cancel
Showing results for 
Search instead for 
Did you mean: 

Disconnections and slow speed

Pettitto
Plusnet Alumni (retired)
Plusnet Alumni (retired)
Posts: 6,346
Fixes: 5
Registered: ‎26-11-2011

Re: Disconnections and slow speed

The line isn't banded, I'd have certainly mentioned that Wink
When you're used to seeing low sync rates at around 180K in speeds, you'll see why I referred to the sync rate being 'lower' Tongue
Anyhow, I've attached full line details to this post.
The line has started to error again...
ratbag
Grafter
Posts: 369
Registered: ‎01-08-2007

Re: Disconnections and slow speed

Thanks Kevin and Chris.
I've installed routerstats. The manual is long, can't go into it at the moment. Feel free to let me know what settings are relevant for the case at hand!
I've replaced the wiring between the main socket and first extension now, can't test at the moment obviously. I have a question though: I bought a new socket box because I was getting frustrated with trying to get the wires to insert in the old box using the insertion tool I bought along with the wire. I got it to work in the main socket front plate though. Anyway, the socket I bought has a large capacitor on the circuit. I did a quick search online and read that these shouldn't really be used as an extension socket. Is this the case?
Thanks again,
Barry.
Townman
Superuser
Superuser
Posts: 22,921
Thanks: 9,538
Fixes: 158
Registered: ‎22-08-2007

Re: Disconnections and slow speed

Hi Barry,
The faceplates with a large capacitor is a master socket and you are quite correct, they should not be used for extension sockets.
Routerstats - I'll get back to you on that directly by listing all of my settings for you.

Chris,
The line figures are "interesting"!
The DP/DN Line lengths (3550 metres) are markedly different to the length implied by the modem reported attenuation (2800 metres).
The AE / BE Capacitance figures vary by more than 1% (BE/AE variance is 1.6%) - does this imply an unhelpful imbalance on the line?
Am I correct in believing that the DS SNRM reported via the exchange is obtained from the user side modem?  I note that the SNRM reported when you collected the stats (5.1dB) is 2dB below the figure reported when Barry collected the stats both at the same synch rate.  Would you agree that this could be indicative of REIN?
Hopefully if I can get Barry sorted out with routerstats we can get a better view of what's happening here.

Kevin

Superusers are not staff, but they do have a direct line of communication into the business in order to raise issues, concerns and feedback from the community.

Townman
Superuser
Superuser
Posts: 22,921
Thanks: 9,538
Fixes: 158
Registered: ‎22-08-2007

Re: Disconnections and slow speed

Barry,
Routerstats - this rather needs to be left running 24x7 which you might find disconcerting!  This program looks more complicated that it really is... once you get the hang of it!  Cheesy  The following is a walk-through of all the setting up bits needed to get you up and running quickly.  Note that I am using a TG router so some settings might not be quite the same / need to be different.  If we encounter some of these, we'll find a solution!
Configuration - Cog Icon opens a window with 5 sub tabs
Router Login & Logout
Router Make and model - select that most applicable to your Netgear router
Login
- Router address "telnet://router-address" for the Netgear I think this should be "telnet://192.168.0.1"
- Login ticked
- Router does not require a username unticked
- Username / password as appropriate
Graphs
Y Axis limits & colours
RX Noise
- Adjust axis auto unticked
- x10 unticked
- Max 12
- Min 0
Rx Synch Speed
- Adjust axis auto unticked
- x10 unticked
- Max 12000
- Min 0
Remember... when closes ticked
Ignore the other two for just now
Common Settings
All graphs
- Sample every 10 seconds
- Points per page 720
Capture properties
Capture graphs (http)
- tick all of them
Schedule (Capture each page)
- Always ticked
- Capture ... when ... closes ticked
Overlay options
- Rx Speed / Rx Noise / Attenuation all ticked
extras
All as per defaults
System
System options
Tick only Start after 10 seconds / Show Rx Noise on icons / Prompt before closing
Right hand side tick all aside from the first two (decimal separator & show time in GMT)
Log
File format
- Enable ticked
- Start new log each day tickec format yyyy-mm-dd
- Tab separated values
- Date/Time format - ddd dd mmm yyyy hh:nn:ss
Schedule
- Log every sample (10 seconds)
Additional
- IP Address ticked
Email
Ignore!
Click "Apply & Save" the "Close"

