Spontaneous disconnections and fluctuating speeds

Dear all,

The more I read around about bb, ADSL, exchanges...the more overwhelmed I get with all the various router statistics and diagnostic figures about noise, attenuation etc. I would be grateful if someone could crunch my stats and let me know if what I am experiencing is normal for my exchange and ISP?

I used to have my line rental with BT and calls + bb with AOL. In Nov I took up AOL's deal of switching over my landline to them too and have one simple bill. So atm I pay £13 in line rental and £7 for unlimited bb and unlimited calls (i've been with them for years, hence the free upgrades to unlimited).

Since doing this, there was a week when there was no dialtone and incoming calls would give an 'engaged' tone...although ADSL would work. Several calls to tech support seemed to resolve that...apparently it was a problem at the exchange and not in my line.

I usually get 5-6 Mbps downstream...but since the switch I get 3-4 Mbps according to speedtests and sometimes I lose connection and speeds dwindle to 0.5-0.6 Mbps. What could be behind this?

Router stats:

Link Information

Uptime: 0 days, 13:40:07
Modulation: G.992.3 annex A
Bandwidth (Up/Down) [kbps/kbps]: 960 / 5,803
Data Transferred (Sent/Received) [MB/GB]: 172.06 / 2.41
Output Power (Up/Down) [dBm]: 12.5 / 0.0
Line Attenuation (Up/Down) [dB]: 24.5 / 43.0
SN Margin (Up/Down) [dB]: 9.0 / 8.5
Vendor ID (Local/Remote): TMMB / IFTN
Loss of Framing (Local/Remote): 0 / 0
Loss of Signal (Local/Remote): 0 / 0
Loss of Power (Local/Remote): 0 / 0
Loss of Link (Remote): 0
Error Seconds (Local/Remote): 62 / 0
FEC Errors (Up/Down): 0 / 2,197,818
CRC Errors (Up/Down): 0 / 5,528
HEC Errors (Up/Down): 0 / 4,638



Exchange details:

BT ADSL broadband availability

You are connected to the Cambridge telephone exchange.

ADSL is available in your area
Your exchange is also enabled for ADSL Max services
Your exchange is also enabled for ADSL2+ services

According to BT Wholesale, houses at your postcode should be able to support

up to 2Mbps via ADSL
4Mbps or greater ADSL connection via ADSL Max
5.5Mbps or greater via ADSL2+
Standard ADSL RAG results

You can receive 2Mbps ADSL
You can receive 1Mbps ADSL
You can receive 512kbps ADSL
You can receive 256kbps ADSL

You are approximately 2.31km from the exchange. Note that this is the straight line distance - the actual cable length will be longer!



Is there any other diagnostic I should run to give you more info?

Many thanks!


PS, running router stats at 1min intervals to show 3000 points per capture....will post tomorrow once it's got 24hrs of stats recorded.

Edited by deleted (Mon 23-Jan-12 18:21:59)

In reply to a post by fsa297:

Modulation: G.992.3 annex A
Bandwidth (Up/Down) [kbps/kbps]: 960 / 5,803

You are running on ADSL2 only. You should be on ADSL2+ (G.992.5) and getting 7 Meg @ 6 dB NM.

Is this a simple matter of AOL activating something on their software to enable me to get ADSL2+? Or is it a hardware thing? My router is ST585v6sl...i think that can handle ADSL2+

Edited by deleted (Mon 23-Jan-12 19:49:31)

No, it's automatic, as a result of your router negotiating with exchange.

Try disconnecting and see if it changes.

Many thanks for your advice. Restarting the router helped...it's connected on G.992.5, but the downstream is still similar:

Link Information

Uptime: 0 days, 0:04:33
Modulation: G.992.5 annex A
Bandwidth (Up/Down) [kbps/kbps]: 965 / 5,589
Data Transferred (Sent/Received) [KB/KB]: 56.00 / 67.00
Output Power (Up/Down) [dBm]: 12.5 / 0.0
Line Attenuation (Up/Down) [dB]: 24.5 / 42.5
SN Margin (Up/Down) [dB]: 8.5 / 9.5
Vendor ID (Local/Remote): TMMB / IFTN
Loss of Framing (Local/Remote): 0 / 0
Loss of Signal (Local/Remote): 0 / 0
Loss of Power (Local/Remote): 0 / 0
Loss of Link (Remote): 0
Error Seconds (Local/Remote): 0 / 0
FEC Errors (Up/Down): 303 / 79
CRC Errors (Up/Down): 36 / 1
HEC Errors (Up/Down): 9 / 1


Does that mean that's probably the best my line can do?

Have you removed ring wire? Tried diff filters? Are you using the standard ADSL cable between socket & router, as supplied? Are you plugged into master socket? What phone extensions have you?

Get the stats with the filter plugged into test socket behind faceplate of master socket, with no phone connected.

SN Margin (Up/Down) [dB]: 8.5 / 9.5

And when you re-connect via the test socket, as XRay has requested, please immediately check you router statistics. Looking at the above, you could be on a 9 dB SNRM target and a 6 dB target margin may be possible -- which could give you an approximate extra 500Kbps download sync.

