Can FTTC handle speeds above 80 Mbps?

The reason I have asked this question is that the Open Reach engineer I saw today said that FTTP is unlikely to come to my area but at the end of the summer FTTC will offer up to 120 Mbps. I thought only FTTP offered this sort of speed. Can the 'fountain of knowledge' people advise!?

In reply to a post by shalom2005:

The reason I have asked this question is that the Open Reach engineer I saw today said that FTTP is unlikely to come to my area but at the end of the summer FTTC will offer up to 120 Mbps. I thought only FTTP offered this sort of speed. Can the 'fountain of knowledge' people advise!?

Depends on the distance. Last I heard on very short cable runs it can do up to 200Mb/s in the lab. Perhaps BT are going to roll out a new profile that increases speeds for those on short lines. There is also something called 'vectoring' that is rumoured to be coming soon. This works in a similar way to noise cancelling headphones to counteract cross-talk between cables in a bundle. In theory it will give almost everyone a boost. There's an Ofcom document floating around that is very speculative but suggests than 1Gb/s might eventually be possible.

Alternatively (and perhaps more likely) the engineer was talking out of his [censored]. That's something BT engineers seem adept at.

Edited by Andrue (Wed 17-Apr-13 18:54:14)

Thanks Andrue - that's why I posted here. I've not read anywhere about what he said & if it was going to launch at the end of the summer I would have thought it would be heavily advertised.

Anyway if I get around 60 Mbps on a 80 Mbps profile I doubt my speed would increase to 90 Mbps

More down to the line cards not being able to handle anything above 17a profile.

In reply to a post by lockyatlrg:

More down to the line cards not being able to handle anything above 17a profile.

Can you explain more fully ?

These are the Huawei VDSL2 linecards, taken from the Smart MA5616 Hardware Guide V800R310:

http://www4.picturepush.com/photo/a/12715522/img/Hua...

Profile 17a is limited to an Actual Net Data Rate (ACTNDR) of 100Mbps Down 60Mbps Up. And Openreach imposes an IP Profile of 80/20 on top of that.

Profile 30a potentially supports an ACTNDR of 200Mbps Downstream. That is only on the very shortest loops though. And the port density of the 30a linecards is only 16, so they're not very cost-effective.

SFAIK, the highest port density of VDSL2 card is currently 48, for both the ECI M41 DSLAM and the Huawei MA5616.

cheers, a

A year ago the max attainable speeds on my line were 102/32 Mbps on an 80/20 profile. As FTTC take up has increased, my max attainable speeds have fallen to 93/28 Mbps. Home to cabinet distance is about 300 Metres. If BT has plans to roll-out higher speeds then my guess is that they will also roll-out vectoring.

Quote: In vectoring, the VDSL signal is optimized for the subscriber line by compensating for interference between adjacent lines. Even when many subscribers use the line, this allows a transfer rate of over 100 Mbit/s for each connection. Unquote Source AVM Berlin

AVM Berlin is already producing vectoring capable modem/routers for German consumers. To work, the network also has to be vectoring-enabled.

Ah - I think I'm qualified for this one

Yes - FTTC can go over 80Mbps in 3 different circumstances:
- Short distances to the cabinet - probably within 200 metres.
- Medium distances to the cabinet (between 400 and 500 metres), but with vectoring applied to remove crosstalk
- Longer distances with multiple pairs bonded together, probably alongside vectoring

Reading stuff about the recent suggestion that the Aussie NBN should consider FTTC, I saw an article summarising the Openreach rollout. In there, Mike Gavin said that Vectoring would be deployed later in 2013.

There were no other details, in particular the headline packages that they're going to offer. But I suspect they'll be keen to match Virgin's offers.

Alcatel-Lucent have a self-aggrandizing white paper on the current state of Vectoring/Bonding, with the more recent trials suggesting that 100Mbps is a straightforward goal to distances of 400 or 500 metres - presumably depending on matters such as the diameter of the copper used in the access network - and that 120Mbps may be possible to around 300 metres.

Of note is the fact that the more recent trials are getting better speeds out of vectoring, suggesting lessons are being learnt.

The use of Vectoring generally means that more people can get closer to the theoretical speed for their distance, giving Openreach a bit more licence to set a higher headline package. I was expecting them to set 100Mbps packages, rather than 120Mbps.

The use of bonding alongside vectoring can give speeds of 200Mbps out to 400 metres. I wonder if we'll ever see BT making use of this in a reasonable fashion, rather than with a BET-like kludge.

Yes - Deutsche Telekom have just been given approval by their Ofcom-equivalent to roll out vectoring.

It is a contentious decision for a regulator, as it means the end of meaningful physical sub-loop unbundling.

In reply to a post by WWWombat:

It is a contentious decision for a regulator, as it means the end of meaningful physical sub-loop unbundling.

Do we have any co-located SLU ? Openreach evaded a Rutland cab deployment, I guess SYDR is the most likely. So was it "meaningful" in the first place.

Mind there's enough about cross-vendor or multi-DSLAM vectoring to suggest that it may not have the suggested consequence.

In reply to a post by yarwell:

Do we have any co-located SLU ? Openreach evaded a Rutland cab deployment, I guess SYDR is the most likely. So was it "meaningful" in the first place.

They're the obvious cases here, but I haven't heard much contention here either - perhaps we're being more accepting of a virtual/wholesale undbundling setup instead.

There have been a few more complaints in Germany - but seemingly from the FTTP network owners.

Mind there's enough about cross-vendor or multi-DSLAM vectoring to suggest that it may not have the suggested consequence.

I saw some figures suggesting you need over 20 Gbps of data to make the right decisions in vectoring. That's a lot going across a backplane, never mind out of the cabinet and into another one from a different vendor.

In reply to a post by shalom2005:

I've not read anywhere about what he said & if it was going to launch at the end of the summer I would have thought it would be heavily advertised.

When Openreach turned on the 17a profile, and the packages jumped from 40/10 to 80/20, there was little advertising until it actually happened.

Yes, we all knew it was coming - NICC had approved the frequencies, and Openreach had set notifications about the new packages.

However, consumer advertising was certainly totally absent.

Now, as far as vectoring has been concerned, BT have only mentioned it in their quarterly financial reports. We haven't yet heard anything concrete - so it was strange that the first mention I've seen was when I came across Mike Gavin's quote earlier today while reading about the Australian situation.

There's a series of blogs on ECI's web site to do with vectoring which make interesting reading. The one comparing linecard level and shelf level vectoring is quite interesting, Here.

The bottom line is don't bother with linecard level vectoring but I can't really see Openreach replacing all the M41's with V41's where ECI kit has been deployed somehow.

Many thanks for the answers. So, it may happen....look forward to hear more.

from all documents I have read it looks a no brainer to me that the obvious speed gains (and stability) are from vectoring. Profile 30A will boost short lines (including mine as I see I am using the highest frequencies on 17a) but those same short lines will still probably see pretty good boosts from vectoring because of crosstalk and in addition vectoring also helps longer lines that wouldnt be helped by profile 30a. Also vectoring removes most of crosstalk variance so lines will be less like a lottery with performance.

If BT are planning to boost speeds further on copper the question is will they do it the absolute cheapest way profile 30a or even just a new 100mbit product on 17a with no vectoring. Or will they do whats most effective but with possibly more cost aka vectoring. I can see personally how bad crosstalk can be but I am also aware not all lines are affected as bad as mine, the average effect on short lines in vendor documents suggests around 40% speed loss due to crosstalk. To me that means if the cost of vectoring is under 40% of the cost of the original FTTC deployment, it makes it cost effective. I expect it will be way under 40% of the original cost as its just upgrading dslam equipment be it software or hardware upgrade.

ECI's shelf idea if I udnerstand it right is actually potentially cheaper than a per card solution if you consider a per card solution potentially needs every card upgrading whilst the shelf variant if I understand it right just needs a new master card to manage all the individual cards.

Edited by Chrysalis (Thu 18-Apr-13 08:29:02)

But will it make any difference to lines like mine, I presume not. I was the first on the cabinet and initially got a 50 meg sync, but this quickly dropped to around 40 to 45 depending on interleaving depth.

I think my line is affected by bad line quality rather than cross talk.

If you are suffering from some alien noise, then no - it isn't going to solve that for you. You ought to get the 50 back, but not necessarily more.

However, it seems that the O&M bods have gotten their hands on vectoring too. The amount of data being generated for vectoring seems like it can be used as input into analysis tools to monitor the line quality. I've also read that it becomes possible for the cabinet to identify lines that are performing poorly without intervention of the end-user.

