Lumpy QLN graph = crosstalk ?

[#1]

OK that should help when hopefully we can get to a point where there is (a) clear fault(s) to pursue. before I pick up on some of the things you mentioned, I think it might be worth asking @Chris if he could post a Vis.rad and an xDSL Status report on this thread which might help build a picture.

[#2]

 

	Upstream DSL Link Information	Downstream DSL Link Information
Loop Loss:	31.1	50.5
SNR Margin:	7.7	6.6
Errored Seconds:	0	1
HEC Errors:	0
Cell Count:	63	206
Speed:	864	4470

Maximum Stable Rate (KBPS):	3424	Fault Threshold Rate (KBPS):	2739
Mean Time Between Retrains (Seconds):	7164	Mean Time Between Errors Upstream (Seconds):	14329
Indicative Line Quality:	A	Mean Time Between Errors Downstream (Seconds):	112

[#3]

Thanks very much Chris.

[#4]

I mentioned in message #12 about some of the possible causes of noise etc that could trigger a drop in connection - I didn't mention which you have rightly considered, that the modem/router itself may be suffering an issue and causing the drops. Whether it be thermal or otherwise can only be proven by substitution. Things like filter problems may also be a cause, but the pattern seen on the SNRM graphs are not suggesting that - filter issues can often be accompanied by unusually high levels of HEC errors.

Whilst individuals can easily think that there are separate issues involved, it best to post all the information in one post. Often things can be related and when all the information is in one post, it's easier to get a complete picture. Even if it turns out that there are separate problems, they can still be dealt with individually in the one thread, there's nothing to stop that happening.

On the subject of the graph sampling, if you can set the sample time to say 10 seconds, or 15 seconds if the modem/router complains and a graph duration of 2hrs, that's usually a happy compromise to show the relevant detail. As already mentioned, this doesn't look like DLM causing the drops and if you've managed to get a custom stability setting as per your message #14 then there shouldn't be any problems with the Target SNRM being raised.

It's now a case of trying to pin down the likely causes. If you have a fan you could try cooling the modem/router to see if it has any effect at all, but there doesn't seem to be a consistent particular pattern of drops looking at the Vis.Rad graph Chris has posted, but there are a few around the middle of the day - would this be anything you are doing - rebooting perhaps - likewise for any of the others?
As you can see from the xDSL Status, the DS MTBE isn't too clever.

Was there anything going on/not going on for the 4 days that there were no drops?

Probably worth checking some of the obvious things like sockets and wiring. Do you have an NTE5a, do you have any extension sockets and what's normally plugged in where?

What I'd suggest is that at a time where there appears to be no interference or the likelihood of any for say 15 minutes, but nevertheless in daylight hours a good hour before sunset or  a good hour after sunrise, is power down your modem/router. Wait about a minute before unplugging it from the line. Whilst you are off-line swap the filter for one you know/believe to be good. If you have a filtered faceplate, use a filter and plug into the test-socket behind the plate. These are hoops you'll have to go through anyway before you'll be able to get a formal fault off the ground.
After ~15 minutes power up again. This will also have fully reset the stats, so it will be easier to pick up on the error changes that are occurring.

Do you have a spare modem/router at all? Do you know/believe it to be good and if so, what is it?

[#5]

ok ok accepted i was trying to be clear about the different faults and yes like everybody sometimes frustration given the state of pnet suppt these days gets the better of us users!

Well first of all whilst physically re-arranging the router this afternoon the power socket at the back must have momentarily d/c I didn't notice til lI looked at DSLstats, realized what had happened and imediatly went and cleaned the power socket resulting in the 2nd outage shortly after.  Normally the modem is never powered down except by power failures we have here maybe every month in winter and a few times in the summer.

The line drops I have recorded this year starting from January in uk format (day/month) are 21/1, 23/1, 6/2, 28/3, 10/5, 2/6, 4/6, 24/6, 25/6, 1/7, 2,7, 4/7, 12/7, 16/7, 25/7, 9/8, 10/8, 11/8, 14/8 & today, on several occasions there were several a day.

