Print

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

[waltergmw]

Thanks yet again Kitz,

I was about to update you with the latest situation and was trying to describe it as a highly intermittent white noise problem.

This (Sunday) afternoon there was another noise storm but this time it was apparently in the middle of a phone conversation:-

"At that time I was on phone and there was a lot of noise on the line which I could hear but
she couldn't.  It  started after I'd been speaking for about 10 mins.  There were no electrical appliances running at the same time."

As I didn't want the bRAS to be reduced I decided to risk re-syncing the modem which did indeed restore the situation to almost pre-incident settings.

Would it be reasonable to postulate that:-
a)  the phone and house wiring are unlikely to be directly involved as it was clear before and after ?
(Remember this is the house where I had to get BT to install an Openreach filtered faceplate and feed the phone connection back down another pair of about 4 metres of the cable to an outside junction box.)
b)  it is a remarkably wide frequency band as it includes the first 3 kHz and it also causes the modem to truncate the ADSL signal by 45 tones as well as reducing the high tone capacities almost uniformly.
c)  the line is normally very stable indeed with only a loss of 1 or 2 dB overnight and then up agiain the next morning.

When I discovered the anomaly, and after I remembered that I had to Stop and Start the DMT monitoring to see the current settings, the bit rate was down to 1024 and the tones in use had dropped to 97 and their capacity had been "hedge trimmed". I used DMT to resync the modem and it returned to nearly the previous bit rate of 2784 and 142 tones.

The three pictures below show the noise burst and my intervention, the lower DMT performance after the anomaly and the restored DMT after I had resynced it.

Any further ideads Kitz and / or Ezzer et al ?

Kind regards,
Walter




[attachment deleted by admin]

[waltergmw]

I spoke too soon!

We were "bRASed down" to 750 kbps within 11 minutes of the storm.

Kind regards,
Walter

[waltergmw]

.... BUT by some miracle we were returned to a bRAS of 2000 kbps WITHIN ONE HOUR !

[kitz]

>> describe it as a highly intermittent white noise problem.

Yes very probably :/  The problem is going to be tracking down something like that

From everything youve said it doesnt appear to be the line - it could possibly have been a HR fault, that causes symptoms when wet or if the wind blew the line against tree branches etc...  but you said that this has already been checked and the D side replaced?

The line itself seems to be in very good condition for its length...  just the 'noise problem' letting things down.

>> tones in use had dropped to 97 and their capacity had been "hedge trimmed". I used DMT to resync the modem and it returned to nearly the previous bit rate of 2784 and 142 tones.

Typical noise burst symptoms through EMI.  Start of the storm knocks out the dsl..  then whilst still there vastly effects the bit loading.

Source of problem may not even be at the EU's home could be being picked up from nieghbouring properties.  We've had quite a lot of these of this forum over the past year or so.  Culprits have been neighbours faulty TV or music centre.  Neighbours Faulty Monitors and Sky boxes have been quite common of late.

>> BUT by some miracle we were returned to a bRAS of 2000 kbps WITHIN ONE HOUR !

ahh  so the 'new' system does work how it should on occasions.

[Azzaka]

>> BUT by some miracle we were returned to a bRAS of 2000 kbps WITHIN ONE HOUR !

ahh  so the 'new' system does work how it should on occasions.

Occassions 

The line is now being investigated again. The e-side and d-side will be the first to be looked at and from there we'll head to the line card. Looking at the Graph I have to agree with your synapsis that the white noise is EMI. The question remains, where is the noise coming from...

[waltergmw]

The D side wasn't actually replaced but re-wired to a new filtered Openreach Master Socket and the modem is plugged directly into it.
Overnight the noise levels have remained almost flat.

Kind regards,
Walter

[waltergmw]

Just to update you all, the saga continues !

