SNR reset

[#1]

Well I suppose it could be worth a try, but it won't fix the fundamental problem of a bad joint on the phone line. Who owns the conifer tree?

[#2]

I don't expect the suggested set-up to fix any underlying external fault but would help rule out any chance that the cordless phone is introducing any noise onto the line, although I would imagine a DECT phone wouldn't cause a crackle and I have heard crackling when the cordless DECT phone was unplugged.
The conifer is situated at the bottom of a neighbours property 7 houses along and don't think any approach regarding the pruning of the tree will get a favourable response, although his trees do overhang the communal private back and side lane. the telephone pole is situated right on the corner of his boundary.
I would imagine that the tree was much smaller when the telephone pole was installed but has since grown around the line and pole.

[#3]

Well, in which case he could be the one that gets a bill from BT OR is they feel it's the tree causing all the problems! I doubt that your cordless has anything to do with the crackling that you have heard. It just a nice safeguard for yourself to be able to say I have a corded phone plugged into the test socket, can you hear that noise on the line!

[#4]

Drop wire from the pole to our house, casing and insulation worn through by a willow tree in our garden, was replaced free with no question less less than ten years ago. That was before they started to get really stroppy about charging though :(.
The old cable itself was of a lower standard than that ruling at the time, being over 30 years old, so maybe that helped.

[#5]

It certainly would have helped. For good broadband the cable need to be the current spec. CW1411, been in use for a few years now.

[#6]

I meant in terms of getting it done free.

[#7]

Am still waiting for a BT engineer appointment, but made a couple of changes yesterday but left behind the microfilters so could connect upto the test socket as I had hoped to do.
I disconnected the cordeless phone and replaced with a corded phone and also swapped the router power supply over.  Since these changes I have ad only 2 drops connection has now been up for over 11 hours and the SNR graph has been an almost perfect but am unsure if this is conincidental.
I have Routerstats full version and the DMT Tool, can I monitor the effect of switching the cordless phone back or any other changes by using error counting or a bits/tone graphs that would show whether the changes have improved things, ideally something that could be measured over a short time period of say 15 minutes.  If so what errors should I be looking at CRC HEC FEC?
The other thing I noticed yesterday was that the boiler situated around 10ft away from the router was occasionally making a few unusual noises that I was concerned might be causing interference in some way.
P.S.  I have attached a copy of the SNR graph that shows several drops from 30dB to 0dB over short periods but the router keeps the connection up which I initially assumed to be temporary connection issues while running routerstats remotely over the internet, but found the same thing happened while moniotring on location with a wired connection.  Does this indicate anything?

[#8]

The wider one's I'd have expected to be loss of sync, but the narrow ones just where RouterStats has been unable to retrieve the data at that instant, BUT a lot depends on the sample rate etc you are using. If there is an identical spike on all graphs and absolutely no change in sync speed, it's usually the polling issue, but with your low capped speed you'd probably get the same DS sync speed each time. On 21CN the US sync speed may be a better indicator if sync was lost in these circumstances.
Selecting sampling of 8 secs and 900 points per page would give good detail and individual graphs of 2hours. Also on the Graphs >Common Settings tab - make sure the "Pause graphs when router doesn't respond" box is ticked.
All the errors can be of interest, the CRC, HEC and ES can be the most significant, depending on the modem/router the FECs might be, but just do a copy and paste of the stats & error figures into a post. It's much easier to see at a glance what's what and easier to interpret than a text translation!
Whilst you may (or may not) see interference problems from things like the boiler or the cordless phone, they are unlikely to be contributing much to the current problem with the current noise margin. I'd guess that if you see a burst in the CRC/ES figures and a big spike on the SNRM graph when the boiler starts under current conditions, then perhaps you should investigate further.

[#9]

The line has just dropped after 12 hours and will be on my way down to the property now.
I have a sampling of 8 secs but with 300 points per page, most other settings are as per one of your useful posts from the past I came across yesterday, although hadn't spotted the "Pause graphs when router doesn't respond" box
Sync was not lost during the strange spikes so asume it is a polling issue.  As you correctly stated when sync is lost the DS Sync speed remains unchanged but the US sync speed changes, I also check the WAN Uptime which gets reset to 0:00 when the line drops.

