Print

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

[konrado5]

There is just one possible explanation . . . it is a rather odd explanation but would certainly fit with your observations. Let us suppose that your CP/ISP has notch-filters, tuned to reject the 160 metre amateur radio band, available to deploy on circuits that are affected by such transmissions.

But it is not consistent with the fact the other man in Poland who uses ADSL service from other ISP and lives in other town has similar problem and slight higher SNR:

Code: [Select]

417          14.0000
   418          12.4375
   419          9.7500
   420          8.8125
   421          9.1875
   422          11.8125
   423          10.7500
   424          12.0000
   425          11.5000
   426          11.6875
   427          11.8125
   428          11.3750
   429          11.3125
   430          10.5000
   431          11.3125
   432          11.5000
   433          10.5000
   434          11.1875
   435          10.9375
   436          10.3125
   437          10.2500
   438          10.4375
   439          10.4375
   440          11.3750
   441          10.8125
   442          11.0000
   443          9.1875
   444          10.0000
   445          10.1875
   446          8.9375
   447          9.7500
   448          10.0000
   449          9.3750
   450          8.8750
   451          10.3750
   452          9.6875
   453          10.6875
   454          10.1250
   455          10.2500
   456          9.7500
   457          10.3125
   458          8.9375
   459          9.5000
   460          9.3125
   461          8.3125
   462          9.8125
   463          9.1250
   464          10.3125
   465          9.6875
   466          11.3750
   467          12.8750
   468          13.6875
   469          12.6250
   470          15.7500Similarly, this gap is noticable in Hlog. However, it is not seen in QLN (my QLN shows this gap).

Best regards
konrado5

[kitz]

Following there is CRC error graph on SNR margin 0.8 dB.

What do you think about this? I have fast path.

Best regards
konrado5

Erm...  I think that it's not good to tweak so low.  Even good lines need a minimum of 3db to keep stable, I'm actually surprised that errors aren't through the roof or the line isn't disconnecting...  Credit to the router that it's keeping you connected, but by gawd that router is going to be working hard.

I'm not on a pc so can't see figures on your graph clearly, more only spikes, but it's fair to say that I bet with the amount of spikiness that real time speeds would be rubbish.  The errors will cause retransmits so viewing some web pages may feel laggy.   I'm not sure what you are trying to achieve by tweaking the snrm that low.

I really can't understand the hangup of circa 40kbps.   There could be a million valid reasons why you lost that tiny but of speed.   Have you heard of crosstalk btw?  It's entitley possible that the addition of one more line on the dslam could have triggered something.   

As an example when my dslam was shiny and new I could sync at 24mbps annex m.   As time went by and more people joined the msan, that went down by a few mb.  Yes megabit not kilobit.  This is an expected feature of adsl..  As more lines join the dslam, they cause crosstalk and bring down the speed of any other neighbouring lines.    I've lost nearly 20mb now on vdsl in just 6 months  there is naff all we can do about it.   Ask bald eagle, or just about any experienced dsl person..  we've all seen it happen to our lines to some extent, it's just more noticeable on vdsl but mark my word it happens on adsl2.

I'm really not sure what you expect us to say, you are seemingly wanting an advanced electronics and dsl lecture that is WAY even above degree level.   I can't think of any forum that would go into as much detail as what we have.   The specialist forum for stuff like that would be thebroadbandforum but you'd have a job getting in because it's designed for specialists in the field.  I'm not being rude when I say this, but they would probably just laugh and tell you it's normal.

Excuse any typos as I can't check, as often long posts may be lost....  I only normally check in to read via this device.

[kitz]

my ISP didn't say it is delibaretely disabled. My ISP only said it is typical inteference on ADSL2+ long lines.

Ah, I see. I am duly corrected. 

However I know it is not normal. Have you any new idea about this gap given D-Link stats?

No, I am sorry. I can deduce absolutely nothing new. 

If I'm looking at the correct graph that's not what I'd call normal long line...  I agree with bcat it looks more like some sort of radio interference or ISP intervention.

