Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[general disquiet]

My village is served by a Secondary Connection Point (SCP) some 3.4 km from the Primary Connection Point (PCP) in the next village.  The PCP is itself 1.4 km from the exchange and all the cable is aluminium, rather than copper and most of the run is un-ducted.  Not surprisingly, the broadband service has always been pretty ropey with 95% of subscribers getting speeds well under 2Mbps.   Back towards the end of last year, BT began to upgrade the exchange for superfast broadband for some customers.  Installation of a new DSLAM and new cabinets have been undertaken to replace some of the PCPs - but not ours.

At the same time work commenced on the exchange, folks around the village experienced a drop in speeds, to about a third of the speeds they had been receiving for the previous 7-8 years and the QoS has also dropped markedly with long TCP delays on upload and download.  A few unlucky individuals have lost their service completely.   BT initially reported that ALL the VPs at the exchange were running hot and that the CJ cable back to the next exchange was under capacity.  We believe more capacity has been added, but for most folks, this has not resulted in any improvement in speed or QoS.  A few of the more vociferous villagers have had their ciruits "lifted and shifted", mostly to little effect.

BT have told the County Council, relayed to the Parish Council, that there is absolutely no way that improvements to the exchange could have a negative impact on on some customers some distance from the exchange.  My question is quite simple.  Are BT correct in this assertion?

[kitz]

Highly unlikely.  In fact as more users move over to VDSL, in theory that should slightly improve speeds for those customers remaining on adsl.  So the short answer is No.

vdsl uses much higher frequencies than adsl, and any vdsl lines have PSD masks and Power cut back heavily applied to lessen the possibility of crosstalk from vdsl to adsl.     Crosstalk is there and always has been for for adsl/adsl2 only lines just not as noticeable.

I noticed it though when on adsl2+ (before vdsl even came to my town).  I was one of the first users on a brand new shiney MSAN, but over the course of a few years as more people got broadband, then crosstalk from other adsl lines decreased my speeds by 3-4Mbps.

So some of it will be crosstalk, but it seems odd that you should experience a sudden drop. Installation of vdsl cabs certainly doesnt affect QoS or TCP delays.

However, TCP delays could point to interference of some sort, it depends on the type of those delays.
If its permanent then it could be that the lines have become noisy so the DLM has applied interleaving to stabilise the line.
If its say worse at certain times of the day, then random noise could be causing packet loss.
REIN type noise could be localised - say for eg in a street. Weve seen several cases of REIN impacting localised areas.

Some other randomish thoughts:
Permanent delays could also be an effect of some sort of backhaul routing change.
Congestion on the backhaul VP/SVLAN seldom causes increased latency.   BTw has always been able to control latency even when speeds are very low.  Many years ago our exchange suffered very heavily from hot VPs and despite atrocious speed loss down to under 100kbps not one of us saw any increased latency.   Increased latency from congestion is usually more indicative of it being with the ISP or perhaps the Interconnects.   

Back to ISP, some BTretail customers seem to have been seeing increases in latency, yet not sure exactly what is causing it.   You only have to look at a tracert from a BTretail customer then compare it to say Plusnet or Zen etc and their average latency is around 18ms compared to about 11ms with the others.  This isnt all lines and its not interleaving dependent, but they do appear to be doing some additional looping around at the RAS. 

[general disquiet]

Kitz, many thanks for your comprehensive response.  The reason I asked the question was that I had come across the paper attached; fairly old, but mentioning the crosstalk vulnerabilities of ADSL lines.  I guess BT would have been beavering away to find ways of minimising this issue, but as far as I can see, all their efforts have been devoted to doing this for copper lines, not the much more vulnerable aluminium.

You mention REIN.  7-8 years ago we had a major REIN issue that BT spent months trying to solve - even dragging a REIN expert from Cornwall to Wiltshire to try and find the source - unsuccessfully.  As a result of this, we know that our cable is very sensitive to any sort of RF interference and crosstalk.

The drop was over a few hours and has affected the whole village.  Problem is 24/7 - packet loss is also 24/7.

I agree that the latency issue is much more likely to be down to hot VPs and we already know that every VP at the exchange, to quote BT, was running hot.  Odd when a couple of hundred subscribers had already been moved over to the new DSLAM and cabs.  I would have expected them to have released a bit of capacity by doing so.

