts will affect data throughput speeds due to the need for retransmission.

I'm not sure where you get your FEC rate from (other than it being the average over a long period) which will appear very high if the period includes a large burst of noise or a line problem. 10Kfec/sec IS very high! Figures bandied about suggest less than 14000 FECs in a 5 minute period is considered acceptable, not that it's going to make any difference to DLM. That said, not everything about DLM is in the public domain, so exactly how it behaves in complex situations is not clear. We've seen cases where problems with high errors on the US have caused DLM to ultimately raise the Target SNRM on the DS Shocked

Tweaking the Target SNRM (in your case presumably to get a bit more speed) can come with problems of course - increased error rates, but sometimes a happy(ish) balance can be achieved.

Now to a bit more nitty gritty of what's happening, that SNRM graph in message 4. What could be important to the ultimate interpretation, is what is the sample rate for the plots?

That drop certainly isn't due to Openreach or DLM! There is some source of interference turning on and it's taking a while to "warm-up" or "get to full steam" so to speak - getting on for about 3½minutes or so for the SNRM which was around 4.5dB to get to zero and the connection drops. The line resyncs and we see the SNRM gradually rising (sharply at first) then tailing off more gradually until the SNRM gets to ~7dB ie. the interference level gradually reduced from it's peak and has now stabilised at a steady rate. A few minutes later the source/cause turns off which is why you see the sharp rise in SNRM by about 3dB. This rise by ~3dB can also be seen on the SNRM graph in your OP at ~1924 where the cause stopped/switched off.

Obviously the question is what's causing this? It could be something in the house or near-by. A length of untwisted drop-wire will not help (whether it's running close to electricity cables or not).
Try and eliminate anything in the house. It could be anything from Discharge lighting to a switch-mode power supply in any bit of equipment. Consider the timing, does it fit with anything switched on or off around the house. This could also include any CRT device (eg. monitor or TV), things like cordless phones or mobile phones and their chargers etc. Once you've eliminated anything around the house it then gets more tricky and it's often a case of observation. Sometimes using a MW radio can help if you can find a noise that comes and goes with the pattern you see on the SNRM graph.

The other observation about the SNRM plot that I don't like is that repetitive cyclic pattern clearly visible on the downstream plot that seems to have a periodicity of around 3 minutes. There seems to be something similar on the US but the periodicity isn't as clear. The magnitude seems to vary a bit as well at different times suggesting whatever it is that is being picked up varies, which might mean it moves slightly or something affecting the amount of pickup does, it might even be your movement that affects it slightly.

I think that's enough to be going on with. This repetitive pattern is bugging me, I've a feeling I've seen something like this before and trying for the life of me to remember when/what/where etc.

Oh, as far as audible line noise goes, it's not impulse noise specifically, it's any noise - crackling/fizzing or other such noises. The line should be consistently silent apart from maybe a very very faint steady hiss.

Anotherone
Champion
Posts: 19,107
Thanks: 457
Fixes: 21
Registered: ‎31-08-2007

Re: Lumpy QLN graph = crosstalk ?

Ah, nearly forgot. I asked about Line Rental. If the line rental is not with Plusnet, then it might be that this has to be "played" in a slightly different way to try and ultimately get it resolved.

hitchhiker43
Aspiring Pro
Posts: 201
Thanks: 30
Fixes: 4
Registered: ‎06-07-2016

Re: Lumpy QLN graph = crosstalk ?

Good morning and thank you for your reply, yes I have become aware not all router settings are supported by BT but I included them anyway for completeness, you will have to excuse the fact that many settings have been trialled over the years in an attempt to improve this line. I cannot unfortunately reset the stats without re-booting the router but it does clearly show the stats since the last linedrop and that usually takes place at least once a day anyway, I will include them at the end of this reply. You are right that a lot of people do not know much about the DLM however with persistence you can find one! Mine set MTBE/R to 86,400/0 respectively and locked tsnr to 6dB on Friday night and things have improved as a result but still getting the occasional dlink re-sync like last night (see vindictive post).

I may have confused minutes and seconds in my FEC statements as dslstats reports in minutes and the dlm uses seconds, to confirm my normal rate is 10K/MINUTE.