[#10]

Have just spent 1 1/2 hours monitoring the stats using both Routerstats and the DMT Tool but unable to get clear graphs of errors like I can on my home connection which gives lovely graphs of CRC error resetting every fifteen minutes, but at the property with the problem the erro reporting seems to be zero, is that likely to be because Plusnet have switched on Interleaving.  Is there an alternative to check for errors, I have mountains of screenshots from DMT and Routerstats Telnet summary page.
In summary I now suspect the old Netgear DG834GT or it's power supply is dodgy.  The DMT graphs looking very strange with the DG834GT but when swapped for two DG834G v4 routers the graphs on both look much more normal and neither suffered the polling issues that the DG834GT displayed.  The sync rate also improved from 239 Kbps to 283 Kbps and the noise margin reduced from 30dB to 12dB
I have now left one of the new router plugged into the test socket via an Excelsus Z-Blocker Z-420UKP2J microfilter to see if it can hold the connection for 24 hours.
The router is now also plugged into a different socket in a different room.
I have attached a few graphs and data for reference.
During the 90minutes of testing the following took place (to help understand the grahs)
11:46am started monitoring - DG834GT (old dogy router) plugged into filtered faceplate.
11:50am - Quiet Line Test
12:01pm - Quiet Line Test
12:12pm - swapped over to DG834G v4 (latest firmware) and microfilter No.1 via test socket
12:19pm - Quiet Line Test
12:27pm - swapped over to alternative microfilter No.2
12:43pm - Swapped router to Netgear DG834G v4 (virgin media firmware) and back to original microfilter No.1
12:51pm - Cordless phone plugged back in
12:54-12:56pm - Quiet Line Test
1:06pm - Swapped router back to Netgear DG834G v4 (not virgin media) as first tested at 12:12pm as no obvious differences between the two routers.
1:06pm - Cordless phone disconnected - will now leave off for another 24-48 hours
1:11pm - Finish
EDIT: DMT data/graphs to follow.

[#11]

More data attached.
P.S.  Line looking stable at 12dB and 283 Kbps uptime over 2 hours

[#12]

Current SNR graphs attached.

[#13]

With the sort of line conditions you've got, I'd expect few errors and if a fault wasn't intermittent (ie permanent) I'd expect apparent stability!
As you've deduced, there seemed to be very little if any difference between the two DG834Gv4's, but what is of concern, is when you swapped from the GT + Filtered faceplate, to v4 + microfilter, where the noise margin dropped from ~30dB to ~9dB. There was no difference when you swapped filters. So whether that first swap triggered a change in the fault conditions, or both v4's have a similar fault or both filters have the same fault, I can't be certain, both seem unlikely but possible
The change in sync speed that occurred just means that there was a slight improvement in the noise at the time which allowed the speed to reach the banded rate rather than just below it, as it was. But nevertheless, I would still expect a much larger margin (around the 30dB that you had) with this low banded speed.
Looking at all the graphs, I think it would be wise just to check that there is definitely nothing else connected to the line especially something without a filter and if the SNRM remains about the 9dB in the morning, is to see if one of the DCT could organise a Copper Test.

[#14]

The strange thing about the graphs is that the GT only allocated 1,2 or 3 bits per tone, but the v4 appears to have allocated a normal number of bits per tone (up to 11), enough so that the v4 should have been able to achieve a normal speed. That might explain the difference in noise margin. I'm not sure what the v4 is doing with all those extra bits, since it's still connected at a tiny speed.

[#15]

There is definitely nothing else connected to the phone line other than one corded phone.  The sky box is not connected to the phone line.  I did have a 5 port Ethernet Switch (Netgear FS605) which was connected to the router via a network which supplies cables to each of the 4 bedrooms for anyone requiring a wired connection to the router but disonncection this yesterday which was added after the router was moved from the hallway to the back of the house where the BT line comes in and I didn't want to reroute 4 network cables.
Since changing the router over yesterday I have only 2 drops:
6:56pm - Caused by the tenant moving the router power supply from one socket to another.
7:21pm - Connection dropped but unable to automatically resync and required the tenant to reboot manually
The connection has been up for 13 hours since - Overnight graphs attached.
While the connection is up I am reluctant to disconnect.
I am unsure if the netgear DG834GT was giving reliable information or was dodgy, I might try it at my home later when I get time to see how it performs.
The V4 routers seemed to be giving better data but disappointed that the noise margin is so low for such a low sync speed.
There were also the strange polling issues that seemed to coincide with periods of high latency on the thinkbroadband latency graph.
P.S.  The large red band on the Latency graph between 12 noon and 7pm was because I hadn't set the new router to respond to pings.