Now working through the Graph Tabs - Select each tab and then individual graph in turn - right clicking on the graphs exposes a configuration options list... tick those identified below
Noise Margin
Downstream (Rx) - This is the one we are most interested in
Overlay Rx Speed (this means you can ignore the "Synch Speed" graphs
Scheduled Capture
Make available to webserver
Include in log
Highlight Gaps
Upstream (Tx)
Ditto
Bit loading Options tab
NB: This might not apply to your router!!!
Advanced Telnet Options
- Refresh every (ticked) 10 seconds
- Restore ticked
- Telnet port 23
- Telnet timeout 2000 msec
- Router timeout 15 minutes

You are now ready to rock and roll!  Just click the green "play" icon.

What we now need to watch is the "Noise Margin Downstream (Rx)" graph.  The program will monitor the modem's status every 10 seconds and plot the current values - the only thing you are going to see change outside of a resynch event is the SNRM - which is exactly what we are looking for.
The active dynamic graph shows you the most recent two hours of history.  Every two hours the program "snap-shots" all of the graphs to the RouterStats directory, so you can go and look at the graphs you've missed to see what happened during the night.
What are we looking for?  I have attached below three of my graphs - the one dated 9th July is nice and clean and what we hope to find for you - rock steady SNRM.  The one dated 11th Jan shows a line having lots of jittery noise with some rather nasty spikes (in this case caused by passing electric trains) and finally the one dated 21st Jan shows some disconnects and resynch attempts - note the different SNRM levels and synch speeds.
HTH
Kevin

Superusers are not staff, but they do have a direct line of communication into the business in order to raise issues, concerns and feedback from the community.

ratbag
Grafter
Posts: 369
Registered: ‎01-08-2007

Re: Disconnections and slow speed

Thanks Kevin, you're very helpful.
I set it up as you described and left everything else as default. I didn't have the bit loading options tab but you said it might not apply anyway. I didn't see an 'overlay speed' option on the Tx tab context menu. I don't like leaving stuff switched on especially when no-ones in, so hopefully we won't need to keep this running too long. I changed the laptop power options so it wouldn't turn off the hard drive or go into standby. It's running as we speak.
I ordered a new socket online, wasted £4.20 on the other one. The packaging said nothing about being a master socket, all there was was a MS logo on the front which I thought was just some brand logo. Should have just got one online to start with (much cheaper) but wanted to get the job done.
Thanks again,
Barry.
Townman
Superuser
Superuser
Posts: 22,921
Thanks: 9,538
Fixes: 158
Registered: ‎22-08-2007

Re: Disconnections and slow speed

Quote from: ratbag
I didn't see an 'overlay speed' option on the Tx tab context menu.

Barry,
You are very diligent in following guidance - that was an error on my part in trying not to repeat text where not required.  This is one nice to have which the tool does not do.  You can combine SNRM and synch on DS (Rx) but not US (Tx).
What do your graphs look like?  To attach graphs, cluck the "Additional Options and Attachments" link below the reply window.

Technically fitting your own master socket is illegal, which makes selling them questionable.  You might try arm twisting along these lines to get your money back.
Kevin

Superusers are not staff, but they do have a direct line of communication into the business in order to raise issues, concerns and feedback from the community.

ratbag
Grafter
Posts: 369
Registered: ‎01-08-2007

Re: Disconnections and slow speed

Hello Kevin,
It has saved some image files in the program folder. I hope I've attached the one you're interested in. Is it okay if I set it to stop creating the 'user defined' ones?
Many thanks,
Barry.
Townman
Superuser
Superuser
Posts: 22,921
Thanks: 9,538
Fixes: 158
Registered: ‎22-08-2007

