Connection is dying....

AlanM · Jul 31, 2009, 20:11:13

I'm hoping someone on here can help, and I appologise now for the long post.
Since early new year, my sync has dropped from a heady 1700ish (yeh, long line) with 6db margin to somewhere around 1100 with a 9db margin. I'm also now getting disconnects which have taken my profile down to 500 on some occassions.
So far, I've tried 2 different BT 2700hgv routers, another PSU suitable for the 2700's, a netgear dg834, removed my ADSL Nation faceplate and tried 2 different brand new BT filters, in the Test socket on the NTE5, a shielded ADSL cable to the router and probably other things I've forgotten about. I only have 1 phone connected directly into the filter.

My suspicions are that while I know I have a long line, there must be some really bad connections along it's length back to the exchange. Last year, BT were in the pit on the opposite the house remaking corrodded joints. I talked to the guy and as it turned out, he had to redo mine twice as after the first remake of the joint, it broke again!

I've posted my stats gathered over 17 days. A couple of reboots are due to me rebooting the router. I have enough RouterStats and SysLogs to last a lifetime and can put these up if required as well.

I've spoken to IDNet a couple of times about this but they put all this down to a long line with high attenuation. Surely the attenuation will be even higher than it should be with poor joints in the line?? Plus, and sorry for going on and on, I have a GPO box on the outside of my house where the undergrond cable terminates. The cable between the NTE5 and this box is old grey 2 core (thick bellwire type) cable installed by the builder. I've had the cover off and can clearly see corrossion on the connections (jelly crip type) to the u/g cable.

General Information
DSL Line (Wire Pair):   Line 1 (inner pair)
Downstream Rate Cap:   8128 kbps
Downstream Atten. at 300kHz:   73.6 dB

Uncancelled Echo:   -27.9 dB   Ok
VCXO Frequency Offset:   -11.4 ppm   Ok
Final Rx Gain:   34.5 dB   Ok
Impulse Noise Comp. Tones:   0   Ok
Excessive Impulse Noise:   0   Ok
Impulse noise protection:    1.20
Delay of latency path:    8.00 ms

Training History
   Downstream
Time   Line      Rate   Max1   Max2   Max3   Mgn1   Mgn2   Attn   Pwr   CRCs   FECs   INP   DLY
2009/07/14 20:02:59 BST   1      1184   1196   1144   1144   9.2   6.0   72.9   16.4   1112   4379   1.17   8.00
2009/07/15 11:33:01 BST   1      1248   1264   1076   1076   9.2   5.0   73.1   16.4   885   9860   1.12   8.00
2009/07/16 22:10:39 BST   1      1152   1168   1160   1160   9.2   7.0   73.1   16.2   2319   13993   1.20   8.00
2009/07/18 13:43:48 BST   1      1248   1272   1124   1124   9.3   5.0   73.1   16.4   1511   27084   1.12   8.00
2009/07/21 07:21:35 BST   1      1216   1224   1016   1016   9.2   6.0   73.0   16.4   2450   113548   1.14   8.00
2009/07/21 19:21:59 BST   1      1120   1136   1024   1024   9.2   9.0   72.9   16.3   1533   15883   1.23   8.00
2009/07/21 19:27:56 BST   1      1088   1116   1028   1028   9.3   3.0   73.0   16.1   143   365   1.26   8.00
2009/07/23 17:49:19 BST   1      1152   1164   1188   1188   9.2   10.0   73.0   16.1   3986   18526   1.20   8.00
2009/07/25 18:20:23 BST   1      1248   1256   1044   1044   9.2   6.0   73.0   16.6   2739   55531   1.12   8.00
2009/07/25 20:06:01 BST   1      768   796   996   996   9.5   11.0   73.0   14.5   8   122   1.19   8.00
2009/07/26 21:21:37 BST   1      1152   1180   1064   1064   9.4   6.0   73.0   16.2   1423   8174   1.20   8.00
2009/07/28 21:18:42 BST   1      1152   1152   1044   1044   9.0   6.0   73.0   16.2   1781   17803   1.20   8.00
2009/07/29 17:59:35 BST   1      1152   1156   1148   1148   9.1   7.0   73.0   16.3   3268   12448   1.20   8.00
2009/07/30 20:47:26 BST   1      1152   1168   1072   1072   9.2   6.0   73.0   16.4   1221   6663   1.20   8.00
2009/07/31 18:42:14 BST   1      1152   1180   1044   1044   9.4   9.0   72.9   16.2   359   3871   1.20   8.00

