Drop-outs on 1MB connection

I've got a 1MB business connection from Bulldog (BT line) and I've been
noticing a lot of drop-outs during the day for a number of months, so I
finally decided to gather some stats.

Sampling the connection figures at 15 second intervals over a period of 24
hours from 16:11 yesterday afternoon to 16:11 this afternoon, I've found
that over that period, the average SNR Margin was 13.7 and the average line
attentuation was 57.4. The lowest SNR Margin was 9.4 and the highest 16.4
while the highest line attenuation was 59.2 and the lowest was 54.9.

Only one drop out occured (at 22:30) between 19:32 and 08:25, but during
office hours they were more frequent (up to 20 on the day in question),
often clustered together in groups of 3 or 4, occuring every couple of
minutes or so. The telephone line in question doesn't get used much during
the day, and all equipment on the line is filtered - line usage definitely
does not coincide with the drop outs.

This only really started when the line was upgrade from 512Kb after the
limits were relaxed last year. However, given the line is well within
limits for 1MB I don't understand why drop outs should be so frequent. My
next step is to sample the stats over a 7 day period, to see if the
drop-outs occur at the same time every day.

For the record, the modem is a Zoom X3, with an old firmware - there are
dire warnings on Zoom's website about not upgrading to the newest firmware
on top of the firmware version I have, but I haven't been able to find
version 2.4.x anywhere on the web as an interim stage. It is plugged in
directly to the master socket.

Grateful for any comments or advice on how to proceed

Iwan

On Wed, 11 May 2005 21:31:00 +0100, Iwan Davies
<NOSPAM-(E-Mail Removed)> wrote:

>It is plugged in
>directly to the master socket.

is it any better in the test socket at the back ?
http://www.clarity.it/telecoms/adsl_faceplate.htm#part3

may be more interfeence in the office in the day with more kit running
etc.

Phil
--
Tiscali - dialup speeds at Broadband prices.
AOL - the unlimited ISP of choice for heavy downloaders.

On Wed, 11 May 2005 23:14:23 +0100, Phil Thompson wrote:

> On Wed, 11 May 2005 21:31:00 +0100, Iwan Davies
> <NOSPAM-(E-Mail Removed)> wrote:
>
>>It is plugged in
>>directly to the master socket.
>
> is it any better in the test socket at the back ?
> http://www.clarity.it/telecoms/adsl_faceplate.htm#part3
>
> may be more interfeence in the office in the day with more kit running
> etc.
>
The master socket is located in a small room which is largely unused during
the day, but there is a portable TV in there that is used occasionally. The
other thing was that the cable to the modem was a long 10m extension cable
(proper modem cable), so I've replaced this with a shorter cable - now to
find a longer piece of Ethernet cable to run around the wall to the
ethernet face plate (wny the builders couldn't have put an ethernet point
adjacent to the master socket is beyond me...). Unfortunately, moving the
modem to another room would mean plugging it into an extension, which
probably isn't a good idea.

Iwan

On Thu, 12 May 2005 09:19:50 +0100, Iwan Davies
<NOSPAM-(E-Mail Removed)> wrote:

>The master socket is located in a small room which is largely unused during
>the day, but there is a portable TV in there that is used occasionally.

TV might interfere I suppose if its close to modem ??

tried the test socket or is there not one to try ? that isolates the
extensions which otherwise are part of the picture.

Phil
--
Tiscali - dialup speeds at Broadband prices.
AOL - the unlimited ISP of choice for heavy downloaders.

On Thu, 12 May 2005 10:43:38 +0100, Phil Thompson wrote:

> On Thu, 12 May 2005 09:19:50 +0100, Iwan Davies
> <NOSPAM-(E-Mail Removed)> wrote:
>
>>The master socket is located in a small room which is largely unused during
>>the day, but there is a portable TV in there that is used occasionally.
>
> TV might interfere I suppose if its close to modem ??
>
> tried the test socket or is there not one to try ? that isolates the
> extensions which otherwise are part of the picture.
>
Not yet had a chance to try the test socket - or even investigate if one
exists, but I'll give it a go tonight to see if it improves anything.
Thanks for your advice.

Iwan

In article <1iy677tja44qi.1ljl1odihqu5w$.(E-Mail Removed)>, Iwan Davies
<NOSPAM-(E-Mail Removed)> writes
>I've got a 1MB business connection from Bulldog (BT line) and I've been
>noticing a lot of drop-outs during the day for a number of months, so I
>finally decided to gather some stats.
>
>Sampling the connection figures at 15 second intervals over a period of 24
>hours from 16:11 yesterday afternoon to 16:11 this afternoon, I've found
>that over that period, the average SNR Margin was 13.7 and the average line
>attentuation was 57.4. The lowest SNR Margin was 9.4 and the highest 16.4
>while the highest line attenuation was 59.2 and the lowest was 54.9.

Isn't that SNR rather low especially the 9.4, could that be your
problem. Does it drop out more when it's at that level?

