Internet Drops Out - Opinions on my options please

[#1]

Every so often my internet connection stops responding. The router says it is still connected but I cannot ping news.bbc.co.uk nor 212.159.8.2 which is a Plusnet computer. Sometimes it comes back after 5 or 6 minutes and sometimes I have to tell the router to disconnect and then reconnect which usually brings the connection back. Repeating that a few times has always succeeded.
Sometimes this happens a number of times within a few hours and sometimes things are good for several days.
Usually there is a note in the router log saying "2 loss of sync"  at just about the time the problem starts. (Hard to be precise as what with cached content it can take a good few minutes to discover that the connection is lost and so far this is not a scientific experiment.)
I believe I have recently been converted to 21CN - Plusnet sent me a mail saying it was going to happen, I have seen nothing in writing to say it has been done but my upload speed is now 928Kb/s (faster than ever before) and my download speed seems faster than before (currently 3616kbps and its been higher recently cf anything between 2000kbps and 3200kbps historically). I am not clear why changing to 21CN should change the local connection speeds though.
For completeness, download attenuation is 42db and noise margin is 15db. Upstream attenuation is 13db and noise margin is 11dB. Historically the line has been very variable and previous investigations have concluded that the BBC or somebody is broadcasting near to me and interfering with the line. But if the transmitter caused the lost sync's then why is the noise margin only a few minutes later so good.
The router is plugged straight into a filter plugged directly into the master test socket. I have not raised a ticket yet as I would like to understand a bit more about what is happening before I do that.
Do these random bits of information build into a sensible story to those more technically knowledgeable than me?
If so - what course of action do you suggest I should  take to stop it happening?

[#2]

Hi there, are you still using a DG834 and if so which model is it?
Can you post the full ADSL stats?

[#3]

Yep - I don't spend money easily so the DG834 is still at work. Its one of the original V1s running V3.01.38 firmware.
I hope these are the ADSL stats that you requested. If not, give me a few clues and I will try again.
AR7 DSL Modem Statistics:
--------------------------------
[DSL Modem Stats]
        US Connection Rate:    928    DS Connection Rate:    3616
        DS Line Attenuation:    42      DS Margin:              14
        US Line Attenuation:    26      US Margin:              11
        US Payload :            5172384 DS Payload:            57692592
        US Superframe Cnt :    334779  DS Superframe Cnt:      334780
        US Transmit Power :    0      DS Transmit Power:      0
        LOS errors:            0      SEF errors:            0
        Errored Seconds:        0      Severely Err Secs:      0
        Frame mode:            3      Max Frame mode:        0
        Trained Path:          1      US Peak Cell Rate:      2188
        Trained Mode:          3      Selected Mode:          1
        ATUC Vendor Code:      4946544E        ATUC Revision:  1
        Hybrid Selected:        1      Trellis:                1
        Showtime Count:        3      DS Max Attainable Bit Rate: 3616 kbps
        BitSwap:                1      US Max Attainable Bit Rate:    n/a
        Annex:                  AnxA    psd_mask_qualifier: 0x0000
        Power Management Status: L0    DS HLINSC: 0
        US ACTPSD:              -345    DS ACTPSD: -358
        Total init. errors:    1      Total init. timeouts: 0
        Showtime init. errors:  0      Showtime init. timeouts: 0
        Last showtime init. errors: 1  Last showtime init. timeouts: 0
        ATUC ghsVid:  b5 00 49 46 54 4e 71 be
        T1413Vid: 00 00        T1413Rev: 00            VendorRev: 00
        ATUR ghsVid:  b5 00 54 53 54 43 00 00
        T1413Vid: 00 00 T1413Rev: 00    VendorRev: 00
        [Upstream (TX) Interleave path]
        CRC:    96      FEC:    6392    NCD:    0
        LCD:    0      HEC:    340
        [Downstream (RX) Interleave path]
        CRC:    7773    FEC:    102969  NCD:    0
        LCD:    0      HEC:    0
        [Upstream (TX) Fast path]
        CRC:    0      FEC:    0      NCD:    1
        LCD:    0      HEC:    0
        [Downstream (RX) Fast path]
        CRC:    0      FEC:    0      NCD:    0
        LCD:    0      HEC:    0
[ATM Stats]
        [Upstream/TX]
        Good Cell Cnt:  107758
        Idle Cell Cnt:  12348547
        Tx Packets Dropped Count:      0
        Tx Bad Packets Count:  84
        [Downstream/RX)]
        Good Cell Cnt:  1201929
        Idle Cell Cnt:  47249211
        Bad Hec Cell Cnt:      85565
        Overflow Dropped Cell Cnt:      0
        Rx Packets Dropped Count:      0
        Rx Bad Packets Count:  0

