Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[waltergmw]

Hello everybody,

No soner do we fix one line problem, than another very strange one appears !

This one might have a "Likely metallic bearer fault" on another very long line that has just been de-DACSed.
Yes, I have checked that the two output lines from the DACS unit have both been removed although the input cable hasn't.
(@ezzer - this was a nice one for BT as we ceased the other line at the same time.)

Three times in 24 hours we've had an unexpected BT Speed test result as shown below.
On other occasions, immediately afterwards, the BT test logic works perfectly.
Can anybody shed any light on what the BT error might be?
Three, different, times in 24 hours the line has dropped and re-synced at the same 704 kbps within 15 to 30 seconds.

Noise margin values are quite good for this distance usually staying constant at 7 dB + / - 0.5 dB and often being absolutely constant for 5 or 10 minutes. Line attenuation is said to be 62.5 dB but as the house is another 30 m on from the last one which had an attenuation of 82 dB at 300 kHz we know the 62.5 figure is grossly under-stated. We're using the standard Thompson 585V7 modem.

BT speed test results vary between 34 kbps to about 350 kbps - however these results must be observed with care.
I'm doing the tests remotely using logmein and logging off immediately I've started the test.
I also don't know if the end user is using the line at the same time.

Kind regards,
Walter



[attachment deleted by admin]

[orainsear]

Could be something as simple as a poor joint/damaged cable along the route.  What's the weather been like over the period of dropouts?

[waltergmw]

The weather has been absolutely perfect in sunny Surrey !
The fact that the line seems so stable including the near-constant noise margin suggests some configuration over-sight within the exchange or possibly the deDACSing in the exchange was only partially completed ? (It wan't done at all to begin with and we had to raise a telephone fault to get it fixed.)

Kind regards,
Walter

[kitz]

You do sometimes see some weird results if the tester is busy.  Ive seen similar message on my own line before now, yet it later worked ok.

[jeffbb]

Hi
I had same last week some time .was there for the whole evening . Yet OK next day .
Regards Jeff

[waltergmw]

Hi Gentlefolk,

Here's another anomaly on this problem. Any ideas would be welcome.

The line reports quite a stable performance but it drops and returns usually within 15 to 30 seconds.
As it had dopped overnight down to 608 kbps but that the noise margin line was absolutely straight I decided to remotely re-boot the modem. It returned at 640 kbps but the noise margin line continues straight at the same 7 dB level.

I assume this must be down to small changes not being seen. (Thompson 585 V7 Routerstats line sampling at 15 secs and the 585 only showing + / - 0.5 dB.)

The first picture is the reported noise margin and the second the speed.

Kind regards,
Walter



[attachment deleted by admin]

[roseway]

7 dB is a slightly odd value, unless the user is tweaking the target noise margin with DMT or something like that. In any event, the router will always start up from a re-sync with the same noise margin. You're probably right about the .5 dB reporting increments explaining why the noise margin doesn't change over time.

[waltergmw]

@ Roseway,

Here's another strange but opposite pattern I've observed several times on this line. The speed stays constant at 640 kbps but the noise curve steps down before returning. We do know the line is behaving erratically and dropping for no apparent reason. We're trying to eliminate the problems as the line continues with a bRAS of 350 kbps when the reported speed never drops below 608 kbps (i.e. above 576 kbps so it should eventually go to 500 kbps) yet it's next door to the other long line where we eventually achieved a bRAS of 750 kbps.

Kind regards,
Walter

[attachment deleted by admin]

[roseway]

I think that the apparent line drop in the middle is just a telnet glitch - from time to time Routerstats is unable to get a sample because the router is otherwise occupied at the time, so it reports zero values. That's why the speed doesn't change - it didn't actually re-sync.

[waltergmw]

@ Roseway,

That seems a very good theory. However I'm still puzzled by the likes of this next graph where there are two apparent drops around 12:26 to 12:28 both lasting somewhere over 15 seconds to perhaps 30 or even towards 45 seconds. The second one clearly was a re-sync where the speed will have dropped down to 608 kbps, but not the first.

(The second set at 12:55 does agree with your theory as I did a remote re-boot to regain the higher speed and the second spike probably is a spurious report whilst the initialisation software was still busy.)

I'd very much like to find out what causes the real drops but perhaps Routerstats is just too crude a tool?
I've also run DMP concurrently with Routerstats with some trepidation and very occasionally I've seen DMT report data unavailable.

Can anyone guess how often you could reliably sample on Routerstats Lite on Netgear and Thompson modems ?

I do realise there are almost certainly race conditions and when there's a lot of bit-swopping or many FEC and HEC errors being recovered the processor idle time will drop significantly.

Kind regards,
Walter




[attachment deleted by admin]

[roseway]

I would only be guessing, but the apparently random dropouts could be caused by something like a dodgy fridge/freezer thermostat or even a poor plug/socket connection somewhere.

[orainsear]

Is there anything in the router logs that might help?

[jid]

I used to have this problem and it was because the time interval at which RouterStats was getting data was too quick for the router to handle.

I set it to sample the stats every 7seconds.

Try changing that to 7 or higher say 10secs and see how that goes.

Otherwise, as Eric says, its most likely a form of interference.

[waltergmw]

Re the above discussion I've kept the sampling interval at 15 seconds and run DMT tools at the same time. The Routersts drops do correlate with the DMT tools No. of resets so I'm hopeful that the Routerstat drops in noise margin to zero are real ones.

Sadly the fault rectification process still drags on with no positive result yet. However we have discovered one cause of the drops as described below. The woosh test does also report a possible metalic bearer fault. Bearing in mind the ADSL service was reported as being fulfilled whilst the DACS equipment was still connected I do wonder if some aspect of the de-DACSing process has been overlooked ?

Can Ezzer shed any light on this problem as the local BT engineers are quite obviously struggling with this fault ?

Kind regards,
Walter

_________________________________________

There is more than one trigger for the sporadic loss of sync as the modem sometimes drops in the early hours when no calls have been received.

We have definitely proven both from the BT O SSFP, and from the test socket with a known good filter, that the modem drops on every incoming ring including option 1 ring back, whether or not the phone is taken off-hook. Tests were done from an ordinary corded phone and a DECT phone.

When the quiet line test was done directly into the test socket without a filter a fast low volume ticking can be heard sometimes. With the separate filter in the test socket a very low volume aproximately 50 Hz hum can sometimes just be discerned.

[waltergmw]

BT have been trying to eliminate this problem but are still having difficulty.
They have done a "Lift and Shift" in the exchange but ringing the number still clearly demonstrates that the fault is unresolved.
Sadly it looks as if they will have to examine the 7 km of this particular line for some sort of high resistance fault.
Some parts of the cable have reasonable charasteristics as the next door neighbour (The subject of my last posting on this topic here) is obtaining a constant bRAS of 750 kbps.

Kind regards,
Walter