Kitz Forum

Broadband Related => Broadband Technology => Topic started by: UncleUB on January 17, 2012, 10:23:39 AM

Title: Not too bad
Post by: UncleUB on January 17, 2012, 10:23:39 AM
For a long line

(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.speedtest.net%2Fresult%2F1712299222.png&hash=064dbbd40e3ae0ebf021343cb1e9c585) (http://www.speedtest.net)


Quote
DSL Type:   G.992.5 annex A

Maximum Bandwidth (Up/Down) [kbps/kbps]:   863 / 4,620

Bandwidth (Up/Down) [kbps/kbps]:   871 / 3,580

Output Power (Up/Down) [dBm]:   12.5 / 17.5

Line Attenuation (Up/Down) [dB]:   41.0 / 65.5

SN Margin (Up/Down) [dB]:   6.0 / 7.5
Title: Re: Not too bad
Post by: roseway on January 17, 2012, 10:29:45 AM
Amazingly good ;D
Title: Re: Not too bad
Post by: burakkucat on January 18, 2012, 01:15:24 AM
For a long line

A long line, which is how long, Uncle?  :-\  (I like to consider all the information, together.)
Title: Re: Not too bad
Post by: roseway on January 18, 2012, 07:15:51 AM
It's this long:

Quote
Line Attenuation (Up/Down) [dB]:   41.0 / 65.5

:)
Title: Re: Not too bad
Post by: UncleUB on January 18, 2012, 08:41:20 AM
This is the actually(approx) distance quoted from my house to the exchange

Quote
Exchange: Attercliffe   BT Code: SLAC
Location: Titterton Street,   S9 3TE
Distance:-   Direct:         1.85 km
    (appx)*   By Road:    3.7 km
Title: Re: Not too bad
Post by: Oranged on January 18, 2012, 09:06:54 AM
This is the actually(approx) distance quoted from my house to the exchange

Quote
Exchange: Attercliffe   BT Code: SLAC
Location: Titterton Street,   S9 3TE
Distance:-   Direct:         1.85 km
    (appx)*   By Road:    3.7 km

That's strange, my attenuation is 38dB and this is my exchange distance result :
Quote
Distance:-    Direct:          2.05 km
     (appx)*    By Road:    3.38 km

I would have expected your road distance to have been significantly longer than mine.
Title: Re: Not too bad
Post by: camallison on January 18, 2012, 11:39:10 AM
Ah yes, but Unky's are Yorkshire miles, and they are a heck of a lot longer than your British Standard mile  ;)

Colin
Title: Re: Not too bad
Post by: burakkucat on January 19, 2012, 03:09:40 AM
It's this long:

Quote
Line Attenuation (Up/Down) [dB]:   41.0 / 65.5

:)

b*cat gives Eric a quick hiss.  >:(  The attenuation can only be related to the physical line length when standard gauge copper is used, with no narrow gauge aluminium, an average number of joints, no "blue beans", gel crimps only, etc, etc, as you know. If a line is in incredibly poor condition, the physical length could be 0.5 mile but still have that 65.5 dB downstream attenuation!  :P

And, as we also know, a modem that reports 65.5 dB is probably saying "I'm unable to quote a proper value for the attenuation".  :-X

b*cat lowers the fur on his back and comes out of grumpy-mode. There now follows the data for my line. So where does it fit into the scale of very short, short, medium, long and very long, please?

(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fspeedtest.net%2Fresult%2F1702430006.png&hash=42f70796325a559f952781d44daf958e)

Code: [Select]
DSL                         Down         Up

Current Rate:                 5305 kbs 1008 kbs
Max Rate:                 5305 kbs 1024 kbs

Current Connection:
Current Noise Margin: 3.2 dB 6.8 dB
Current Attenuation: 45.7 dB 27.3 dB
Current Output Power: 19.4 dB 12.4 dB

Having measured my D-side cable length (with a 301C tester) I know that there is exactly one quarter of a mile of cable (440 yards / 400 metres) between my home and the PCP.

