And who is going to pay?
This ADSL consumer pays under £6 per month for an unlimited adequate stable connection
The FTTC alternative looks to be around £18 per month upwards
Forgive me if I opt out

ADSL might well be cheaper now but there will be a tipping point in future where price of ADSL and VDSL based products are similar as they try to "persuade" customers to "migrate" so they can decommission ADSL hardware. If the operators margins are as tight as they would have you believe they wont continue to support the two systems forever.

In reply to a post by MrSaffron:

Of course until the decision is made on the future of Openreach everything is a might, and if independent we can expect the plans of Sky to dominate as they are calling for the total split so heavily i.e. they may want to focus away from xDSL totally and start a 10-15 year FTTH roll-out plan instead and walk away from any CityFibre plans.

As Sky have dived headlong into making their LLU assets and full-loop copper redundant in the name of enhancing UK broadband you mean?

In reply to a post by mr_mojo:

The [power] masking on that BT does to allow FTTC & ADSL to coexist is having a really serious penalty on people with long lines from the cabinet.

It's [fair] to say many people would see a gain of at about 20mbit down if the VDSL2 band plan could use the frequencies that ADSL2+ does.

This whole thread starts on a premise that I haven't seen challenged yet, and both of the statements above need to be challenged, at least in part.

1) First, power masking is not even across all lines; it depends how far the cabinet is from the exchange.

The ANFP uses a CAL value to denote this electrical distance, and CAL values of 0, 2, 4, ... 50, 52dB. A best estimate is that each CAL step (of 2dB represents a real distance of approx 200m).

For cabinets that are very close to the exchange, there is no masking at all. From the ANFP graphs, a CAL of 0 applies no masking at all - and looks like it is used for cabinets within 100-200m of the exchange.

Even CAL values up to 10 apply limited masking (in the sense that the spectrum is limited, and the power restriction is small).

CAL values of around 20dB (cabinet-exchange distances of 2km) seem to give the maximum amount of masking - being both the widest spectrum and deepest power restriction.

Beyond CAL values of 20dB, the power masks start to shrink again (in this case, the affected spectrum narrows, but the power restriction stays deep).

Looking on MyDslWebStats, you can see examples of these lines.
Here are a couple that have almost no masking; the first looks to be a cabinet close to the exchange, while the second looks to be far away, but both are on 80/20 speeds:
http://postimg.org/image/9fg2c8pk9/
http://postimg.org/image/rlwve8mal/

The conclusion is that power masking *can* have an effect on long lines (as well as short lines), but that it isn't the same on all lines. For some, there will be little effect at all.

Unfortunately, we have little idea how many lines are on near or far cabinets vs those on middle-distance cabinets, so it is hard to work out the total possible impact.

Statistically, you'd expect there to be few cabinets close-in to the exchange, a medium number in the middle (with the worst impact) and more cabinets at further distances.

I'd guess that there's a fairly even spread between the "no gain" locations, and the "full" gain" ones.

2) Would it be fair to say that many people would gain 20Mbps by getting rid of the masks?

2a) How much speed could be gained in the best possible case - a line with maximal masking, but where the D-side is short enough for full 80/20 speeds?

Here is an example of a line with around the maximum masking possible:
http://postimg.org/image/jmyy2o001/

The "missing bits" caused by the power mask can be roughly calculated from three shapes:
- A triangle, tones 104-440, bits 13-3, plus
- A rectange, tones 440-512, bits 12-3, plus
- A triangle, tones 512- 531, bits 11-3

I make that a total of approx. 2,500 missing bits.

Meanwhile, the bits that *are* present (using a similar process) add up to 21,000 bits. So we can gain 2,500 bits to the 21,000 that are already in use: about a 12% gain.

I calculate the *best case* gain available to therefore be 10Mbps. This probably applies to lines up to 400m from their cabinet.

2b) What about lines that are far from the cabinet?

Here is an example of a line with around the maximum masking possible, but this time with speeds of 30/8 (around 900m-1km):
http://postimg.org/image/mw1xmag19/

In this case, the missing bits comes out somewhat lower, as the longer line isn't capable of reaching a full 15 bits ... I calculate around 1500 bits are missing, which would be around 6Mbps.

2c) And longer lines?

Much harder to quantify, as examples are harder to come by on MDWS.

This line runs at 17/1, and has a medium level of masking:
http://postimg.org/image/7qjb5gqn3/

In this case, the missing bits comes out to be around 800, so about 3Mbps.

Conclusion
a) Removal of power masks could improve some lines, but certainly not all.
b) For fast lines, around 80Mbps, the best improvement is around 10Mbps
c) For slow lines, around 30Mbps, the best improvement is around 6Mbps.
d) On average, though, the gain will be around 3Mbps (slow lines) to 5Mbps (fast lines).

When the gains are reduced to realistic figures like this, is it still worth persuing?

Considering the pressure politicans are putting on BT right now to get more of the country up to the 10mbit/sec this seems like a good course.

To do so requires stopping all exchange-based services, and deploying enough cabinets to cope with the extra demand ... which most locations do not already have.

The kind of gains mentioned - 5Mbps for short lines, 3mbps for longer lines - might be better achieved by the deployment of vectoring to cabinets, without bothering to turn off exchange-based services. Vectoring can be deployed tactically - only on cabinets that have lines where a USC benefit can be seen.

Would it be cheaper to deploy vectoring to existing cabinets - and just those cabinets which could gain - than deploying whole new ADSL-replacement cabinets on every exchange?

Clearly - they are racing to invest hundreds of millions to go to FTTH, or are they hoping someone else will if the conditions are right and they can reap the benefits?

The odd few hundred millions pounds aren't going to make much of a dent in the infrastructure at the national level.

In reply to a post by Seansmit17:

Personally I think that ADSL should be phased out.

No thanks, I'm quite happy with not having to rely on my connections all being powered by the local cab and all vulnerable in the same way. It's nice having lines that carry on working when the cab loses power or gets mowed down.There are distinct benefits to a little mini diversity in the real world.

Perhaps some further deployment of FTTP (or in some cases actually finishing the half rollout BT have abandoned - such as in Exeter - there is stackloads of half rolled out FTTP service!)

Edited by therioman (Mon 07-Dec-15 19:37:56)

In reply to a post by TheEulerID:

The odd few hundred millions pounds aren't going to make much of a dent in the infrastructure at the national level.

It'd make more of a dent than either what Sky have stumped up so far or the precisely zero that, for all the bluster about how it'd lead to a new age of outside investment, they've committed to if Openreach are separated from BT Group.

In reply to a post by craski:

ADSL might well be cheaper now but there will be a tipping point in future where price of ADSL and VDSL based products are similar as they try to "persuade" customers to "migrate" so they can decommission ADSL hardware. If the operators margins are as tight as they would have you believe they wont continue to support the two systems forever.

But that will not be for a few years and also some people are happy with the speed they get from ADSL, certainly if they live near the exchange.

I know a few people on ADSL and have been having letters and emails from their supplier, mainly BT to change to Fibre, but they won't as they don't require it.

In reply to a post by TheEulerID:

Most people have either VDSL or ADSL, not both and for many rural cabinets ADSL will really not be an option. Aluminium is an irrelevance. It will affect both VDSL and ADSL anyway.

But aluminium's affect on VDSL is far more significant than ADSL. My connection syncs at 47 Mbps on a 300m line due to a 120m section of aluminium and deteriorating Cu/Al joints on either end of it.