Some information here, from the broadband forum, plus both this article and this one from Alcatel.

It seems like BT would be able to respond better to people with poor lines - if they chose to improve their current attitude of "you get what you're given".

The 17a colud handle over 80Mbps but not my much... you would need to swap to 30a profile which is being trialed at the moment which is up to 200Mbps. Though if you're on the lower end of a line you are less likely to benefit from vectoring compared to someone who is 50m away. The trials have started for vectoring... very recently.

While looking at the Broadband Forum's document An Overview of G993.5 Vectoring (May 2012), I saw a couple of interesting graphs (figures 7 and 8) - Vectoring on the 17a profile and on the 30a profile.

Those show that the effect of the 30a profile (with vectoring) runs out to 550 metres, just.

So 200Mbps out to 350 metres, on 0.4mm lines. Don't we have 0.5mm lines, which ought to be better?

I'd always thought of the 30a profile as being for lines that were too short to make it worthwhile in reality, save perhaps as FTTB for MDU. Perhaps it is worthwhile after all.

Openreach vectoring trials start May 2013 for a period of between 3 and 6 months.

In reply to a post by Andrue:

In reply to a post by shalom2005:

The reason I have asked this question is that the Open Reach engineer I saw today said that FTTP is unlikely to come to my area but at the end of the summer FTTC will offer up to 120 Mbps. I thought only FTTP offered this sort of speed. Can the 'fountain of knowledge' people advise!?

Alternatively (and perhaps more likely) the engineer was talking out of his [censored]. That's something BT engineers seem adept at.

I kind of agree with Andrue's opinion.

Lots of people talking about vectoring and new profiles, but no one seems to have mentioned that the BT Openreach HG612 has a 100Mb bottleneck.

vectoring is getting more advanced for sure.

I also read some documents where they can prioritise resources.

eg. a line owner pays higher premium so he gets higher priority vectoring, they can set min max target sync speeds, so if pay more vectoring gets rid of more crosstalk.

In reply to a post by thompsbd:

Openreach vectoring trials start May 2013 for a period of between 3 and 6 months.

how do I get on? is staff only?

In reply to a post by wanapoo:

In reply to a post by Andrue:

In reply to a post by shalom2005:

The reason I have asked this question is that the Open Reach engineer I saw today said that FTTP is unlikely to come to my area but at the end of the summer FTTC will offer up to 120 Mbps. I thought only FTTP offered this sort of speed. Can the 'fountain of knowledge' people advise!?

Alternatively (and perhaps more likely) the engineer was talking out of his [censored]. That's something BT engineers seem adept at.

I kind of agree with Andrue's opinion.

Lots of people talking about vectoring and new profiles, but no one seems to have mentioned that the BT Openreach HG612 has a 100Mb bottleneck.

which isnt really a bottleneck when without vectoring 100mbit is a dream.

The HG612 is rather old as all new cabs are ECI and are given ECI modems. The majority will already have ECI modems so, obviously Openreach have taken note. Though I'm not sure if ECI modem has a max limit either so...

Is the 100Mbps firmware capped or processing power bottle necked?

The vectoring trials are taking place on high number of connections within a cabinet.

Edited by deleted (Thu 18-Apr-13 14:34:40)

I am on a fairly new ECI cab and was given a hg612.

I got no idea how its 100 capped, I assume the person who said its capped knows why.

So the field trial is limited to one single cabinet right now? I wonder if its cornwall or wales.

Is it ECI or HG trial? I would expect at least both vendors to be trialled.

In reply to a post by Chrysalis:

I got no idea how its 100 capped, I assume the person who said its capped knows why.

I thought they meant it only had 10/100Mb/s Ethernet ports.

Edited by Andrue (Thu 18-Apr-13 14:57:08)

It will be tested on more than one cabinet, I don't which type I would assume both or just ECI and replacing the HG cabinets in future, though that is very unlikely.

You were lucky, I was given an ECI which you should have been and had to buy my own HG612.

Is this link relevant?

http://community.plus.net/forum/index.php/topic,1142...

HG612 is bottlenecked because it has a 100Mb ethernet LAN1 connection to the router.
(I think it does; just checked the specs at the back of the user guide, and it doesn't say one way or the other).

LOL - Yes. That would do it!

In reply to a post by ryant704:

The vectoring trials are taking place on high number of connections within a cabinet.

Presumably it is only worth trialling on cabinets with a high take-up, and where almost all subscribers can be persuaded to take part - whatever ISP they belong to.

I assume that BT will want to have reasonable control of the NTE being used by all the triallists, and will have different firmware versions from the norm. That might mean we have difficulties getting stats out of the modem while the trial is going on.

The only other prediction would be that the trial won't be in South Yorkshire.

Agreed WWWombat though if they mess something up it means a lot of users will be affected and rather pee'd off.

Patience I will ask for details

As the 612 supplied with a single usable Ethernet port, yes. However, it can actually have 4 ports installed so in theory could provide over 100Mb provided more than one port was downloading.

thanks.

what?

we are talking about the wan port only being 100mbs.

No mount of installing more ports will change this.

Edited by lockyatlrg (Thu 18-Apr-13 19:49:55)

Get an unlocked HG612, select Routed mode instead of Bridge and it becomes a router. Having one with four ports (and they are around - I've seen a few but missing the case and BT logo) will turn it into a four port router with 100Mbps on each.

In reply to a post by MHC:

Get an unlocked HG612, select Routed mode instead of Bridge and it becomes a router. Having one with four ports (and they are around - I've seen a few but missing the case and BT logo) will turn it into a four port router with 100Mbps on each.

Any idea if the CPU can cope with 400 Mbps of throughput ?

In reply to a post by MHC:

Get an unlocked HG612, select Routed mode instead of Bridge and it becomes a router. Having one with four ports (and they are around - I've seen a few but missing the case and BT logo) will turn it into a four port router with 100Mbps on each.

ie a HG622 - same chip as 612 but with 4 LAN ports

If you take a 612 apart, the board is designed for 4 ports, so it could be the same board fully fitted.

The HG612 has a broadcom chipset, which has a maximum capability of 100mb downspeed and 50mb upspeed on the modem side.

The actual wan side has 1000mb bandwidth, but is restricted by only having a 10/100mb switch and 1 working port (2 if hacked).

The modem with the 4 ports which MHC is talking about is the HG610V, which uses the same circuit board.

Actually no, I am referring to what were 612s (that is what the labels said) and they had 4 RJ45s fitted.

Any idea what Huawei hardware upgrades would be needed?

I've seen Huwei refer to their best efforts as "Node Level Vectoring", but no mention of what underlying platform changes are needed to support it.

Swisscom have gone with Huawei in their "FTTS" project (S=Street, with copper lengths of up to 200 metres). That'll start as a 100Mbps system, but I guess they're aiming at G.Fast with speeds of 400Mbps.

So is NLV needed for our lesser FTTC environment?

On a different note... about how much further we can go with copper, beyond an FTTC rollout:

I've also seen a couple of presentations relating to FTTdp and gigabit copper: One UK-centric (and dumbed-down a little), and one US-centric (with more technical content).

G.Fast: The final hurrah for copper after our current FTTC rollout? Standardisation coming in 2014?

given that BT have already done a modem recall. Its certianly not impossible new modems can be issued to people signing up to a product the hg612 cannot deliver. These hg612, are very light small and I suspect are not costing BT that much per unit anyway.

I was part of the Coventry Virgin Media proof of concept speed trails. Initially, I was on a 200Mb FTTC connection. This constantly gave me over 190Mb. Towards the end of the trial I was given a much faster connection. On this I achieved 385Mb.

This, of course, depended on me having a PC powerful enough to handle flash-based speed tests smoothly and a gigabit router. Also I was the only premises on that connection. I'm not sure how good my performance would have been if many others shared my segment. However, currently, I achieve 104Mb 24/7 on a regular paid-for shared connection. With VM's coax technology the distance from the street cabinet makes no difference. I get Virgin TV with 3D and my landl ine down the same length of coaxial cable. I am expecting 120Mb by the end of May.

Edited by deleted (Fri 19-Apr-13 06:38:29)

VM is not FTTC.

In reply to a post by roughbeast:

I get Virgin TV with 3D and my landl ine down the same length of coaxial cable.

In my area the phone runs as twisted pair alongside the coax.