Different filters yield similar results. I'm using the standard ADSL cable as supplied. Router is directly connected to the master socket...no phone extensions in use. (phone is connected to the filter).

The master socket is the old type without the faceplate...which is strange considering the line was fitted when I moved in last year. Am I allowed to remove the ring wire since this is the old style socket?


This is with a different filter and no phone line:

Link Information

Uptime: 0 days, 0:00:19
Modulation: G.992.5 annex A
Bandwidth (Up/Down) [kbps/kbps]: 1,020 / 5,677
Data Transferred (Sent/Received) [KB/KB]: 1.00 / 1.00
Output Power (Up/Down) [dBm]: 12.5 / 0.0
Line Attenuation (Up/Down) [dB]: 24.5 / 42.5
SN Margin (Up/Down) [dB]: 9.0 / 9.0
Vendor ID (Local/Remote): TMMB / IFTN
Loss of Framing (Local/Remote): 0 / 0
Loss of Signal (Local/Remote): 1 / 0
Loss of Power (Local/Remote): 0 / 0
Loss of Link (Remote): 0
Error Seconds (Local/Remote): 1 / 0
FEC Errors (Up/Down): 176 / 0
CRC Errors (Up/Down): 32 / 0
HEC Errors (Up/Down): 9 / 0



On the last upgrade they provided a new dsl 2640r, but since my ST585v6sl (also from them in the previous year) seemed to be working fine, I didn't bother changing. Is the D-Link one supposed to be better at ADSL2+ than the ST one?

Edited by deleted (Mon 23-Jan-12 21:58:30)

In reply to a post by fsa297:

The master socket is the old type without the faceplate...which is strange considering the line was fitted when I moved in last year.

That shouldn't be if it was installed by BT. Think you can get BT to fit a modern NTE5 for free to regularise it, particularly as they failed to do it last time.

The error count would seem to be high, indicating that there might be something else going on, when using the phone at the same time as the router is in sync, do you get any noise on the line ??

Would they still do it even though now I've switched line rental to AOL?

No noise....seems clear to me.

That's 1 of the many disadvantages of moving your line away from BT. Now it may be AOL's responsibility, but the work is still done by BT Openreach.

But is your line still a BTw line (i.e. Partial LLU)? Does your phone # still work in here: http://www.dslchecker.bt.com/adsl/adslchecker.welcome ?

In reply to a post by XRaySpeX:

That's 1 of the many disadvantages of moving your line away from BT. Now it may be AOL's responsibility, but the work is still done by BT Openreach.

But is your line still a BTw line (i.e. Partial LLU)? Does your phone # still work in here: http://www.dslchecker.bt.com/adsl/adslchecker.welcome ?

Or do you get the menu read out to you if you call 17070. If yes, select option number 2 and listen, critically.

From the statistics you have just posted, we can definitely say that your target SNRM is set at 9 dB.

I guess I really should've explored things more thoroughly than jumping into the lure of the apparent savings. My no doesn't work in that link and I get my phone no read out to me repeatedly when I dial 17070...no menu. So that means my landline is fully under AOL LLU?

In reply to a post by fsa297:

I guess I really should've explored things more thoroughly than jumping into the lure of the apparent savings. My no doesn't work in that link and I get my phone no read out to me repeatedly when I dial 17070...no menu. So that means my landline is fully under AOL LLU?

ACK. You have a fully LLU'd line. And as AOL is now owned by TalkTalk, I will assume that you are connected via the TalkTalk circuits.

Oh cr*p. I really should do more reading around before committing. Are the TalkTalk circuits known to be worse than BT?

How can I reduce my SNRM? I just read somewhere that increasing it improves line stability at the expense of bandwidth....but who decides what my SNRM is set at? If it's the ISP, how do they determine how unstable my line is? Unless it's automatically done by the router...

No. Depending upon circumstance, it can be an advantage.

(For the record, I am a TalkTalk user with a fully LLU'd line.)

The target SNRM value is set by the ISP (the CP). The line profile is adjusted by the DLM software, which constantly monitors the line.

To get an idea of how currently stable / unstable your line is behaving, it is recommended to install routerstats and allow it to monitor the line for a couple of days.

In my case, as an actual TalkTalk customer, I was able to request that the DLM was turned off, that the target SNRM was set to 6 dB and that fast path (no interleaving) was enabled for my line. Obviously I need to keep a check on my line's behaviour but, to date, it has remained stable with the best performance possible for the physical condition of the line.

The default Target NM is set by the ISP and the router and exchange negotiate it from that (using DLM = Dynamic Line Mgt). TalkTalk's default is 12dB and user's can request changes to it (also Interleaving/Fast Path) on their Control Panel at TT site. Don't know, & never heard, if AOL operates such a system.

