Constant unexpected Internet Drops

Hi,
I my ISP is AOL UK on BT phone line. I use a Belkin N600 router. The internet connection seems to constantly drop through out the day. I have found these ADSL stats on the router. Is there any problem with these?

ADSL
Type:Interleave Path
Status:Show Time
Downstream Upstream
Data Rate (kbps) 13253 448
Noise Margin (dB) 5.0 36.1
Attenuation (dB) 27.3 13.9
Output power(dbm) 20.4 12.2

Thanks,
Red_Dragon23

Based on http://www.coolwebhome.co.uk/calc/calculator.php

Your connection is just below expected range, and I guess its AOL LLU due tot he slow upstream.

So your speed does suggest there may be some instability, first step is to eliminate the phone wiring in the home by using the test socket that is accessed by removing the lower half of the master socket, which then disconnects all extension wiring in the property.

See the stats when you do that may help us to advise on next course of action.

The internet has still been dropping since and the SNR got down as low as 3dB when I check up on it.
Here's the results using the test socket.
Screenshot of ADSL Stats using test socket
After reinserting the front plate again and here's the stats.
Stats using socket - Start
And Hour later...
Stats using the Socket - 1 Hour Later

Red_Dragon23

The link to test socket stats did not work, the other two did

In reply to a post by Red_Dragon23:

Here's the results using the test socket.
Screenshot of ADSL Stats using test socket

Don't have your login for this.

Just copy & paste ADSL section; that is all we need.

You should be getting 17 Meg sync. I get 19 on same attenuation but lower 3 dB NM.

Oh here's the stats:

ADSL
Type Interleave Path
Status Show Time

Downstream Upstream
Data Rate (kbps) 15615 448
Noise Margin (dB) 9.0 33.1
Attenuation (dB) 26.9 13.7
Output power(dbm) 20.4 12.3

NEW LINK TO IMAGE

In reply to a post by MrSaffron:

Based on http://www.coolwebhome.co.uk/calc/calculator.php

I am confused, puzzled and unconfident by your calculator.

I enter 20143 K, 26dB Attn, 3dB NM and get the result:

Estimated connection speed using ADSL2+ mode (maximum 24 Mbps)

Normal speed range at 26dB attenuation is 15000 Kbps to 18000 Kbps
Target Margin Estimate
9dB 9456
6dB 13801
3dB 20143

1. The Normal speed range is patently untrue, as I'm getting faster, unless it means the range of the population of users at this attenuation connecting with the standard NM of 6 dB. In which case, it should make that clear.
2. On the Estimates at various NMs: The one @ 3dB is obviously a restatement of my input figures. The others @ 6 & 9 dB seem far too low. In my case I have got about 17 Meg @ 6dB NM ant 15 Meg @ 9 dB NM.

So a sample of one makes a tool useless?

A 20143 sync at 26dB attenuation is pretty exceptional and the calculator is about giving mid range expectations, not the odd exceptional line. Perhaps your line is not normal, and the 3dB margin can sometimes be worth a lot more, particularly as the fall off of ADSL2+ is very steep.

Sorry, but I really think it is suspect.

Granted my line may be exceptional , I put in the Kitz figures, which I understand is meant also to cater for mid-range expectations, for my 26 dB attn. line: 17600 K, 26dB attn., 6dB NM and get the results:

Estimated connection speed using ADSL2+ mode (maximum 24 Mbps)

Normal speed range at 26dB attenuation is 15000 Kbps to 18000 Kbps
Target Margin Estimate
9dB 12059
6dB 17600
3dB 22800

Slightly low

This predicts a fantastic +/- 5 Meg gain of speed per -/+ 3 dB delta in NM. Furthermore it predicts my actual speed @ 3 dB NM as slightly below its estimate and so can no longer be regarded as exceptional.

I would not expect them to be identical, but I would be extremely surprised that 2 reputable estimators, yours & Kitz's, would produce vastly different results.

Sorry again, but I think you really need to look at it.

Where on Kitz did you get the figures quote and wording looks very like my site?

Kitz for 26dB says a 17600 sync speed? No playing around with noise margin. At least where I can easily spot

I'm not sure what you are asking.

I got Kitz's estimate for my attn. from: http://www.kitz.co.uk/adsl/max_speed_calc.php and fed them into your calculator which produced the figures I quoted. I am querying those figures.

Kitz assumes as standard:

The calculator is based on a default Target SNR of 6dB.

Kitz does not exactly commit to how much speed increment is due to a 3dB delta in NM other than:

Each 3dB of SNR is worth anywhere between 400 to 1200kbps of speed

but no-one else has ever suggested a fantastic 5 Meg change per 3dB delta in NM.

You calculator and Kitz's fulfil 2 diff functions.