This seems quite an unusual problem so it's caused me to evaluate what I'm doing.
I am using empirical observations to investigate a highly intemittent problem on an otherwise nearly perfect line.
Quite reasonable speeds are apparent for most of the time but it is relying on the valid communication of the 585V7 modem and the corresponding equipment in THCN.
When things go wrong they do so quite dramatically and the exchange equipment responds.
There is sometimes a smaller excursion just before the crash as can be seen in the Storm brewing picture.
There is also a "horrible 160 kbps" picture showing a much reduced but symetrical result; this does not seem to be the normal sort of problem where a few bins are unusable due to particular set of noise frequencies. Hence my earlier description of white noise.
On one fault occasion when the householder was actually talking on the DECT phone, and had been doing so for about 10 minutes, he stated that the speech "suddenly sounded all fuzzy, but the other person couldn't hear it". This suggests that whatever the event is, it affects the entire frequency spectrum down to the 3 kHz audio levels. (Yes we have changed filters.)
Yesterday there was one event at just about the time the washing machine started but other events have happened in the middle of the night when the house is quiet and also during the day when the house is unoccupied.
There is a refurbished 3 phase 11 KV overhead power line and transformer in the vicinity but nothing unusual was observed during the severe storm we had last Tuesday night.
Another ADSL user within about 300 m of this house has a good stable connection, nor am I aware of any other similar problems in the village.
The house wiring is of varying ages and includes a double solenoid bell but the master socket is within 4 m of the cable entering the house and the modem is connected directly to a new BT Openreach filtered faceplate.

One question I'm asking myself - is there actually a problem or is the modem or the exchange equipment wrongly detecting an anomaly ?
We have noticed that the 585 can clock up astronomic FEC errors which might suggest there's a logic fault ?
To help assess the situation yesterday I replaced the 585V7 with another 585V7 I've bought direct from Thompson and am still waiting for another crash.

I have posted a set of recent screen dump pictures below.

I have also included one showing the rapid change of bRAS rates, both up and down, when large sync speed changes are detected at THCN.

All comments and suggestions welcomed !

Kind regards,
Walter


http://s694.photobucket.com/albums/vv301/waltergmw/?action=view&current=TwodropsResync09-3507Mar093A.png

http://s694.photobucket.com/albums/vv301/waltergmw/?action=view&current=Horrible160kbpsDMT09-18Pict2.gif

http://s694.photobucket.com/albums/vv301/waltergmw/?action=view&current=00-17sat07Mar09spike.gif

http://s694.photobucket.com/albums/vv301/waltergmw/?action=view&current=DMTafterre-boot23-305Mar09.gif

http://s694.photobucket.com/albums/vv301/waltergmw/?action=view&current=Stormbrewing10-43to11-05.gif

http://s694.photobucket.com/albums/vv301/waltergmw/?action=view&current=Stormarriving11-45to12-48.gif

http://s694.photobucket.com/albums/vv301/waltergmw/?action=view&current=StormaftermathDMTscreen.gif

http://s694.photobucket.com/albums/vv301/waltergmw/?action=view&current=bRASasrecordedbyZen.gif

[kitz]

s there actually a problem or is the modem or the exchange equipment wrongly detecting an anomaly ?
We have noticed that the 585 can clock up astronomic FEC errors which might suggest there's a logic fault ?

A large amount of FECs indicates low level noise....  FECs meaning that data has been recovered/repaired, that without RS encoding would  become more serious errors and cause the line to drop sooner.  The line should be able to cope with quite a lots of FECs, but too many and its an indication that at times the line's SNR is being affected causing what would otherwise be corrupt data packets.

If its racking up tons of FEC's the router may even get a bit warm as the processor will probably be working overtime correcting and reconstructing data packets. 

[waltergmw]

Kitz et al,

I'm pleased to say the problem has probably been eliminated.
BT O fitted a RF3 filter and the line is much more stable now.
SNRM varies between about 12 dB and 14 dB (night to day).

