Kitz ADSL Broadband Information
adsl spacer  
Support this site
Home Broadband ISPs Tech Routers Wiki Forum
 
     
   Compare ISP   Rate your ISP
   Glossary   Glossary
 
Please login or register.

Login with username, password and session length
Advanced search  

News:

Pages: [1] 2

Author Topic: Long Line Noise Problem  (Read 8788 times)

waltergmw

  • Kitizen
  • ****
  • Posts: 2776
Long Line Noise Problem
« on: April 11, 2009, 10:24:03 AM »

Hello everybody,

This is a continuation of the TG 585 V7 saga you can see at:-
http://forum.kitz.co.uk/index.php?topic=4399.msg110417#new

Some of you with time to spare might like to look at some notes I've made of my observations whilst hunting an intermittent noise problem on a long line.
However the document is a PDF and is 6 pages long.

http://www.scribd.com/doc/14144754/Modem-Observations-Iss-A

There are both DMT Tool and Routerstats pictures.
The DMT tool shows the line performance using v7 with a TG585V7 and using v8 with a DG 834V4.

The problem is not yet solved so any thoughts would be much appreciated.

Kind regards,
Walter
Logged

kitz

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 33888
  • Trinity: Most guys do.
    • http://www.kitz.co.uk
Re: Long Line Noise Problem
« Reply #1 on: April 11, 2009, 11:24:06 AM »

>>  One possibility, as the max sync. line speeds do not exceed 700 kbps, might be to change the service to an ADSL 512 fixed speed service.)

Theres also something called Max capped rate 500 profile which the ISP should be able to easy provide.   This may help avoid some of the fluctuations.   Because this is a remote configuration change at the RAS, perhaps zen could try it for a period to see how it goes.

Max capped works similar to a fixed rate line, and it stops the DLM getting involved... and theres no messing with target SNRs and SNR margins.
If it doesnt work out then they could easily swap you back again.

>>> The reason for the thinner lines around tone 56 is unclear to me.

Not sure what you mean about thinner lines? Is that not just the shading on some versions of DMT?  Tone 64 is the pilot tone so you wont see anything on that subchannel.


>> I am yet to understand the significance of the lower channel characteristic panel.

Me neither but the suspicion is that its something to do with power output and/or gain.
One thing to note is that if you actually look at the channel characteristics direct from your router they show as negative and not positive as reported by DMT.

I'm not an expert in electronics so maybe someone else can explain this better than me.. but gain and attenuation can be related and the graph maps the loop loss per tone.
By looking at my own stats this would certainly appear to be the case that its mapping loop loss (attenuation) at each downstream frequency.
Logged
Please do not PM me with queries for broadband help as I may not be able to respond.
-----
How to get your router line stats :: ADSL Exchange Checker

waltergmw

  • Kitizen
  • ****
  • Posts: 2776
Re: Long Line Noise Problem
« Reply #2 on: April 11, 2009, 11:53:03 AM »

Many thanks indeed Kitz,

I'll keep you up to date with our progress.

Re the shading, I do remember seing one set of "pencils" on V7 where the max tops were greyed with the newer and shorter ones still in full colour.

However if you don't know of a reason, it might well be a video-painting glitch as I'm viewing the screen remotely on my Apple Mac, then taking screen dump pictures which I'm implanting in a document which is then "Printed" to a PDF. So there's plenty of scope for some aberrations !

Kind regards,
Walter
Logged

kitz

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 33888
  • Trinity: Most guys do.
    • http://www.kitz.co.uk
Re: Long Line Noise Problem
« Reply #3 on: April 11, 2009, 03:46:32 PM »

Ahhh I think I know what you mean.

V7 + V8 display the graphs differently, because of the different information available from the relevant routers that they support.... and what information can be obtained from that router.

On some versions of DMT8 (depending on the router) you will see shading such as this



where the graphs are shaded simply to make the sub-channels stand out better.

>> where the max tops were greyed with the newer and shorter ones still in full colour.


This is what I think you mean, which happens on DMT v7



In the 2nd graph...  Grey shows the actual SNR, whilst yellow shows the SNR Margin.
Logged
Please do not PM me with queries for broadband help as I may not be able to respond.
-----
How to get your router line stats :: ADSL Exchange Checker

waltergmw

  • Kitizen
  • ****
  • Posts: 2776
Re: Long Line Noise Problem
« Reply #4 on: April 11, 2009, 04:44:16 PM »

Thanks yet again Kitz. I think I now have as much understanding of DMT as I need.

