[help] DSL Stats. Normal? in AT&T Northeast http://www.dslreports.com/forum/r20370391 en Sat, 26 Jul 2008 02:04:59 EDT Sat, 26 Jul 2008 02:04:59 EDT Re: [help] DSL Stats. Normal? http://www.dslreports.com/forum/remark,20416800 NormanS :
said by usweefive See Profile :

The attenuation was low due to a bad connection.
I would expect an open to raise the attenuation, not lower it. An effective impedance of infinite ohms should provide perfect attenuation. It was my S/N ratio which was low.

I believe the reason the DSL worked into an open circuit is that it operates at an RF freqency, so an extremely short gap, while looking "open" to DC current, still looks like a usable path to AC.
--
Norman
~Oh Lord, why have you come
~To Konnyu, with the Lion and the Drum]]> http://www.dslreports.com/forum/remark,20416800 Thu, 01 May 2008 14:30:21 EDT Re: [help] DSL Stats. Normal? http://www.dslreports.com/forum/remark,20415276 usweefive : There you go ! The attenuation was low due to a bad connection. There is a reason why you had partial success: the differential input of the DSL modem. Twisted pair systems use common-mode rejection similar to an Op Amp where common noise on a pair of wires is canceled out. The DSL modem's phone line input was operating single-input instead of dual and there was no common-mode rejection possible... ergo the very low noise figure incoming.]]> http://www.dslreports.com/forum/remark,20415276 Thu, 01 May 2008 09:26:51 EDT Re: [help] DSL Stats. Normal? http://www.dslreports.com/forum/remark,20408100 NormanS :
said by usweefive See Profile :

What I meant to say is that in traditional magnetic theory that two parallel conductors in close proximity to the other will have about -45 dB crosstalk between them.
I am reasonably certain that the numbers reported by the modem have no direct relationship with that figure.
This implies that the actual desired signal be above this -45 dB 'noise floor' to be seen as something usable. Manchester encoding is used to ensure valid data streams in DSL data transfers to mak sure packets do not have missing or dropped bits.

Electrically a signal that has 12 dB of signal to noise margin shows that the actual useable signal is only 4 times bigger in amplitude than the noise. ( 6 db = x 2 difference in amplitude ). The signal attenuation of -51dB means very low signal amplitude overall.
In my experience (watching a line deteriorate due to problems in the physical plant) ADSL will be usable down to 9 db noise margin, reported by the modem. Others have reported usability at 6 db; but I have my doubts. Clearly getting too close to the crosstalk interference at that level. I do know that my SS4100 kicked over to a "noise profile" (Interleaved path) at 6 db S/N. But at 3 db it fell apart (constant loss of synch). BTW, phone repair handled that problem; I had lost dial tone due to an open on the "Ring" side of the loop.
--
Norman
~Oh Lord, why have you come
~To Konnyu, with the Lion and the Drum]]> http://www.dslreports.com/forum/remark,20408100 Tue, 29 Apr 2008 21:33:44 EDT Re: [help] DSL Stats. Normal? http://www.dslreports.com/forum/remark,20407004 usweefive : What I meant to say is that in traditional magnetic theory that two parallel conductors in close proximity to the other will have about -45 dB crosstalk between them.

This implies that the actual desired signal be above this -45 dB 'noise floor' to be seen as something usable. Manchester encoding is used to ensure valid data streams in DSL data transfers to mak sure packets do not have missing or dropped bits.

Electrically a signal that has 12 dB of signal to noise margin shows that the actual useable signal is only 4 times bigger in amplitude than the noise. ( 6 db = x 2 difference in amplitude ). The signal attenuation of -51dB means very low signal amplitude overall.]]> http://www.dslreports.com/forum/remark,20407004 Tue, 29 Apr 2008 17:41:50 EDT Re: [help] DSL Stats. Normal? http://www.dslreports.com/forum/remark,20399991 NormanS :
said by usweefive See Profile :

I am an Electrical Engineer and the downstream value of 35 dB is right next to the approximate 40 to 45 dB levels where twisted pair parallel wires have crosstalk.
His noise margin is only 15 db.

The 34.7 db is attenuation. Are you saying that attenuation of 40 db to 45 db is too high?
From an SS4100. So, I should be severely impaired by crosstalk? Only 3,394 CRC errors since March 11, 2008. My connection is fine.

--
Norman
~Oh Lord, why have you come
~To Konnyu, with the Lion and the Drum]]> http://www.dslreports.com/forum/remark,20399991 Mon, 28 Apr 2008 13:55:55 EDT Re: [help] DSL Stats. Normal? http://www.dslreports.com/forum/remark,20396677 wb8imy : In my instance, the dropout problem appears to have been the modem. After posting the message, I replaced my SpeedStream 5260 with a 2Wire AT&T modem/router combo. Ever since installing the 2Wire device, I've had no dropouts. (Fingers cross)

Steve]]> http://www.dslreports.com/forum/remark,20396677 Sun, 27 Apr 2008 20:09:38 EDT Re: [help] DSL Stats. Normal? http://www.dslreports.com/forum/remark,20396504 usweefive : It just isn't you. My DSL here in East Hampton has been doing the same thing for over 3 weeks at night. Your noise margin levels are the same and the incoming (downstream)dB levels are a bit too large.

I am an Electrical Engineer and the downstream value of 35 dB is right next to the approximate 40 to 45 dB levels where twisted pair parallel wires have crosstalk. I see that outgoing upstream is 16 dB - but in this case the modem is the transmitter outward and the encoding etc. has less overhead so it will look better.

Take a look around and see what other devices you have connected to the phone line, if a connection to a junction box is poor, or perhaps try the Noise and Attenuation tests with just the DSL modem and filter / no filter. Make sure there is on other devices connected - just the 2wire to the phone line. Also, try the Tools section here and put in a request for a Line Test and get the info a few days later. I have similar issues with a DSL line in Wallingford at work and it uses a Speedstream 4100 modem and it also has had these provider disconnects.

Lastly since now I know I am not insane about these drop outs, I will check into DSL support in Texas and ask some questions because I have been thinking it was my 2Wire's issue with a wireless laptop and now I see it on this wired PC at night so something is definetly WRONG. ]]> http://www.dslreports.com/forum/remark,20396504 Sun, 27 Apr 2008 19:27:52 EDT Re: [help] DSL Stats. Normal? http://www.dslreports.com/forum/remark,20370637 DrStrange : Those numbers look OK. What do they look like when it's disconnected?]]> http://www.dslreports.com/forum/remark,20370637 Tue, 22 Apr 2008 19:42:33 EDT [help] DSL Stats. Normal? http://www.dslreports.com/forum/remark,20370391 wb8imy : According to the stats generated by my 2Wire DSL modem/router, I am getting the following noise margins and attenuation on my DSL line. However, I don't have a basis for comparison. Are these figures "typical" or do they indicate a problem? (I've been plagued with regular disconnects, especially late at night.)

Current Noise Margin: 15.0 dB (Downstream) 17.0 dB (Upstream)
Current Attenuation: 34.7 dB (Downstream) 16.0 dB (Upstream)

--Steve
]]> http://www.dslreports.com/forum/remark,20370391 Tue, 22 Apr 2008 18:56:49 EDT