BT have messed up my broadband connection, is there anything I can do?

[#1]

Quote from: Anotherone

I wouldn't swap it over without knowing what the current SNRM is and it needs to be stable at 6 or above with no spikes!

I was thinking that myself.  That means that I cannot really do anything until the training is over.  In fact even if the SNRM were good it would be risky to plug in an unknown modem during the training in case it is not as good as the one I am currently using.  Is there any way I can tell when the training is complete?

[#2]

Yes precisely what I was bearing in mind. Training is 10 days from the reset. The consolation is that a further reset could be done if a disaster happened but I wouldn't try anything tonight, it's getting a bit close to dusk. I'm glad it all held up overnight. If you've got any SNRM plots of last night and/or today it may be interesting to have a look, but otherwise post up a plot in the morning and we could see whether it might be worth trying anything late morning/midday. Which reminds me, In RS go on the Config >Graphs >Capture Properties tab and in the Box labelled Overlay Options, tick the two boxes Overlay current Rx Sync Speed and Rx Noise Margin. Click Apply & Save. This will put the current values at the time of capture at the top of the graph. Test it out by right clicking on the SNRM graph and select Capture now.
I'll reply to your 2nd PM.

[#3]

Attached is noise graph for this morning.  A lot of the noise is still caused by the phone, but it is not affecting the level between the spikes.  If I switch it off it sits mostly at 6dB.  I can switch the phone off while connecting and it would be unlucky to catch a glitch then.  I am hoping the ferrite will come today so will see if that makes a difference.  I am going out this afternoon and whether the ferrite is fitted or not I will switch the phone off to get a couple of hours sampling with it off to check what the 'base' level without the phone looks like over an extended period.

[#4]

OK, all noted. Whilst I think about it in preparation for (just before) changing modem/routers when you close down RS and power down the v3, make a copy of the Routerstats.ini file (in the folder where RouterStats.exe is) and rename it say Routerstats.ar7 so that if you need/want to swap back to the v3 you have all your current settings intact. Then open RS (don't run it) and open the Configuration setup and change all the Login and Telnet stuff for the v4. It should then be ready to run when you power up the v4.

[#5]

I have fitted the ferrite core on the phone lead and it did not make much difference, possibly a small amount but not much.  Having done an extended test with the phone off I am confident that there is not now a lot of noise coming from the phone, I am confident that most of it is coming from outside my environment.  I will see what it is like in the morning but I suspect that I may have to wait till after the training period is complete before doing anything else in order not to jeopardise the 1344 it is currently running at.

[#6]

I have found my notes from a couple of weeks after the broadband was installed (using Metronet then) in 2006.  I recorded the ADSL connection speed but not the SNRM.  The broadband was installed on 15/11/06.  At that time I was switching the modem off overnight as I did not know any better.
28/11/06  2240
29/11/06  2272
30/11/06  2304
01/12/06  2336
02/12/06  2240
04/12/06  2336
05/12/06  2368
I had forgotten that it used to be that fast.

[#7]

You (or indeed someone else) might be feeding RF interference down the mains wires.

[#8]

Mains borne interference is always worth looking at and that possibility certainly would be worth eliminating in the near future, so I might just as well quote exactly what I've put in another thread -
Some potential noise sources to try and eliminate.  Set RouterStats running, leave it going a bit to see what sort of level it's at, then switch off your monitor and if separate, its power supply. Leave it like that a while then switch on again and see if there was any step changes to the noise levels.
Do you have a second computer, if not could you get a friend with a laptop to visit? Put Routerstats (Lite would do) on it. First with only your computer on, run Routerstats and look at the noise level. Then turn yours off and unplug it from the mains. Run the other computer and Routerstats and see if there's any significant difference. You may have to do this check in daytime when the noise levels are more stable, as you are no doubt aware the noise levels after dark can change quite wildly at times.
Reason for the checks - switched mode power supplies can turn noisey and generate a load of trash.
Other thinks to think about are an assortment of wireless devices, fluorescent lights, compact fluorescents, low voltage halogen using switch mode supplies, even LED lights. Boilers starting up, arcing thermostats, motorised things such as washing machines, vacuum cleaners, drills and so on.
Well, a good job that you've got those speed notes Colin. Re the ferrite, try it on the modem/router line (close to the back of it) and  worth a go on it's mains lead as well.
As regards when to swap to the v4, maybe wait until your twisted pair screened modem cable arrives, subject to SNRM conditions at the time.

