Print

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

[DeanB]

My question concerns my 22dB attenuation and why I can't get more than 13,309 kbps sync from my line, I thought I might get 19,520kbs...

A bit of history about my ADSL service 3 months ago I moved from Demon to Sky Connect.  I though this useful as it would have been
the same line but different? exchange wiring. The line stats were:

==============================================================================
Monday 13th July 2009 21:13
SKY CONNECT STATS
ADSL Link        Downstream     Upstream
Connection Speed    7328 kbps    448 kbps
Line Attenuation    22 db       5 db
Noise Margin       6 db       23 db
Test1 comprises of Best Effort Test:  -provides background information.
    Your DSL connection rate: 7328 kbps(DOWN-STREAM),  448 kbps(UP-STREAM)
    IP profile for your line is - 5500 kbps
    Actual IP throughput achieved during the test was - 638 kbps
===============================================================================


The router is a DG834v3 and this was master test socket connected. The low IP throughput was due to throttle.  Soon after that the Sky DSLAM had capacity and I asked to be migrated to Sky Max.

Start of DLM process Mon 27 Jul 2009  21:47:35,    Noise Margin= 15.8dB Sync Speed= 4094 kbps
..end of DLM process Tue 28 Jul 2009  23:01:40,    Noise Margin= 7.2dB  Sync Speed= 12220 kbps

When the DLM had finished I spoke to a tier 2 at Sky who said that their estimation of line speed had been 14 Mb/s (at the time it was 12,220 kb/s) who tried to knock it up starting at 15 mbps and became unstable immediately.  After about 10 minutes on the call, the best managed was 13,309 kb/s, but they did try!


The current stats - router DG834v3 test socket connected
================================================================================
Sunday 20 Sep 2009  11:01:45
SKY MAX STATS
ADSL Link        Downstream     Upstream
Connection Speed    13309 kbps    1023 kbps
Line Attenuation    22 db       10 db
Noise Margin       7.2 db       7.2 db

IP Throughput Speedtest to Maidenhead gives
Ping 28mS Download 11.37Mb/s  Upload 0.83Mb/s
================================================================================


I have a stable connection, can't fault the actual IP throughput and it's been consistent in terms of noise, ADSL link speed any time of day, so no complaints with Sky's service     My normal set up is not from the test socket, it runs through the active filter which gives identical line sync figures, however I've eliminated the house wiring by connecting to the test socket (inside the master).  My incoming (black) cable arrives overhead and goes straight into the NTE5.  There are two pairs in the cable (BTO side orange/white connected and green/back not) - eliminated any user wiring.

The line length to exchange SSCMN Cheltenham - Kitz distance tool says 1.1 km direct, 1.7km by road and 22dB is 1.6km so that all looks OK.  The voice element of the line is perfect - quiet line test is totally silent [using BT8756G which has a wide frequency response and loud].  I've used an AM/MW radio and there is nothing that concerned me - we do have halogen lights on a SMPS which can cause lots of noise, but that wasn't on during the test and indeed makes no difference anyway to the noise margin.  Router stats shows 7.2dB all day and night, all week and month.  The Sagem Fa@st gives a slightly worse result.

BTO capacity report for SSCMN says: no interframe ties.  Long term Closure. Site visit arranged to review options. extensive building works required to increase site capacity further.  There is a proposed solution for Cheltenham but it is only at the conception stage so there is no date yet. So perhaps it's all a bit packed in there!

Bit loading, seems fairly even across the range.

Right, sorry for all that detail, but thought it useful.  So here are my questions:


* Should I expect anything more than 13,309 kb/s out of this line? 
* I'd have thought the cause of the problem is noise - but not at this end? But where?  Any ideas of what else it could be?
* How would I convince Sky or BTO to do anything, after all I have 11.37Mb/s IP throughput!  If so any advice?
* Have I missed anything and can I check or change anything more?
* Guess the last one is have I become obsessed with this - I know the answer!

Many thanks to Kitz and all the contributers as I'd never have got this far without your help.


[attachment deleted by admin]

[jeffbb]

Hi
Looking at the Bit loading it looks as if there is some general noise down in the ADSL1 region . Your bit loading looks a little low over the whole band with some little dips . I would have expected a more gradual slope ,with decreasing loading . your maximum bit loading is only eleven dropping to 6 on the ADSL1 section. Difficult to explain in words . As an example my line on ADSL1 with a 30db attenuation has a block of 12 bits near the beginning then ithe bin allocation gradually drops to 8db by bin 254 .The slope is almost a straight line . my guess is that your missing synch speed is due to that loss of between 2 and 4 bits for the whole of the ADSL1 section.