VM is fibre to the node (FTTN), which is a much larger area than FTTC - so you share the coax with a LOT more properties. If you look at some towns where lots of people in an area are maxing out the connection, the latency for all other users rises. This doesn't happen with FTTC as its point to point.

In reply to a post by lockyatlrg:

VM is not FTTC.

Virgin Media in most parts of urban Coventry except CV4 have fibre to the street cabinet followed by coaxial to the premises. What do you call that if you don't call it FTTC?

VM FTTC

In reply to a post by roughbeast:

In reply to a post by lockyatlrg:

VM is not FTTC.

Virgin Media in most parts of urban Coventry except CV4 have fibre to the street cabinet followed by coaxial to the premises. What do you call that if you don't call it FTTC?

VM FTTC

Generally FTTN - Fibre To The Node. It depends on the context. If we were debating broadband architectures you'd have a point. However it seems pretty obvious that we are discussing BT's FTTC so comments about VM are not really relevant.

Bit more info on vectoring (not much) here:

http://www.ispreview.co.uk/index.php/2013/04/bt-reve...

In reply to a post by roughbeast:

In reply to a post by lockyatlrg:

VM is not FTTC.

Virgin Media in most parts of urban Coventry except CV4 have fibre to the street cabinet followed by coaxial to the premises. What do you call that if you don't call it FTTC?

In South Manchester where VM started as Nynex, the fibre is to a Node cabinet with the cooling systems, then coax from that to many passive slave cabinets and coax from them to the premises.

That should be the same scenario for all VM, as the coax can only carry signal so far, and if you can convert optical to electrical without any electricity then you would be a very rich man.

In reply to a post by asbokid:

Profile 17a is limited to an Actual Net Data Rate (ACTNDR) of 100Mbps Down 60Mbps Up. And Openreach imposes an IP Profile of 80/20 on top of that.

It'd be nice if a brick was applied to the side of the head of the person(s) ion OpenWound who decided to set that profile.

In reply to a post by lexden16:

Quote: In vectoring, the VDSL signal is optimized for the subscriber line by compensating for interference between adjacent lines. Even when many subscribers use the line, this allows a transfer rate of over 100 Mbit/s for each connection. Unquote Source AVM Berlin

Yes, but it only works if the lines are clean. Many (most) openreach cables aren't, because they operate a cut-to-clear policy instead of fixing network faults.

DC faults, etc don't cause problems between voice circuits, but DSL is radio frequency and has to be treated differently. ANY faulty pair in a cable has the potential to badly mess with adjacent circuits, even when disconnected.

Not much more information, but as I understand it, vectoring involves playing about with the channeling (FDM) structure AND QAM structures to basically interleave on adjacent circuits.

It'll be interesting to see how well this works, given the rotten state of a lot of street cabling in the UK.

Vectoring certainly works when the copper is clean. The question is how much it continues to work when the connections degrade somewhat.

I guess we'll find out from the trials just how much real-life "dodgy" cables compare to lab tests that attempt to simulate such reality.

Luckily, one of the outputs from the vectoring data is an analysis that highlights those lines creating excessive noise. Of course, everything depends on Openreach taking proactive activity to fix things then.

In reply to a post by Stoatwblr:

Not much more information, but as I understand it, vectoring involves playing about with the channeling (FDM) structure AND QAM structures to basically interleave on adjacent circuits.

I'm not sure how much interleaving is involved - at least how we understand the term interleaving as it applies to VDSL2. And I don't think the basic FDM structure gets changed either

The simple overviews tell you that vectoring works by calculating which other circuits affect yours (at individual frequencies), and by how much (as a measurement of the error). This is fed back to estimate the future FEXT, then pre-subtracted from what gets transmitted.

However, from the look of what an "error measurement" is, it certainly looks like "pre-subtraction" fiddles with the QAM modulation.

It'll be interesting to see how well this works, given the rotten state of a lot of street cabling in the UK.

Aye.

In watching the posts from subscribers related to faults on FTTC, I've been surprised by how much less hassle it seems to have been compared to the ADSL2+ roll. Perhaps our D-side is in relatively better shape than the E-side.

In reply to a post by WWWombat:

Vectoring certainly works when the copper is clean. The question is how much it continues to work when the connections degrade somewhat.

I guess we'll find out from the trials just how much real-life "dodgy" cables compare to lab tests that attempt to simulate such reality.

Luckily, one of the outputs from the vectoring data is an analysis that highlights those lines creating excessive noise. Of course, everything depends on Openreach taking proactive activity to fix things then.

yeah if vectoring gets rolled out we will find out if the dodgy speeds are due to crosstalk or cabling faults. As I expect its easy at the moment to pass of everything as crosstalk.

The zdnet article linked to from the TBB story is a bit embarassing for the uk, USA already announced vectoring last year (so like I said it isnt as new as make believe just BT late to the table as usual), most other countries in that article stating they using FTTP or at least mostly FTTP whilst BT is bragging about how they still use copper cables from 1920 and that those cables are apparently in perfect condiotion because BT have cared for them .

http://www.zdnet.com/nbn-fibre-to-the-world_p3-70000...

Normally one advantage of FTTP is the hugely reduced maintenance costs over copper, sine the cost of fixing faults and maintaining the network plummets, every isp who has rolled out FTTP or who has started to roll it out is reporting they are glad they are doing it because they see it already paying for itself, so the question of course why doesnt BT see this same business case, especially as we have a small land mass which should make FTTP even more viable, the only ansers I can think of are BT just go for the absolute cheapest solutions at the time to satisfy shareholders even if they not the best long term solutions or that because BT dont maintain their local loops and ignore faults which means if they ignoring copper faults then the cost savings for moving to FTTP is reduced as BT are artifically reducing copper maintenance costs by not fixing copper faults (unless very severe or it happens to be the COO of a major business isp). http://www.ispreview.co.uk/index.php/2013/03/timico-... In this case his line would have passed openreach tests and could have been marked as in spec. Makes you wonder how bad a line has to be for the tests to fail.

Also it seems BT see FTTC as a mid teens length of investment, whether that includes a FTTP deployment or if they expect FTTPoD on its own with FTTC will suffice for over 10 years.

Edited by Chrysalis (Sat 20-Apr-13 06:27:42)

In reply to a post by Chrysalis:

whilst BT is bragging about how they still use copper cables from 1920 and that those cables are apparently in perfect condiotion because BT have cared for them .

Considering that those same lines can currently handle speeds of up to 80Mb/s I'd say it was worth bragging about. That's way more data than they were originally intended to carry. Being able to claim that cables that are nearly a hundred years old are still providing all the services a reasonable user would require (and let's be honest almost no-one needs higher speeds than a typical FTTC connection provides) is very impressive. What other hundred year-old infrastructure is still doing its job?

Edited by Andrue (Sat 20-Apr-13 08:04:11)

In reply to a post by Stoatwblr:

In reply to a post by asbokid:

Profile 17a is limited to an Actual Net Data Rate (ACTNDR) of 100Mbps Down 60Mbps Up. And Openreach imposes an IP Profile of 80/20 on top of that.

It'd be nice if a brick was applied to the side of the head of the person(s) ion OpenWound who decided to set that profile.

Why? What's wrong with it?

In reply to a post by Andrue:

What other hundred year-old infrastructure is still doing its job?

Many of the sewers in London that were built as big as tube tunnels

In reply to a post by Chrysalis:

yeah if vectoring gets rolled out we will find out if the dodgy speeds are due to crosstalk or cabling faults. As I expect its easy at the moment to pass of everything as crosstalk.

When I look at posts on here, and on the Plusnet forum, I'm struck by how much less hassle we seem to get on FTTC compared with upgrades done onto ADSL2+.

Perhaps the D-side+Tie cables are in better condition, relatively, than the E-side + Distribution frames. Perhaps there are just considerably fewer joints to interfere.

The zdnet article linked to from the TBB story is a bit embarassing for the uk

Embarrassing to the UK? Did you read the actual details in that story, or were you too embarrassed to take it in?

The US "announcement" plans to use vectoring to take speeds to 75Mbps - so less than we already get before considering vectoring.

The article states that U-verse pricing for 4Mbps was $41pm (£28) and 24Mbps was $66 (£44pm). I wonder what that 75Mbps package would cost? It doesn't seem to be available on their website yet.

The headline for the US project shows it aims at passing 16 million homes by 2015, with the target being 75% of 22 states - about the same total that can already get fibre-based access here, but with no match for the BDUK top-up coverage.

I bet the Americans are glad they have a non-embarrassing plan.