[SAR AAL5 Stats]
        Tx PDU's:      3121099
        Rx PDU's:      4450956
        Tx Total Bytes: 268609661
        Rx Total Bytes: 1242957872
        Tx Total Error Counts:  0
        Rx Total Error Counts:  9106

[OAM Stats]
        Near End F5 Loop Back Count:    0
        Near End F4 Loop Back Count:    0
        Far End F5 Loop Back Count:    0
        Far End F4 Loop Back Count:    0
        SAR OAM Ping Response Drop Count=0
#
#

[#4]

Well, I was watching Routerstats trundle along when suddenly the spike shown in the attachment happened. Hard to be sure but I reckon I lost communication to the internet while the noise spike was at its worst but it recovered by itself by 22:50. No loss of synchronization messages in the log at this point.
(Hope something is attached - we will see when I press post!)

[#5]

My understanding is that DG834v1's with that firmware should run happily on 21CN ADSL2+, but you really need someone who has used and played with them to give more informed comment, some opinion says they are not as good on ADSL2+.
That said, I think you have other issues which is giving you poor performance. Those ADSL stats are what is useful, but when looking at errors knowing the Uptime/Showtime would be useful. Also the stats from [ATM Stats] onwards aren't necessary.
To answer your question about why 21CN (ADSL2+) should "change the local connection speeds" - in simple terms a wider frequency band is used for ADSL2+ (more tones) so more bits of data and so a faster speed, until you get to very very long lines (6+ km).
Your line attenuation puts you about 3km from the exchange and frankly your sync speed (DS Connection Rate) is poor. My attenuation is virtually identical and I get speeds of just over double that (on 20CN ADSL). Your noise margin is 15dB which is high (not good) giving you a slower sync speed.  Also having an uncapped upstream speed will lose you some downstream speed, so unless you upload a lot it may be better to cap it.
I'll take a look at the Routerstats you've just posted, I'm expecting the spike and loss of sync to be part of your performance problems.
Edit:  It seems you didn't lose sync,  I'll comment more in a few minutes.

[#6]

Those are some fairly large swings in the Noise Margin, but the mean looks pretty level,  so I don't think it's due to atmospheric propagation.
Can you turn on the Average Noise plotting (Graphs >Extras tab  - previous 100 samples)? Could you also adjust the sampling to 8 seconds if currently longer, & 900 points per page will give 2hrs, unless you want it less.
When the "spike" occurred, were you aware of anything switching on or off, did you use the phone or did it ring?
Can you tell us something about your installation?
Let's start with the BT cable - is it overhead (just to the nearest pole or for some distance) or underground? Is the incoming cable of round cross-section, black sheath just over 5mm in diameter and connected to an NTE5 Master socket ? Are there any other connection boxes between the Master socket and the incoming drop-wire?
Do you have any fixed extension wiring and sockets, if so how many? If so, what colour are the wires connected to the NTE5 Faceplate and to the extension sockets and which terminals are they connected to?
Which socket is your modem/router and it's filter plugged into? What type of phone(s) do you have - corded or cordless and where are they plugged in? What is plugged into any other sockets and are they all connected by filters?
Are all the filters this type as shown in para.2?
When you use your phone, can you hear any crackling or other noises on the line?

