High Target SNR Margin

[#1]

Re Question 31791292
An engineer arrived 22nd January to fix a Voice problem and said he had patched my line to a different twisted pair between the nearby cabinet and the exchange.  He asked if everything was ok. I said Voice was ok but that my router now synced at only 384 kbps.  This compared with a high of 2784 kbps with 55Db of line attenuation and 6Db of Noise Margin achieved on the original pair on 18 Nov 2009 and was way below the FTR of 1817 kbps.
 
The engineer said that I should arrange a broadband engineer visit.
PN arranged the engineer visit (thanks PN!).  He switched my line to a better twisted pair and my line immediately synced at a more respectable 1856 kbps.
Netgear DGN200 d/s router stats (4.2.10 @ 1:15am):
Sync Rate: 1792
Attenuation: 61
SNR Margin: 12.4
Line Stats (advised by PN on 28.1.10)
SNR Margin: 13
ErroredSecs: 0
HEC errors: 0
Speed: 1856
Current Line Rate: 1500  (consistent with SpeedTest IP Profile of 1500)
Max Stable Rate: 2272
Immediately after rebooting the router the Noise Margin was 13.1Db, presumably the Target SNR Margin advised by the DSLAM.
I ran RouterStats for 36hrs. This showed that the line was stable varying between 13.1 and 12.0Db except for one spike down to 9.8Db and one automatic re-sync 13 days later today 4th February (re-connected at 1792 kbps, 61Db attenuation, settled down at 12.4Db Noise Margin).
QUESTIONS:
1) Given that my line is stable, why does my line have such a huge Target SNR Margin (13Db)?  Surely this could easily be reduced to 6Db with stability allowing a faster sync rate and ultimately a higher IP Profile permitting faster downloads.
2) Will the Target SNR eventually reduce? If so, how long does it take? If not, should I take up PN's offer to re-train my line and what pain would that involve?
Thanks.

[#2]

Your target noise margin has probably been set to 12dB by the DLM because it regarded the line as unstable during the period of the fault. Raising the target noise margin is one way it tries to improve connection stability.
Left alone the target might reduce again, however this can take 4 weeks or more with a solid connection. Rather than waiting - in hope - if you take up PN's offer to retrain your line (justified by correction of a fault) the target will immediately go back to 6dB. Hopefully demonstrated stability of the line will allow that to continue. If you do opt for this in my opinion it is best to leave your router on 24x7 during the 10-day training period. A day-time re-sync once per day is OK if you want to, though personally I wouldn't even do that.

[#3]

Thanks for the reply spraxyt! 
PN kindly initiated line retraining at my request and as you anticipated, target noise margin got reset to 6dB allowing my modem to sync at 2464 kbps (c.f. 1792 previously with 12dB target noise margin).  Attenuation was unchanged at 61dB.  24Hrs of RouterStats showed very little change in SNR margin, nailed at 6dB +/- 0.2dB. 
I am hopeful that such stability will allow my SNR margin to be further reduced to 3dB at the end of the training period giving a commensurately higher sync speed and allowing a higher IP profile/faster data throughput.

[#4]

Bad news 6dB is the lowest it goes on standard ADSL

[#5]

> Bad news 6dB is the lowest it goes on standard ADSL...
Damn!
But I've just discovered that my local exchange has recently been updated to support ADSL2+ and that it is available as part of PN's Premium package. I know very little about ADSL2+ (only just figured out how ADSL Max works!).  From the little I do know and a comparison of PN's published download speeds for each product I can see ADSL2+ offers several advantages but would ADSL2 + be able to help in this respect?
My DGN200 router supports ADSL2+.

[#6]

To be honest there is almost zero chance that your line would be stable with a 3dB noise margin

[#7]

With such a high attenuation (61dB) it is unlikely that an ADSL2+ connection could be achieved, the router falling back to ADSL2. A very small increase in sync speed might be seen because a minimum of 1-bit per tone is used on ADSL2 (and 2+) whereas the minimum on ADSL1 is 2-bits per tone. However the BT profile is likely to be unchanged so download speeds will not change.
Reducing the target noise margin to 3dB should increase sync speed, but increased numbers of errors might (at best) make actual throughput worse. At worst sync could be lost, and in either case the DLM is likely to increase target noise margin again pronto.

