Print

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

[GigabitEthernet]

Yet more...

[les-70]

   I don't know what others think but to me your really odd graph is SNR over time.  The 5 day MyDSLWebStats of snr may be worth high lighting.

[GigabitEthernet]

Thanks les-70. I have sent that graph to TalkTalk.

[GigabitEthernet]

Some strange behaviour I have noticed is that as the downstream SNRM decreases, the upstream SNRM increases. Very odd.

[Bald_Eagle1]

I have collated the various snapshot montages into an animated gif file to see what has been changing.

A copy is attached for reference.

Your ongoing SNRM changes over the period plotted in the FULL__Monty montage is very concerning.

As are the Hlog changes.
Regardless of sync speeds/SNR/SNRM etc, Hlog data shouldn't really change much (unless there is some sort of line problem - possibly an intermittent HR issue?).

See the attached Hlog graphs from my connection.
The important parts (where attenuation isn't too high for my connection to use) have hardly changed in 3 years.
Ignore the absence of US data from the 2012 graph.
That was from an older modem firmware version that didn't actually report US Hlog data.
 


You have a reasonably highish DS interleaving depth & for US, the depth of over 180 is VERY high.

Would you mind zipping a copy of your modem_stats.log & posting it here for me to examine (if it's not already too large for a zipped copy to be uploaded)?

[GigabitEthernet]

Thanks BE, much appreciated

I've got two modem stats logs, as like I say, I migrated to a new system. There might be some overlap or gaps therefore. Please find the most recent one attached.

Regarding the upstream, I'm getting SES on it so that may explain the high interleaving?

[GigabitEthernet]

And here is the earlier log.

[Bald_Eagle1]

Looking at your raw data log made me realise that the wrong data was being plotted for upstream SES.

I have corrected that now & attach an updated version of graphpd.exe

This version also now includes the DS & US SES data in the FULL__MONTY montage.
(See the attached).


Something seemed to have gone wrong with the times in parts of the log in that some of it included data supposedly from the previous hour.
This caused some 'spurious' spikes in the graphs & showing more than 60 minutes per hour for a short period, which I removed by manually editing my copy of your log.

Did you change PCs or synchronise the PC's time during the period you have been logging the data?



I'll have a proper look at the raw data after work tonight to see if I can spot anything particular that might point to the cause of your issues.

[GigabitEthernet]

Yes I did change PCs. Sorry about that.

[GigabitEthernet]

OR came to move the master socket today.

I think the Hlog is perhaps slightly better but not much.

[Bald_Eagle1]

That still doesn't look 'right'.

If everything is indeed O.K. internally, it does suggest some sort of external cabling issue.

Do you know if your D-side includes any aluminium.


Regarding detecting patterns of interference, it's hard to spot any due to the number of recent resyncs & intermittent logging.

Would it be at all possible to not force any resyncs & leave the logging running 24/7 for say 8 days in order to obtain a full set of stats?


FWIW, I have attached one of your SNRM graphs for approximately 3.5 hours back in May 2012.
It didn't look too great back then either.

[GigabitEthernet]

Computer is now being left on as requested. Will not toifh the the modem either

[GigabitEthernet]

TalkTalk wish to send out another BT Openreach engineer. Is this a good move?

[les-70]

 I guess it may help.  If your lucky they check everything.  To some degree engineers differ in their response and another won't do any harm. 

  I remain puzzled by your stats.  To me the Hlog variations are minor and may just reflect the small changes in the reported attenuation.  That attenuation is probably based on Hlog. The actual Hlog plots vanish when they reach the upstream tones that don't have any bit loading and then reappear -but only show noise- on the downstream tones that also don't have bit loading.  Maybe the upstream tones simply report differently?  The changes in attenuation or Hlog look much too small to account for the SNRM changes which are I think are the main issue. 

  As noted early on by burakkucat I would be suspicious about the AC balance.  My limited impression is that either engineers or Openreach don't like dealing with or  raising AC balance faults and only do so if your line gets a definite fail for AC balance on their tests. I suspect dealing with such faults can be a lot of work and a pair swap will be the solution if it is available.

[GigabitEthernet]

Got many days of stats recorded now. Would some graphs be useful?