The sample rate of the dslstats plot is 30 seconds, yes I have always been confused by the almost gradual change in snr at disconnect events, I would have expected a much sharper change if the dlm/dslam dropped the connection and also an immediate rather than gradual rise to the new snr at re-connection, if i force a disconnect/reconnect by disconnecting and reconnecting the line then the changes in snr are immediate, this is my method of restoring my rate after an event and I think thats what you are referring to as the interference source going away.  The other reason I am beginning to think it's not the DLM disconnecting is when I restore using a disconnect it returns to the previous tsnr!!

All activity in the house has long been eliminated including some astonishing chinese laptop psu's that although marked CE/FCC contained no filtering components at all and resulted in a 3dB snr loss!!

I am very badly hit by power line noise as my figure8 untwisted drop wire is actually run immediately alongside the two power cables feeding my house for a length of 5meters(15feet) in a place high up that I cannot reach, nice air cored transformer!

It's good to talk about this stuff as it helps to clarify what might be going on, its that gradual change in snr at the dropout and returning to the same tsnr when i force a drop that gets me, its as if the MODEM is faulty!! Actually I just checked it's case temperature, 43degC god knows what that makes the chip temperatures inside, I noticed the case has gone brownish over the years instead of it's pristine white! and it's on a fairly long line chucking out 0dbM 24/7, maybe it's thermal design is not all that it could be hmmm

The only thing I can say I have heard on the quiet line test is some faint power line hum, but nothing else that I have noticed.

Here are the stats, last reset was last night for the usual reason

adsl info --stats
adsl: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 1
Max: Upstream rate = 864 Kbps, Downstream rate = 5820 Kbps
Path: 0, Upstream rate = 864 Kbps, Downstream rate = 4920 Kbps

Link Power State: L0
Mode: ADSL2+
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 3.9 7.6
Attn(dB): 53.5 31.1
Pwr(dBm): 0.0 12.8
ADSL2 framing
MSGc: 60 64
B: 151 13
M: 1 8
T: 1 2
R: 12 16
S: 0.9820 4.0000
L: 1336 256
😧 32 8
Counters
SF: 3332192 209120
SFErr: 20615 1
RS: 219924694 2451644
RSCorr: 9281335 302
RSUnCorr: 32444 0

HEC: 16130 1
OCD: 1 0
LCD: 0 0
Total Cells: 626550734 883773
Data Cells: 15711101 639
Drop Cells: 0
Bit Errors: 0 0

ES: 54585 1
SES: 5354 0
UAS: 806 4294966700
AS: 53992

INP: 1.14 2.00
PER: 16.20 17.50
delay: 7.85 8.00
OR: 32.58 32.00

Bitswap: 10212 14

Total time = 1 days 17 hours 47 min 55 sec
FEC: 171996608 0
CRC: 370853 0
ES: 54585 1
SES: 5354 0
UAS: 806 4294966700
LOS: 20 4294967268
LOF: 79 4294967268
Latest 15 minutes time = 2 min 55 sec
FEC: 22640 0
CRC: 7 0
ES: 7 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Previous 15 minutes time = 15 min 0 sec
FEC: 133240 0
CRC: 56 0
ES: 52 1
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Latest 1 day time = 17 hours 47 min 55 sec
FEC: 11053612 0
CRC: 30753 0
ES: 16928 1
SES: 160 0
UAS: 72 4294967254
LOS: 2 4294967293
LOF: 9 4294967293
Previous 1 day time = 24 hours 0 sec
FEC: 13500945 0
CRC: 8329 0
ES: 6958 0
SES: 18 0
UAS: 26 4294967272
LOS: 1 4294967295
LOF: 9 4294967295
Since Link time = 14 hours 59 min 51 sec
FEC: 9281335 302
CRC: 20615 1
ES: 16079 1
SES: 0 0
UAS: 0 4294967182
LOS: 0 4294967293
LOF: 0 4294967293
>

Tp-link W8960n on longgggggggg line!
hitchhiker43
Aspiring Pro
Posts: 201
Thanks: 30
Fixes: 4
Registered: ‎06-07-2016

Re: Lumpy QLN graph = crosstalk ?

No the line rental IS with Plusnet Smiley

Tp-link W8960n on longgggggggg line!