Print

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

[Mick]

Still stuck at the 15dB target margin. 

I am waiting on BT now to reduce it again.  Meanwhile, as the days grow longer I swapped my modem to an old 2WIRE 1800HG and I can sync higher than the Netgear.  Unfortunately, this sync figure is higher than my fault treshold (this had been set at 4992Kb/s during training, when the ring wire was still connected to my house telephone wiring).  Therefore I doubt it that it is worth me chasing my ISP to get BT out to check the line.  They'll just resync at the master socket and file a no-fault-found.  The fact that I was sync'ing at 7626 kb/s would be irrelevant for them.   

Could you please have a look at the attached patterns and tell me if you see anything suspicious there.  There is a dip just before 12 noon every day and another between 16:00 to 17:00.

I also got these errors (over a period of 2 days, 5:51:51):

Code: [Select]

Collected for 2 days 5:51:51    Since       Current            Current
ATM                Reset 24-Hour Interval  15-Minute Interval   Time Since
Last Event
Cell Header Errors: 58 4 0 1:11:32
Loss of Cell Delineation:   0 0 0 0:00:00

DSL
Link Retrains:           0 0 0 0:00:00
DSL Training Errors:   0 0 0 0:00:00
Training Timeouts:   0 0 0 0:00:00
Loss of Framing Failures:   0 0 0 0:00:00
Loss of Signal Failures:   0 0 0 0:00:00
Loss of Power Failures:           0 0 0 0:00:00
Loss of Margin Failures:   0 0 0 0:00:00
Cumulative Seconds w/Errors: 89 7 0 0:22:47
Cumulative Sec. w/Severe Errors: 0 0 0 0:00:00
Corrected Blocks: 13809 231 8 0:01:36
Uncorrectable Blocks:          129 11 0 0:22:48
DSL Unavailable Seconds: 21 0 0 2 days 5:51:29
[attachment deleted by admin]

[jeffbb]

Hi

Quote :"I am waiting on BT now to reduce it again. " . It isn't BT as such that will reset the snr target (although they can if pushed hard enough )  its when the DLM has seen a stable line for long enough ,there is NO fixed time for that . My target dropped by 3 DB after about 4 weeks of very stable conditions and very LOW errors .

Regards Jeff

[Mick]

Yes, I meant it metaphorically.  I've read elsewhere that it takes up to 30 days assuming that the line has become stable, but a blip here and a dip there can make it take for ever.

Did you reboot at all over those 4 weeks?

[Mick]

Today, it seems that there is more noise in the . . . air.  The SNR won't reach 15dB as it has been doing comfortably over the last week and of course the maximum potential sync speed is less.

How do you explain this?

[attachment deleted by admin]

[jeffbb]

Hi Mick

I  rebooted after about 16 day run to see if It would resynch at higher speed with lower SNR . No joy But about 5 days after that the Synch suddenly shot up by 1.3Mbps to 7168 Mbps. and the snr dropped by 3db. (comparing with records for similar time of the day) the BITS TONE graph also filled up along most of the spectrum. My router DID NOT register a resynch ?.
I am puzzled by your SYNCH graph showing a constantly changing synch ? or am I misreading it . I thought that once you had Synch that was it . ? or are you with an ISP that supports SRA?

Jeff

[Mick]

2WIRE routers report both the initial sync rate and the sync rate that you would achieve at any given moment thereafter.  I assume that it calculates what that would be based on the current SNRM.  Of course if I were to actually resync at that moment, then prevailing circumstances could affect the SRNM differently and the new sync speed could be different than that indicated.  It gives you a ballpark figure and I have found it to be reliable enough in the past.  I hope the attached snapshot is not too small for you to have a look at what I mean.

[attachment deleted by admin]

[jeffbb]

Hi Mick

OK I got it .So you would expect the synch graph to follow the SNR graph . So according to the graph when your snr drops 3 db then your synch rate if you resynched at that time would drop around 400Kbps presumably with an SNR of 15 (your target ? ) ,which from what I have read would be expected .

Regards Jeff

[kitz]

>> I assume that it calculates what that would be based on the current SNRM.

Logical explanation


>> The SNR won't reach 15dB as it has been doing comfortably over the last week and of course the maximum potential sync speed is less.

If Im reading the graph right - youre at about 1dB lower than the 15dB?  Like you say just more noise around.
My own line often has what I call 'castle effect'...  but one of the other things I see is that for some periods of time it will often 'loose' a couple of dB for a period of time that may last several days at a time.

Over the past year Ive noticed its always worse during the weekends and at half-term. 
I'm pretty sure that my own SNR pattern is related to my neighbours as it doesnt tie up with anything I do..  but some of the timings are pretty regular.. and also the 'worse when kids are off school'.

Wouldnt mean a thing to my neighbour though and they dont care..  I went in once to look at something when there were some local problems. 
They really dont give a toss.. to give you some indication I can sync at the full 24Mb... yet they are perfectly happy to get a crappy 400kb throughput on a TT connection.  "It works so it doesnt matter".