         Upstream

Rate   Max   Mgn   Attn   Pwr   CRCs   FECs
448   544   10.0   31.5   12.4   247   421
448   524   9.0   31.5   12.4   173   407
448   524   9.0   31.5   12.3   560   1016
448   540   10.0   31.5   12.4   262   434
448   544   10.0   31.5   12.4   390   842
448   536   10.0   31.5   12.4   312   669
448   564   10.0   31.5   12.2   65   54
448   544   10.0   31.5   12.3   369   629
448   540   10.0   31.5   12.3   262   654
448   536   10.0   31.5   12.3   0   0
448   536   10.0   31.5   12.3   297   413
448   536   10.0   31.5   12.3   490   857
448   532   10.0   31.5   12.3   396   622
448   548   10.0   31.5   12.3   279   533
448   544   10.0   31.5   12.3   36   153

Mode   Vendor   State   Exit Code   Echo   VCXO   Rx Gain   INC Tones
G.DMT Annex A   TI   230/105   ERR_LOS_LIMIT   -27.9   -12.4   34.5   0
G.DMT Annex A   TI   230/105   ERR_LOS_LIMIT   -28.1   -11.0   34.5   2
G.DMT Annex A   TI   230/105   ERR_SIG_INTERRUPT   -28.8   -11.5   34.5   0
G.DMT Annex A   TI   230/105   ERR_LOS_LIMIT   -27.9   -11.9   34.5   0
G.DMT Annex A   TI   230/105   ERR_SIG_INTERRUPT   -28.3   -11.8   34.5   0
G.DMT Annex A   TI   230/105   IDLE_REQUEST   -27.2   -11.5   34.5   0
G.DMT Annex A   TI   230/105   ERR_LOF_LIMIT   -28.8   -11.8   34.5   0
G.DMT Annex A   TI   230/105   IDLE_REQUEST   -27.9   -11.5   34.5   0
G.DMT Annex A   TI   230/105   ERR_SIG_INTERRUPT   -27.8   -11.4   34.5   0
G.DMT Annex A   TI   230/105   IDLE_REQUEST   -25.6   -11.3   34.5   0
G.DMT Annex A   TI   230/105   ERR_SIG_INTERRUPT   -29.1   -11.3   34.5   0
G.DMT Annex A   TI   230/105   ERR_SIG_INTERRUPT   -27.5   -11.7   34.5   0
G.DMT Annex A   TI   230/105   ERR_LOS_LIMIT   -28.8   -11.1   34.5   0
G.DMT Annex A   TI   230/105   ERR_LOS_LIMIT   -28.1   -12.0   34.5   0
G.DMT Annex A   TI   229/103   N/A   -27.9   -11.4   34.5   0

rireed3 · Jul 31, 2009, 21:32:45

QuoteMy suspicions are that while I know I have a long line, there must be some really bad connections along it's length back to the exchange.

There are two reasons BT, and consequently, IDNet (I'll explain) might not think you have a problem.

73db attenuation for average line characteristics should mean 5.3 km length. If you were on a 45-degree diagonal of a road grid to the exchange, that would mean the exchange is 3.75 km away 'As The Crow Flies' (ATCF). That's a friendly route for your line. Often they backtrack at some point, meaning that the exchange could be even closer ATCF for that line length, and often they are made of variable thicknesses of copper, or even aluminium, all of which affect attenuation, meaning that for that attenuation, your exchange could be closer yet again as the crow flies -- (< 3 km?)

Reason one: your line really is that long, and has average characteristics. How far is your exchange ATCF?

Reason two: BT Wholesale don't have to care. The Universal Service Obligation for fixed-line telephones doesn't include broadband at all. In the UK it includes 28 Kbps dial-up. If your line doesn't have an audible noise fault that interferes with conversation, they don't have to make your broadband better. They will, if it's not too much trouble, and they'll make you pay if it's any trouble at all.

Also, I don't see what you mean by dying, it looks like you keep syncing at 1152 Kbps. Lines that long often re-sync once a day, but please get the opinion of someone with a long line -- mine isn't.