Regards Jeff

[DeanB]

Jeff

Thanks very useful comparison.   Your description of your line makes sense, with a longer line the capacitance increases, frequency response then drops and has the effect you describe (ie tails off at the top end).  Where as the bit loading on my line is poor at the low frequencies, and gradually falls, but no steep fall off as it starts off low in the first place.  The fact the bit loading is so low in the lower frequency range compared to your line, would perhaps seem to indicate some cross talk with another line (with the slight jump up at the next band of frequencies).  Perhaps the router is being told to reduce output power level overall due to fairly low attenuation level....  (although I don't think I can get power level from DG834v3 terminal as it says 0).    I've seen at least one BT Openreach ADSL REIN specialist on the forum, be interested in his view.

[HPsauce]

That's very odd, my attenuation is almost identical at 21.5dB and I get over 20mbps (on Be).
My bit loading is VERY different, starting at 15 on the lower downstream frequencies, dropping gently to 11 in the mid-range, then a small rise to 12 and a gentle tail-off to about 6 at the top end. With the odd gap/valley.

Mind you, I have a 3dB target and interleaving is off, but your graph still looks like there is something restricting the downstream signal across the board.

[HPsauce]

Another thought.
Certainly on Be the DG834Gv3 has some incompatibility issues, so have you tried another router - the v4 and GT work well with Be but I don't know what DSLAM's Sky use.
And have you got the latest firmware? Or use DGTeam's?

[DeanB]

HP, Thanks - and looking at the bit loading on other lines shows they have much higher bit loading at the lower frequency end.  Router is the AR7/DG834v3 with DGTeam software although the ADSL module is original code, so should be the same as the Netgear code.  Sky supplied the SAGEM F@st which is Broadcom based and gives a slightly worse result!

[DeanB]

I've managed to find the result of bitloading on the same line with the DG834 on skyConnect, which is over the lower range of frequencies.  Extracted from the router then via exel (30 mintutes....) before I knew router stats could produce the bitloading graphs!  Anyway as you can see the bit loadinig was higher at lower frequencies but would have been via the BTw DSAM.  Not sure if this helps, but thought it would be useful.

[attachment deleted by admin]

[HPsauce]

I spoke to a tier 2 at Sky who said that their estimation of line speed had been 14 Mb/s (at the time it was 12,220 kb/s) who tried to knock it up starting at 15 mbps and became unstable immediately.  After about 10 minutes on the call, the best managed was 13,309 kb/s, but they did try!

Just thinking about that comment, it sounds like Sky have some detailed control in place and that could well be set at a level that is limiting the bit loading.
Now I'm no expert on what they can do, but hopefully someone else here is.

[jeffbb]

Hi
I thought bit loading was sorted during the synch process ,and not under control of the ISP?
quote The better the SNR at that frequencies in the sub-channel range, then the more bits that can be allocated to that particular carrier bin.

see http://www.kitz.co.uk/adsl/adsl_technology.htm#bit_loading

Mind you I might have misunderstood 

Regards Jeff

[kitz]

>> Looking at the Bit loading it looks as if there is some general noise down in the ADSL1 region Looking at the Bit loading it looks as if there is some general noise down in the ADSL1 region

yep..  hmmm..  crosstalk?
I know see similar on my own line as the MSAN has got busier.  The tones dont improve until the higher regions... I strongly suspect crosstalk and so far its cost me about 2Mb of sync and nothing I can do about it

HP is with the same ISP than me.. and has double the attenuation, but I now struggle to see the same sync as him 

--------

Another thing sky wont turn interleaving off - which will be costing the OP a fair bit of sync speed too.

>> I thought bit loading was sorted during the synch process ,and not under control of the ISP?

It is.

What they can do is control the target SNR..  and also the depth of Interleaving... both will have an effect on max sync as well as the stability.

[DeanB]

Once again thanks as this started a brand new idea - This forum is great.

Previously on this forum http://www.nicc.org.uk/nicc-public/Public/interconnectstandards/dsltg_spec/nd1602_2005_08.pdf gives the details for PSD masks

up extra short the customer NTP where electrical distance from the MDF is 21 dB or less
up short the customer NTP where electrical distance is 26 dB or less but over 21 dB
up medium the customer NTP where electrical distance is 29 dB or less but over 26 dB
up long the customer NTP where electrical distance from the MDF is over 29 dB
Table C.1 – ANFP Interface Category Definitions

So perhaps I am on the edge of and probably have extra short applied limiting my output power an SNR, which leads to lower bit loading which also makes sense because the mask at the lower (ADSL1) frequencies are 10dB appart and closed as the higher frequency.