[jeffbb]

unbelievable

[dizzy4528]

I often wonder how many people are paying for SUPER FAST BROADBAND
 and only recieving speeds just above dial up, because they are blissfully
ignorant or unaware of the problems created in there own home ie:-  incorrect
 wiring , electrical noise or just bad practice.

[Mick]

I saw this error message today but I am not sure what it means:

Code: [Select]

Troubleshooting – DSL DiagnosticsGeneral Information
DSL Line: 1
Downstream Rate Cap: 8128 kbps
Downstream Atten. at 300kHz: 35.0 dB

Uncancelled Echo: -17.8 dB Ok
VCXO Frequency Offset: -86.8 ppm Ok
Final Rx Gain: 14.8 dB Ok
Impulse Noise Comp. Tones: 60 Suspicious - impulse noise detected
Excessive Impulse Noise: 0 Ok
Meanwhile the CRC/FEC errors have shot up significantly today, especially for the Upstream.

Any ideas what is this "impulse noise detected"?

[kitz]

Nice reporting from your router
remind me again which it is please.


Probably mean some sort of noise burst occurred at tone 60 (258.75 kHz ) which meant your router was unable to load any bits into that sub channel bin.  If the SNR gets too low on that frequency the router is unable to use that tone.

[Mick]

I see ... the scale in bitloading shows up to 256 and no more.  Is this noise frequency off scale?  (EDIT: nope, it must be tones rather than frequency).  The last 5 bins are showing zeros and have been since that blasted SNRM was raised by the DLM to 15dB.

The modem is a 2WIRE.  This one is an old 1800HG.  The resync's started when I was using a 2700HGV and was not sure if it was because of the modem developing a fault.  As a rule the 2700HGV is much more fault tolerant than anything else I have tried on my line, but have read the odd report of faulty 2700s so I swapped it over first for a DG834v2 and then for the 1800.  There were no suspicious impulse noise reports for the last week that I have been running the 1800HG and I have to say this is the first time I have noticed such a report on my line, or CRCs/FECs shooting up like this.

In +006 days 17:11:21 since the last resync I have:

Code: [Select]

Downstream                      Upstream
CRCs  FECs                     CRCs   FECs
654  1843114                   1383  1318523
I haven't yet configured mrtg to capture errors (my perl skills are non-existent, anyone care to help?) but I would think that two thirds of these Upstream errors happened over the last 48 hours.  In my mind there is a fault on the line.  However, I am not sure I can convince BT that this is the case and avoid a charge, since at least during day light I can still sync above the fault threshold of 4992.

Kitz, you wouldn't have any friends in BT to take a look at this miserable performance - would you?

[Mick]

Quick update:

Since my last post things deteriorated overnight and I noticed an audible noise on the telephone.  The ADSL filter was ready to die - which is rather annoying since it was only 2 weeks old!  I am assuming that all these stats about suspicious impulse noise must have been related to the ADSL filter.  Next time I'll know what to do as soon as these show up!

Anyway, I replaced the filter with a new one and refitted the 2WIRE 2700HGV as a test, which I had been using when the original fault and resyncs occured.  To my surprise it has been holding the line like a trouper for the last 24 hours (with a slightly higher resync than the 1800HG).  This is probably because it has a more advanced chipset which actively manages noise variations.  In addition, I fitted a ferrite filter on the power lead before it enters the router.

Noise count is now much lower and for Upstream is almost non-existent.  After 24 hours I only have 2 CRCs and 4 FECs on Upstream, compared with the thousands it was counting as the filter was collapsing.  Hoping it holds the line, I am now waiting for the DLM to improve the snrm target.  If it does hold the line without mishaps it would indicate that there was nothing wrong with the 2700HGV router and the fault must have been telephone line, rather than home equipment related.

[kitz]

>> Is this noise frequency off scale?  (EDIT: nope, it must be tones rather than frequency).  The last 5 bins are showing zeros and have been since that blasted SNRM was raised by the DLM to 15dB.

Yes youre correct it will be bins - 256 is the maximum subchannels for adsl1. 
Theres a calculater which converts the tones to frequencies if it helps here
http://www.kitz.co.uk/adsl/adsl_technology.htm#frequencies

Its not unusual for the last few tones not to be in use.

>> The ADSL filter was ready to die - which is rather annoying since it was only 2 weeks old!  I am assuming that all these stats about suspicious impulse noise must have been related to the ADSL filter.
>> If it does hold the line without mishaps it would indicate that there was nothing wrong with the 2700HGV router and the fault must have been telephone line, rather than home equipment related.

Very likely the filer from what you say

>> Noise count is now much lower and for Upstream is almost non-existent.  After 24 hours I only have 2 CRCs and 4 FECs on Upstream, compared with the thousands

Excellent.   Hopefully it stays that way for you