I don't suppose we'll ever find out what was causing the difficulties.
There was a 9 hour power outage last week so it's just possible that an arcing component has been replaced.

It's interesting to note that the bRAS rate adjusted very rapidly when there were major sync speed changes but it has taken 6 days to return to 2,500 kbps whilst the sync rate has remained between 2624 and 2912 kbps. (See the picture below.) The tone count has varied between 151 with 6 gaps and 145 with four gaps which has provided the fastest sync speed. Line attenuation has remained constant throughout at 55 dB and Tx power has returned to 18.5 dBm.

I'm now intending to leave sleeping dogs well alone !
We have been chasing this highly intermittent fault since December and we are all relieved that it seems to have been resolved now.
Thank you Kitz et al for your help and suggestions.
The ISP is also to be congratulated at staying with this problem for so long !
I know of many that would not have been interested, nor have the systems in place to monitor and record the anomalies.

Kind regards,
Walter

[attachment deleted by admin]

[roseway]

That's good to hear, even if there isn't a full explanation.

[kitz]

BTo usually fit RF3s in an area where there is a lot of radio EMI inerference.. usually where its from a radio/tv mast which causes widespread interference across a whole range.  I'm not sure exactly which frequencies an RF3 filters out.. but despite the fact it is supposed to let adsl frequencies pass through unhindered, it can sometimes (but not always) filter out some of the higher adsl frequencies and may occasionally in these instances give a slightly higher attenuation reading.

It looks like it may have done some good though on that particular line.  
Out of interest - sorry I cant remember - but did that line have any bell wires still in place?  RF3's tend to work best where the bell wire is acting as a large aerial.

I'm just glad that all now seems to be sorted for you

[waltergmw]

Thanks Kitz,

No there are no bell wires at all, but there is a "prehistoric birdsnest" including a double solenoid bell acting as another junction box, a DECT phone on the ground floor and a wireless fuel tank level indicator, but no WiFi. The modem and master socket were within 4 metres of the drop wire in the second floor attic.

The attenuation has remained constant at 55 dB on a line of about 4 km in length.

We do also get the occasional high level aircraft from LGW and lower Chinooks etc. from Odiham. I started in this house by getting BT to remove a star connection and back-feeding the phone line from the new BT O filtered faceplate back down another pair to the external junction box.

I'm afraid this one is going to remain a mystery as BT O are not usually forthcoming with their exchange modifications. I'm reasonably sure the problem, which was highly intermittent - i.e. about once in every 4 to 7 days, had little to do with a constant public wireless broadcast nor any other actual human-operated radio device such as CB radio. It happened several times in the early hours in a sleepy village with only one house occupied by septogenarians closeby and the main road 300 metres away. As you've seen the effects were across a very wide frequency spectrum right down to the 3 KHz audio level.

I've attached a Noise margin picture from this afternoon. The night time ones drop a couple of dB and are a bit more spikey by 0.5 dB

I do hope we can just say RIP !

Kind regards,
Walter

[attachment deleted by admin]

[Azzaka]

It appears the issue was the line card. This was replaced after they fitted an RF3 filter.

[kitz]

>> It appears the issue was the line card.

Thanks for the report Azzaka.. 

... although Im not 100% convinced about it being a line card fault as such, due to all the classic symptoms of intermittent REIN.

One plausible explanation though could be that a neighbouring line on the dslam was on occasions getting a good SNR and power cut back on that line was taking a while to kick in/or/ powercut-back didn't occur until that lines next resync.   This would cause NExT on the dslam and affect all users on that particular line card.  With you perhaps mentioning REIN/EMI in the fault report to them, some engineer may well have twigged where to look for the REIN.

The main thing is that this is now fixed for the EU and thats good news... and well done to Zen for sticking with a fault that is often very hard to track down.

[Azzaka]

So you DID see my " It appears the issue was..." 

No other notes from the engineer apart from that one, Now its time to fix another one. Same exchange, little further down the line.