SNR and Loop ATT

Running a Draytek 2600 into Plusnet ADSL, i have an SNR of +/-41, with a
loop attenuation of circa 23.5

Good, bad, could be better, or nothing to worry about?

On Tue, 24 Aug 2004 08:43:54 +0100, in article
<(E-Mail Removed)> Tx2
<tx2newscollection-invalid-@hotmail.com> wrote:

>
> Running a Draytek 2600 into Plusnet ADSL, i have an SNR of +/-41, with a
> loop attenuation of circa 23.5

I have Loop Att 44.0dB with SNR margin of 30 - 35 dB with occasional dips
to lower values. Approx 3.8km from the exchange on 512kbps.

>
> Good, bad, could be better, or nothing to worry about?

Very good, the modem has a sufficient SNR to be a huge amount above any
line noise so you have *absolutely* nothing to worry about.

--

Brian Morrison

please observe reply-to address

From Brian Morrison on 24/Aug/2004 09:09:

> On Tue, 24 Aug 2004 08:43:54 +0100, in article
> <(E-Mail Removed)> Tx2
> <tx2newscollection-invalid-@hotmail.com> wrote:
>
>
>>Running a Draytek 2600 into Plusnet ADSL, i have an SNR of +/-41, with a
>>loop attenuation of circa 23.5
>
>
> I have Loop Att 44.0dB with SNR margin of 30 - 35 dB with occasional dips
> to lower values. Approx 3.8km from the exchange on 512kbps.
>
>
>>Good, bad, could be better, or nothing to worry about?
>
>
> Very good, the modem has a sufficient SNR to be a huge amount above any
> line noise so you have *absolutely* nothing to worry about.
>

SNR 13.0, Att. 62.0 so really close to the "limit" but I get no problems at all.

--
Julian Knight, http://www.knightnet.org.uk/
Sheffield, United Kingdom
Security, Directory, Messaging, Network & PC Consultant
Yahoo! IM=knighjm, Skype Internet Phone: callto://j.knight

On Tue, 24 Aug 2004 10:32:36 +0100, in article
<412b0b2e$0$13410$(E-Mail Removed)> Julian Knight
<news003@[127.0.0.1]> wrote:

>> Very good, the modem has a sufficient SNR to be a huge amount above any
>> line noise so you have *absolutely* nothing to worry about.
>>
>>
> SNR 13.0, Att. 62.0 so really close to the "limit" but I get no problems
> at all.

Indeed, most modems are better than expected and BT's previous 60dB limit
was quite conservative.

At long distances from the exchange provided that the attenuation is low
enough to provide enough signal, the noise will be the determining factor
as to error rate.

--

Brian Morrison

please observe reply-to address

In article <412b0b2e$0$13410$(E-Mail Removed)>,
Julian Knight <news003@[127.0.0.1]> writes
>From Brian Morrison on 24/Aug/2004 09:09:
>
>
>SNR 13.0, Att. 62.0 so really close to the "limit" but I get no problems at all.
>
Is it 'really close' ??

- from previous correspondence on this subject people have suggested the
SNR can get down below 6db before the modem starts losing sync.
Anyone comment?
And, even though BT won't offer you 1Mbps on this attenuation, would you
expect success??
And what about SNR of 13dB and atten of 58db??
Again, anyone comment (preferably from some knowledge :-))
--
robert w hall

In article <(E-Mail Removed)>, Brian
Morrison <(E-Mail Removed)> writes
>On Tue, 24 Aug 2004 13:36:05 +0100, in article
><(E-Mail Removed)> robert w hall
><(E-Mail Removed)> wrote:
>
>> In article <412b0b2e$0$13410$(E-Mail Removed)>,
>> Julian Knight <news003@[127.0.0.1]> writes
>>>From Brian Morrison on 24/Aug/2004 09:09:
>
>Please trim the attribution more carefully, I wrote none of the quoted
>material.
>
>
my apologies

>>>
>>>SNR 13.0, Att. 62.0 so really close to the "limit" but I get no problems
>>>at all.
>>>
>> Is it 'really close' ??
>
>No, clearly not,

why 'clearly' - not to me or I wouldn't have made the point ('clearly')

>but it would be if the SNR margin was below 6dB or so.
>That would imply a high noise level rather than low signal level.
>
>
In what you (I think :-) are saying below I find I'm (I think) getting
confused by Your (I think) differentiation between SNR and noise margin.

SNR= signal to noise ratio (yes?) and is the ratio of two powers

BUT

it is usually expressed as a log. (for many reasons, not the least that
that's the way it comes in Shannon's theorem and similar, and so many
detection systems depnd directly on it)

Since its log goes through zero when the ratio is unity, in my usage the
terms SNR (expressed as a log.) and margin are synonymous.

what's yours that's different, please

When we've clarified this point perhaps we can discuss the remainder!
--
robert w hall

On Tue, 24 Aug 2004 13:36:05 +0100, in article
<(E-Mail Removed)> robert w hall
<(E-Mail Removed)> wrote:

> In article <412b0b2e$0$13410$(E-Mail Removed)>,
> Julian Knight <news003@[127.0.0.1]> writes
>>From Brian Morrison on 24/Aug/2004 09:09:

Please trim the attribution more carefully, I wrote none of the quoted
material.


>>
>>SNR 13.0, Att. 62.0 so really close to the "limit" but I get no problems
>>at all.
>>
> Is it 'really close' ??

No, clearly not, but it would be if the SNR margin was below 6dB or so.
That would imply a high noise level rather than low signal level.


