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 3 4

Author Topic: Auto reduction in target noise margin  (Read 29910 times)

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43472
  • Penguins CAN fly
    • DSLstats
Auto reduction in target noise margin
« on: February 07, 2009, 10:50:20 AM »

This morning about 10:30 I believe that I saw my target noise margin drop automatically by 3 dB.

Some time back I had a lot of problems with an intermittent line fault, and my target noise margin was bumped up to 15 dB. But for the last few weeks my connection has been exceptionally stable and with low error rates. I'm using a DG834GT with the DGTeam firmware, and I tweaked the target noise margin down to the lowest value it would go to, which was about 9.5 dB. After two weeks of uptime with only very small variations in the noise margin over the day and night, this morning the connection suddenly dropped and it immediately reconnected with a noise margin of 6 dB. The only explanation I can see is that the target noise margin has been lowered by 3 dB.

You can see a routerstats snapshot below.

[attachment deleted by admin]
Logged
  Eric

HPsauce

  • Helpful
  • Kitizen
  • *
  • Posts: 2606
Re: Auto reduction in target noise margin
« Reply #1 on: February 07, 2009, 11:00:22 AM »

I'm sure you're on the case, but after a day or two (assuming it stays like that) you need to change the % setting in DGTeam to be "gentler" so that it syncs with a slightly higher margin (say 7, 8 or even 9dB) as the next 3dB step may take you into unstable territory and undo everything before you notice.
Logged

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43472
  • Penguins CAN fly
    • DSLstats
Re: Auto reduction in target noise margin
« Reply #2 on: February 07, 2009, 11:45:02 AM »

Yes, that was my plan. I'll probably set it back to 9 dB, which should hopefully give me the same stability as I've had for the last few weeks and eventually trigger another reduction.
Logged
  Eric

HPsauce

  • Helpful
  • Kitizen
  • *
  • Posts: 2606
Re: Auto reduction in target noise margin
« Reply #3 on: February 07, 2009, 12:10:23 PM »

I'm particularly interested in how you get on as it's very relevant to a good friend of mine who I'm helping.
He had truly awful internal wiring, probably some from the 1930's and was running below 2mbps.
After sorting this all out he now gets about 6mbps but has a "stuck target margin" at 15dB.

Despite repeated calls to BT (and internal assistance from a forum member elsewhere who works for BT) they have failed/refused to reduce the target. They did however agree to turn interleaving off which got some extra speed.

So he then took to using DMT to force a faster sync and gets over 7mbps, though obviously it starts at a lower figure.

However, a few days ago I flashed his router (DG834GT) with DGTeam firmware and it now syncs straight off at about 7700kbps and typically just over 9dB. Whether this will ever improve we don't know, hence the interest in your progress.  ;)
Logged

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43472
  • Penguins CAN fly
    • DSLstats
Re: Auto reduction in target noise margin
« Reply #4 on: February 07, 2009, 12:39:24 PM »

I'll certainly let you know what happens.
Logged
  Eric

sevenlayermuddle

  • Helpful
  • Addicted Kitizen
  • *
  • Posts: 5369
Re: Auto reduction in target noise margin
« Reply #5 on: February 07, 2009, 03:21:19 PM »

All right for some, then?  ;)

After too many resets one day late last summer, I fought tooth-and-nail with Demon to get mine reset from 15dB to 9dB and, eventually they did, last September or thereabouts.  48 hours later (possibly to the minute, curiously), it retrained itself at 15dB again and has stayed there ever since.

I now tweak it using the CLI of the DG834GT, which seems to allow up to -6dB relative to target.  Although it's happy tweaked to -6 (9dB at sync), I've settled on a -3dB tweak (12dB at sync) for extra stability, and to ensure that if DLM did reduce the target, I'd still sync at 9 which I know to be stable (I'm not so sure 6 would be stable).   I live in hope, week on week, and month on month, that one day I'll check and find it's resynced at 9, but it just doesn't seem to happen.

On theory I have is that, because my line is so much improved (just by cutting the bell-wire) since the 10-day period, DLM can increase the target without me getting anyware close to my FTR.  And just maybe, because I'm syncing at a long way above my recorded FTR, DLM is less inclined to give back my margin.

Logged

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43472
  • Penguins CAN fly
    • DSLstats
Re: Auto reduction in target noise margin
« Reply #6 on: February 07, 2009, 03:31:58 PM »

Quote
On theory I have is that, because my line is so much improved (just by cutting the bell-wire) since the 10-day period, DLM can increase the target without me getting anyware close to my FTR.  And just maybe, because I'm syncing at a long way above my recorded FTR, DLM is less inclined to give back my margin.