[#7]

<img src="http://ccgi.psmith12.plus.com/visradius/generated/image13364398848914.png" />
Very odd, it wouldn't appear that it is the broadband dropping out, it could potentially be the Wireless Connection giving you a few problems. It might be worth trying to change your <a href="http://www.plus.net/support/broadband/hardware/thomson_wireless.shtml">Wireless Key</a>

[#8]

I will try to answer the questions and then investigate altering the settings you suggest.
No, the phone was not in use and did not ring at the time of the spike. I am not aware of anything turning on or off at about that time in my house - its pretty quiet at that time of night.
Installation details:- I am guessing a bit but there is a cabinet at the top of the road say 400yds away and from there the cable runs underground to the house. There is a length of overhead (thick) cable from the cabinet running back towards the exchange which as you say is about 3km away but I need to go for a walk to see whether all of that 3km is overhead or not. I suspect not.
Yes, its a black circular cable about 5mm in diameter that comes up the outside wall to a little grey cover on the wall about a foot off the ground. Inside there, the black wire ends and is joined to a short length of wire that goes through the wall to the NTE5 box. Again, I am guessing but the black wire seems to contain 2 pairs only one of which is connected through the wall.
The variable noise level is a long standing one as you know (not the spikes or loss of sync though) so for the last 18 months I have had the faceplate off the NTE5 and a filter plugged directly into the test socket. One arm goes to the router, the other goes to a phone splitter which feeds an adjacent totally standard phone and a Dect phone base station. Cables could be shorter but they are typically the ones supplied with that type of kit so 6 or 7 feet. There are a number of power supplies (for the router, Dect etc) nearby but nothing has been changed recently that I am aware of.
The router is not a wireless router - it just contains the modem and a 4 port wired router. Plugged into the router is a CAT5 cable that goes off to another router in the loft that feeds out using CAT5 cable to wired sockets around the house. The computer I am using is plugged directly into that CAT5 house wiring and I am not using wireless for my connection. There is a  separate wireless router plugged into the DG834 which is used for various family laptops around the house. There is also a NAS box plugged into the DG834.
The filter is of the type shown on the page you linked to. It bypasses the filter in the NTE5 box because using that would reconnect some extension wiring that I would need to improve before reconnecting.
There is no crackling on the phone line when I listen.
Do you notice how the trace on the Routerstats plot attached earlier is quieter after the spike than before - I know there is only a short length to see but it has continued like that through the night, 14db with an occasion tick to 13db or 15db. Nothing like the variation seen before the spike. I will keep monitoring that and see whether/when the noise comes back.
Chris: I do not think your comment applies as I am on a wired connection and other computers in the house suffer the "lack of response" at the same time. If I understand your graph, then the two connection closes and restarts at about 6pm and  7pm on Monday could be the problem in action. Or is a close/start not the same as a broadband dropout? I keep the router permanently on and when I restart it I do so at midday, so the session start on Sunday 29th is probably the same problem - noticeably at about the same time of day. I know that the times do not seem to quite match but I could have missed the first one if it reconnected automatically and the second one could just about be the problem that triggered this topic.
Before anyone says that 2 breaks in 7 days is not bad, I agree. But this problem occurs in bursts and the last burst (probably the 29/4/12 one) involved a lot of losses of connection all occurring during a 5 or 6 hour period, some of which took a lot of reconnecting from. I nearly phoned Plusnet then to ask if their system was down. So I am jumping at these latest two breaks wanting to head off a potentially bigger problem. I would also like to get the noise problem sorted (separate problem maybe) as my connection speed has been appalling at times in the past - probably due to the noise.
Final question for the moment. When Plusnet said they wanted to change me over to 21CN, they did not say anything about changing to ADSL2 as well. I thought they were separate elements, one from the exchange to Plusnet (in simple terms) and one from me to the exchange, and could/would be changed separately. I am now assuming that not only was the 20CN to 21CN change made but that the change in speeds shows that I was also changed from ADSL1 to ADSL2. Is that correct?
Thanks for the help. I will now look at changing the routerstats settings and get back here if anything interesting happens.

[#9]

From the stats you posted I think you are on ADSL1 - but you have to compare "Trained Mode: 3" with the values in /proc/avalanche/avsar_dsl_modulation_schemes .

[#10]

@ejs. Against Trained Mode 3 it says Selected Mode 1 which /proc/avalanche/avsar_dsl_modulation_schemes translates to Multi_Mode.  Have I understood what you intended me to do correctly and can I take this further?

