Print

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

[Mick]

I have noticed some changes in my attenuation which I cannot explain. 

My attenuation used to be rock solid.  No matter what time of the day I resync'ed I would get 39.8dB.  The only time I would get a different reading is if I used a different router.

Since about 9 months ago things have been more variable.  Most of the summer I had 40.0dB and occasionally I would get 39.9dB.  Two days ago I resync'ed and noticed that my attenuation had increased to 40.0 after a few weeks at 39.9dB.   Last night there was a lot of noise and eventually the line dropped.  I resync'ed this morning and to my surprise the attenuation has decreased to 39.8dB.

Perhaps 0.2dB is not much of a difference, but why is this happening lately when previously the attenuation was so much more stable?

[roseway]

These really are very small differences. The attenuation is a sort of weighted average of the attenuations of the individual frequencies which make up the ADSL signal. If your connection has recently been experiencing some interference, then it's quite likely that the frequency bins will be allocated somewhat differently after a re-sync, and this can result in a slightly different reported attenuation.

[Mick]

Thanks Roseway,

I didn't know that changes in bitloading affect attenuation.  I thought that it was the other way around:  good attenuation = good SNR = little bitloading.  If noise increases after initial sync the bitloading increases to compensate at the given attenuation, while attenuation itself stays static until the next resync.  Did I have all this wrong in my head?

[roseway]

Bitloading isn't a quantity which increases and decreases. It refers to how many data bits are allocated to each bin. As noise conditions vary, the router will swap bits between bins to maintain the best noise margin it can. In theory this could result in small changes in reported attenuation, but I think in practice the router probably only calculates the attenuation at the time of sync, and after that it stays the same until the next resync.

There's a much more complete description of bitloading on the ADSL Technology and DMT page.

[Mick]

Bitloading isn't a quantity which increases and decreases. It refers to how many data bits are allocated to each bin.

Right, which in my less than eloquent description was stated as a little/more bitloading.  I think we are saying the same thing.  So, the router stats only illustrate what it gets at the time of sync, although in reality attenuation varies as noise fluctuates.

[jeffbb]

Hi
quote So, the router stats only illustrate what it gets at the time of sync, although in reality attenuation varies as noise fluctuates.

Not quite right. Noise fluctuation may not cause any change to bit allocation.

when you look at your stats they are correct for that time. What might mislead a little is that it is said that attenuation does NOT change That is true as far as it goes . During a session the total bitloading does not change , you do have bitswaping the effect on the calculated attenuation would be very small. Each tone has its own attenuation , the average of all tones  is the calculated attenuation. **

If you resynch then a new bit allocation is made ,it is then possible that because of line condition at that time for a slightly different loading say favouring tones with a lower attenuation so fractionaly changing the calculated  attenuation. Also things like temperature will alter attenuation of the line . There is also the rounding effect .
31.44 calculated goes to 31.4  31.45 goes to 31.5  as an example . when it is said that attenuation does not change it is generally understood to be say <  0.5 db or thereabouts
That is unless you go to adsl2+ from ADSL where attenuation will increase because of the higher frequencies in use .

** I have tried averaging the attenuation and have not had the same figures as those reported ,so maybe its not a simple averaging ?
Also found that SNR margin was not an average of bits used ,all bits or a particular bit .??
Regards Jeff