Re: Disconnections and slow speed

Hi Barry,
NICE  Cheesy  You did well!  For the time being we can switch off the user defined graphs, though there might be merit in looking at using these for plotting error counts later on.
OK now you've got it working, we need to attempt some detective work.  When the SNRM line goes up, some source of interference has gone away, when it goes down a source of interference has "woken up".  A sharp change such as the one at 10:15 indicates something which is a black and white noise generator being switched on / off (in this case off).  A spike / dip such as the one at 11:13 either indicates a source of noise which has a start-up burst and then settles down (typical of anything drawing a large start-up current) of possibly a passing external source.
Are you anywhere near an electric rail or tram line?  
Can you think of anything switched OFF at 10:15 / 10:30 / 10:58 and / or ON at 10:50 / 11:13 - for me the time pair 10:50 / 10:58 as the noise level at 10:58 returned to what it was at 10:50.
I think there could be a couple of noise issues here (0.6db - 5.7dB to 6.3dB / add 0.5dB -6.5dB to 7.2dB / 0.2dB sinewave modulation between 11:15 and 12:05),  in addition to any line problem - indeed the noise pick-up could be caused by phone line problems...
Quote from: Townman
Chris,
The line figures are "interesting"!
The DP/DN Line lengths (3550 metres) are markedly different to the length implied by the modem reported attenuation (2800 metres).
The AE / BE Capacitance figures vary by more than 1% (BE/AE variance is 1.6%) - does this imply an unhelpful imbalance on the line?

To help see if there is a correlation to anything you are doing in the home, you could set up an alarm on SNRM variance.  I suggest you set the variance to 5% on both increase and decrease and the sample count to 6 (i.e. within the last minute.  Then set the alarm action to Audible (test the sound) - the klaxon can be heard right through my home much to SWMBO's disapproval!
Kevin

Superusers are not staff, but they do have a direct line of communication into the business in order to raise issues, concerns and feedback from the community.

ratbag
Grafter
Posts: 369
Registered: ‎01-08-2007

Re: Disconnections and slow speed

Hello Kevin,
I've set up an alarm (I think). I checked the bottom 2 boxes and set to 5% each and 6 samples and turned on 'audio alarm' in 'alarm actions'.
There's no tram or rail anywhere near here. The only thing that comes to mind is the fridge and the freezer. I turned my main computer on to do some work, that's upstairs. Nothing else is on that I can think of.
EDIT: I turn the laptop screen off when not using it. I noticed the noise margin go up with the screen off and go down with it on. Perhaps a coincidence. Will try a more 'scientific' experiment by making a note of the time I switch it off then on again.
EDIT2: Yes, switching the monitor on/off causes an alarm and you can see on the graph that it's higher when off. I suppose I should leave the screen on as well. Going to be hard because it's a force of habit!
Barry.
Townman
Superuser
Superuser
Posts: 22,921
Thanks: 9,538
Fixes: 158
Registered: ‎22-08-2007

Re: Disconnections and slow speed

Barry,
Following that observation through - if you confirm the correlation, can you try doing the same with the laptop running on its battery?  Try just switching off the laptop PSU and see if that makes a difference / shows anything.
Current Switching Power Supplies (aka switch mode PSU) have a habit of "going south" after a while, some of the electronic components degrade and cause them to emit RFI, especially when that have warmed up.  To all intents and purposes they appear to work but in doing so they are pushing out lots of RFI.  Two personal experiences.
1. I was noticing a marked dusk - dawn SNRM shift coincident with the street lights switching on and off.  The inbound phone wire crossed (touched) the OH street lighting power lines.  Having that sorted rectified a good proportion of the problem, however not all of it.  Having then spent weeks looking for logical sensible sources of RFI - fridges, freezers, coolers, air conditioning, pumps, compressors fans, fluorescent lights... - I was left with the "just could not possibly be" devices, one of which was a CCTV system, which was switched on all of the time and thus could not be the culprit.  TO say the least I was shocked when I disconnected it to see that there was an instant rise in SNRM.  Powering it up lead to a slow decay of SNRM as the PSU got warm again.  Further investigation identified that the dusk / dawn variation was due to the additional current required to drive the night time IR emitters.  The PSU was going marginal under the increased current drain.  Replacement of the PSU eliminated the issue.
2. I'd been trying out a NetGear router in place of the PN TG582n for the benefit of the former's dual band WiFi.  There were a number of problems so I was switched back and forth between the two.  Due to DNS issues, I ditched using the NetGear... but due to difficulty in accessing the socket, I left the PSU connected.  Some weeks later when convenient I removed the unused, but powered up NetGear PSU - I noted a 1.5dB rise in SNRM as a consequence.
Your laptop PSU could be getting stressed - the extra current required to drive the display might be causing interference.  Running on the battery would help to test that theory, though it might be the screen light electronics themselves causing the issue.
Kevin

Superusers are not staff, but they do have a direct line of communication into the business in order to raise issues, concerns and feedback from the community.