Latency - well - at best it is 50-60ms - At its worst, anywhere between 3 and 7 seconds - seriously.  I've added a second file which gives a better idea of the QoS we are getting.

[Weaver]

> all their efforts have been devoted to doing this for copper lines, not the much more vulnerable aluminium.

Why is aluminium “more vulnerable” compared with copper?

[licquorice]

Because it oxidises and corrodes.

[NewtronStar]

Because it oxidises and corrodes.

Copper is a far better conductor than Aluminium and as the pairs have a voltage/electricity going through them it makes a difference.

https://en.wikipedia.org/wiki/Electrical_conductor

[Weaver]

I read the op's "more vulnerable" to mean more vulnerable to _crosstalk_, that's why I was mystified.

[licquorice]

I read the 'more vulnerable' to mean more prone to faults due to bad joints as a result of corrosion.

[Weaver]

I read the 'more vulnerable' to mean more prone to faults due to bad joints as a result of corrosion.

Quite. You were being sensible, I was being the opposite. Blame all the pills. 

[general disquiet]

@ Weaver & liquorice - I hate to stir the pot, but I was actually suggesting both things.  First, that old, oxidised, poorly jointed aluminium cable, laid directly into the ground without any ducting was a bit more prone to developing faults than nice juicy, well-jointed and ducted copper.  Second, that crappy aluminium is even better at pulling in all sort of external interference whether it be RF from military comms here on Salisbury plain, FEXT or NEXT (from VDSL ) at the exchange end than would be good (is that possible in the context of broadband?) aluminium or copper of the same length.  If I've got this last bit wrong it's because I'm a mere biologist, not a physicist! 

I guess my question is whether BT have developed appropriate PSD masks and Power cut back strategies for both copper and aluminium to minimise VDSL crosstalk with ADSL, or whether thay've applied a "one size fits all" strategy to doing this which works with copper, but is inherently unsuitable for long and failing aluminium lines.

[licquorice]

I don't think aluminium is more susceptible to interference than copper per se, but rather it is susceptible to high resistance joints which can then act as diodes and pull in interference.

[Black Sheep]

From memory, I think our Ali cables have less 'twists per metre' than our Copper cables. This will have a negative impact on the Ali cables 'AC Balance' reading, as opposed to the Copper cable.

The AC Balance is an indicator of the circuits integrity at keeping out interference ..... the higher the better. In my experience, long lengths of Ali are far more sensitive to 'Noise' than their equivalent Copper cables. 

[Black Sheep]

@ Weaver & liquorice - I hate to stir the pot, but I was actually suggesting both things.  First, that old, oxidised, poorly jointed aluminium cable, laid directly into the ground without any ducting was a bit more prone to developing faults than nice juicy, well-jointed and ducted copper.  Second, that crappy aluminium is even better at pulling in all sort of external interference whether it be RF from military comms here on Salisbury plain, FEXT or NEXT (from VDSL ) at the exchange end than would be good (is that possible in the context of broadband?) aluminium or copper of the same length.  If I've got this last bit wrong it's because I'm a mere biologist, not a physicist! 

I guess my question is whether BT have developed appropriate PSD masks and Power cut back strategies for both copper and aluminium to minimise VDSL crosstalk with ADSL, or whether thay've applied a "one size fits all" strategy to doing this which works with copper, but is inherently unsuitable for long and failing aluminium lines.

They use Cabinet Assigned Loss (CAL) values, worked out separately for each individual Cabinet. This is done at the planning stage and not something I get involved in, but I imagine there's a complex equation involved somewhere ?? 

[Weaver]

I can't see for the life of me how aluminium might be more susceptible to crosstalk, and I _am_ (or was) a physicist. (Not that that counts for much, as firstly, I was a very idle one, and secondly, classical electromagnetism isn't really my thing, I'm ashamed to say.) 

Others will be much better qualified to comment especially since they've probably thought about it for rather a lot longer than one minute.

[burakkucat]

I can't see for the life of me how aluminium might be more susceptible to crosstalk,

B*Sheep has given a very good reason in reply #11 --

From memory, I think our Ali cables have less 'twists per metre' than our Copper cables. This will have a negative impact on the Ali cables 'AC Balance' reading, as opposed to the Copper cable.

The AC Balance is an indicator of the circuits integrity at keeping out interference ..... the higher the better. In my experience, long lengths of Ali are far more sensitive to 'Noise' than their equivalent Copper cables.