[#11]

It wouldn't matter two hoots which ADSL mode you may or may not be on, the speed should be double or more. The noise and whatever is causing the sudden change in Noise Margin is what will have caused the Target Noise margin to go up and limit speed.
Yes, I did notice how flat the Noise margin trace went, I deliberately didn't comment at the time as I didn't want to hazard any guesses without more information, but I would say that at least gives hope that once the cause of this noise is fixed, then things should be more stable.
I'll comment more shortly, I just want to re-read your detailed reply about you comprehensive set up.
Only two questions at this point, do you have another modem or modem/router and filter to do some testing with if needed?
At any time when the noise margin is varying as per the first part of the graph rather than when it is flat, have you listened to the phone line?

[#12]

No, I do not have another modem/router. I am sure I can find another filter if needs be - I could plug the faceplate back in after disconnecting the extensions for instance.
I have not scientifically listened to the the phone while watching the graph for noise level but my impression is that it is always quiet. I will check more carefully next time it happens.
The significance of the ADSL2 thing is that the drop outs have only started recently, possibly since the conversion from whatever to whatever. Some form of noise is a long standing problem.
I attach a plot of the noise level from this morning with the settings changes your suggested. Uptime on the wan is 15:37 hours and on the lan it is 215 hours 38 minutes.
I notice the screen display from the DG834 says 13db upstream attenuation while the printout says 23db. I guess its a Netgear bug.

[#13]

Let's just clear up a couple of points - I'm not sure why Chris thought there may be a wireless problem, perhaps he didn't spot which modem/router you were using. The session start/stops aren't necessarily drops in sync, but a lot of users don't stop/start their "ppp" sessions so very often it does indicate a sync drop. But it does suggest on Sunday, you didn't come online until around 6 pm.
Also, you are definitely on ADSL2 - your upstream speed is clear proof.
I doubt that switching to ADSL2 will be anything to do with what is happening, just the "noise issue" has maybe got worse and therefore having a more noticeable impact.
The 23dB upstream attenuation is near the correct value (consistent with the downstream). The wan uptime is mysterious, suggests something happened at around 0430 this morning!
On the Noise margin plot, can you reduce the number of points to say 900 (2hrs) the plot will be less cramped, also check that the capture image size is set to 100% and if you maximise the window (before minimising to tray), a capture should then be maximised and so clearer.
On the phoneline, do a "Quiet line test" dial 17070 select option 2 (no others!), if the noise is still present, can you do it now?
I'm guessing you've only started running Routerstats again in the last 24 hours, so it's too early to say if you can spot any pattern about when the noise is present or not.

[#14]

But it does suggest on Sunday, you didn't come online until around 6 pm.
Its most unlikely that that is the case. We would normally be using the internet most of the day on a Sunday. I can believe we had a burst of problems that Sunday and that the 6pm restart was the last of the day. Why the previous ones do not show up, I do not know - probably something to do with the graphing.
The wan uptime is mysterious, suggests something happened at around 0430 this morning
I do not understand your comment. WAN uptime of 15:37 surely ties in with the last restart at about 8pm last night. 15 hours after that is about now.
On the Noise margin plot, can you reduce the number of points to say 900 (2hrs)........
It is at 900 points, 2 hrs in the last plot I posted. I will try to do better with the screen capture but am using windowed wine so not quite straight forward.
I have just done a quiet test - though the graph is quiet at present as well. No noise that I can hear.
You are right that there is not enough data to detect any sort of pattern to the coming and going of the noise. I am wondering if there is a command line version of routerstats that I can run on the NAS. I will investigate.

[#15]

Quote

We would normally be using the internet most of the day on a Sunday. I can believe we had a burst of problems that Sunday and that the 6pm restart was the last of the day

The plot doesn't suggest a restart, it suggests you weren't on until then. I wonder if it's the right chart?
You don't have another account under your forum username that may mean the wrong one was looked at?

Quote

WAN uptime of 15:37 surely ties in with the last restart at about 8pm last night

Woops, use calculator or engage brain - take 4 1/2 away from 12 instead of add it on  sorry!

Quote

I will try to do better with the screen capture but am using windowed wine

It wasn't a criticism, I understand about running under wine may complicate things, don't worry.
Try the quiet line test again as sson as the noise variation re-appear.
I wouldn't waste your time looking for a command line version of Routerstats for your NAS box because of what I'll be suggesting, but I don't think one exists, but  you could check if Routerstats Lite will run under Wine if all your PC's are Linux based.
Are you currently on a desktop or laptop, and do you have another laptop you could maybe put Routerstats Lite on?