Connection drops frequently... REIN?

[#1]

We don't need any bigger picture - don't run before you can walk!!
Actually, the proximity to the electrics should not be such a big issue if it weren't for the bell wire connection to the extension socket!

Quote from: Timmymac

It's not easy to see but it just looks like two wires from the line coming in and 2 going out to the only other socket in the house.

You are right, it's not easy to see, which is why I asked, there did look like the possibility of the bell-wire being connected to terminal 3.

Quote from: Timmymac

There is a third, white wire attached to T3 from the extension. The incoming line is just orange to to T2, white to T5. They are indeed screw terminals.

So, initial description not exactly matching reality :mand:  The lesson to be learnt here for readers, is that pics are always a good idea to supplement descriptions so that anyone who doesn't have a strong technical background can be guided correctly  
It is precisely the extension cable and phone line that needs to be worried about first cedlor - [Censored] in, means [Censored] out
IP profiles & Current Line Speeds are a secondary issue with a dropping connection, forget about them for now, lets just find out if we have an external line issue to resolve.
Don't worry about the cordless phone for now Tim, just follow my previous guidelines for disconnecting the extension and cleaning wires except add the bell wire terminal 3 to be disconnected and leave it disconnected from now on - and also at the extension socket.

Quote from: Anotherone

......... the best way to proceed would be in daylight hours, power off the modem/router. Unscrew each wire from the terminals, and leave the extension wires disconnected. Make sure each wire from the BT incoming cable is clean and shiny and reconnect the incoming cable to terminals 2(white) & 5(orange). - Plug the phone into the filter at the master socket. Check that the line is free of noise (Quiet Line Test) and then power up the modem/router and post the new stats.

I won't ask about the rest of that stuff on the left for now, it will only distract from the primary issue, the mention of landlord makes the picture relatively clear   The stats when you've done the clean up and any SNRM variation will be of the greatest interest. The last posted stats with the 6dB SNRM look a lot more promising, so there is light at the end of the tunnel.

[#2]

Better?

[#3]

Hugely better
It's now a case of keeping an eye on your SNRM variabity. But also that phone line for intermittent noise which will continue to disrupt you much improved BB should it have an intermittent fault.
Your BT IP Profile should now be around 12.3Mbps, but it will take time for your Current Line speed (Login required) to catch up and match it before you'll see download speeds approaching that figure.
But note, if you have any drops in connection, that will affect your sync speed and IP Profile and the knock on effect to the Current Line Speed.

[#4]

Well thanks again for your help. I'll keep an eye on it for a while and try it with the extension wired back in. Line speed is listed as 10Mb at the moment and SNRM hasn't changed.
Assuming we've cracked this, would it be fair to say that it is indeed REIN but that the bell wire is just an unnecessary conductor acting almost like an 'antenna' receiving the interference from the EM field around the electrics?
Tim

[#5]

 

[#6]

Hmm, disappointing, what would be best would be some detailed monitoring on the SNRM to try and discover the type of issue that is dropping the SNRM. Exactly which modem/router model/version do you have?
The bell-wire does indeed act like an antenna as it's not part of a balanced pair and is connected to one side of the line via the bell capacitor if it's connected.
Now whether this is REIN, or something like cross-talk is difficult to say at this stage. Obviously you have a lot of other lines & circuits coming in there, if one (or more) of them is suffering problems (faulty or maybe their own bell-wire issues) the cross-talk could be quite severe.
The proximity to the electrics might not be helping if there's a lot of garbage on it, maybe from some of the poorer quality Home Plug type devices or other iffy equipment. The rule of thumb is not to have the phone line cables running parallel to electrics if closer than 6" - ideally a foot away when a very short parallel run doesn't usually give an issue.
Not knowing what's to the left of all the bundles of cables on the pic in reply #11 it's difficult to make a definitive suggestion, but at this stage I'd be inclined to unscrew the Master socket box from the consumer units board and let it dangle by the phone and (is it ethernet?) cables on the left as a temporary measure. I'll also unscrew that Netgear switch from the board and put it on the floor for now, again try 6" to a foot away from the major electrics stuff.
What is the room that all this is in? I'm guessing you are in a basement flat? Is this lot in your space or a common area other have access to?
Edit: Forgot to mention, do the obvious check that the SNRM doesn't drop when the phone is used and go back again at the end of the call.

[#7]

It's a very small 'utility room' which is essentially a cupboard adjacent to the kitchen. The oven, fridge etc are all on the other side of the kitchen. There's not much in there really. The control box for the underfloor heating is on the opposite wall, about 8' away. This does have wireless thermostats associated with it. Oh and an ironing board but I don't do much ironing.. 😉
The cables are a mixture of things, largely unused. There's cat6 to each room in the house but only one has anything on it (a Netgear Wireless AP). The black stuff is coax from the tv antenna. The rest I don't know much about but I believe it will be for some sort of security system one day. Currently not connected to anything.
I have had a separate modem and router for years. Used to run an old linksys router with tomato fw installed and so had a separate modem and just stuck with it I suppose.  Currently the modem is a TP-LINK TD-8616 and router TP-Link Archer C7.
I'll move those things as suggested.

[#8]

Ok, so as we stand at the minute, I've got that master socket hanging a foot or so away from the major electrics, amongst some largely unused wires. The extension has been removed totally. I've got a filter in place and the phone/modem both plugged in to the master socket.
Things seem much the same unfortunately.
I've also just bodged together a little perl script to scrape the stats from the modem every 10mins and squirt them into a spreadsheet so we can look for some patterns in the SNRM variability.  Results are here:
https://docs.google.com/spreadsheets/d/1ysShHe3TjG8iB48uU-Po6aofQYJBj8ZiEVwtCTg98og/edit?usp=sharing

Tim

[#9]

OK until a pattern developes can you post the reslt of a BTW speed test and the  PN line speed test  - link below

[#10]

Quote

Estimated line speed: There's no speed estimate currently held on your account. Current line speed: 10 Mb

Seems pretty reasonable! I don't always get that though....  🙂
Latency a bit high?

