Print

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

[kitz]

Oh dear.... perhaps not the Be DSLAM afterall.   

Here we go again with the weekend wee small hours thing.

Graph 1 shows how rock steady the line normally is
Graph 2 shows the decline in upstream starting at about 05:30am
Graph 3 shows now and how the line is slowly recovering back to normal (11:45)

I dont have any record of attenuation, but it clearly shows losses of up to 10dB of SNR on the upstream. 
Because my SNRm now is just shy of 16dB, this doesnt seem to have caused any ill effects at present,  but its no wonder that when on adsl2+ my line couldnt handle such losses.

So it seems this intermittent problem has followed me to fttc

[burakkucat]

So it seems this intermittent problem has followed me to fttc

Very puzzling. 

So, at present, all we can definitely conclude is that the 'disruptor' is coupled into the D-side and that it is affecting the lower-frequency, upstream tones . . . 

[kitz]

yep :/

[Bald_Eagle1]

I may have missed a graph or some data somewhere, but how do we know it is only affecting the lower frequency US tones now that Kitz has moved to a FTTC connection?

[burakkucat]

I may have missed a graph or some data somewhere, but how do we know it is only affecting the lower frequency US tones now that Kitz has moved to a FTTC connection?

By virtue of the fact that it is the US SNRM only that is being depressed by up to 10 dB when the 'disruptor' becomes active? 

[Bald_Eagle1]

I agree that if US ADSL2+ Annex M tones 6 to 56 were affected, it is highly likely the same tones will also be affected on Kitz's VDSL2 connection.

However, if there was evidence that all the VDSL2 US tones are being affected, leaving DS tones each side unscathed, it could point toward something other than just an 'interference' issue at just certain tones/frequencies.

We have seen evidence from other VDSL2 connections where only the US U2 tone band was affected to a much higher degree than the U0 & U1 bands, suggesting 'interference' at specific frequencies only (see attached example graphs).

The point I'm making is that any data reported/graphed from the HG612's 'collective' values for the connection don't always depict the whole picture & that it is often useful to examine what actually happens within each tone band.

[burakkucat]

Thank you for spelling it out. 

I trust you are prepared and ready to explain to Kitz how to obtain the relevant graphical information? 

(Oldham to Blackpool is but a few flaps of the wings, in a westerly direction . . .  )

[kitz]

Thank you for spelling it out. 

I trust you are prepared and ready to explain to Kitz how to obtain the relevant graphical information? 

(Oldham to Blackpool is but a few flaps of the wings, in a westerly direction . . .  )

I think this may be the information I was trying to gather the other night from various sources. 
I (think) I now have HG612_Modem_Stats running, but its not been up long enough to provide much info so far.

Below (1) is what I have from running GraphPD manually.   
I assume that this should plot every 6hrs automatically with out me having to do anything further. (2) Settings

[Bald_Eagle1]

I think this may be the information I was trying to gather the other night from various sources. 
I (think) I now have HG612_Modem_Stats running, but its not been up long enough to provide much info so far.

Below (1) is what I have from running GraphPD manually.   
I assume that this should plot every 6hrs automatically with out me having to do anything further. (2) Settings

Not quite.
Ongoing stats are automatically plotted once per day.
Your settings are for that to be scheduled at 06:00 each day.

Snapshot logging/graphing (current data such as bitloading etc.) can be scheduled to start at a given time & repeat throughout the day (see my settings in the attached).
I have snapshot data logged/graphed each day at 06:00 (a quiet time) & repeated each 8 hours i.e. at 14:00 (when SNRM is likely to be at its peak) & at 22:00 (usually quite a 'noisy' time).

[kitz]

Not quite.
Ongoing stats are automatically plotted once per day.
Your settings are for that to be scheduled at 06:00 each day.

Thank you
I think I already have that set.  Everything Ive done, Ive done from within the GUI, do I need to do anything else?
The only thing Ive run manually is the set date/time batch file. I havent set any scheduled tasks aside from the 1st one that the program sets up for you.

[kitz]

Okay this problem has now been here for the past 3 days, each time starting at around 5am..  the periods of instability seem to be lasting longer too.   But overall its so erratic that I cant say for sure of a pattern yet.

21/07/2013   05.30 to 11.45 am   up to 10dB fluctations
22/07/2013   05.10 to 11.10 am   up to 7dB fluctuations


23/07/2013   decline started at 3am - still ongoing at present.  Losses of up to 13.5db  & upstream SNRm was down to 2.5dB a short while ago.

Graphs below show
(1) the gradual decline that started at 3am
(2) Present


If I didnt have a 16dB SNR margin my line would be dropping constantly this morning

[kitz]

Adding my stats from HG612_modem_stats.


These show that this morning all 3 upstream bands were affected to various degrees.
Downstream is showing slight losses of 2dB

[ColinS]

Kitz,

Any chance you could post BE's Full_Monty plot from 1 day's Onging stats (If you have a complete one yet demonstrating this issue)?
I would be helpful to be able to see what else is going on at the same time, e.g. error rates of one kind or another and bitswapping.

[kitz]

Does this help?

[ColinS]

Thanks Kitz. 

Well that just about does it for me - so to state the obvious, only the US bands are affected (significantly) with a reduction in SNR margin, just like you said.  Don't have the bitswap rates/tones affected in this version of the FM plot, but Eric's program will have them.  It would be interesting to see that nothing much is happening in the US bands there.  If not, then you're getting a sudden change in SNR with (so far) no obvious signs of the effects of impulse noise on the line. Which might leave broadband EMI if it wasn't that all 3 DS bands are affected but none of the intervening DS bands.

So, to my suspicious mind the line card in the cabinet is now sus.  What are BS's thoughts?