USA already announced vectoring last year (so like I said it isnt as new as make believe just BT late to the table as usual),

The US announcement, 6 months ago, is for an intention to increase speeds, and doesn't itself mention vectoring. It was a financial announcement, not a technical one, and is in no detailed plan yet. Other technical websites seem to show that it is more about upgrading to an "IP DSLAM", which would be rather similar to BT's 21CN WBC. Even there, statements about vectoring and bonding are tacked on - "merely" an eye on the future.

BT has been making the same sort of announcement - financial eye on the future - since at least a year earlier. Here is November 2011 (2 mentions on Page 12, one by name and one as "other technologies").

Further than that, here is a BT technology teach-in from October 2011 that details BT's plan for speed increases (ie band plan change, vectoring and bonding), and specifically says that it is trialling vectoring. Considerably more detailed than AT&T. Embarrassing? Not to me...

most other countries in that article stating they using FTTP or at least mostly FTTP

2 countries with predominantly FTTP rollouts, and 2 (more populated) countries with predominantly FTTC rollouts. One country that states FTTP but with unknown plans. Not exactly conclusive.

But look at the costs of FTTP in those places:
Australia: 100/40 packages cost £75-£80pm
New Zealand: 30/10 packages at $100, 100/50 packages at $134 (£55-£75)
Canada: 80/30 alone for $100pm, or 50/20 incl phone and TV for $160. (£67 - £105)
US: $66 for 24Mbps (£44)

That is quite a hike for the privilege of having fibre into the home. The packages certainly contain no more (or little more) than we get here.

whilst BT is bragging about how they still use copper cables from 1920 and that those cables are apparently in perfect condiotion because BT have cared for them .

Those'd be mainly the E-side cables, and he's actually pretty correct there. These larger cables tend to be pressurised to keep water out, and do tend to be in good condition.

The problem is almost never in the cables themselves, but in the joints - and the older PCP boxes date back more to the 50's and 60's. The cables that are a problem tend to be the final drop cables - especially when overhead.

Normally one advantage of FTTP is the hugely reduced maintenance costs over copper, sine the cost of fixing faults and maintaining the network plummets, every isp who has rolled out FTTP or who has started to roll it out is reporting they are glad they are doing it because they see it already paying for itself, so the question of course why doesnt BT see this same business case

BT *do* see the same business case, and we know they pay attention. One look at their financial statements show just how much they are juggling Opex against Capex. Major justification for their 21CN rollout was the tradeoff between the two, and the reduction in future maintenance cost.

The answer is that installation of FTTP is *hugely* intensive, making it hugely expensive. The reduction in maintenance does indeed offset this... but obviously not enough, yet.

especially as we have a small land mass which should make FTTP even more viable

We are also a small land mass with a well developed copper network, with the option of using the most recent technology to make good use of it. It happens to be that the topology of the copper network is well suited to that technology too.

The problem is that the option of using that copper comes at about a quarter of the cost, and a third of the time. Problem? It is an advantage... and one we should make use of!

or it happens to be the COO of a major business isp). http://www.ispreview.co.uk/index.php/2013/03/timico-... In this case his line would have passed openreach tests and could have been marked as in spec. Makes you wonder how bad a line has to be for the tests to fail.

You picked a story of "line fails, line gets fixed" as an example of why we must go for FTTP?

The line may have passed the standard fault-finding tests, so the fault would have remained unidentified. That doesn't mean it would have been marked "in spec". All that means is that it requires more human investigation rather than automated... which is precisely what happened.

The real problem at that point, for more normal mortals (or harder problems), is that it takes 2 or 3 (or more) engineer appointments, to get there. That isn't a problem with the copper... that's a problem with the way that Openreach structure the appointment system.

Also, did it occur to you that "a CTO for a major business ISP" is rather likely to have (and be paying handsomely for) business level service response? In fact, I vaguely recall an article of his showing how you could get faster Openreach response by ordering that higher service level shortly before reporting the fault. [That anecdote might easily apply to a different ISP blogger, so don't hold it as gospel].

Yes - these are the cases that start the drive towards FTTP being worthwhile. But are enough of them happening yet? I'm not sure - in particular I'm not sure that cabinet-based FTTC is having a worse problem than exchange-based ADSL2+.

Where the failures *do* happen, they show our one real failing here with our (UK) enforced separation of concerns through so many retail/wholesale layers. It is hard to get someone to take ownership, and get a problem thoroughly investigated and sorted. There are probably better problem cases than Trefor's to illustrate this - just not as high-profile. (sorry BaldEagle).

Also it seems BT see FTTC as a mid teens length of investment, whether that includes a FTTP deployment or if they expect FTTPoD on its own with FTTC will suffice for over 10 years.

I reckon that BT sees FTTC with a largely 20 year lifetime. Right now, it is creaming the top 15% of users away, and I reckon it will suffice for the first 5-10 years for most heavy users (particularly if vectoring does add ~50% to speeds). As those people move over to FTTPoD, they'll be replaced by the light users, for whom the 10-20Mbps is no longer enough. FTTC will be enough for them for the following decade.

In those 20 years, BT has the opportunity to build out the network based on FTTPoD, and to contract the rural edges using FOX. This, I suspect, is the gist of the decades-long strategy.

As for the end-user in this story?

Right now, I can get 80/20 - and I expect I'll get 100/20 within a year - for the sum of £35pm, with 250GB (unlimited if I chose to renew contract) and telephony included with all calls.

The best that Australia's NBN can offer at the moment is 100/40, even on fibre, with the next step down at 50/20. Getting a similar GB allowance (not unlimited) and a telephone (but no call allowance) will cost in the region of Aus$100 for the lower package to Aus$120 for the higher one. Roughly £67 - £80 per month.

I'm saving £35-45 per month, for the same service.

The bottom line is that I can get 80/20 now, 100/20 shortly, for £35pm. If I saved the extra £35 for 4-5 years, I'd have enough to order FTTPoD in 2017-18... and get faster speeds than NBN. And I'd still be paying less per month (probably, no retail price plans known here yet).

I get faster speeds and greater allowance now. I'll get faster speeds then. I'll pay less for it throughout. And I'd have finished my minimum contract term on FTTPoD before NBN has been fully built.

And I'm far from alone. That story could be repeated by half of this country in that same time.

I'm *not* embarrassed, except for our final 10%.

/RantOff

Very nice post...

FTTC has a projection of 20 years which I would think is about right to be honest.

Though the entire network isn't made up of copper as we know, ali is used on some connections as well. I wonder if we could finally see this be replaced, does ali work with vectoring? I'm assuming it will but have the same affect as it does currently.

The technology that is available now would easily suite the FTTC network.

We have 4K TV's coming this/next year which are said to use up to 60Mbps (Remembering off the top of my head). You have 3D advancing, it's very possible we could this come available over the Internet in the future though how far is beyond me...

FTTC will easily be enough for the future (10-15 years) especially if we see the upgrade to the 30a profile and vectoring is introduced.

'some' lines can handle it yes.

Although we dont know which ones.

The way he was bragging is as if the entire local loop is maintained and in perfect condition.

As an outsider I expect people are more impressed with investment in new infrastructure.

I find the vdsl technology impressive, it acheives a lot but I dont find BTs attitude and quality impressive. To me its the technology (which was developed by BT) impressive not their copper.

How well would vdsl be working for BT if none of the copper was replaced? all done from exchange.

that seems logical to me.

It seems logical E sides will be older and more higher density cabled as cabinets would have been placed as networks expanded to new housing areas been built.

Likewise its my view vdsl is far less problematic than adsl.

In terms of that news article yes the US has much lower speeds on at&t, although I was talking about vectoring as a technology rather than speeds. Of course The USA doesnt just have 1 provider owning all its lines, it does also have verizon rolling out FIOS and google FTTP.

Edited by Chrysalis (Sat 20-Apr-13 17:27:24)

In reply to a post by Chrysalis:

The way he was bragging is as if the entire local loop is maintained and in perfect condition.

Well it's not bad overall. Most people seem to be getting a service they are happy with. We have one of the highest take-up rates for the internet in the world and have had for some time. In a 'bums on seats' 'fit for purpose' view the UK is one of the world's great success stories I'd say. Considering it's mostly built on an copper it suggests there isn't anything fundamentally wrong with the network.

As WWombat stated in his excellent post - the problem is with the BT fault process not the copper. My own ongoing problems (started again yesterday as predicted) are not due to copper faults. They appear to be due to network faults and the issue my ISP and I have is trying to get BT to understand that and get the right people to look into it. It's likely your problems would be resolved if only BT would get a decent engineer to spend a proper amount of time looking at it.

