Interferences only at first minutes of connection?

What can be cause low SNR margin for upstream only at few first minutes of connection? Is it erronously measured SNR margin? It happens since my ISP changed DSLAM. To avoid misunderstanding, I have in mind SNR margin about 3 dB lower than normal on upstream during first minutes of connection. It seems I have some interferences only during some first minutes of new synchronization. Sometimes my router reports for a one second strange upstream SNR margin: 96.2 dB. At one connection SNR margin jumped around: 0 dB and 102.9 dB and I became disconnected after 20 minutes.

However except first minutes of connection, my upstream is very stable. I had at some time above 10 days period with 0 CRC errors. My SNR margin is continously 5.7-6.3 dB from 20 days. However at first three minutes it was: 3.7 dB.

I will be grateful if someone explain me this crux. Is it any line fault?

Best regards
konrado5

Edited by konrado5 (Tue 04-Feb-14 16:58:29)

In reply to a post by konrado5:

Sometimes my router reports for a one second strange upstream SNR margin: 96.2 dB. At one connection SNR margin jumped around: 0 dB and 102.9 dB

Your router is faulty, at least as far as its display of stats is concerned, and therefore cannot be relied on for the type of analysis you are attempting.

You sure it is not reporting in units of 0.1db? Some do, you know?

In reply to a post by XRaySpeX:

]Your router is faulty, at least as far as its display of stats is concerned, and therefore cannot be relied on for the type of analysis you are attempting.

My router reports it since my ISP changed DSLAM.

You sure it is not reporting in units of 0.1db? Some do, you know?

My router reports SNR margin properly in 0.1 dB accurracy except from first minutes of new connection for upstream.

Still wrong figures. Maybe the router has a bug that only shows up with that specific dslam config

In reply to a post by MrSaffron:

Still wrong figures. Maybe the router has a bug that only shows up with that specific dslam config

And only during first five minutes of connection? Morever, it takes place usually but not always. At the first minutes of connection there are many more bitswaps reported by router and more errored seconds (only on upstream).

Edited by konrado5 (Tue 04-Feb-14 21:44:31)

In reply to a post by konrado5:

My router reports SNR margin properly in 0.1 dB accuracy

I never said 'accuracy', I said 'units'! Do you understand the difference? You should do if you are to attempt this mathematical analysis.

Most routers report in 1 dB units, which is the same as 0.1 B(el) units, usually to an accuracy of 0.1 dB, e.g 'Noise Margin (dB): 6.3'.

A few reuters report in 0.1 dB units, probably to an accuracy of 0.01 dB, e.g 'Noise Margin (0.1 dB): 63.5'.

My router reports in 1 dB units.

Edited by konrado5 (Tue 04-Feb-14 22:33:24)

So maybe the code bitswaps for a while and then settles done.

You the router microcode to know for sure what is happening.

In reply to a post by MrSaffron:

So maybe the code bitswaps for a while and then settles done.

You the router microcode to know for sure what is happening.

Thank you for reply but I don't understand. Does bitswap activity causes lowered SNR margin measurements?

Best regards
konrado5

More often called centiBels !

No doubt by scientists, but not by router manufacturers, as this typical one here:

In reply to a post by Loading:

SNR Margin (0.1 dB): 22 70
Attenuation (0.1 dB): 635 315
Output Power (0.1 dBm): 161 124

Tho' not as accurate as I imagined; only to the nearest CeeBees .

Edited by XRaySpeX (Wed 05-Feb-14 00:15:20)

In reply to a post by XRaySpeX:

No doubt by scientists, but not by router manufacturers, as this typical one here:

In reply to a post by Loading:

SNR Margin (0.1 dB): 22 70
Attenuation (0.1 dB): 635 315
Output Power (0.1 dBm): 161 124

Tho' not as accurate as I imagined; only to the nearest CeeBees .

That is because they either do not understand the concept of Bels and deciBels &c or they cannot be bothered to code correctly to change the reported figure into dBs ...



Would you define a length in terms of 0.1cm units or would you use mm ?

You don't have a line fault. If your aeroplane is steady in the air, but the altimeter needle is all over the place, what would you do?

Have you tried a different modem? The upstream snr is retrieved by the modem from the ATU-C using the extended operations channel. I've seen modems before suffer various glitches in this operation - one popular model flashes zero at regular intervals for example.

In reply to a post by MHC:

More often called centiBels !

In reply to a post by MHC:

That is because they either do not understand the concept of Bels and deciBels

They are bels, decibels and centibels, actually.

Bit swaping can and does change the SNR margin displayed, but how much will depend on the bins played with and the algorithm for how the SNR margin is displayed

You don't have a line fault. If your aeroplane is steady in the air, but the altimeter needle is all over the place, what would you do?

Sound advice.

But then given how many people end up following a GPS blindly many seem to believe the computer rather than employ some common sense, or heaven forbid look at a road sign

Yep, I'd agree with that too, you don't learn nowt following a sat-nav.