I've just heard some gossip indirectly from the last engineer who called when doing the D side swop, which could just be the beginning of some better news. At least it's something to aim at.

He said that "there was still a redundant DACS connected which would not help" ** and that he "noticed a significant drop in noise when another line out of the same wall box was disconnected".

** I assume this just means the DACS unit was still jumpered on to the ADSL pair, but that all E side "DeDACSing had been done, as otherwise the ADSL just couldn't work at all. Is he correct in saying that a jumpered DACS can contribute to noise ? (It would be interesting if Ezzer could say what's involved in deDACSing, and perhaps why it's such an unloved operation?).

Kind regards,
Walter
Logged

Ezzer

  • Helpful
  • Kitizen
  • *
  • Posts: 1713
Re: Long Line Noise Problem
« Reply #5 on: April 11, 2009, 06:09:53 PM »

DACS  :no:

dsl will not work through a dacs either through an exchange unit (although why on earth anyone would jumper a dsl and still have it going through a eu dacs. It's not a mistake any one could make). or through a remote unit.

To provide dsl on a line which is on a dacs means it will have to be de-dacs'ed.

Just to explain what a dacs is, it referd to as a "pair gain device". In circumstanses where ther are not enough pairs of wires in the network to get all the working lines in demand for a route a dacs digitaly encodes 2 lines at the exchange via the exchange unit (EU) and sends this signal down the same pair of wires. this is then decoded by the remote unit (RU) at a reasonable point near the 2 customers for the 2 lines which then feeds off to the relavent nte's as with a normal line.

The technology pre-dates dsl so wasn't designed to allow the dsl signal to pass through it.

The only way an end user might know if a line was running via a dacs is on normal dial-up internet the speed peaks at 28kbps.

The problem with de-dacs-ing a line is that the dacs is on there for a good reason, so we need to find another good pair from the exchange to the ru to put the line on. Unless there's either only one working line on that dacs (then the dacs can be removed completely) or if there's another line by the ru which isn't dsl, isdn, or has regular digital traffic which can be swapped with the number your trying to de-dacs. then the job can become a nightmare.

If an engineer gets a tack on their laptop with the description de-dacs and with out the 2 easyer options, they usually cringe throw a hissy fit, use a few words not to be repeated in front of their mother etc etc.

They normaly tend to be found at the end of long routes in the country side or some times in very high density of line areas like shopping roads/streets. It's a black box slightly narrower than a dp as found as the top of poles a give away is a 30mm white square at the bottom right of the face of the ru, internal ru are in all white, on an incomming call you can hear it clicking inside as it generates the ring.

Thanks to dsl having both internet and normal voice telephony running on the same pair this has feed up a lot of demand for pairs where end users had a second dedicated line for their internet connection in the past.

One delevopment of dacs was know as Homehighway isdn, here you have 2 64kbps isdn channels down the same line, one can be voice one internet, when not using the voice connection the 2 64k channels combine to give a 128k connection, this is defined in europe as mid band. the most parts of the US this would you belive is called broadband ! I think in 1994 there where more 512+ lines working in the old 9 nations of the eec than in the whole of the usa.
Logged

orainsear

  • Reg Member
  • ***
  • Posts: 635
Re: Long Line Noise Problem
« Reply #6 on: April 11, 2009, 06:58:54 PM »

If a DACS device becomes redundant, is it normally left in situ or should it be removed?
Logged

waltergmw

  • Kitizen
  • ****
  • Posts: 2776
Re: Long Line Noise Problem
« Reply #7 on: April 11, 2009, 07:00:09 PM »

Thanks Ezzer,

Around the Ewhurst outliers almost every pole has at least one DACS box although, as you say, many have been decommissioned but just left on the pole.
I can certainly understand that it would cause a lot of grief trying to re-route lines where there was a major shortage and then hurtle back around the lanes to the exchange 6 km away - and no doubt all to be completed in the "statutory" time slot.

Kind regards,
Walter
Logged

Ezzer

  • Helpful
  • Kitizen
  • *
  • Posts: 1713
Re: Long Line Noise Problem
« Reply #8 on: April 13, 2009, 10:59:07 AM »

Time slot only applies to broadband faults, De-dacs dosn't come under this
Logged

waltergmw

  • Kitizen
  • ****
  • Posts: 2776
Re: Long Line Noise Problem
« Reply #9 on: April 13, 2009, 03:15:45 PM »

@ Ezzer,

I've examined the cable route for the noisy line. It comes from a road pit up a garage wall into a BT 66 where there are four pairs made off, two to each of two houses.
The second house is being refurbished and both lines are open ended although still officially live. The drop wire from the BT 66 goes around the garage in close proximity to an alarm system cable. From the garage and another BT 66 two overhead lines are routed to different parts of the house where they are just left dangling.
We believe these currently open ended pairs are injecting noise into our pairs.