We don't need a better copper loop or replacing copper with FTTP. We just need BT to do a better job of investigating and fixing faults.

Edited by Andrue (Sat 20-Apr-13 18:47:34)

In reply to a post by Andrue:

...We just need BT to do a better job of investigating and fixing faults.

I totally agree with that, my line performs very badly for the distance, yet we have others on here that can almost achieve the full 80Mbs for the same distance.

When I did manage to get an engineer out, all he did was check the wiring in the house, he wasn't permitted to check the PCP, swap the pairs, or anything else, it was a complete waste of his time and mine!

In reply to a post by Chrysalis:

'some' lines can handle it yes.

Although we dont know which ones.

I agree here - aluminium can, at best, be treated like longer copper lines.

But aluminium does get corroded, so runs the risk of suffering more than copper would. It makes aluminium lines (or lines that are part aluminium) more of a lottery.

So far, Openreach have shown themselves to be very reluctant to dig and replace aluminium. I don't think FTTC or vectoring is going to persuade them to treat it differently.

In fact, I think they're probably more likely to route around it with FTTP. Whether they get to it first, or a customer gets fed up and requests FTTPoD first is another matter.

The way he was bragging is as if the entire local loop is maintained and in perfect condition.

I didn't mean to come over as bragging, nor am I defending the entire local loop. But while it isn't perfect, it seems that a lot of it is doing a fine job with FTTC - and will continue to do so with vectoring.

The fact that some people have problems doesn't mean we should consign it to the bin in its entirety.

I'll come back to the issue of non-perfect lines in a separate post.

As an outsider I expect people are more impressed with investment in new infrastructure.

As a spectator of how many people have taken broadband because of the way it gets bundled by the likes of Sky and TalkTalk, I expect the vast majority of people are more impressed by getting it cheap.

I find the vdsl technology impressive, it acheives a lot but I dont find BTs attitude and quality impressive. To me its the technology (which was developed by BT) impressive not their copper.

On attitude, I'll explain in a different post.

Let me be explain my position (on technology and connectivity) here...

I'm not invested in keeping copper nor in replacing it wholesale. I'm interested in new technology, but not a gadget freak for gadgetry's sake.

I spent years wanting "more" out of network connectivity for the home in the 90's. Virgin (nee NTL) ignored my newly built area (and still does). I looked at all manner of things to get internet connectivity, including packet radio and WiFi (including some schemes that let you use Linux WiFi to be part of an ad-hoc fixed-wireless chain back to some PoP that *did* have access - but can't remember the name of that one). Interesting "hacking" stuff, but it was never mainstream.

So I was massively taken to get into the early ADSL trials in 2000. I took a 2Mbps connection, even though there was no way I could justify the £50+ cost. Back then, it is fair to say, I was desperate to get an always-on, semi-decent-speed connection.

If fibre was the only way to get an internet connection today, I'd be ordering FTTPoD the day it came out - the minute it came out - and money would be almost no object. Or I'd have already moved to somewhere with native fibre. And we wouldn't be having this discussion.

But it isn't the only way. Now I have the choice - and I will pick the technology at the cost sweet-spot that has some flexibility above what I need. I'm seeking neither the cheapest nor the fastest/latest. I'm seeking good enough, but still good. But I am seeking reliable connectivity.

We work from home. We use VoIP, and we both upload and download. I am fairly impatient. We expect things to work, and to not give us excessive delays.

I know that some people will seek the fastest or latest, but that most people don't need more than I do. I know that some will pay more, but that most will want to pay less. I know some people want all-out speed or all-out low latency, but that most just want sane stability. I'm pretty confident in those beliefs.

Right now, copper (with FTTC) hits the sweet-spot for me - whereas 14 years ago the copper access network wasn't even close. I have indeed changed my position as technology has come in, though now it does restrict my property choice to about 50% of the UK.

If it hits my sweet-spot, I am pretty confident that this is a high point for most customers in the country - less speed and less money, with equal stability.

The only way that I don't fit in with the majority of people is that I don't watch as much pay-TV. We don't have Sky, and haven't felt the need to go to Virgin for TV. We *do* timeshift a lot of TV on a Freeview PVR though, and we're quite versed at iPlayer and Youtube, and do want HD video to stream without buffering. I'm sure we'll try out Netflix or LoveFilm at some point this year.

So I do have to make allowances for the fact that there is a growing portion that places more demands for video over the internet.

Right now, that allowance doesn't change my assessment.

How well would vdsl be working for BT if none of the copper was replaced? all done from exchange.

For a lot of people, about the same as ADSL2+, but for the furthest people (beyond say 3km, so a lot of them) it'd be considerably worse, to non-existent. Its a matter for technology horses for courses, and this filly would be in the wrong race.

In reply to a post by Chrysalis:

In terms of that news article yes the US has much lower speeds on at&t, although I was talking about vectoring as a technology rather than speeds. Of course The USA doesnt just have 1 provider owning all its lines, it does also have verizon rolling out FIOS and google FTTP.

I agree - it is hard keeping track of who covers what, where. Telecomms in the US seems to always be very "city" oriented, in that most people/businesses only care about the one city where they are located.

In reply to a post by R0NSKI:

In reply to a post by Andrue:

...We just need BT to do a better job of investigating and fixing faults.

I totally agree with that, my line performs very badly for the distance, yet we have others on here that can almost achieve the full 80Mbs for the same distance.

When I did manage to get an engineer out, all he did was check the wiring in the house, he wasn't permitted to check the PCP, swap the pairs, or anything else, it was a complete waste of his time and mine!

A lot of this seems to be as a consequence of the way Openreach has organised itself to manage the "equivalence" requirements, and to hit more appointment targets.

There has to be something better they can do that actually gets a fault fixed better, and gets feedback through to the ISP (and along to the next engineer in line). It is letting them (and us) down, and the ISPs must be sick of taking the brunt of the flack for it.

It'll certainly be interesting to see if they do end up using the extra analysis tools for Vectoring, and start fixing lines before the owner even knew there was a problem.

It'll also be interesting to see when the FTTPoD rollout starts to be used as an alternative to fixing copper. At some point, for some customers, it will become cheaper to just install fibre than to investigate and the copper. I guess it'll be a few more years yet.

In reply to a post by R0NSKI:

In reply to a post by Andrue:

...We just need BT to do a better job of investigating and fixing faults.

I totally agree with that, my line performs very badly for the distance, yet we have others on here that can almost achieve the full 80Mbs for the same distance.

Thirded - I'm at 400m, and others get 60+meg, with nearly 20meg upload. I get 40/45 and 8. At least mine seems to be 100% reliable as I used to get from 2006/2010 with Bulldog then BE, but not from 2010 to late 2012 after "something changed".

In reply to a post by R0NSKI:

When I did manage to get an engineer out, all he did was check the wiring in the house, he wasn't permitted to check the PCP, swap the pairs, or anything else, it was a complete waste of his time and mine!

Last time my connection was playing up my ISP contacted their account manager and explained that they needed the network throughput investigating, not the line. So what did BT do? They sent out an SFI engineer :-/

In reply to a post by WWWombat:

Where the failures *do* happen, they show our one real failing here with our (UK) enforced separation of concerns through so many retail/wholesale layers. It is hard to get someone to take ownership, and get a problem thoroughly investigated and sorted. There are probably better problem cases than Trefor's to illustrate this - just not as high-profile. (sorry BaldEagle).

On a positive note, the lack of ownership & TBH, complete denial from both my ISP & BT that anything was actually wrong with my connection did drive us on to develop firstly the logging scripts & more recently the more efficient program versions so that users can at least see how well their connections are performing (or not).

FTTC & copper D-sides was/still is the only technology available in my area that allows me to achieve more than the 1Mb available via ADSL due to distance from the exchange.

When it works, it does work quite well.

Getting the powers that be to acknowledge when it genuinely needs repairing is the harder part, made even harder by what appears to be the rather poor training provided for engineers regarding 'proper' fault diagnosis etc.

Even getting DLM reset to clear 'stuck' capping following repair works was/still occasionally is a major battle.

I might be late to this concept, but I only just did the maths...

On FTTP, the basic architecture is for 32 properties to share a single GPON - a single optical network - back to the exchange.

This GPON can carry 2.4Gbps down, and 1.2Gbps up.

The maths tells you that each property gets 75Mbps down and 37Mbps up.