Edit: I said I'd explain why IDNet might not think you have a problem. IDNet can only ask BT Wholesale to address things that are broken. I'll try to guess what that might look like. Say you can see in the windows of the exchange. I'm not sure that would get their attention, but it would mean there is almost certainly a different pair wires that would work better to your house, for whatever reason. That's the kind of situation that IDNet might be able to make a case for to BT.

Richard

dujas · Jul 31, 2009, 22:29:06

I assume you've tried disconnecting the bell wire? Really the only way I can see BT doing maintenance for free is to claim noise on the line is affecting voice calls. If the corrosion on the wire leading to the master socket is noticeable, a friendly engineer may replace the joint and/or the wire and use it as the reason to avoid charging you. AAISP have their broadband fix promise but I'm not sure if they will consider this a fault or just the reality of being on the end of a very long line.

dujas · Jul 31, 2009, 22:40:51

Who fitted the NTE5 btw?

AlanM · Jul 31, 2009, 22:51:08

Hi Richard, I said it was dying because it just seems to be getting worse over a period of time. When I joined IDNEt 2 years ago, I was getting what I thought was a reasonable sync rate of around 1700 which gave me a profile of 1500. Also, the connection was pretty solid with only a few disconnects. However, since around Jan time, the sync has been dropping in stages to where it is now, the noise margin has increased by 3db, and now in the last few weeks I've started getting disconnects at various times of day.

By the way, my nextdoor neighbour is syncing at 2000. They recently had broadband installed and the BT engineer replaced their cable which runs from the box outside throught to the NTE5. That's the cable which I think would be a good starting point to replace to see if that improves things.

SamKnows gives my ATCF distance to the exchange as being 2678m, so I guess 5.3km line length would be about right.

I've done the quiet line test loads of times but nothing really conclusive.

I understand BT only have to provide a voice service but it's frustrating when I know what the line was capable of before and I'd be happy if I had that again. Just need to figure out how to go about getting BT to replace the cable between the outside box and NTE5. I can hear crackles on the quiet line test when the crimp joints in the outside are moved slightly and funnily enough, thats exactly where the corrossion is.

Hi dujas, just seen your post and thanks for your input. Yeh, bell wire has been off almost since day one. BT fitted the NTE5 way back in 1988 I think. All they did was replace the old "Choc-Block" type we had so that we could plug in our own phone. Didn't do anything with the cabling. Definately, have to find some way to get the cable replaced as it's the only thing which hasn't been replaced at least twice!. Just don't think that I should pay to have BT replace a piece of cable which they are responsible for.

dujas · Jul 31, 2009, 23:08:52

Persistently report a voice noise fault (even if there isn't one) to whomever you pay your line rental to, accept you'll have to pay a call-out fee if they can't find anything, then when the Openreach engineer turns up clearly direct them to the joint as a potential cause of the 'problem'. You'd have to practically lose the ADSL signal altogether before an ISP could get BT to do something.

rireed3 · Aug 01, 2009, 09:31:40

AlanM,

Sorry, I missed what you said about 1700. There are two possibilities. It's possible that your attenuation has increased because of some kind of deterioration, assuming you have stayed on the same pair, but you'd have to have seen it happen gradually to know. This is usually unlikely, but your attenuation is in the worst-case range of what it should be, I think.

What's more likely is that there are more broadband users in your cable. Many of us have seen noise increasing in stages as you describe. It seems that at the same time your line is having more trouble holding sync, and that has caused BT DLM to raise your target noise margin, which is one reason you have lost sync rate.

There may also be an external source to the increase in noise, but unless it changes in the course of the day it will be quite difficult to find.

I'm afraid dujas is right about the likelihood of getting BT involved, unless someone else has a better idea. The audible crackling is the best hope of getting an improvement.

Richard

Rik · Aug 01, 2009, 10:08:35

Hi Alan

The answers you've already been given are spot on (thanks, guys

). You don't have a fault at this point, just an ageing line, and it's very unlikely that BT are going to do anything about it, so getting them out is a £160+ gamble for you. As Richard says, many of us (I do have a long line) have seen speeds fall, partly because of the increase in BB takeup and partly because of degradation of lines. I get 3Mbps on a 56-63db attenuation (depending on which figure I use from the router), but my neighbours get 2M, 1M and 512k. I did benefit from a friendly engineer, as an early adopter, who made sure he found the best pair for me that he could.

If you happened to start hearing other people's conversations, and reported this to BT, the standard practice is to swap the exchange pair, but I'm not sure that would help if the corrosion is so local.