[#11]

Looks like something is holding it back -did you get the  IP linespeed from BT as well - sometimes it shows as higher than it does at PN in which case it may be stuck
or need longer to update at PN.  The  10 at PN may be holding it back.or maybe its just evening noise or a contention issue. The results in daytime on your spreadsheet
might give more clues.

[#12]

There's nothing holding it back, just the Current Line Speed doesn't appear to have updated still. You can't rely on the latency figures from the BTw speedtest as an accurate assessment of the line either. The results will depend entirely on how busy everything was between the computer and the test server at the time and can be quite variable.  However they do suggest that Interleaving is turned on.
@Tim
Well done for the perl script.
I wouldn't worry about speedtest results atm, if your sync speed changes so will the IP Profile and eventually the Current Line Speed.
I see the sync speed has increase fractionally since your previous stats which I assume was maybe due to a disconnect whilst you moved your kit etc.
I had in mind some more intensive monitoring of the SNRM and your perl script is very useful. I was going to point out that where you see significant changes in SNRM, in daylight especially, remember to check for phone line noise - use the Quiet Line Test if need be. Expect some variability between dusk and dawn but how much you get will depend on how much your line is susceptible to the increase in MW/AM propagation at these times.
Having said all that, the few results in the spreadsheet at this time suggest two things -
1) your line is not especially prone to MW/AM propagation interference - the figures after dark are generally too stable - I assume the the line is probably totally underground from the exchange, and other than any possible intermittent connection it would seem well balanced.
2) there looks as though there could be a pattern and the SNRM seems to drop by a fairly consistent amount when it does.  Apart from eliminating the possibility of line noise/faulty line (regular checks with the phone) I'm guessing this does look like a source of REIN whether directly affecting your line (and possibly others) from something very local, or maybe due to some faulty equipment on another (faulty) line - ie crosstalk - I'm not thinking this latter possibility is the most likely at this point.
As you've mentioned, looking for patterns will give the first clues as to possible causes and hence tracking it down. Those few results do suggest something happening at specific times for a possibly defined period.
I don't know how CPU intensive your perl script is, but you've obviously appreciated that a computer has to be on to do this sort of monitoring, but results so far suggest the sampling needs to increased, at this time I'd suggest 1 minute intervals to see if there is more of a defined pattern.
I initially had in mind the possibility of monitoring with programs such as RouterStats/RouterStatsLite.  However, your modem is not listed in the RSLite setup (the one that can be plug&play) but the 8816 is listed and the stats in your modem may be in the same location so using that setup may be worth a quick try. If it doesn't work, you can choose a User-defined setup.
Setting up RSL is far less complex than the full RouterStats, but won't work with all modems so you may have to try the full RS. I'm sure setting one or other up won't phase you in light of you writing perl scripts, you know about locating the position of the various stats on the relevant page.
The advantage of RS/RSL is the graphical output which can give a much easier to interpret picture than a spread sheet - but it's no real issue at present - increasing the sampling rate is more important right now.
If you give RSL/RS a try and have any queries about setup, please ask away. Initially I'd go for 1 minute sampling and 240 Points per page - ie a 4hr graph being saved when it's running.
Depending on what we see, the sampling rate (and points per page if using RSL/RS) can be adjusted accordingly.
(On RSL/RS a right click on graphs gives additional options - on the SNRM graph on RSL plot the US noise as well as DS).

[#13]

Sampling rate increased. Will take a look at RS/RSL this evening if I can...
Thanks again

[#14]

Looks like the sampling has stopped  at 07.09  pinged in once at 09.13 then stopped again.
@AO      "There's nothing holding it back,"      not sure how you can say that    then      " just the Current Line Speed doesn't appear to have updated still."
That was what I thought I was suggesting , my understanding  (or the lack of it) being that when the IP line speed on the BTW test and the PN IP Line speed dont match
the  actual line speed can not be higher than the lower of the two? which was why I was asking to see both.
Suggestion from PN seemed to be that they get updated every few hours from the dlm? at BT, yet in other posts it has been said it can take up to 5 days!!  - just thinking
maybe that is 5 days to return to full potential  because it is incremental assuming no noise or disconnections.

[#15]

Quote from: Timmymac

I've also just bodged together a little perl script to scrape the stats from the modem every 10mins and squirt them into a spreadsheet so we can look for some patterns in the SNRM variability.  Results are here: https://docs.google.com/spreadsheets/d/1ysShHe3TjG8iB48uU-Po6aofQYJBj8ZiEVwtCTg98og/edit?usp=sharing

Quote from: Anotherone

The advantage of RS/RSL is the graphical output which can give a much easier to interpret picture than a spread sheet - but it's no real issue at present - increasing the sampling rate is more important right now.

@Tim,
A very impressive solution.
@AO,
I was impressed that when looking at the spread sheet Tim posted, Google apps automatically offered to plot the data for me.  A bit rough and ready, but it clearly showed a number of periods of lower SNRM.  I do not know much about the auto graphing (something worthy of research) - possibly something which is only available to those with a Google account login - which I have.
A more frequent sample time and a quick look at the auto Google graphing (Explore IIRC) might deliver a quick win if RS/RSL will not work with Tim's router.