Perhaps the question on this thread should be: Can FTTP handle speeds above 75Mbps?

I got to wondering when I looked at ISPs offering services on FTTP network (like the NBN in Australia). They wanted exorbitant prices (as you'd expect for FTTP) but still only seemed to want to offer services at a maximum of 50-100Mbps.

It's a little tongue-in-cheek. as burst speeds are a lot higher, but the average is quite low.

Obviously, one reason BT is deploying all of this fibre (of both kinds) is so they can start offering TV services over it. Ironic then, that when the nation is all watching Eastenders or Corrie at peak evening hours, it is the FTTC houses that could have the most spare bandwidth.

I confess I know nothing about fibre....

Does this ties in with the data rates for Gigaclear's offering in those villages where it is putting in FTTP.
They seem to be only offering 10Mbps symetrical guaranteed while putting lots of PR into the "bursting speed" of up to 1GB.
I think they do offer higher guaranteed rates but at much higher prices

your post makes no sense to me.

If I understand you right you saying because its built on copper it must mean its in good condition.

To me if we see hugely varying results from people it suggests somethign isnt right, and reports of faults whether it be slow speed or instability or both are too common to suggest its all in generally perfect condition.

Certianly as well alot of misleading information on the status of the local loop is openreach's approach to faults, where they generally assume a problem in the property first and then if the JDSU test passes they consider it job done.

From where I sit we can all only speculate on the state of the copper due to the fact BT have shut their eyes to faults. Any figures from BT will be inaccurate because of their refusal to accept faults exist.

My experience on ADSL was in short a joke, to call that acceptable service would be a lie. The only reason now my service is not like that anymore I suspect is the lengthy E-side of copper is no longer involved.

BT have just done a better job at managing expectations with VDSL, hiding the sync speeds, under estimating speeds, also I expect most FTTC customers are on 40/10 or 40/2 so a line doesnt have to be particurly great and that if you consider many are coming from adsl even a dodgy vdsl service is going to be a decent improvement, of course generally people will be happy.

In general I am happy as well. This service is superior to VM cable, its most defenitly superior to adsl and of course over 3G. Even if the line declined further and i only got 40mbit I would still consider it superior to all those other services but that doesnt make it perfect as BT are claiming.

It seems we differ in that I think there is a clear difference between acceptable/adequate and perfect (or even great).

Edited by Chrysalis (Sun 21-Apr-13 10:39:43)

I think the copper/ali network is generally in a poor state.

About 10 years ago when our new workshop was built a BT engineer told me the lines to the estate were in bad condition.

My brother last year had a fault which took months to sort out, one engineer told me that there were no available pairs left as there were so many dead pairs. He said that BT would rather keep patching, than replace cables. He said that he had recommended cables to be replaced, but no they would rather patch it, and then he would be back a few weeks later to repair another bit of the same bunch of lines.

Obviously this is only what I've heard, but I suspect it's indicative of a lot of the older network.

In reply to a post by zom22:

Does this ties in with the data rates for Gigaclear's offering in those villages where it is putting in FTTP.
They seem to be only offering 10Mbps symetrical guaranteed while putting lots of PR into the "bursting speed" of up to 1GB.
I think they do offer higher guaranteed rates but at much higher prices

I know nothing about the technical details of Gigaclear's offering, but that kind of price & throughput structuring suggests they are indeed using a shared GPON setup. It doesn't tell us how many individual properties share the same ultimate fibre back to the head-end though - they might just set the low limit so they have the opportunity to sell higher priced packages as a way to recoup their investment.

In reply to a post by Chrysalis:

If I understand you right you saying because its built on copper it must mean its in good condition.

I read it as "it is built on copper and mostly appears to be working well so it must be in good enough condition on the whole"

That doesn't mean that it is *all* working well, or that it is all in good condition.

To me if we see hugely varying results from people it suggests somethign isnt right,

First, we have to recognise that self-crosstalk from other FTTC customers (and in this cobversation, we were only talking about the copper portion in the D-side) is the predominant noise, that it is random, and that it alone can cause hugely varying results from different people.

One table (from a vectoring report) that shows un-vectored performance (ie what we have right now), shows that the unluckiest (most disturbed) 200 metre line can get the same 80Mbps speed as the luckiest (least disturbed) 600 metre line.

It shows that someone on a 400metre line could get anywhere between 55Mbp and 90Mbps, depending on luck.

With all this variance comes extra noise, which triggersDLM to turn on interleaving and FEC, where speed is deliberately sacrificed for stability. That adds more variance.

So, right now, we *expect* huge variance - and there is nothing that BT can do about the laws of physics to change these.

and reports of faults whether it be slow speed or instability or both are too common to suggest its all in generally perfect condition.

On balance, I find there are less reports of real problems in FTTC than when ADSL2+ was introduced. There are still some - and I agree there are indeed problems in the loop.

While there seem to be less problems, when we do have one, the end-user has a lot of problems getting it dealt with readily. That *is* an attitude problem that BT need to deal with.

Right now, we are seeing a lot more people reporting a drop in speed on FTTC, and questioning why. We know why, and we expect it - but it isn't an indication of any additional problem in the copper loop.

Certianly as well alot of misleading information on the status of the local loop is openreach's approach to faults, where they generally assume a problem in the property first and then if the JDSU test passes they consider it job done.

I suspect their resolution statistics tell them that the most likely source of faults *is* the property.

I think the problem is really that the engineer is only allowed to spend a limited amount of time investigating, and can broadly only get to deploy the JDSU at the property, then give up.

The lack of resolution at this point is worsened because of the lack of continuity when the next engineer finally gets involved.

Getting things fixed in one go has to be a priority...

From where I sit we can all only speculate on the state of the copper due to the fact BT have shut their eyes to faults. Any figures from BT will be inaccurate because of their refusal to accept faults exist.

As you say, we can only speculate. Your reasoning is speculation too.

My experience on ADSL was in short a joke, to call that acceptable service would be a lie.

It's hard to disagree, and it is infuriating to have to go over the same issue again & again with engineers that don't appear to learn, and don't (in any case) have much authority to go re-arranging the physical network to sort your problem out.

The only reason now my service is not like that anymore I suspect is the lengthy E-side of copper is no longer involved.

Almost certainly true.

If that experience is repeated throughout the country, then we're concentrating our broadband onto the better part of the copper network. And it does seem to be repeated.

BT have just done a better job at managing expectations with VDSL, hiding the sync speeds, under estimating speeds, also I expect most FTTC customers are on 40/10 or 40/2 so a line doesnt have to be particurly great and that if you consider many are coming from adsl even a dodgy vdsl service is going to be a decent improvement, of course generally people will be happy.

I agree with all of that.

The "under-estimating speeds" part is required, because of the variance we are now seeing with crosstalk. As the cabinet fills up, there will be *some* people who end up down on those speeds.

But the rest is true. Choosing headline "up to X" values that are available to a significant portion was a way better choice than describing ADSL2+ as "Up to 24".

In general I am happy as well. This service is superior to VM cable, its most defenitly superior to adsl and of course over 3G. Even if the line declined further and i only got 40mbit I would still consider it superior to all those other services but that doesnt make it perfect as BT are claiming.

I agree. I don't think BT are claiming perfection though, and I do see that they're running programmes to improve their resolution times.

... I think there is a clear difference between acceptable/adequate and perfect (or even great).

Same here. I also try to distinguish the faults into cases where the copper is deficient from those where the support & engineering procedures are deficient.

In reply to a post by WWWombat:

The "under-estimating speeds" part is required, because of the variance we are now seeing with crosstalk. As the cabinet fills up, there will be *some* people who end up down on those speeds.

Is anyone actually able to confirm the symptoms of crosstalk in as few words as possible, perhaps linking to sections of a technical document for further detailed reference?

Yes, we know the ultimate symptom is reduced sync speed, but what exactly causes that?

Presumably line attenuation should remain static, but perhaps signal attenuation is affected.
Likewise, no doubt SNR (not SNRM) is reduced as the 'noise' increases.

What about errors such as Error Seconds, CRC, HEC & FEC errors?
Do these ALL increase with increased crosstalk?

QLN at DSL frequencies may also be affected.

e.g. does this look like crosstalk or some other problem?:-

QLN - Before & After animated GIF




Interleaving depth hasn't varied much other than when it was first applied (as fully expected)following repairs & a DLM reset 26/05/2012, but sync speeds have gradually reduced from 33 Mb to currentl;y 23.7Mb.