<snip>
--
Paul

"Iwan Davies" <NOSPAM-(E-Mail Removed)> wrote in message
news:1iy677tja44qi.1ljl1odihqu5w$.(E-Mail Removed).. .
> I've got a 1MB business connection from Bulldog (BT line) and I've been
> noticing a lot of drop-outs during the day for a number of months, so I
> finally decided to gather some stats.
>
> Sampling the connection figures at 15 second intervals over a period of 24
> hours from 16:11 yesterday afternoon to 16:11 this afternoon, I've found
> that over that period, the average SNR Margin was 13.7 and the average
> line
> attentuation was 57.4. The lowest SNR Margin was 9.4 and the highest 16.4
> while the highest line attenuation was 59.2 and the lowest was 54.9.
>
> Only one drop out occured (at 22:30) between 19:32 and 08:25, but during
> office hours they were more frequent (up to 20 on the day in question),
> often clustered together in groups of 3 or 4, occuring every couple of
> minutes or so. The telephone line in question doesn't get used much during
> the day, and all equipment on the line is filtered - line usage definitely
> does not coincide with the drop outs.
>
> This only really started when the line was upgrade from 512Kb after the
> limits were relaxed last year. However, given the line is well within
> limits for 1MB I don't understand why drop outs should be so frequent. My
> next step is to sample the stats over a 7 day period, to see if the
> drop-outs occur at the same time every day.
>
> For the record, the modem is a Zoom X3, with an old firmware - there are
> dire warnings on Zoom's website about not upgrading to the newest firmware
> on top of the firmware version I have, but I haven't been able to find
> version 2.4.x anywhere on the web as an interim stage. It is plugged in
> directly to the master socket.
>
> Grateful for any comments or advice on how to proceed

Assuming the line attenuation measure is accurate then, at least as it was
explained to me when I encountered similar problems, the attenuation
shouldn't fluctuate by that much. Perhaps there is some sort of line
problem?

Also whilst the modem is logging line noise at a minimum of 9 perhaps it is
unable to log the actual lowest line noise which leads to the loss of synch?
My own Netgear modem was unable to log anything below 4 because beyond that
there was no connection left to log (actually, for most modems, line noise
is measuring the difference between the actual line noise and the minimum
noise the modem needs to function iirc).

I had no problems at 512Kb but as soon as my line was upgraded to 1Mb the
problems started - they were eventually traced to a line fault that has
never actually been fixed but has been patched up (literally with black
masking tape!) Consequently I am unable to get a 2Mb service despite people
living very locally being able to obtain such a service with no problems. It
seems that a 512Kb line is more forgiving of certain line problems - in
terms of loss of synch - than a 1Mb line.

Gareth.

On Thu, 12 May 2005 19:00:21 +0100, "Gareth" <(E-Mail Removed)>
wrote:

> It
>seems that a 512Kb line is more forgiving of certain line problems - in
>terms of loss of synch - than a 1Mb line.

it only needs half the useable channels, or less SNR if it puts half
the data into the same number of channels. Either way, 512k is easier
/ more robust and hence goes further than 1M.

Phil
--
Tiscali - dialup speeds at Broadband prices.
AOL - the unlimited ISP of choice for heavy downloaders.

On Thu, 12 May 2005 10:43:38 +0100, Phil Thompson wrote:

> On Thu, 12 May 2005 09:19:50 +0100, Iwan Davies
> <NOSPAM-(E-Mail Removed)> wrote:
>
>>The master socket is located in a small room which is largely unused during
>>the day, but there is a portable TV in there that is used occasionally.
>
> TV might interfere I suppose if its close to modem ??
>
> tried the test socket or is there not one to try ? that isolates the
> extensions which otherwise are part of the picture.
>
> Phil

Bought a 2m modem cable rather than the 10m job I was using and tried the
test socket - initially it registered over 60dB line attenuation, but with
21dB SNR margin! (The modem autosensed modulation to T1_413). Rebooting the
modem got it back to normal ADSL as opposed to RADSL. But now it's settled
down, there doesn't really seem to be much difference to before. The SNR
margin is maybe a wee bit higher for longer, but I've still had 12
drop-outs in 2 hours! I'll leave the stats gathering script running for the
next week or so and then I'll try to establish a pattern before reporting a
fault.

Ta

Iwan

Iwan Davies wrote:
> I've got a 1MB business connection from Bulldog (BT line) and I've been
> noticing a lot of drop-outs during the day for a number of months, so I
> finally decided to gather some stats.
>
> Sampling the connection figures at 15 second intervals over a period of 24
> hours from 16:11 yesterday afternoon to 16:11 this afternoon, I've found
> that over that period, the average SNR Margin was 13.7 and the average line
> attentuation was 57.4. The lowest SNR Margin was 9.4 and the highest 16.4
> while the highest line attenuation was 59.2 and the lowest was 54.9.
[...]
> Grateful for any comments or advice on how to proceed

I suffered similarly with a different ISP suffering BT.


All BT's 'tests' showed the line to be fully ok. After hours on the
phone to their ops, we arranged for them to do their 'extended pings'
tests. This is where you relogon the ADSL modem to their test account.
Again, their tests showed nothing wrong.

Another 12 hours of calls to various people I finally get some details...

Their ping tests are just to send a few pings that only shows that there
is a connection. Their extended tests are to send just 1000 pings!


Myself running similar tests to yours showed periods where the error
rate went very high and latency increased to over 2 full seconds. After
yet another long and determined phone call going through their various
call centres at a great waste of time and hassle and...

BT finally sent a line engineer out who checked the line, was surprised
it worked, and then swapped over to another pair in the cabinet at the
road. We got over twice the signal level and a lot less interference.

Strangely enough, it works now.


Be determined, insist it is not working, and argue your case from a few
million ping timings or traceroutes. You can flood ping their gateway at
up to ten pings per second to check for line consistency.

(Otherwise, be considerate for whatever external target servers you use
for testing.)

Good luck,
Martin


--
---------- OS? What's that?! (Martin_285 on Mandriva)
- Martin - To most people, "Operating System" is unknown & strange.
- 53N 1W - Mandriva 10LE GNU Linux - An OS for Supercomputers & PCs
---------- http://www1.mandrivalinux.com/en/concept.php3