But then again both of those were discussed months ago...  Iirc it was one of the first things I said, comparing it to what BE did in the uk on their MSANs.    The notching is another possibility ...     Also another is that the ISP may not disclose if they are doing this,  it took BE quite a while for them to admit it, and even longer before they gave the actual reason why they did it... and BE we're supposedly a technical ISP with techie users who won't be fobbed off easily.

I have absolutely no idea what your ISP is like, but it's not unusual for CS staff to not have any idea how adsl actually works and come up with an excuse that seems half way valid.

[konrado5]

Have you not excluded ISP intervention?

The decline just before and after the blocked tones gives the impression that it is EMI/radio interference.

I've had very low SNR nearly the unused tones and I have negative SNR on some unused tones. Is it proof that it is not ISP intervention?

  Iirc it was one of the first things I said, comparing it to what BE did in the uk on their MSANs.

Had they disabled determinate tones? I have slight differences between synchronizations and router eg. on D-Link I have 415-473 gap, on TP-Link 410-477, sometimes I had 410 tone disabled, sometimes not.

Following is significant:

All three graphs, Hlog, SNR & Qln, have an anomalous region at tones 411 - 476.

  The errors will cause retransmits so viewing some web pages may feel laggy.

I have not noticed any lags. I have about 80-100 errored seconds per hour on fast path with no INP. Perhaps high stability with very low SNR margin is additional crux on my circuit.

Best regards
konrado5

[Loading]

The decline just before and after the blocked tones gives the impression that it is EMI/radio interference.

I've had very low SNR nearly the unused tones and I have negative SNR on some unused tones. Is it proof that it is not ISP intervention?

I had the same with certain negative tones, and it's due to the rfi from the radio stations, which are worse at night. As Kitz had previously said.

[konrado5]

I had the same with certain negative tones, and it's due to the rfi from the radio stations, which are worse at night. As Kitz had previously said.

But I have it continously (radio should be disabled on the moment since October 2013). Moreover, nearly the unused tones I have very low but also very stable SNR. Changes doesn't exceed 1 dB.

If I'm looking at the correct graph that's not what I'd call normal long line...

What would you reply to ISP when they said it it normal long line? I'd like to explain it to my ISP. Perhaps they don't take me seriously because I had reported attenuation rasing about 0.5 dB.

Best regards
konrado5

[Loading]

My snr jumps 1db every couple of seconds, and do you not have any am stations where you live?

[konrado5]

My snr jumps 1db every couple of seconds, and do you not have any am stations where you live?

As far as I know i don't have any am stations.

[kitz]

If I'm looking at the correct graph that's not what I'd call normal long line...

What would you reply to ISP when they said it it normal long line? I'd like to explain it to my ISP. Perhaps they don't take me seriously because I had reported attenuation rasing about 0.5 dB.

This is going to be short because just right now Im not in the best of moods and Im knee deep in cables and crap.... and sick of running round like a blue arse fly trying to sort stuff for others when I have a zillion and one things of my own to do.   It doesnt help that I knocked my own PPP connection into touch whilst trying to do something else. *


But
1) Your line isnt long
2) Your line doesnt show the symptoms of tail off of a long line like this  (look at the grey bars behind the yellow)



oh and whilst on that page...  look at the other graphs...
http://www.kitz.co.uk/adsl/adsl_technology.htm

such as this one



Does something look familiar?

Note the gaps and decline in SNR (yellow lines) at around tones 203 - 211.
This is indicative of additional noise/interference in the frequency ranges 875kHz - 914kHz
The decline in SNR at these frequencies has had a corresponding effect on the bit loading so that either side only 2 bits per bin have been able to load.

Read more: http://www.kitz.co.uk/adsl/adsl_technology.htm#ixzz2yFG4JSsX

In that particular case some of it was resolved by checking all the internal wiring..   but some of it was just local background noise..
 

Theres LOTS of lines that can see the same symptoms - its just the way adsl works.