Back in December 2012, dynamic SNRM reduced on my connection & QLN in my D1 band changed when the connection resynced at a lower dpeed after a few days.

Friday 19th April, we experienced 3 or 4 split second power outages where the router disconnected/reconnected but the HG612 modem remained in sync.

A couple of hours later, I did reboot the modem & have noticed that QLN in the D2 band has changed for the worse, along with sync speed dropping to below 24 Mb..

Running my HG612 from a fully charged car battery & using a battery powered laptop with the mains completely switched off yesterday, confirmed that there were no measurable 'noise' problems within my own mains power wiring or from any electrical appliances.

Could this be the effect of increased crosstalk that I'll just have to live with or could it be a 'dodgy' underground mains supply cable problem?

I would really like to understand the differences in symptoms between them.

If it is felt that these questions are off topic, I'll be happy to start a new specific thread.

ahh yeah, your post reminded me of when a engineer told me the same when I was on ADSL pairs had ran out, the situation was so bad BT even built a new cabinet round the corner from me which I assume also had a new E side feed as there was no other cabinet anywhere near it. I may check if that cabinet has a FTTC cab also as I am curious now.

As I understand it but only based on articles and posts I have read.

The symptoms 'may' be but not necessarily.

Decreased snrm/sync speed without any affect on attenuation.
Increased QLN.
Increased CRC/FEC errors.

Of course what also could be happening is its possible openreach are messing with power masking, bringing good lines down to estimated speeds to reduce crosstalk to struggling lines.

Two separate answers...

In reply to a post by Bald_Eagle1:

Is anyone actually able to confirm the symptoms of crosstalk in as few words as possible, perhaps linking to sections of a technical document for further detailed reference?

How about this:
Broadband Forum TR-197, August 2012 "DQS: DSL Quality Management Techniques and Nomenclature"

Scope:

TR-197 provides a unified nomenclature for the techniques and strategies available to address DSL line stability and quality issues. TR-197 defines a common language to refer to parameters, functionalities and typical network practices in the problem domain specified above

Section 5 lists the different kinds of noise - Crosstalk, INP, REIN, PEIN and SHINE and RFI. I guess that can teach you a bit.

Section 6 talks about all sorts of techniques for controlling the noise, including many that we don't come across in the UK. It is more complicated, and may be more than you want to know.

Yes, we know the ultimate symptom is reduced sync speed, but what exactly causes that?

You want a simple introduction to the principles of "DMT", "QAM" and "Water-filling", here...and I found this page at Kitz after I wrote the stuff below. Great description - go read it... this'll wait.

You already know that DSL works by diving the spectrum into a number of separate frequencies, at 4.3ish kHz intervals, right?

Consider just one of them. The modem, during negotiation, determines how much signal can be heard at that frequency, and how much noise (all forms of noise) can be detected. The difference between the two (less the target margin) dictates how many bits can be carried at the frequency point: A bigger difference can carry more bits, while a smaller difference can carry fewer bits.

The overall throughput, the sync speed, comes from summing the bits carried at all N different frequencies.

So, we get to make some choices, and see some consequences, at this point...
+ If we increase the power, we see more signal and carry more bits. But we interfere with other users
+ If the noise increases, we run the risk of it overcoming the signal, and the receiver getting the wrong set of bits.
+ If we use a larger margin, we end up carrying fewer bits. However, we're more immune to the noise

The decisions made here result in the bit-loading graphs we usually see: A number of bits at each of many tones (frequencies).

If, before the next sync, we change the target margin, then the negotiation determines it can carry fewer bits at all frequencies across the board.

If noise increases in a narrow band, then the next negotiation will causes fewer bits to be allocated to just the affected frequencies.

So... at the next synchronisation, speeds change.

But every change comes about because of a change in signal, noise or the target margin.

Presumably line attenuation should remain static, but perhaps signal attenuation is affected.
Likewise, no doubt SNR (not SNRM) is reduced as the 'noise' increases.

Attenuation shouldn't change - but I haven't learnt if there is a subtle difference between "line" and "signal" yet.

SNR is (strictly) the ratio between signal level and noise level, but you can just think of the SNRM as the difference between the two. It is measured constantly, so it can fluctate over time.

What about errors such as Error Seconds, CRC, HEC & FEC errors?
Do these ALL increase with increased crosstalk?

No - not necessarily, as it depends on how crosstalk impacts your system:

- If crosstalk is relatively steady, seen and measured as increased noise at the time of sync, then the noise is already taken into account when the sync speed is set - fewer bits are carried at the frequencies where crosstalk impacts are felt.

Result: you run at a slower speed, with the modem running at the target SNRM value (if you are below your package speed), but don't then tend to get any errors because of the noise.

- If crosstalk isn't measured at the time of the sync, the sync speed will not take it into account, so be higher. It might be because the noise is random or intermittent, and is missing at the time of the sync, or because the crosstalk has just gradually increased in the period (possibly long, possibly short) since the last sync. Or perhaps a neighbour just came back from holiday.

When the crosstalk *does* start to impact your line, the extra noise is detected, and it will reduce the SNRM at the appropriate frequencies (you just see a summary average on the modem GUI). The noise makes it hard for the receiver (of a particular frequency) to read the signal and identify the correct values for the N bits it is expecting. When the receiver makes a mistake, it becomes an error in the bitstream.

The impact of the error depends on whether FEC is running: If so, it attempts to fix the error, and the appropriate counters change.

If FEC isn't running, or couldn't fix the error, it gets noted in the CRC and ES counts.

Such noise becomes visible to you because the counters change, and because the SNRM reduces, for the duration that crosstalk affects you.

However, the other forms of noise have the same impact on counters and SNRM. You cannot use these to determine the type of noise alone.

QLN at DSL frequencies may also be affected.

To determine the kind of noise, you'd want to look at:
- The SNR-Tone graph while noise is currently present, compared with one where the noise wasn't present
- The Bitloading graph after a sync when the noise was present (even if it isn't present now), and compare it to a bitloading graph of a sync when noise wasn't present.
- As you guessed, the QLN graph after a sync when the noise was present (even if it isn't present now), and compare it to a QLN graph of a sync when noise wasn't present.
- Maybe the Hlog graph (which I can never remember how to interpret)

and compare with the descriptions of the different kinds of noise from the original link.

I'll deal with your example in a separate post.

Yup - all of those are possible symptoms, and possible alternate causes.

Thanks for the link & your detailed comments.

I have downloaded the document & will wade through it in due course - perhaps not as bedtime reading matter though

The first snippet I read in section 5 when having a quick glance could possibly explain why a lift & shift is sometimes recommended - i.e. not all cables in the same bundle are affected by the same amount of crosstalk.

Edited by deleted (Sun 21-Apr-13 22:26:11)

Look at WDM and other GPON architectures for how you can boost capacity without a lot of rejigging

FTTC only has a 30 Mbps allocation on the Openreach segment so FTTP beats that based on this simple sum. 75 Mbps would allow each property to view several HD streams too.

Every connection on Gigaclear is Gigabit capable, but you pick the price point you want for the guaranteed bandwidth you want.

So yes its about ensuring that people simply don't eat all the bandwidth, e.g. getting 10 Gbps to a village of a few hundred properties is not cheap, and I think many people would be shocked if Virgin, BT and others were clear about the allocations in their retail products. Some say that with TalkTalk fibre services it is just 1 Mbps but no proof.

Gigaclear could do like some overseas providers have apparently done, and offer different speeds for local, regional, national and international bandwidth, reflecting the costs of providing capacity to different areas.

In reply to a post by MrSaffron:

Look at WDM and other GPON architectures for how you can boost capacity without a lot of rejigging

I agree there's plenty of room in theory, but I don't think they're planning on WDM in the too-near future.

Are they planning on, say, having smaller PON groupings for businesses, so there is less contention?

FTTC only has a 30 Mbps allocation on the Openreach segment so FTTP beats that based on this simple sum. 75 Mbps would allow each property to view several HD streams too.

Ta - I was wondering what the shared backhaul would be like. Is this upgradeable with multiple fibres?

Sorry, I haven't had a chance to come back to this before....

In reply to a post by Bald_Eagle1:

e.g. does this look like crosstalk or some other problem?:-

QLN - Before & After animated GIF

This looks very much like crosstalk.

The noise spreads a long way across the spectrum, but not all the way, covering tones 250-1600.

At less than 250, the power maps reduce the transmit power to prevent interference from ADSL. It means that there tends to be less interference from a noisy VDSL2 neighbour, but relatively more ADSL2+ noise from the exchange. So in your before/after graphs, I'd expect little change.

