Looking at copper, still the most common broadband carrier but rapidly becoming obsolete.
Copper wire for telephone and ADSL comes as a twisted pair, the wires twisted around each other. The reason for that is that if the wire passes some strong electric field it would shift the voltage of the wires but since they are twisted this interference affects the voltage in both lines the same and it is the difference in voltage between the lines carries the signal.
If we want to increase the data carrying capability of the copper—increase its bandwidth—we can do that by increasing the frequency of the signal, or to put it another way use a shorter wavelength. That will work, within limits but eventually as the frequency is raised higher the signal interferes with signals in adjacent pairs and “cross talk” happens. Early in my working life this happened even with voice—I would ring a number—and find myself connected to a phone call between two other, unknown people. As we raise data signal frequencies this cross talk results in muddled data being received which meant the receiver had to tell the sender (electronic receiver/sender I mean, not people) to re-send the data which obviously slows the transmission of data.
So ADSL has a fairly limited spectrum (range of frequencies) and so a limited range of speeds.
Copper also has a fairly high impedance, needs power sources etc to transmit the signal over distance. The higher the speed (56K modem/ADSL/ADSL2/VDSL/VDSL2/GFast) the higher the frequencies the higher the need for better line and the shorter the distance the signal will travel before it degrades (over a Km VDSL2 is barely better than ADSL—there are charts of this in the main NBN thread.
To get the signal through with least degradation over distance the lines and joins need to be impeccable—one bad joint will seriously degrade the signal. The copper should be low impedance and for Australia there is some bad news: Whereas most countries that have run out FTTN (i.e. VDSLx) have wires of .6mm or more, Telstra wires are a mere .3 or .42mm. Nobody (bar Telstra management and the current NBN Co board) will call Telstra wires impeccable either.
To minimise crosstalk
vectoring was developed:
Quote:In a typical copper loop, Peeters explains, the signal is transmitted over a copper pair to minimize the electromagnetic interference. ”But still, part of the signal of one customer’s wire often leaks into another customer’s, and that’s when you get what’s known as cross talk,” he says. Cross talk limits bandwidth by corrupting the signals that are transmitting information.
Cross talk is theoretically easy to completely eliminate, as long as you can estimate how much one line is leaking into another. Then you can cancel out that interference with calculations performed by hardware installed in the cabinet, Peeters says. Cross talk estimation uses what’s called an error vector, which is sent from a user’s home equipment back to the cabinet. The receiving equipment there knows what the signal should look like, and the difference is a good measure of the amount of cross talk between a wire and all the wires around it, Peeters says.
In the new VDSL2 scheme, the cross talk for every line in a cabinet is estimated using these error vectors and then processed to minimize the interference for them all.
”Computation for VDSL2 vectoring can be demanding,” Peeters says, because there’s a large matrix of cross talk measurements to consider. ”For 48 lines, it requires as much processing power as a PlayStation 3,” he says. But due to the decreasing cost of silicon and the high cost of digging for fiber, he estimates that the cost of installing VDSL2 with vectoring is still no more than one-third the cost of running fiber all the way to the home.
http://spectrum.ieee.org/telecom/internet/copper-at-the-speed-of-fiberCan we use vectoring with Telstra copper? With its relatively high impedance, poor quality wiring and joints I am not sure vectoring would actually be introduced here. (note the article quoted puts vectoring at 1/3 the cost of FTTH—beginning to be not worth doing here!
Sortius is not hopeful of vectoring working here:
Quote:First, I’ll  look at AT&T’s vectoring & what their plan is. AT&T have slated US$3b to upgrade 1 million customers to either pair bonded/vectored services or to FTTH/FTTB. The upgrade is minimal at best, servicing a paltry 1.7% of their customer base, yet the per service cost is around US$3000, not exactly cheap for something they’ve already spent billions on deploying, & will have to spend billions upgrading soon. There’s also the whole bonding with vectoring, something that just wouldn’t be possible with Australia’s copper network: there just aren’t enough high speed data capable pairs for people to run bonded services.
http://www.sortius-is-a-geek.com/why-youll-never-get-vectoring/(Sortius spent a fair bit of his working life working with the actual copper—sure check what he says but will find he is spot on normally.)
GFast won’t happen here, needs like 8 bonded good pairs per premises, a lot here don’t have ONE good pair—the dreaded pair gain!
Forum
Pages: 1 