[#8]

> With such a high attenuation (61dB) it is unlikely that an ADSL2+ connection could be achieved, the router falling back to ADSL2...
So my connection would fall back to ADSL2 if ADSL2+ won't fly? - That's interesting.
> A very small increase in sync speed might be seen because a minimum of 1-bit per tone is used on ADSL2 (and 2+) whereas the minimum on ADSL1 is 2-bits per tone...
Presumably because some tones at higher frequencies containing 2 bits may not be available on account of unacceptable error rates detected by DLM, but that many  these additional tones will have acceptable error rates if only only 1 bit stands to get corrupted, hence a possible small increase in sync rate?
> However the BT profile is likely to be unchanged so download speeds will not change.
3 days into ADSL Max line training and my line now syncs at 2432 kbps, 61dB attenuation, 6dB Noise Margin.  BT's Speed Test at www.speedtester.bt.com reports an IP Profile of 2000 kbps with a download speed of 1766 kbps.  I understand my line will have to sync at 2848 to get a 2500 IP Profile.  Not far off!  Even if my IP Profile remains unchanged, presumably I should get a small increase in data throughput on account of the additional 1-bit tones available..
It's a great pity that in solving the Voice problem, the OpenReach engineer ended up patching my line to a new twisted pair at the Cabinet with a 61dB attenuation.  Before the Voice problem arose my line had an attenuation of only 55dB, about 3.9Km (real distance) from the exchange.  The engineer said that there were only 3 or 4 twisted pairs left capable of a decent broadband connection so I imagine I got left with one of the dog's bodies.  He also said that many users connected to that Cabinet were syncing at close on 4000 kbps so as I am situated only a short distance from the Cabinet, I might be lucky (or not!). 
I think the line is still within BT's tolerance level so there's not much I can do about it. 
> Reducing the target noise margin to 3dB should increase sync speed, but increased numbers of errors might (at best) make actual throughput worse...
Understand.
I move to the PlusNet Premium product at the start of my next billing period, 27th February, so look forward to what I will get with interest.  I understand a new line training period will be required for this product.

[#9]

Quote from: siddalr

> A very small increase in sync speed might be seen because a minimum of 1-bit per tone is used on ADSL2 (and 2+) whereas the minimum on ADSL1 is 2-bits per tone... Presumably because some tones at higher frequencies containing 2 bits may not be available on account of unacceptable error rates detected by DLM, but that many  these additional tones will have acceptable error rates if only only 1 bit stands to get corrupted, hence a possible small increase in sync rate?

Not quite. With ADSL1 any tone which can support only 1 bit will not be used, whereas ADSL2 will use it. So potentially the bit-rate capability is increased resulting in perhaps 60kbps higher sync speed, albeit useful to have if you are close to an IP Profile switch-point - as you said 2848kbps sync for an IP Profile increase from 2000 to 2500kbps.

Quote from: siddalr

I move to the PlusNet Premium product at the start of my next billing period, 27th February, so look forward to what I will get with interest.  I understand a new line training period will be required for this product.

Changing product to Plusnet Premium will not immediately change anything, but the follow-up change to 21CN will. I think the main lesson with 21CN is to let the new DLM take its course, trying to 'help' by restarting the router can turn out to be counter-productive. You might find the reported attenuation increases after the change to 21CN because it is measured differently (average value rather than value at 300kHz). Hope the change proves beneficial for you.

[#10]

You use a Netgear DGN2000 router. If you are adventurous  you could try using the DGTeam firmware which allows you to tweak your own SNR.
I use a DG834GT with the DG Team firmware, and my router is stable at 4-5dB - just enough to raise my BRAS profile from 5000 to 5500.
I don't know if routerstats will let you tweak the SNR on the DGN2000 as an experiment - if it will you could try that - the problem being that changes applied by routerstats don't survive a router reboot (or a power glitch)