I don't really think that DLM is that devious. But automatic lowering of the target noise margin is not a well documented process, and certainly depends on a number of factors including error rates, not just connection stability. I think it's far more likely in your case that the error rates are above the level needed to trigger the reduction.
Logged
  Eric

sevenlayermuddle

  • Helpful
  • Addicted Kitizen
  • *
  • Posts: 5369
Re: Auto reduction in target noise margin
« Reply #7 on: February 07, 2009, 03:52:08 PM »

I'm not covinced, I see about 500-800 CRCs, and maybe 3-4 SES errors per day.   It's not reconnected once since well before Christmas, bar the very occasional time I force it manually, it's certainly had several runs of 2-3 weeks continuous sync.   That's not too bad, is it?

And there is that BT patent application I link to some time ago, which may or may not describe the current DLM.  That patent suggests FTR is taken into consideration by DLM in management of target margin...

 http://www.freepatentsonline.com/EP1953959.html 

BTW - if anybody ever plucks up the enthusiasm to read that paper, the main thrust of the patent seems to be the process of identifying whether the router has retrained ten time in an hour.  Put another way, they've patented an algorithm for counting to ten. :lol:
Logged

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43472
  • Penguins CAN fly
    • DSLstats
Re: Auto reduction in target noise margin
« Reply #8 on: February 07, 2009, 04:03:59 PM »

:)
Logged
  Eric

kitz

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 33881
  • Trinity: Most guys do.
    • http://www.kitz.co.uk
Re: Auto reduction in target noise margin
« Reply #9 on: February 08, 2009, 03:13:15 PM »

Thanks for reporting those observations eric.  Do you know an approx timescale how long it took overall?
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

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43472
  • Penguins CAN fly
    • DSLstats
Re: Auto reduction in target noise margin
« Reply #10 on: February 08, 2009, 03:57:13 PM »

The router uptime was just about 14 days, and prior to that I had changed routers and done a bit of fiddling which included some reboots, so I think that the relevant period was just the 14 days of continuous sync with a low error rate.
Logged
  Eric

kitz

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 33881
  • Trinity: Most guys do.
    • http://www.kitz.co.uk
Re: Auto reduction in target noise margin
« Reply #11 on: February 08, 2009, 04:18:27 PM »

Thanks eric
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

sevenlayermuddle

  • Helpful
  • Addicted Kitizen
  • *
  • Posts: 5369
Re: Auto reduction in target noise margin
« Reply #12 on: February 08, 2009, 05:22:38 PM »

It's certainly not the first time I've heard suggestions of 14 days being a significant threshold (with error rates relevant too, of course). I've tried monitoring mine on 15 day boundaries (to remove ambiguities about whether the 14s inclusive or exclusive), but no joy - as Eric points out, it 's most probably down to my error rates.  Then again, I've set a relatively tame margin of 12dB compared with the 9dB it was to begin with, so my error rates ought to have improved, yet still it doesn't change.

Ooops, I butted in again. Sorry if I always seem to seize upon any debate about target SNRM, it's become a bit of a special interest for me.  In fact it was after noticing my actual margin (I didn't know about 'targets' then)  had suddenly increased that fateful day summer that I embarked on a crusade to find out why, and to learn a bit more about how ADSL actually worked, and that's what brought me to the kitz forums in the first place. 

Now that I'm here of course, I reallise there's so much more for me to enjoy, as well as target SNMR debates  ;)
Logged

kitz

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 33881
  • Trinity: Most guys do.
    • http://www.kitz.co.uk
Re: Auto reduction in target noise margin
« Reply #13 on: February 08, 2009, 07:51:49 PM »

>> Ooops, I butted in again.

Not at all - you have some interesting observations of your own which you validly bring into the conversation :)
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

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43472
  • Penguins CAN fly
    • DSLstats
Re: Auto reduction in target noise margin
« Reply #14 on: March 01, 2009, 10:39:19 PM »

At 10:30 this evening my target noise margin dropped by another 3 dB. There have been a couple of random re-syncs during the last couple of weeks, so clearly a continuous connection isn't needed; but the error rate has been consistently low (about 2 errored seconds per hour) and this appears to be the main requirement.
Logged
  Eric
Pages: [1] 2 3 4
 

anything