Code: [Select]
Exchange: Bury St Edmunds BT Code: EABSE
Location: Whiting St, IP33 1NS
Distance:- Direct:    1.8 km
  (appx)* By Road:   2.4 km
Title: Re: Not too bad
Post by: Oranged on January 19, 2012, 09:58:32 AM
So the conclusion that I draw when comparing those 3 sets of data is that there is no reliable relationship between attenuation and road distance to the exchange.
Title: Re: Not too bad
Post by: roseway on January 19, 2012, 10:14:28 AM
Quote
And, as we also know, a modem that reports 65.5 dB is probably saying "I'm unable to quote a proper value for the attenuation".

I'm not sure that's quite right. Many modems won't report anything over 63.5 dB, because this is the highest value specified in the G.992.1 standard. See http://www.kitz.co.uk/adsl/linestats.htm#attenuation_63dB . But if a modem is reporting more than 63.5 dB, that implies that it's not using the 63.5 dB 'cut-off' and is reporting a correct value.
Title: Re: Not too bad
Post by: CurlyWhirly on January 19, 2012, 03:55:34 PM
For a long line

(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.speedtest.net%2Fresult%2F1712299222.png&hash=064dbbd40e3ae0ebf021343cb1e9c585) (http://www.speedtest.net)


Quote
DSL Type:   G.992.5 annex A

Maximum Bandwidth (Up/Down) [kbps/kbps]:   863 / 4,620

Bandwidth (Up/Down) [kbps/kbps]:   871 / 3,580

Output Power (Up/Down) [dBm]:   12.5 / 17.5

Line Attenuation (Up/Down) [dB]:   41.0 / 65.5

SN Margin (Up/Down) [dB]:   6.0 / 7.5
That's excellent as the attenuation on my line is 'only' 55 dB and I only get around the same bandwidth as you !

Mind you, I live on a main road and with a nearby street light and at night when the street light comes on, this affects the broadband signal as my line is only around a foot away from the street light at one point.
Title: Re: Not too bad
Post by: burakkucat on January 19, 2012, 09:25:53 PM
Quote
And, as we also know, a modem that reports 65.5 dB is probably saying "I'm unable to quote a proper value for the attenuation".

I'm not sure that's quite right. Many modems won't report anything over 63.5 dB, because this is the highest value specified in the G.992.1 standard. See http://www.kitz.co.uk/adsl/linestats.htm#attenuation_63dB . But if a modem is reporting more than 63.5 dB, that implies that it's not using the 63.5 dB 'cut-off' and is reporting a correct value.

Ah, that you for that correction, Eric. b*cat bows, in apology.  ;)
Title: Re: Not too bad
Post by: burakkucat on January 19, 2012, 09:29:11 PM
So the conclusion that I draw when comparing those 3 sets of data is that there is no reliable relationship between attenuation and road distance to the exchange.

Correct. Other than a very broad "rule of thumb", to be taken with a big "pinch of salt".  :-X
Title: Re: Not too bad
Post by: waltergmw on January 20, 2012, 12:48:54 AM
@ Eric,

Quote
Many modems won't report anything over 63.5 dB, because this is the highest value specified in the G.992.1 standard.

I expect you know that the 2700 has the best of both worlds with the restricted value in the normal user area but TWO actual values on the diagnostic page.

Kind regards,
Walter
Title: Re: Not too bad
Post by: roseway on January 20, 2012, 08:12:28 AM
@Walter,

Yes, and it also reports the specific value at 300 kHz, which might possibly be useful in some cases.
Title: Re: Not too bad
Post by: Oranged on January 20, 2012, 11:03:42 AM
So the conclusion that I draw when comparing those 3 sets of data is that there is no reliable relationship between attenuation and road distance to the exchange.

Correct. Other than a very broad "rule of thumb", to be taken with a big "pinch of salt".  :-X

So why even begin to associate straight line distance and road distance with attenuation......it may as well be the number of lamp posts between you and the exchange.

Logically the further you are from the exchange, the more lamp posts, cats eyes in the road (apologies bc), telephone poles etc etc. there will be.
Title: Re: Not too bad
Post by: Bald_Eagle1 on January 20, 2012, 11:21:57 AM
So the conclusion that I draw when comparing those 3 sets of data is that there is no reliable relationship between attenuation and road distance to the exchange.

Correct. Other than a very broad "rule of thumb", to be taken with a big "pinch of salt".  :-X

So why even begin to associate straight line distance and road distance with attenuation......it may as well be the number of lamp posts between you and the exchange.