I have doubled the DSLstats sample rate though this reduces the chance of capturing an event, let's hope Since this afternoon there is no snr spoofing set anymore so I am running 4470kbps @6.4dB however my speedtest is down to 3000Kbps as something must have not liked seeing the modem powered down twice in short succession (cannot remember the name of that bit of software).

The wiring in this house was a nightmare, it had an old BT era master socket but not in the original location then a few star extensions, none of the wiring actually used pairs!! The old drop cable had been removed from the front door/hallway and coiled up in the loft! anyway the wiring is now properly paired and I use two filter in series between the test socket and the rest of the house (the modem being in the one nearest the test socket) these are the old speedtouch filters that I found outperformed many others I have tried.

I will have to ask around for a spare modem but most people around here are on BT, I have an old thompson but I already know they are very noisy, also no wifi.

I am not sure about powering down the modem, look what its done to my speedtest already!

I think I might wait a few days and see if there is another event without touching anything

Many thanks for your help and interest, ohh and yes I did get a "custom" stability setting so perhaps things will settle down especially if it had anything to do with modem thermals as its now verical with a nice big airgap underneath and the case is almost 10C cooler....

 Ohh I almost forgot, I will change filters tomorrow as I have some more and check the contacts are clean on the test socket etc!

[#6]

No events overnight replaced filters and cleaned contacts this morning, here is the nasty old master socket with filters between test socket and remainder of house, as noted modem is in filter attached to master socket. It's buried behind furniture so its messy look doesn't bother me

2nd pic is an example of non-paired use of CW1308 when I eventually discovered the original drop cable termination in the loft/attic ora/wh and blu/wh hahaha how does someone do that when you strip the cables the pairs are obvious even if you are color blind lols

3rd picture is my key suspect for where the damp is getting in, unreachable joint box high on the gable that probably hasn't been touched since the house was built 40 years ago, I have been told by a lineman they often corrode!

[#7]

Another event about 11:57am slow reduction in snr till lost connection, came back when snr slowly recovered to 1.5dB (no spoofing), chart sample rate 15 seconds. Quit line test 50hZ hum in background but no discernible crackles etc, radio usual powerline? noise at ~500-800Khz. no change in bitrate, still 4590Kbps but all high tones gone. pdf doc contains stats as near the event as I could get and after. Still running at 12:42, 4.5dBsnr 3K/min fecs. Will leave dslstats on over lunch and see what happens.

[#8]

14:25 there back from lunch ?? another event and this time the dlm did intervene as the rate has dropped dramatically, snr graph is 15 second resolution, no snr spoofing. Dont know if the high tones are back as new rate 3286Kbps doesn't need them.

pdf file is best effort to capture before and after stats

I actually drove a few miles to check if there was an OR van at my cabinets or along the route...none!

[#9]

15:36 whilst I was out digging potatoes it's shot up another 3dB, fast, no other change! weird.

[#10]

Just a quick reply so you know I'm not ignoring all this as I have been short of time and will need to take a longer look later. First I did mean to mention in my earlier reply about understanding that frustration can sometimes get the better of us, that's all best put to bed now I reckon.

A couple of quick things, the CW1308 from the loft, have you now got that as a pair (Blue/white & White/blue) to the Master socket?

Next, that won't have been DLM intervening, it's because all the interference is present, some of the tones are unusable so when it resyncs with the target SNRM (6dB) you end up with a slower speed. The interference is coming from something that takes time to settle once it starts - you see the quick rise in SNRM at the point of resync after the drop. The level of interference is now decreasing a bit, hence the gradual rise in SNRM until it settles, and then when the source stops, you get the sharp rise in SNRM. It's after this point that a resync would see your speed recover.

Not sure why you think that 15 seconds sampling might miss something, quite the reverse, which is why in some cases 10 second sampling is better. Stick with it as it is for now.

Powering off the modem/router is far better than just unplugging it from the line if you are going to swap anything about, check connections etc. It's a "clean" death of the xDSL and some modem/routers send a dying gasp signal, however there aren't too many line cards about the take any notice of that these days, BUT if you just unplug from the line it isn't necessarily a "clean" break and you could get loads of errors before you are actually disconnected. When you already have a problematic connection, it's just that sort of thing that can get DLM to act.