ratbag
Grafter
Posts: 369
Registered: ‎01-08-2007

Re: Disconnections and slow speed

Hello Kevin, thanks a lot, I appreciate all the time you're spending on this.
First, I don't know if you saw my second edit before posting your reply, but I confirmed the monitor affect the results.
The laptop is an old one I was given, probably about 9 years old, but good enough for everyday use. I had to buy a power supply for it and, of course, it's Chinese. Just unplugging the power supply made the noise margin jump up to about 8dB and has stayed around that point since. I will try the monitor test again with the battery only and report back. Not sure how good the battery is, I don't use it in battery mode generally
Barry.
Townman
Superuser
Superuser
Posts: 22,921
Thanks: 9,538
Fixes: 158
Registered: ‎22-08-2007

Re: Disconnections and slow speed

Barry,
Well I guess I must be pushing your sceptical tolerance to the limit - first the impact of disconnecting the bell wire - now identifying a suspect laptop PSU.  You are becoming an expert in sorting out internal sources of poor ADSL performance.  Cheesy Grin
You said the jump was up to 8dB - from what previous level?  Permanently eliminating 3dB of noise might deliver up to 1.2Mbps improvement in synch speed.
Do keep an eye on the laptop battery, the last thing we want now  is for the battery to die and the laptop crash.
I rather think that you might need to consider putting a branded PSU onto your shopping list.  The important thing here is to identify RFI sources and to measure each source's contribution to the problem.  As I understand it, you have identified an impact from the screen (on battery as well as mains) and from the PSU itself (was that the same when powered on but not plugged into the laptop?).
For the moment, can you record the difference made by each of these sources please?  After which you should put the laptop back on the mains.
When the laptop is thought to have a good charge, it would be interesting to switch off the PSU and restart the router with routerstats already running.  Capture the routerstats through the web interface, then post back here.  Watch the routerstats plot for say 30 minutes, then put the laptop back on to the mains and see what happens.  From that we should be able to work out if there are any other sources of REIN.
... and get a view from Chris on my question of AE/BE in-balance.
Cheers,
Kevin
PS: I'm not going to be around the forums much at all from later today until Tuesday and then not for a full week.  It would therefore be good to reach a conclusion on where next with DCT by COP today or next Tuesday.

Superusers are not staff, but they do have a direct line of communication into the business in order to raise issues, concerns and feedback from the community.

ratbag
Grafter
Posts: 369
Registered: ‎01-08-2007

Re: Disconnections and slow speed

Thanks Kevin,
It all went a bit pear-shaped. The monitor test failed without the PSU because the power options for battery mode are different and I hadn't modified them and so it put the computer into sleep mode without me realising it. Second, I had another internet disconnection around the same time. Third, since the disconnection the noise margin has dropped down to around 6dB again (where it was before) even though the PSU is unplugged. I'm going to have to put it back on mains, recharge the battery and follow your advice above. Will post back, hopefully in time.
Thanks again,
Barry.
ratbag
Grafter
Posts: 369
Registered: ‎01-08-2007

Re: Disconnections and slow speed

Here's the plot showing the 'sleep mode' and internet disconnection disasters, as well as the drop in dB when the PSU was plugged back in (end of the graph). So it looks like you are right about the PSU, only the thing has dropped a couple of dB or thereabouts since the internet disconnection from where it has been. It's all fun and games isn't it!
P.S. Just noticed on the graph that the connection speed has increased since the disconnection.
Barry.
Townman
Superuser
Superuser
Posts: 22,921
Thanks: 9,538
Fixes: 158
Registered: ‎22-08-2007

Re: Disconnections and slow speed

Quote from: ratbag
Third, since the disconnection the noise margin has dropped down to around 6dB again (where it was before) even though the PSU is unplugged.

Barry,
That is exactly what should happen - the bit you've not commented on is what happened to the synch rate?  Let me explain...
An ADSL line has a target SNRM - in your case 6dB.  When a line synchs up, the modem (if the line is not banded) will synch at the highest speed which still delivers the target SNRM above the then present back ground noise (from whatever source).  Thereafter the current SNRM will vary if there is a change in the background noise (REIN) as you have observed.
Right now, it would be very interesting to know, if the resynch happened whilst the laptop PSU was switched off and if so how much more synch are you getting now than before?  If the resynch was whilst the PSU was switched off, you will now see the SNRM drop markedly lower than 6dB, possibly as low as 4dB.
Kevin

Superusers are not staff, but they do have a direct line of communication into the business in order to raise issues, concerns and feedback from the community.