In reply to a post by Cycle194:

I have previously set up a Raspberry Pi to monitor the number of connection drops I receive in a single day. It only checks every hour and so 24 is the maximum number of drops it will report. As a guide, the connection has been dropped the following number of times recently:-

3rd of December - 7
...

Is your router listed in the dropdown for RouterStatsLite? I used to run that 24/7 with the sampling interval set to the minimum, (5 or 7 seconds), and IIRC only an hour and a half or so per graph, to make sure it properly registered all disconnections. I set it to save the graphs and also a csv file.

Post deleted by Cycle194

It does RobertoS, thanks for the suggestion. What kind of thing would I be looking for? A significant increase in noise before disconnection?

You can see the noise pattern - sudden spikes, or rapid/slow rises and falls. Any regular grouping. Also easy to see if things like turning off the central heating (or at this time of year Christmas lights) have any immediate effect.

Plasma TV's, Sky boxes, DECT phone power supplies when plugged into the mains are frequent suspects. Less common, electrical exercise machines, fluorescent lights. All the above also applying to the same things at a close neighbour's.

Uploading a few graphs for us to see, or copy and pasting relevant sections of the csv file can give us ideas.

Edited by RobertoS (Thu 04-Dec-14 18:42:37)

Hello, Going by your router stats you are a similar distance as me from the exchange, ~4-5kms.
I found my DSL performance was dramatically affected by the house wiring, that was frankly awful. I have also found DSL filters vary from total cr*p to good, there seems no reliable standard. FYI the ones I use are the DSL4132003 that used to be supplied with the speedtouch modem, these have a big enough ring capacitor for several phones and a double section balanced filter, I have never had one fail.
You have not mentioned anywhere if you have tried plugging the modem/filter directly into the test socket at the master socket (so doing disconnects all house wiring) note that you will have to remove the lower part of the faceplate to do this and your phones wont work BUT if you can stand that a while and see how your modem performs.
Also do you know that your master socket is where BT installed it on the end of the drop cable ? in my case I discovered some fool had extended it using split pairs!!

Thanks again for the suggestions all.

I'll set up RouterStatsLite RobertoS and report back here in a few days. Hopefully I'll have some useful data available.

Fourtytwo1, my connection has been plugged into the test socket for some time. Given the issues I was having, I wanted to ensure that there was as little to go wrong as possible during debugging Although I agree with you on the microfilter quality, I also don't think the issue can be related to those as I've tried so many, both branded and unbranded with exactly the same results.

Interestingly though, I was about to write I only have one telephone box in my house as I've only used it and then I remembered I had seen another around that I'd never touched. I just plugged a phone into it and I got a dial signal so it's working. It has the same phone number as the main box, so it's on the same line. I could hear a lot of interference and electronic signals so that's definitely interfering, but there's never been any phone plugged into it and it would have been the same from the start, even when the line was working perfectly.

You did say however that using the test socket would disconnect all other phones? But this one appears to still be working. The split seems to have been done outside too as the second box is wired through an upstairs window and then a rather long distance to another room. Could some degraded wiring to this box be causing the issue?

In reply to a post by Cycle194:

Could some degraded wiring to this box be causing the issue?

Almost certainly and it's wired incorrectly, not being isolatable on the master faceplate.

When you plugged a phone into it and heard noise did you remember to insert a filter before it?

In reply to a post by Cycle194:

The split seems to have been done outside too as the second box is wired through an upstairs window and then a rather long distance to another room. Could some degraded wiring to this box be causing the issue?

This split before the master socket almost always causes a considerable loss of speed, (up to 2.5Mbps), but doesn't normally cause disconnections after a while. From Day 1 if it's going to do that.

It's called "star wiring", which was fine before we had internet connections.

So it now looks as though you have two problems:

1) The annoying disconnections.
2) The loss of speed you weren't aware of.

I'll leave others to continue for now.

Edited by RobertoS (Sun 07-Dec-14 12:06:06)

Interestingly though, I was about to write I only have one telephone box in my house as I've only used it and then I remembered I had seen another around that I'd never touched. I just plugged a phone into it and I got a dial signal so it's working. It has the same phone number as the main box, so it's on the same line. I could hear a lot of interference and electronic signals so that's definitely interfering, but there's never been any phone plugged into it and it would have been the same from the start, even when the line was working perfectly.

You did say however that using the test socket would disconnect all other phones? But this one appears to still be working. The split seems to have been done outside too as the second box is wired through an upstairs window and then a rather long distance to another room. Could some degraded wiring to this box be causing the issue?

Ohhh dear this is the sort of thing I suspected!! all this additional wiring will act as aerials for RFI and cause multiple reflections for the OFDM signals from the modem!! yes it will cause speed loss BUT it will also cause droputs due to fading snr, and there will be particular sensitivity to incoming calls as a consiquence of the ringing signal being on an umbalanced wire on all this additional cabling.

It sounds to me as if your house-wiring needs a complete overhaul, however I understand BT can be rather expensive, if you dont mind that then they are the best resort but otherwise find someone competent who knows what pairs are etc

There should be no junctions (branches) between the street and where the modem/filter is connected to the line, what happens AFTER the filter inwards to the house is within reason not a problem providing a balanced impedance is maintained.

As for fading snr this is best spotted using a tracking program, i use dslstats, its free and very good BUT you will have to check if your modem is supported, it draws a graph of your snr over time so you can see if and when you have fades and glitches then you can tally them to events.

I had two laptop power supplies causing problems that I found this way BUT the poor state of the internal phone wiring was the real cause as that made the line more sensative to the inteferance than it should have been.

Good luck

Thanks for all the information, it's helping me to diagnose the issue. I'm doing some more investigation in the meantime.

I should have mentioned previously that I had not plugged a filter into the second telephone box and so the electrical inference was likely to be the ADSL connection (i.e lots of retro 56k modem type noise from the early-90's . Anyway, I unplugged all devices apart from a wired phone and did a quiet line test. The second telephone box seems to be slightly noisy than the main one, but it's still pretty quiet. I can only hear a very low volume hiss and some occasional clicking noises. I can't hear anything at all from the first box, but the location is slightly noisier, so I can't be sure.

I should say my mobile stats have improved since my initial post. From experience, it's just temporary though The SNR Margin seems much higher at the moment. I tried to trigger an issue by calling the phone several times, but it seemed to have no effect.

Operational Mode - G.DMT
Upstream - 800 - 800
Downstream - 4000
SNR Margin(Upstream) - 11
SNR Margin(Downstream) - 8.9
Line Attenuation(Upstream) - 28
Line Attenuation(Downstream) - 53

Interestingly, the SNR dropped to 6.3 just five minutes after reconnecting, although this was before I started testing calls. It's up to 6.9 now. As a guide, I've seen downstream speeds of 5000 as a maximum in the past.

I've started using both RouterStatsLite and DSLStats so I should be able to post some results soon I hope. I realise they overlap somewhat, but I started with RouterStatsLite and DSLStats seems to do more, so I figured I'd leave them both running.

I'm considering calling out BT again to remove the connection to the second box (only the cost puts me off, which is why I'm hoping to get as much a guide as possible if it'll help or not). Presumably, this would resolve that particular issue? If it's just a matter of disconnecting it, then it should be fairly simple, although it'll require a ladder, the line seems to originate near the roof. I should say that most of the long wiring on the second box is internal to the house and wouldn't need touching I think if it was disconnected from the source.

I'm not entirely sure that the second box is causing much of a problem though apart from a potential loss of speed. But given I can't really explain the problems I'm having now, anything is possible