[#9]

The line was quiet this morning so I decided to go for modem replacement. The 'new' one is a Netgear DG384Gv4, the old one a v3.  It synced at 1184 and around 7.5dB.  The SNRM level on the new one is much less noisy than the old.  See attached SNRM and bits/tone from 1602 this afternoon. 
I have seen a couple of odd effects this evening, the SNRM on the graph drops to zero briefly, but when I look in the sample log I see stuff like
Fri 12 Oct 2012  20:47:52,
Fri 12 Oct 2012  20:47:54, Rx-Noise=0.0, Tx-Noise=0.0, Rx-Sync=, Tx-Sync=,,
Am I right in thinking these are software glitches?
It is a bit noisier this evening, see SNRM graph 2122.  The spikes down to zero all appear to be corrupted data, but the one down to 3.8 is real I think.
Am I right in thinking that these noise levels are now perfectly acceptable?

[#10]

In my experience, a drop to zero that is not accompanied by loss of sync probably just means that the router failed to respond within the allotted time.

[#11]

As was pointed out to me in a recent thread if the SNR drop shown on a routerstats graph is due to a failure of the link between the system and the router the line changes to red. Yours does not do this.
Mine did due to concurrently running inSSIDer

[#12]

I think maybe that is not an infallible rule.  Looking at the last SNRM graph attached to reply #83 most of the spikes are red and so are comms failures as you suggest.  The first one, however, is not red, but when I look in the log I see
Fri 12 Oct 2012  19:06:57, Rx-Noise=6.4, Tx-Noise=15.0, Rx-Sync=1184, Tx-Sync=448,
Fri 12 Oct 2012  19:07:20, Rx-Noise=0.0, Tx-Noise=4.0, Rx-Sync=3, Tx-Sync=4.0,
Fri 12 Oct 2012  19:07:22, Rx-Noise=6.4, Tx-Noise=15.0, Rx-Sync=1184, Tx-Sync=448,
This cannot be real data as there is no way the rx and tx speeds could change from 1184, 448 to 3, 4 and then back in this way.
dick:quote

[#13]

You are certainly right in your last remark there Colin.
Ok lets deal with a fallacy or two, you can't rely on the colour of the line changing to red all this indicates is that the particular plot has gone outside the alarm limits that you have set and there is also the possibility that you may be confused with the fuchsia plot of the average if set up.
I would recommend plotting the average noise margin, but nothing else - totally meaningless for WCB ADSL/ADSL2+.
Reed is correct, but it may not be just the router not responding is may also be the RouterStats program's request or data returned to it, not arriving in a timely manner. This is all dependant on how busy the modem/router or your computer's CPU/HDD is at a given instant. There is a check box setting on the >Configuration >Graphs >Common Settings tab to "Pause graphs when router doesn't respond" but this does only deal with the Modem/router. It may be possible to deal with the CPU/HDD busy situation by raising the priority of the RouterStats task in Task Manager, but as I don't suffer from it very often (and usually not in critical circumstances) I keep forgetting to try it on more occasions.
The rule is, if you get a drop to zero on all graphs and no change in sync speed or margin after the event, it was probably a glitch. It is possible, though not common, to lose sync and resync at exactly the same speed, but remember a resync usually takes around 30 seconds.
As you say, the first drop in the Graph in reply #83 "looks" different, and I wondered whether it was a real spike due to interference, which it seems it wasn't when looking at the log which is why that, and comparison with other graphs you are plotting should be done before making an interpretation. The one to 3.8 does look real. There was static about yesterday, did you hear any thunder at all (you won't always)?
What's the SNRM doing today?
Looking at the remaining graphs in reply #83 there could have been a noise burst from something just at sync as the bit loading is well down over tones 53-70 ish and hence the lower sync speed.
I'd suggest a daylight resync when the SNRM is a nice steady value +/-0.25 at most, preferably +/-0.1 and higher than 6.5, preferably nearer that 7.5 you mentioned.

[#14]

WRT the spurious values on the graph I am running RouterStats on Ubuntu under Wine (which is a Windows emulator).  I have upgraded to the latest version of Wine (1.5) and most of the spurious data have disappeared so I think some of them were due to problems there.  I have only had one spurious today and that was when I opened a browser window on the router admin pages and I think it is a known issue that RouterStats can have problems if there is another client interacting with the router.
I don't think the 3.8 spike yesterday was lightning, the weather was quiet at that time.  It might have been when I switched the dishwasher on, though that is at the other end of the house.  Now that the basic signal is much smoother I may be able to tie down the other events.  Though as long as they only knock a couple of dB off they are probably nothing to worry about too much are they?  As long as they are rare enough that I don't get two at the same time that is.
Although the new router came up at 7.5dB it has not achieved that value again since.  It drifted down throughout yesterday morning and today it has been hovering around 6.5 with no noticeable large spikes so I don't there is much to be gained by re-syncing at the moment.

[#15]

I have just noticed that the router is now showing Rx attenuation at 63.5dB.  Unfortunately I don't know whether this is due to the new router or whether it has changed for some other reason (the v3 Netgear was showing 59 when it was first fitted and the Linksys 58).  At some point I will have to switch back to find out, but I will wait till the training period is complete.  I did not have that value overlaid on the graphs, I do now so that I can watch it.