Logically the further you are from the exchange, the more lamp posts, cats eyes in the road (apologies bc), telephone poles etc etc. there will be.


Just a little food for thought..............

Logic, BT and ISPs don't always fit nicely in the same sentence.

I think attenuation AND physical distance from the exchange/cabinet need to be considered together.

e.g. Using the Kitz ADSL calculator, a DS attenuation of 30dB suggests an Approx Line Length of 2.2 km.

If a given user knows (for a fact) that their own line length is say, only 1000m, but they are seeing 30dB DS attenuation, it could indicate a severe, repairable fault (High Resitance etc.).

It may just be that their connection includes a lot of aluminium cable, or the copper has corroded down in thickness over time (in which cases there would be very little, if anything, the user could do about it).
 
Title: Re: Not too bad
Post by: roseway on January 20, 2012, 11:35:05 AM
The figure of 14 dB per kilometre has always been reckoned to be a reasonably typical figure for the UK (and indeed for many other countries too), but as you say, there are many ways in which the actual value can be quite a bit higher. There's a lot more to it than the electrical resistance of the cable of course - we are dealing with AM radio frequencies being carried over cables designed for low frequency audio. So the largest element in the attenuation is not the resistance, it's the reactive effects (capacitance and inductance), and these will vary depending on the type of cable, its proximity to other cables and metal objects, and so on.
Title: Re: Not too bad
Post by: Bald_Eagle1 on January 20, 2012, 11:56:31 AM
If I'm not mistaken the Kitz calculator uses 13.81 dB per km (I like exactness  ;)) at an assumed SNRM of 6dB, also assuming 0.5mm copper (AWG24 or AWG 26 - can't remember which) for ADSL frequencies.

It's probably a lot different again for the higher VDSL2 frequencies.

Before razpag departed, he mentioned that quite often BT's twisted copper pairs have corroded to less than 0.4mm in thickness, thus throwing yet another spanner in the works to consider.

As b*cat mentioned:-
Quote
Correct. Other than a very broad "rule of thumb", to be taken with a big "pinch of salt".
Title: Re: Not too bad
Post by: Oranged on January 20, 2012, 03:45:28 PM
If I'm not mistaken the Kitz calculator uses 13.81 dB per km (I like exactness  ;)) at an assumed SNRM of 6dB, also assuming 0.5mm copper (AWG24 or AWG 26 - can't remember which) for ADSL frequencies.

Lot of assuming going on for what is a calculation using technical data to 2 decimal places  ;)

Don't get me wrong I've always "assumed" that distance and attenuation could be loosely related but since I posted the query comparing UncleUB's, burakkucat's and my stats, I'm not so sure it's realistic.
Title: Re: Not too bad
Post by: Bald_Eagle1 on January 20, 2012, 08:50:36 PM

Lot of assuming going on for what is a calculation using technical data to 2 decimal places  ;)


Hey, don't shoot the messenger  ;D

From http://whirlpool.net.au/wiki/DSLAM_speeds

Estimating distance and speed

The standard signal attenuation spread for a given speed is somewhere in the region of 15-20dB for ADSL2/2+ speeds and 25-30dB for ADSL1 speeds. That is, we see users at 15Mbit with signal attenuation between 0-48dB at the extremes and the statistically significant around 12-32dB. For a given signal attenuation level we see a statistically significant spread of 10Mbit. And by statistically significant, an even spread over the range for a sample set of ~50,000 connections.

A way to "guesstimate" the line attenuation is the following: (originally posted by Just 'Adam')

Less than 1km should be 23-24Mbit
1.0km = 13.81dB = 23Mbit
1.5km = 20.7dB = 21Mbit
2.0km = 27.6dB = 18Mbit
2.5km = 34.5dB = 13Mbit
3.0km = 41.4dB = 8Mbit
3.5km = 48.3dB = 6Mbit
4.0km = 56dB = 4Mbit
4.5km = 62.1dB = 3Mbit
5.0km = 69dB = 2Mbit

A Microsoft Excel document containing a calculator to obtain these figures can be downloaded http://users.on.net/richard.moulynox/whirlpool/adsl2calculatorv1.xls. Current Version 1.4.

Until last June, when I had an ADSLMax connection, I always felt privileged to get 1Mb (on a good day).

