Will FTTC result in a crosstalk crisis?

The rollout of FTTC across the country is clearly good news. The optical fibre itself will bring huge advantages: increased speeds, very little crosstalk, immunity from electrical noise, etc. But given that some copper is still used between the subscriber's house and the exchange, a thought occurred to me the other day - Could this ultimately result in a crisis of crosstalk on the copper?

Think about it. The "final 100m" in copper is, in practise, going to be more like 500m - 800m. Maybe more, in some cases. So, what with FTTC download speeds being at least five times what they are under ADSL Max, surely crosstalk is going to markedly increase on that 500m stretch of copper? And the more people who switch to FTTC, the greater this problem will become. Okay, crosstalk is a function of length but even taking into account the much shortened copper lengths with FTTC, surely the crosstalk figures are still going to worsen?

Am I raising a non-problem here? Maybe the researchers at BT have done the studies and found this to be a non-issue. But I wonder. 40M bps signals zooming along (typically) 500m stretches of bundled wire-pairs aren't going to be without their problems, are they?

I forget the details but there are several ways to configure VDSL2. IIRC Openreach have opted for "8c", which I think is one of the more conservative ones. They also of course for now have the 40Mbps limit, which so far as I know is a semi-arbitrary cap, presumably to allow the real-life data on this possibility to be gathered at a speed where the cross-talk won't be as bad as at higher ones.

The only copper involved is between you and the PCP (phone) cabinet and between that and the fibre cabinet.

Also on 8c (if I'm correct about that) the actual speed falls off dramatically after 500m or so. I think your 500-800m "typical" is way too high, though I stand to be corrected there by anyone with relevant data.

Can you explain why Openreach has selected 8c? And what exactly does 8c specify? Does it define guideline limits for crosstalk in respect of distance and operating speed?

I don't think my 500m - 800m will be too extreme. One or two contributors in these forums have already mentioned their own path lengths of around 700m. Although I myself am not on FTTC yet, I know the very cabinet to which I'd be connected and I know the route of the copper cable between my house and that cabinet, and I paced it out and estimated around 750m. I've also been singing the praises of FTTC to a photographer friend of mine who is online, situated just a block from here, and she would use that same cabinet. In her case, the copper length would work out to around 450m.

Of course, by the very nature of distribution, there will equally be a good many subscribers who happen to be situated virtually at the cabinet, but I think we should put aside this notion that the copper stretch is the "last 100m". For large numbers of people it simply won't be.

In reply to a post by meditator:

For large numbers of people it simply won't be.

I'm in a small (6) block of flats, which is situated next to another block of 8. All the BT phonelines for both blocks, and the small houses in the street go to the same cabinet (all same postcode).

We're all about 500m from the cabinet; or 1010m from the exchange. I currently get about 16meg ADSL2+ sync.

So FTTC / VDSL2+ might help - but if they fitted a new cabinet in our street that would really help....

many of the technical papers refer to 'disruptor's' and clearly the more traffic the more interference.
I think the biggest problems with crosstalk are with line bonding, (two (or more)) lines taking the (usually) same physical path, and having traffic at <exactly> the same time.
8c compromises power levels, to lower crosstalk.

You've clearly put a great deal of thought into this, but only time will tell.

You might be worrying more than you need too though

As you correctly stated, crosstalk is a function of lengh, but it's also a function of proximity too. Adjacent pairs within a single cable will affect one another with crosstalk because the insulation in thin they are be tightly bundled together, i.e. close proximity. However, adjacent cables within the same duct affect each other less due to their relatively low proxinity (mm and cm as aposed to fractions of mm).

It is unlikely that a pairs feeding a dp a 100m from the cab would be in the same cable as pairs feeding a dp 800m away. They would more normally be on their own 20 pair (possibly 10, 30, 50 pair etc.) cables.

Even with the worse case scenario, where bigger cables are dropped off at dps and pairs carry on to further dps, the pair count, and therefore crosstalk, reduce as pairs are dropped off at each dp. A 50 pair (cable) may only run 200m or so before being split into two 20's and a 10, which then take different routes, losing the parallel influence on one another.

And when all this doesn't work you've got the adaptive power levels already mentioned.

Have a search on google for ANFP.

Add to that - differential twists in the pairs which helps to reduce cross talk and that the induced noise will be much attenuated when it reaches the modem.

So it's only induced noise from BT Vision and other powerline adaptors that people on VDSL need to worry about then?

This Openreach PDF, dating back a while, has a *great* primer on VDSL2 - including aspects that answer almost all of your questions.

In particular, it mentions band plan 997 for symmetric services, up to 12MHz. The basic variants of VDSL2 start by using 8MHz; band plan 997 dictates how the frequencies are split between upstream & downstream.

The PDF then mentions the fact that BT are using a truncated band plan 997 - which cuts the tops frequency to 7.05MHz - because this is the maximum frequency allowed by the ANFP in Openreach's cabinets (defined by Ofcom). The frequency lost would have been used by upstream data, so doesn't affect the overall downstream rate.

It then mentions the fact that the VDSL2 kit in the cabinets must employ extra power-control. This is factored-in, based on distance from cabinet to exchange, and is done to ensure that where the VDSL2 frequencies overlap the ADLS2+ frequencies, they have the same apparent strength to subscribers further out on that cabinet. This stops VDSL2 from swamping the ADSL2+ subscribers.

The PDF then mentions a little about the profiles - which really define the frequency use, and the power levels. Some profiles are intended for use from exchanges (so are higher), while some are intended for use from cabinets (and are lower). 8c is intended to be used from cabinets, so is lower power, but also brings in BT's frequency cut-off of 7.05MHz.

Finally, crosstalk... Yes, it will happen, but profile 8c, and the extra power-control feature, are meant to be measures to reduce the impact.

You can find a number of graphs showing how fast FTTC can be - with some giving estimates of 100Mbps. However, a graph that shows the effect of distance on speed in the presence of crosstalk shows the extreme speeds are only really available within 100M. There is one such graph in this document by Ericsson (page 8), which shows more realistic speeds for profile 8d (8c will probably be similar) with crosstalk from 20 lines.

My feeling is that BT's estimates are making predictions of speed with crosstalk present (and match that Ericsson graph reasonably well), which is why the first people are getting much higher speeds... for now.

The graph suggests a path limit of 1500M - and limited by the upstream, not the downstream!