> - from previous correspondence on this subject people have suggested the
> SNR can get down below 6db before the modem starts losing sync. Anyone
> comment?

It can, but usually this is SNR *margin* that is quoted, the absolute SNR
required for low enough error rates depends on implementation, but is
usually around 6-8dB in most digital modems for sensible modulation
schemes. You may need a lot more for less sensible schemes.

> And, even though BT won't offer you 1Mbps on this attenuation, would you
> expect success??

Yes, but it might be a little marginal.

> And what about SNR of 13dB and atten of 58db?? Again, anyone comment
> (preferably from some knowledge :-))

Yes, it just means that there is somewhat more noise than the previous
case, hence signal level is 4dB up, but SNR margin is not.

--

Brian Morrison

please observe reply-to address

In article <(E-Mail Removed)>, Brian
Morrison <(E-Mail Removed)> writes
>On Tue, 24 Aug 2004 13:36:05 +0100, in article
><(E-Mail Removed)> robert w hall
><(E-Mail Removed)> wrote:
>
>> In article <412b0b2e$0$13410$(E-Mail Removed)>,
>> Julian Knight <news003@[127.0.0.1]> writes
>>>From Brian Morrison on 24/Aug/2004 09:09:
>
>Please trim the attribution more carefully, I wrote none of the quoted
>material.

my apologies - but the correct level of quoting is of course correctly
preserved by the carat marks
>
>
>>>
>>>SNR 13.0, Att. 62.0 so really close to the "limit" but I get no problems
>>>at all.
>>>
>> Is it 'really close' ??
>
>No, clearly

not to me or I wouldn't have asked :-)

>not, but it would be if the SNR margin was below 6dB or so.
>That would imply a high noise level rather than low signal level.
>
>
>> - from previous correspondence on this subject people have suggested the
>> SNR can get down below 6db before the modem starts losing sync. Anyone
>> comment?
>
>It can, but usually this is SNR *margin* that is quoted,

er, is it?

>the absolute SNR
>required for low enough error rates depends on implementation,

the 2nd part of this remark appears true, but the first part puzzles me.
It was my understanding that
Signal to Noise Ratio (SNR), (the ratio of signal to noise powers,
usually quoted on a log scale), and
SNR margin, (as reported say from a Speedtouch 510 router or similar,
or quoted in normal usage),
are the same thing.
(ie the 'margin' is relative to unity ratio, or zero db, NOT the
threshold SNR at which the box unsyncs)

But you seem carefully to differentiate between 'absolute SNR' and 'SNR
margin'.


Please help me here

Bob



--
robert w hall

On Tue, 24 Aug 2004 20:27:48 +0100, robert w hall
<(E-Mail Removed)> wrote:

>But you seem carefully to differentiate between 'absolute SNR' and 'SNR
>margin'.

one can envisage some devices reporting the SNR and that is clearly
defined as signal to noise ratio.

however several devices report "SNR margin", now the problem is that
we don't know if SNR margin = SNR or if SNR margin = Actual SNR -
minimum SNR required

so an SNR of 12 dB on a modem needing 6 dB to achieve the specified
error rate may equal an SNR margin of 6 dB.

the proof of the pudding is in the eating - if it works at a margin of
zero (as some do) then it is a margin and not an SNR, if it stops
working below about 6 then its reporting SNR.

How does that sound ?

Phil
--
spamcop.net address commissioned 18/06/04
Come on down !

In article <(E-Mail Removed)>, Phil Thompson
<(E-Mail Removed)> writes
>On Tue, 24 Aug 2004 20:27:48 +0100, robert w hall
><(E-Mail Removed)> wrote:
>
>>But you seem carefully to differentiate between 'absolute SNR' and 'SNR
>>margin'.
>
>one can envisage some devices reporting the SNR and that is clearly
>defined as signal to noise ratio.

yes agreed
>
>however several devices report "SNR margin", now the problem is that
>we don't know if SNR margin = SNR or if SNR margin = Actual SNR -
>minimum SNR required

for that 'trained state of the modem' - see below)

>
>so an SNR of 12 dB on a modem needing 6 dB to achieve the specified
>error rate may equal an SNR margin of 6 dB.
>
>the proof of the pudding is in the eating - if it works at a margin of
>zero (as some do) then it is a margin and not an SNR, if it stops
>working below about 6 then its reporting SNR.
don't forget when SNR gets low, the correct factor is (in theory)
log(1+SNR)
>
>How does that sound ?
>
>Phil

Yes, that certainly seems to be a plausible interpretation of what he's
saying.
And there are articles on the web (google on 'SNR margin' & find eg
www.aware.com/support/x200/faqs.htm) which DO differentiate 'SNR margin'
from SNR, pointing out how that SNR margin changes if you change the
number of signal levels in a channel, and so the data rate (eg a change
of 6db if going from a binary system to a 3 level system).

But I had not previously thought that this margin, which thus varies
depending on how the modem has trained itself, was what bog standard
Alcatel modems reported, rather 20.log(SNR).
(Indeed I asked a similar question about 3 weeks back, and was then
reassured that the 'Noise Margin' reported by an Alcutel 510 was indeed
just 20.log(SNR).
Ideally you'ld expect a good diagnostic to report both, or give you
enough data to deduce both.

For one thing, one is a property of your wires and the other is a
property of wires+(setup state of modem), so you'd expect some
uniformity to be needed.

Doh! I tort I unerstood this last week
--
robert w hall