Above 1600, there seems little difference. Whoever is interfering may not be using those channels.

Between 1200 and 1500, the effect decreases dramatically, perhaps as the other user uses these frequencies less - and possibly showing you their distance from the cabinet too.

Between 250 and 1200, the effect seems to be a fairly linear drop.

I suspect that such a widespread drop is an indication of interference from someone using all those frequencies - such as another VDSL2 user, rather than a pinpoint frequency (such as a radio station).

Given the time difference, do you suspect that a single user turned his modem on? Or is this the power problem?

Interleaving depth hasn't varied much other than when it was first applied (as fully expected)following repairs & a DLM reset 26/05/2012, but sync speeds have gradually reduced from 33 Mb to currentl;y 23.7Mb.

That would be explainable by crosstalk too. Do you have similar before/after graphs?

A couple of hours later, I did reboot the modem & have noticed that QLN in the D2 band has changed for the worse, along with sync speed dropping to below 24 Mb..

I'd have expected a power issue to exist in a narrow frequency band somewhere, rather than wideband.

I did find something else to help you:
JDSU doc with some troubleshooting, but not a lot to help this specific problem.

And more here on comparing SNR/tone with QLN.

In reply to a post by WWWombat:

Sorry, I haven't had a chance to come back to this before....

In reply to a post by Bald_Eagle1:

e.g. does this look like crosstalk or some other problem?:-

QLN - Before & After animated GIF

This looks very much like crosstalk.

Thanks for your comments & links to the various documents that have added some clarity.

Brief history of my connection:-

FTTC installed June 2011 Plusnet 40/2 Profile 8c service.
Throughput speeds 32Mb - 33Mb for almost 1 month (no detailed stats - modems still locked).

Loss of phone & broadband services for 4 days.

Services reinstated at lower speeds.

September 2011 onward able to unlock the modem & start harvesting/graphing stats.

Many unforced disconnections & sync speeds occasionally as low as 8Mb DS until late May 2012.

Many engineering visits - issues seemed to worsen during warm & dry weather, stability & speeds improving during cold & wet weather!

October 2011 Profile 17a introduced.
Further slight DS speed reduction (D2 band moved slightly higher up the frequencies - due to distance (1000m or so), my connection can't make use of tones higher than 1750 as attenuation is by then too high).

November 2011 - Drop wire from pole mounted DP replaced.
Connection 'seemed' stable & speed had increased, but the issues returned after a few days.

February 2012 - Engineers working on a different DP on the same pole unwittingly caused disconnections & lowered sync speeds.

May 2012 - faulty SSFP replaced (had been recently causing connection drops when phone used - diagnosed by using dangly filter in the test socket which stopped the issue).
Faulty joint between underground cable & pole mounted DP was finally repaired (the engineer had to be coaxed to inspect it, but he said it just fell off in his hand as soon as he touched it).
Connection finally repaired & stable at 33 Mb sync speed - was initially almost 36Mb until interleaving applied.

During the 'issues' one engineer TDR tested the connection back to the PCP & located a double jumper at the PCP.
This 'bridged tap' was removed - very slight speed increase.

Experimented with Plusnet, switching between 40/2 & 40/10 services a number of times.
When US speed increased by around 3 Mb, DS speed decreased by 3 Mb.

I stuck with 40/10 as I work from home a lot & upload speed over VPN to the office is important for transferring large files.

Sync speed remained around 30 Mb / 5 Mb, very few resyncs & quite lowish error counts.

December 31st 2012 - SNRM suddenly lowered & the connection resynced a few days later at a lower speed (approx 26Mb), SNRM back at the target 6dB.
D1 band shows increased QLN 'interference'.

March 2013 - suddenly start seeing increased Error Seconds & also increased HEC, CRC & FEC errors.

Tried the new BLOB firmware - approx 1Mb speed increase to around 27Mb.

April 19th 4 momentary power outages (each lasting less than 1 second).
Modem remained in sync, but router disconnected/reconnected.
D2 band shows increased QLN 'interference'

Throughput seemed low, so I rebooted the modem.
Sync speed reduced to 23.7 Mb.


April 20th, ran the connection from a car battery & a battery powered laptop (all house power switched OFF at the incoming mains consumer unit) to eliminate internal power/appliance issues.
No real change to QLN/SNR graphs (still showing increased 'interference' in D1 & D 2 bands.



April 22nd - Connection had resynced 'on the fly' overnight - Retrain Reason 2 - at just 23 Mb.
Reverted to original BLOB firmware. Sync speed 23.48 Mb.
The modem is still reporting increased errors.

Reading the various documents does indeed indicate increased crosstalk, but can it really be pure coincidence that a sequence of very brief power outages appear to have caused another sustained reduction in sync speed & increased 'interference' in the D2 band, despite only the ROUTER disconnecting/reconnecting?

I do have a spare unlocked modem which I will try in a few days.

Just as a final check of things at my end, I will try a dangly filter in the test socket a few days after swapping modems, although use of the phone is NOT causing any issues & the audible QLN test via the phone is still quiet.

I have not seen frequent random disconnections since the DP joint was repaired in May 2012, so I suppose my connection has been & still is genuinely stable, but at lower speed due to increased crosstalk (or possibly interference from a now 'noisy' underground mains cable that had caused the random power outages the other day?)

The difficulty in finding the faults had been that they were intermittent.

Crosstalk presumably is more constant, apart from other users maybe switching their ADSL or VDSL2 routers/modems off overnight or only switching them on whenever they use the internet.

This is all way off the original topic, but it may help us & other users in understanding the effect of crosstalk etc.

Vectoring may assist (to an extent), but it would appear to be more limited for longer connections where the opportunity of crosstalk to have an effect, from differing types of services such as ADSL in the same D-side bundle(s), would increase, so traditional interleaving & INP would still be required.

after reading ignitions post, are you sure its M41's already deployed? and were V41's available when BT started deployment? It be a bit said if BT were a bit shortsighted and V41's were already available.

Thanks Simon,

That is an interesting series of blogs. There is the added (unmentioned) complication of the non-vectored lines in a cable binder. Those will be the ADSL lines served by central office kit in the exchange. Since there is no way of quantifying and compensating for the "alien crosstalk" that those lines produce, it will continue to degrade the service on the VDSL2 pairs.

It doesn't look like the Huawei DSLAMs will need to be replaced to support system-level and node-level vectoring. Just a firmware upgrade for the controller board. So the system architecture of the ECIs and the Huaweis is perhaps very different in this respect. ECI appears to have loaded more logic onto its linecards. There's mention of a vectoring daughter board for the linecards as well. Huawei seems to have kept its linecards simpler, with the logic focused on the central controller unit (CCUB / CCUC) boards.

cheers, a

Thinking longer term here:
Maybe the complication of existing ADLS2+ users would lead to a program of moving all such users on to VDSL2 and with it aligning the pricing structure.
In much the same way as ADSLmax/20CN customers were/are being automatically moved onto 21CN....though that was easier in that it was seemless for most users as they did not need a new routermodem unit.

Then BT could remove all the ADSL equipment from the exchanges with associated cost benefits to them.

At the moment the name of the game is getting more money out of those wanting to swap to fibre and willing to pay the price premium. Once the takeup rate starts to approach a saturation level of some % of broadband enabled lines, then one could expect the above scenario starting to play out.

Perhaps the ADSL2+ alien crosstalk issue will not be too bad as it's all concentrated at the lower end of the vdsl spectrum....

In reply to a post by zom22:

Then BT could remove all the ADSL equipment from the exchanges with associated cost benefits to them.

But what about LLU?

If the LLU DSLAMs are still in the exchange then they would still cause crosstalk.

Does vectoring work with ADSL? Run the ADSL lines from the FTTC CAB?

But this also assumes that "all" CABs have FTTC, in practise many don't.

In reply to a post by zom22:

Thinking longer term here:
Maybe the complication of existing ADLS2+ users would lead to a program of moving all such users on to VDSL2 and with it aligning the pricing structure.

You need to remember that not only would such a scheme impact LLU ADSL it would also remove ADSL from the significant number of EO lines who currently have no access to either FTTC or FTPOD

Good point about EO lines - I had forgotten about them. Anyway I'm talking long term: 5yr+ when indeed all the cabinets on an exchange have been FTTC'd - by then a solution for EO hopefully should have been found.

As regards LLU - it seems to me that FTTC is sounding the death knell for the concept of LLU.