Kitz Forum

Broadband Related => ADSL Issues => Topic started by: renluop on October 28, 2010, 08:10:03 AM

Title: To this odd mind.... something puzzling from Kitz speed estimator.
Post by: renluop on October 28, 2010, 08:10:03 AM
When I was not able to do anything constructive I listed all the ADSL1 and 2+ speeds listed by Kitz for the various attenuations.

I had expected to see either a fairly smooth  arithmetical or percentage decrease in speed the more the attenuation, but as can be seen this is not the case.

Can anyone explain, please?

[attachment deleted by admin]
Title: Re: To this odd mind.... something puzzling from Kitz speed estimator.
Post by: roseway on October 28, 2010, 08:20:27 AM
No doubt the figures are the result of a formula, but I don't have that to hand at present. One thing to note, though, is that dB are logarithmic values, not linear ones, so you can't sensibly talk about percentage changes in their values - this has no meaning mathematically. A fairly accurate rule of thumb is that a 3 dB increase in attenuation equals a doubling of the arithmetic value.
Title: Re: To this odd mind.... something puzzling from Kitz speed estimator.
Post by: waltergmw on October 28, 2010, 09:10:21 AM
Hi Renluop,

Ezzer provides a little more explanation of dB ratios for signal to noise here too:-

http://forum.kitz.co.uk/index.php/topic,147.msg132979.html#msg132979

Kind regards,
Walter
Title: Re: To this odd mind.... something puzzling from Kitz speed estimator.
Post by: renluop on October 28, 2010, 10:11:12 AM
But even then,would you not expect the actual speed differences to have some pattern?
Title: Re: To this odd mind.... something puzzling from Kitz speed estimator.
Post by: roseway on October 28, 2010, 11:06:13 AM
I'm sure there is a pattern resulting from the formula used, but I'm afraid I don't have the time to research it at the moment.
Title: Re: To this odd mind.... something puzzling from Kitz speed estimator.
Post by: kitz on November 09, 2010, 06:24:16 AM
No - there isnt any smooth line
The calculation can probably be best explained by looking at the graph on this page which shows how unlinear it is.

http://www.internode.on.net/residential/broadband/adsl/extreme/performance/

(http://www.internode.on.net/media/images/internode-adsl2-dist07.jpg)

Off the top of my head I cant recall the exact formula now, other than it took me about 2-3 days to work it out and plot the figures I use, (thats not including coding the app).  Theres various little known elements than can slightly affect the speed so that its not a straight line.  For eg different cable gauge that may be used on longer lines...  or the fact that BT uses different PSD masks to prevent x-talk depending on the length of your line.

Credits to any relevant sources of reference are given at the bottom of the page.
Title: Re: To this odd mind.... something puzzling from Kitz speed estimator.
Post by: 4candles on July 06, 2011, 03:42:16 PM
For eg different cable gauge that may be used on longer lines... 

As an example, there must be some super heavy metal twixt me and the exchange.   :)

With an attenuation of 53dB, the tables would indicate a line length of around 3·8km and a sync about 4128 on ADSLMax.

In fact the road distance is 4·8km, and the erstwhile 17070 test indicated 5·3 km. Sync is around 4900.
Title: Re: To this odd mind.... something puzzling from Kitz speed estimator.
Post by: NewtronStar on September 26, 2013, 12:12:48 AM
It would be interesting to see an Attenuation Speed calculator on FTTC as the Kitz ADSL calculator was always spot on for me in the ADSL days  ;D
Title: Re: To this odd mind.... something puzzling from Kitz speed estimator.
Post by: kitz on October 14, 2013, 10:56:58 AM
I'd need some sort of reliable figures to work with.  With adsl the internode graph was considered fairly accurate.  I then plotted the figures from the graph, the results are what you see in the pdf, although there was a formula used in there too.  One I had some figures, then I was able to code the program.

I'm not sure if there is anything like that graph, nor figures to use.  The other factor is that on adsl most people had routers that would display their attenuation to give them an idea of line length from the exchange.  With fttc its distance from the cab and that figure could prove difficult to get  :(