Given a suitable router, such as Netgears DG834G v4 and DG834GT (mine), you can change Target NM via CLI or tools like DMT Tool. Have a read of: Tweaking Netgears through the CLI. & Using the DMT Tool and round & about

Many thanks to the both of you for your advice and input.

So even if the ISP has set a higher default target NM, it can be overridden by tweaking the router's target NM?

I'm running routerstats right now and will post the capture once it's got a couple of days' data.

Unfortunately my ST is running at v6.2+ which means I need to flash it to 6143 before being able to use DMT Tool to change the target NM. Any way of finding out if the router is locked to AOL which would prevent downgrading?

Edited by deleted (Tue 24-Jan-12 01:47:50)

and user's can request changes to it (also Interleaving/Fast Path) on their Control Panel at TT site.

XRay,

As a TalkTalk user I have never heard of, nor seen, any reference to a "Control Panel". Do you have a URL that I may check, please?

So even if the ISP has set a higher default target NM, it can be overridden by tweaking the router

For certain ISP / CP provided services, yes. Whether it will work for your AOL (via TT) LLU'd line, I am not sure.

Unfortunately my ST is running at v6.2+ which means I need to flash it to 6143 before being able to use DMT Tool to change the target NM. Any way of finding out if the router is locked to AOL which would prevent downgrading?

I guess it would be a case of waiting for "someone who knows" to post to this thread or to just try it and see.

"Control Panel" was just my name for it as it seemed coded but perhaps its just a TT forum request. Seen it mentioned a few times.

Can't paste link from my Kindle but do a Search in Technical over 3 m for "TalkTalk & Noise & default & 12".

So far I have this:

[IMG]http://img811.imageshack.us/img811/4407/rxsyncspeedk...[/IMG]

So quite a few line drops overnight. Guess the higher SNRM is justified?

Edited by deleted (Tue 24-Jan-12 15:06:38)

Here is the noise margin


Any advice?

http://img811.imageshack.us/img811/4407/rxsyncspeedk...

. . .

http://imageshack.us/photo/my-images/828/rxnoisemarg...

Yuck! is my initial thought.

Could you please configure RS so that both the DS sync and SNRM are both plotted on the same graph? It makes things a lot easier when confirming the co-incidence of spikes.

Certainly there is something inducing noise onto your line. The first thing to do is to determine whether it is from within or from without your home. The first simple test would be to turn off every other electrical item bar the modem/router/'puter and see if those spikes still continue.

You will find plenty of further advice for tracking down that sort of interference on the main TBB site, on RobertoS site and Kitz site.

many thanks for your time.

Unfortunately I closed RS after getting the last screenshot....i assumed it automatically kept a log. Will run it again with the overlay.

I'll have a read of those websites and apply as much as i can. Apart from the light, the only other electrical item near the router is the TV about 3m away. i'll try and see if there's any correlation between the two.

In reply to a post by fsa297:

Apart from the light, the only other electrical item near the router is the TV about 3m away. i'll try and see if there's any correlation between the two.

Is it a plasma TV? Was it on when the spikes were captured in the above graphs? Is it left in standby mode, 24/7?

In a difficult case, I would make up a lead to run the mode/router direct from a car battery, use a laptop running on its own battery and turn the electricity supply off at the main switch.

You need to eliminate the possibility that the noise is being induced into the circuit from within your home. Once that has been done and you have also confirmed that any extension wiring is not responsible, then external sources must be responsible. Once you've got to that stage, it will be time to submit your RS graphs to your CP / ISP. They, in turn, will eventually arrange for an Openreach engineering visit.

Edited by burakkucat (Thu 26-Jan-12 02:14:41)

It's a CRT not left in standby mode. The only times it was on in that graph was around 1930 on both days for 2-3 hours. There aren't any other electrical devices near the router...it's right next to the front wall of the house by the main socket. On the other side of the wall is the front lawn....nearest lampost is about 10m away.

Here is the sync rate and noise superimposed.

The car battery method sounds really long winded...but might have to do that if I can't rule anything out.

I think we can probably rule out that the source of the interference is within your home.

Thinking about it, your graph is similar to what I saw at the end of last year in the Kitz forum. At the end of a number of frustrating months, the cause of the problem -- a defective external joint -- was found and eliminated.

To satisfy your CP / ISP, you should try with your modem/router plugged into the test socket of the NTE5/A via a different micro filter. Carry on collecting data with routerstats.

Once you have performed these basic tests, report the issue to your CP / ISP and provide them with copies of the graphs. They should then arrange for an Openreach engineering visit.

Really appreciate your advice. What if I have the old type main socket without the test socket? Should I still remove the ringer wire?

Edited by deleted (Fri 27-Jan-12 23:03:37)

In that case, try to identify which is the first one fed by the incoming service cable (assuming you have more than one) and run your tests from that socket (having ensured that everything else has been unplugged).

If you do not have a NTE5/A fitted, you should also mention that to your CP / ISP. Again, it will require an Openreach engineering visit to replace the socket and thus regularise the line termination.

Edited by burakkucat (Fri 27-Jan-12 23:11:45)