I Really am at a loss at what to say...  we suggested everything it could be, weve tried to help all we can....Weve given you far more help above and beyond any where else would...   but we cant perform miracles. 


----
*PS
Paul.. if you read this, Im not in any way shape or form mad at you....  just annoyed at myself for not realising the knock on effect it would have

[boost]

I had the same with certain negative tones, and it's due to the rfi from the radio stations, which are worse at night. As Kitz had previously said.

But I have it continously (radio should be disabled on the moment since October 2013). Moreover, nearly the unused tones I have very low but also very stable SNR. Changes doesn't exceed 1 dB.

If I'm looking at the correct graph that's not what I'd call normal long line...

What would you reply to ISP when they said it it normal long line? I'd like to explain it to my ISP. Perhaps they don't take me seriously because I had reported attenuation rasing about 0.5 dB.

Best regards
konrado5

Hi Konrado

I found this... does it help at all?
SpectrumWiki - you've already mentioned something about radio should be off from Oct 13?

ITU Region 1 Allocations
1635 - 1800 kHz
FIXED
MARITIME MOBILE 5.90
LAND MOBILE

[5.92 5.96]

5.92 Some countries of Region 1 use radiodetermination systems in the bands 1606.5-1625 kHz, 1635-1800 kHz, 1850-2160 kHz, 2194-2300 kHz, 2502-2850 kHz and 3500-3800 kHz, subject to agreement obtained under No. 9.21. The radiated mean power of these stations shall not exceed 50 W.

5.96 In Germany, Armenia, Austria, Azerbaijan, Belarus, Denmark, Estonia, the Russian Federation, Finland, Georgia, Hungary, Ireland, Iceland, Israel, Kazakhstan, Latvia, Liechtenstein, Lithuania, Malta, Moldova, Norway, Uzbekistan, Poland, Kyrgyzstan, Slovakia, the Czech Rep., the United Kingdom, Sweden, Switzerland, Tajikistan, Turkmenistan and Ukraine, administrations may allocate up to 200 kHz to their amateur service in the bands 1715-1800 kHz and 1850-2000 kHz.

New DSLAMs, new PSD masks...? Do masked frequencies typically show some SNR or would they be zeroed out or something?
I'm just wondering if 'long line' is getting lost in translation here?
Did they, perhaps, mean at the far end of the ADSL2+ frequencies? 1800KHz~ - 2200KHz~?

I'm still trying to figure half this stuff out myself

[konrado5]

QRG: 145.312.5 MHz. This is possible radio in my town. However it is 3.7 KM away from my home.
http://aprs.fi/info/?call=SP9KLF

Best regards
konrado5

[konrado5]

Is it possible the ham radio 3.7 KM far away causes so much interferences?

Best regards
konrado5

[konrado5]

Sorry, I made mistake. Distance from my home to possible ham radio is 2.1 KM. Distance from MSAN to possible ham radio is 1.6 KM.

Best regards
konrado5

[GigabitEthernet]

Sorry, I made mistake. Distance from my home to possible ham radio is 2.1 KM. Distance from MSAN to possible ham radio is 1.6 KM.

Best regards
konrado5

Nope, no chance.

[konrado5]

Thank you very much for your all help.
Now I'm almost sure  there is 160m band notch-filter somewhere on the line.

Let us suppose that your CP/ISP has notch-filters, tuned to reject the 160 metre amateur radio band, available to deploy on circuits that are affected by such transmissions. Now let us suppose that at the time when your circuit was connected to the other DSLAM, one of those 160 metre amateur radio band notch-filters was inadvertently connected into your circuit. The end result: no xDSL signal will be present on those tones. 

It is noticable on QLN graph. If it was radio interferences QLN would be higher on 1800-2000 kHz than on the other frequiences. On QLN graph there is contrary. The interferences are very small (graph is very low). The conclusion is: on 1800-2000 kHz there are both very weak signal and very weak radio inteferences. Therefore, there is 160m band notch-filter somewhere.

Best regards
konrado5