AR7 DSL Modem Statistics:
[DSL Modem Stats]
        US Connection Rate:     1034    DS Connection Rate:     13309
        DS Line Attenuation:    22      DS Margin:              7
        US Line Attenuation:    10      US Margin:              8
        US Transmit Power :     0       DS Transmit Power:      0
        LOS errors:             0       SEF errors:             0
        Errored Seconds:        96      Severely Err Secs:      0
        Frame mode:             0       Max Frame mode:         0
        Trained Path:           1       US Peak Cell Rate:      2438
        Trained Mode:           16      Selected Mode:          1
        ATUC Vendor Code:       4753504E        ATUC Revision:  2
        Hybrid Selected:        1       Trellis:                1
        Showtime Count:         1       DS Max Attainable Bit Rate: 14316 kbps
        BitSwap:                1       US Max Attainable Bit Rate: 1196000 bps
        Annex:                  AnxA    psd_mask_qualifier: 0x0000
        Power Management Status: L0     DS HLINSC: 0

The Downstream and upstream power DS/US Power - show as 0, which I assume is an implementation issue with the reporting. 
However Power Management status is LO.  Next I notice I have AnxA and psd_mask_qualifier: 0x0000.  I'm unable to find out what shape mask0 is.  Anyone know?

Google the psd qualifer shows some routers (in this case Motorola Netopia) allows "lab technicians" to alter the psd mask.  This is where you can see this is not a great idea as if you could start fiddling around, you may drown out everyone else and end up with a bill.... perhaps that's my problem someone else is too loud on the way to the exchange [x-talk].

Two modems different chip sets same result.  It appears that the subscriber modem negotiates the power management and mask from what I've read?  In which case, can I change it?

Kitz - what can you do about the cross talk?  Live with it because it's just a feature of a multipair cable ? 

[orainsear]

I don't know a huge amount about this, however be aware that those listed attenuation losses are at 100 kHz and are for the upstream PSD mask.

I'd also guess that the psd_mask_qualifier: 0x0000 is the upstream mask that is set by your modem, and depends on those attenuation figures.

Power Management Status: L0 (Full On) means the connection is fine.

L2 is a low power status and L3 is idle with no signal.  Some reading about power modes can be found @ http://focus.ti.com/pdfs/bcg/adsl2_low_power_wp.pdf.

[kitz]

I suspect theres a reporting issue or bug with the router firmware for it to be reporting power as 0.


No I dont think its anything to do with the PSD masks - heres why:


Power cutback is actually what reduces the SNRM, which in turn causes lower bit loading at sync.

I havent yet read your link, but AIUI, PSD masks are applied to certain tones to ensure that crosstalk is kept to a minimum for those specific frequencies. I wouldnt fiddle with the masks (not sure if you can) as different masks i think somewhere come into play depending on the adsl provision.. eg Annex A or Annex M.

Power cutback however can occur across all tones and depends on how hard your line is working and if its being pushed to its limits.

For example this is the bitloading graph for a short line on adsl1 where power cutback has been applied and limiting power output to around 12dBm.


Now bit loading for the same line on adsl2+ for an 'extra short' line.


Notice how the PSD mask doesnt stop tones 40 onwards from getting the full bit loading

Now look at same line again on Annex_M.



The mask is still letting tones 60 onwards get a decent level of bit loading..  but its preventing the upstream tones before tone 56 from loading too many bits per bin.

If you overlaid the Annex_A and Annex_M graphs your will likely find that somewhere around 100kHz is where the effects of PSD masks to prevent crosstalk between upstream and downstream begings to occur.
As the downstream gets more... then the upstream gets less and is shaped. 

The same thing occurs before 56kHz to prevent the upstream tones 12 and below from the effects of xtalk with voice.

The effect of this PSD masking is the familiar inverted U shaped bit loading for the upstream.
Power cutback when the SNR is strong across all tones..  affects the bit loading of all tones.

[kitz]

@ HP

From my latest DMT... can you see what I mean about the detoration of my line?
Used to sync at full 24Mb easy on Annex A.   On Annex_M it used to be happy at just shy of 24Mb.. now its lost 2.5Mb.. and theres naff all I can do about it.
Bins around 230 (100kHz) are loading less than 430 (1855kHz).  The line is dreadful and vastly under performing for its attenuation..  but I doubt I'd get anyone to take notice.

Sorry Dean..  took yours off topic for a while....  back to your line.

[HPsauce]

Yes, mines a bit worse than that, but not a lot. Though you're on Annex M of course.
Mine used to get nearer 21/22mbps, I suspect crosstalk is a factor for me too though less so.


[attachment deleted by admin]