Also with problematic connection issues, it best to power off for a minimum of 15 minutes just to resync - although the difficulty is if you have interference coming and going and you don't have a known time pattern, it could return when you are off-line and when you resync you aren't any better off It's a case of just picking the right time in daylight hours for ADSL.

I'll look at the stats and everything in more detail later

[#11]

No worries, I am sorry about the volume of data! Yup all the remaining CW1308 is on pairs now as per normal colors but some lengths (damaged/burnt) were replaced with cat5 I had to hand! In all cases A/B are a twisted pair, I did all this several years ago.

I think you must be right it's a source of interference, the only other thing I could think of to explain the slow fall was the modem desperately trying to hang onto the connection when it's d/c at the other end of a very long line but that wouldn't explain the slow rise

With the sampling I meant the screen width becomes 1 hour instead of two, but I do see they are independently adjustable (width and rate).

Ahh yes I see what you mean about untold things being accumulated by the DLM if just line breaks are used, interesting about the 15 minutes too, is this to tally with it's 15 minute bins ? I guess so, anyway to get a truer picture of whats going on for a few day's unless instructed otherwise I shall leave well alone for now So anything that happens is entirely down to interference/DLM.  I have not totally exonerated my modem yet but the very slow changes in SNR make it unlikely to be a hardware fault in it IMOP.

Thank you for your interest and I apologize in advance for adding to your load

Roger

[#12]

Well today Wednesday it got past the critical lunch period with no signs in snr or errors that anything happened, given the random nature that could be true BUT I have another theory, it's running so slow (3286Kbps) it's not using any frequencies above 715Khz and has loads of headroom below as its also running almost 12dB snr. Now my wife is complaining bitterly about slow internet (speed test 2800Kbps) so I have to do something but I don't want a return to random dropouts! Knowing the DLM is locked at 6dB and I also know if I let it sync at that all the nastiness will return. So I am going to experiment with spoofing the snr in the opposite direction than usual to increase it above 6dB, the purpose being to find a stable speed. I will leave it 24 hours after each change to get some reliable stats from the router, one day I might be able to return to all my other jobs

Done it, 3766Kbps @ 9.7dBsnr, will leave for 24hrs, feel like a human DLM

[#13]

If it's running at ~12dB SNRM, I would just do the power-off resync.

Remember that although being on 21CN your IP Profile will update immediately with the new sync speed, the BT servers need to propagate the new profile through to PN's system so that your Current Line Speed updates. It should result in a drop in PPP when that happens, sometimes it doesn't so you have to drop the PPP yourself to see the improved speed. I have a horrible feeling those TP-Links don't have a provision for dropping the PPP session, only sync as well

Edit: sorry haven't looked at all your data yet.

[#14]

Hi well I want to try and establish how fast it will go and maintain stability, just like the DLM is supposed to do but it doesn't seem to work well with this intermittent interference problem I have. If I just let it loose at 6dB I think I will be straight back into lunchtime disconnects again, as you have seen due to the slow on/off characteristic of the interference this drastically cuts the line rate. As I am loosing hope of ever improving my line to make it more interference proof for the time being I am trying to find an operating point where at least for a while it can withstand it without dropping out.  Sorry for the long explanation, I may be barking, let's see At the moment I am less worried about the speedtest speed until I can achieve a stable line rate despite this awful interference.

Ohh I forgot to mention it's impossible to change anything on a Tp-link without a reboot, real nuisance Grrr

[#15]

---

@hitchhiker43 wrote:

Well today Wednesday it got past the critical lunch period with no signs in snr or errors that anything happened, given the random nature that could be true BUT I have another theory, it's running so slow (3286Kbps) it's not using any frequencies above 715Khz and has loads of headroom below as its also running almost 12dB snr.

The Interference was affecting plenty of tones below that frequency, so you would have seen some reductions on your SNRM plots if it had occurred. Question is what never got switched on today? - unless it's had other implications/effects elsewhere and has been taken out of service or repaired