Apparently the distance from the exchange is reported as
:
:
:
:
:
:
:
:
:
:
:
5283m (approximately)  :lol:


The attachment shows my best ever DS rate, but just look at the US rate!


Title: Re: Not too bad
Post by: Oranged on January 21, 2012, 12:26:59 AM

Hey, don't shoot the messenger  ;D

No I wasn't....it's just my slant on "21 century technology".....you would have thought by now all the relevant data and stats associated with ADSL and vDSL could be relied upon wouldn't you ?
Title: Re: Not too bad
Post by: asbokid on January 21, 2012, 01:47:52 AM
RichardM kindly replied to queries about his spreadsheet formula for calculating likely connection speeds.

Today, we can probably build on this formula using the attenuation data [Hlog(f)] that is available for each frequency across the entire VDSL2 tone map.

The attenuation (aka insertion loss) data could be used in conjunction with other data, e.g. the constants defined in the technical specifications for BT cable reference models.

A function could be developed to use this data for deriving estimates of unknown characteristics of the loop, e.g. the loop length, its theoretical Shannon capacity and the electromechanical properties of the loop cabling itself.


[1] http://www.ja.net/documents/development/llu/llu-technical-document.pdf
[2] http://www.sinet.bt.com/349v2p3.pdf
[3] http://www.sinet.bt.com/351v4p5.pdf
[4] http://ftp.tiaonline.org/TR-30..ETSI%20Documents/970p02r3%20ETSI%20Cable%20Reference%20Models.pdf (http://ftp.tiaonline.org/TR-30/TR-30.3/Public/0206%20Columbia/ETSI%20Documents/970p02r3%20ETSI%20Cable%20Reference%20Models.pdf)
[5] http://www.niccstandards.org.uk/files/current/nd1601_2005_05.pdf?type=pdf
[6] http://stakeholders.ofcom.org.uk/binaries/research/technology-research/asses.pdf
Title: Re: Not too bad
Post by: burakkucat on January 21, 2012, 03:46:56 AM
You have provided us with links to six reference documents, for our edification. Thank you.

It would be nice to know at which point in your post that each of those documents is contextually relevant.

I am currently obtaining a "Problem loading page" with BT's SIN site (http://www.sinet.bt.com/). Hopefully that is just a temporary glitch.

Perhaps you could persuade Mr Eagle to develop something along the lines of your musing . . .  :-\
Title: Re: Not too bad
Post by: Bald_Eagle1 on January 21, 2012, 08:14:33 AM
Hi b*cat,

I am currently obtaining a "Problem loading page" with BT's SIN site (http://www.sinet.bt.com/). Hopefully that is just a temporary glitch.

It was working earlier this week, so as you say, hopefully that is just a temporary glitch.

Quote
Perhaps you could persuade Mr Eagle to develop something along the lines of your musing . . .  :-\

As we are now moving into the realms of VDSL2 & FTTC etc. I have started another thread in the FTTC section.
FTTC - Line Length and Attenuation Calculations (http://forum.kitz.co.uk/index.php/topic,10566.0.html)

I did have a half-hearted stab at this some time ago, but didn't fully understand matters at the time (still donít) :-
http://forum.kitz.co.uk/index.php/topic,9726.msg198765.html#msg198765
Title: Re: Not too bad
Post by: Bald_Eagle1 on January 21, 2012, 08:31:43 AM
Hi asbokid,


A function could be developed to use this data for deriving estimates of unknown characteristics of the loop, e.g. the loop length, its theoretical Shannon capacity and the electromechanical properties of the loop cabling itself.


Thanks for the links. Interesting stuff.

To avoid clogging up this thread with VDSL2/FTTC specific matters, I have started another thread in the FTTC section:-
 FTTC - Line Length and Attenuation Calculations (http://forum.kitz.co.uk/index.php/topic,10566.0.html)


Paul.
Title: Re: Not too bad
Post by: asbokid on January 21, 2012, 06:15:36 PM
I will try and thresh out a coherent post around the documents.

BT SIN351 can also be downloaded from here: https://issues.asterisk.org/jira/browse/ASTERISK-11859   (The page is amusing. Apparently BT's overnight testing of its Metallic Path Facilities caused Asterisk PBX servers to crash!)

EDIT:  BT SINet is up and running again.