Is it possible to requst BT O to install shielded earthed drop wire cable around the garage, and if so, can you please give me the type number ?
(I recall you suggested than not many engineers carry shielded cable but I assume it's available upon request.)

Failing that is it likely that BT O would install Cat 6 FTP if we  provided it?

Kind regards,
Walter
Logged

Ezzer

  • Helpful
  • Kitizen
  • *
  • Posts: 1713
Re: Long Line Noise Problem
« Reply #10 on: April 13, 2009, 03:55:14 PM »

never come across a sheilded d/w, sheilded internal cable yes.
Logged

waltergmw

  • Kitizen
  • ****
  • Posts: 2776
Re: Long Line Noise Problem
« Reply #11 on: April 15, 2009, 02:11:29 PM »

The noise source seems to have been reduced significantly as, when running on my Netgear DG 834 V4, the sync speed remained constant at 604 kbps. Routerstats lite showed a very gently undulating noise margin line around 6.5 to 7.1 dB with just a few spikes down a couple of dBs. DMT also reported almost identical readings.

We then replaced the original 2Wire modem which caused the sync speed to increase to 830 kbps and strangely the noise margin line to remain absolutely dead straight at 6.0 dB from start-up @ 10:45 right through to the last time I looked at 13:44. Having seen the netgear performance which seems credible, the absolutely constant 2Wire does not ! The real world is bound to have minor noise margin fluctuations.

Does anybody know if there's smoothing logic in the 2Wire / Routerstats noise margin interface or is it likely that there is a bug in Routerstats Lite which perhaps masks changes in real noise margins ?

Kind regards,
Walter
Logged

kitz

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 33888
  • Trinity: Most guys do.
    • http://www.kitz.co.uk
Re: Long Line Noise Problem
« Reply #12 on: April 15, 2009, 09:14:36 PM »

>> a very gently undulating noise margin line around 6.5 to 7.1 dB

Does the 2wire report exact decimal places...  some (many) routers will only report 0.5 dB changes.
If this is the case it could account for why the 2wire reports a more or less flatlined 6dB.
The reality is that the SNRM could be anything between 5.75dB to 6.25dB but will only report a flat 6dB
Logged
Please do not PM me with queries for broadband help as I may not be able to respond.
-----
How to get your router line stats :: ADSL Exchange Checker

waltergmw

  • Kitizen
  • ****
  • Posts: 2776
Re: Long Line Noise Problem
« Reply #13 on: April 16, 2009, 06:58:08 AM »

@ Kitz,

Well done Kitz ! I should have realised but didn't discover until this morning that the 2Wire 2700HG-B Hardware Version 2700-100534-005 and Software version 4.25.19 only shows noise margin in whole dB steps. As we've managed to "quieten" down the REIN, at least temporarily, that explains the dead straight lines, which I found un-nerving !

Around 03:55 to 04:33 this morning the noise margin dropped one and two dB to a minimum of 4 dB.

I've yet to find a way of observing the Tone loading data which I saw here on another 2Wire set out as a matrix of Xs.
Has anyone any tips on this please ?

The errors I have managed to extract don't look too bad at present - see below.

(I also Googled somewhere that someone had managed to get DMT to nearly work on a 2Wire by emulating a different router in a Linux system.
However that looked too difficult and possibly risky to waste time experimenting.)

Kind regards,
Walter

[attachment deleted by admin]
Logged

kitz

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 33888
  • Trinity: Most guys do.
    • http://www.kitz.co.uk
Re: Long Line Noise Problem
« Reply #14 on: April 16, 2009, 11:05:29 AM »

>> I've yet to find a way of observing the Tone loading data

From looking at the info in the 2wire user manual available here
The bit loading information should be on the following page:

From the LH Menu >
Troubleshooting >
DSL Diagnostics

The bitloading graph should be at the bottom

Logged
Please do not PM me with queries for broadband help as I may not be able to respond.
-----
How to get your router line stats :: ADSL Exchange Checker
Pages: [1] 2