Kitz simply estimates the speed at a given attn. for a standard 6 dB NM.

Yours takes in all 3 stats and 1) rates the current speed on a traffic light basis, 2) simply restates the current speed if the NM falls on a 3dB step and 3) estimates the speed at other 3dB NM steps.

It is the results of the function 3) that I am querying as I find them most suspect.

You said "I put in the Kitz figures, which I understand is meant also to cater for mid-range expectations, for my 26 dB attn. line: 17600 K, 26dB attn., 6dB NM and get the results" and then pasted a set of data that looked like it was from my site, so was worried someone had lifted my code.

So much code I have written in the last few weeks I cannot recall the exact logic and algorithm of everything a the snap of my fingers.

So what is the right speed for a line of 26dB attenuation at 3dB?

The only accurate way to do the calculation would be data from every frequency bin, and even then with noise variability you would be wrong at different times. The calculator is not a total replacement for someone casting an eye over the stats and asking questions - it is also not stated to be 100% accurate.

I get a mixture of too slow emails and too fast emails, which normally suggests about right.

Sorry, perhaps I wasn't as clear as I could have been.

In the context of discussing your calculator, I just meant I put in Kitz-derived figures into it.

OK!

No one expects an absolutely accurate estimate; it is an estimate after all, but nor do they expect widely variables ones.

I do wonder what proportion of your emails relate to the current speed at the presently imposed Target NM and how many to the potential speed at other possible Targets. I wager it is more natural to be biased towards the former.

However, it is the latter I am doubtful about and if they are not reasonable estimates then it brings into question as to their worth.

Edited by XRaySpeX (Thu 21-Feb-13 00:26:15)

My stats of the ADSL has seem to have changed again. Here's the latest stats again below. And you said my sync speed should be around 17meg, how will I go about fixing that?


ADSL

Type Interleave Path
Status Show Time

Downstream Upstream
Data Rate (kbps) 13253 448
Noise Margin (dB) 3.5 33.1
Attenuation (dB) 27.3 13.9
Output power(dbm) 20.4 12.5

In reply to a post by MrSaffron:

I guess its AOL LLU due to the slow upstream.

How does that follow (from a 448K quasi-ADSL Max Up Sync)?

Just had a look at the AOL Forum, LLU exhibits this sort of normal ADSL2+ Up Sync:

In reference to forums.thinkbroadband.com/aol/t/4196222-re-me-or-them.html:

Data Rate (kbps) 12403 1020

Not the the use of the word GUESS, implying not a fact but a hunch.

Given AOL uses TalkTalk LLU and thus available on 94% of lines, would not expect them to use WBC which has a smaller footprint

The noise margin is showing some variation.

Are those latest stats using the test socket or your normal socket. The test socket looked better.

The removal of the ring wire from pin 3 can often help improve stability
http://www.thinkbroadband.com/faq/sections/radsl.htm...

In reply to a post by Red_Dragon23:

And you said my sync speed should be around 17meg, how will I go about fixing that?

Your previous stats from the test socket were in line with my estimate but these aren't.

So the obvious way is to use the test socket. This implies that the extension wiring is pulling the speed down. Removing the ring wire as MrSaffron suggests would a good idea for starters.

Yes I did notice word GUESS. Also noted words DUE TO; that was the FOLLOWING that I was querying.

In fact you no need to guess. It is ADSL2+ speeds on AOL. Therefore it is LLU! AOL do not use WBC anywhere, any more than TT does.

Those results are from the normal socket.

Are those latest stats using the test socket or your normal socket. The test socket looked better.

Okay, I have just done some research online as I didn't want to disconnect the ring wire as it may cause some hassle if BT technicians come around in the future. I have just purchased a iPlate for the master socket instead, I will try that first when it arrives through the post and I will also be posting the new stats on here.

You don't need to remove or cut it; just set it aside.

In reply to a post by MrSaffron:

Given AOL uses TalkTalk LLU and thus available on 94% of lines

Andrew,

FYI Only (I'm not getting at you as you always seem to think):

AOL & TT LLU networks are separate. AOL only has under half the # exchanges LLU's that TT has (1253 vs. 2729).

In reference to www.samknows.com/broadband/llu/aol:

Note that even though AOL UK are technically a part of Carphone Warehouse now, they still run separate networks, which is why they appear separately on this site.

They also package their products differently and AOL BB still depends on AOL Inc. for its user IDs/pwds and eMail, presumably as part of their agreement.

I believe you are wrong, AOL Broadband in the UK used to have its own DSLAM but with the purchase by TalkTalk that all changed.

The 1253 is a 2007 figure that was the height of the networks coverage.

Nothing stopping a LLU provider offering two very different products from the same MSAN of course

Post deleted by MrSaffron

Yes, thanks, MrS. Even the notification email looked suspicious.