[LINK] BPL

[Robin Whittle]

Hi Stephen,

I promised myself I would write no more on this, but I am shocked by
your gullibility in accepting the marketing dross which passes for
technically correct information about Broadband Over Powerlines AKA
Powerline Communications (PLC).

Your continued optimism for BPL seems to be driven by a desire to
find a new broadband technology.  However, it also seems to be based
on a complete lack of understanding of the technology and most
importantly the electrical behaviour of power wires at high frequencies.

I have researched this stuff in detail in recent years and I can tell
you BPL is a heap of crap.  My two previous messages are here:

  http://mailman.anu.edu.au/pipermail/link/2009-March/082183.html
  http://mailman.anu.edu.au/pipermail/link/2009-March/082199.html

If you don't want to research it or don't have the expertise, just
take my word for it that your optimism is completely and utterly
misplaced.  The trouble with you writing things like "successful
trial" is that it may give other similarly uninformed people false
hope that BPL could be useful.

Please also read Bruce Arnold's excellent round-up:

  http://www.caslon.com.au/powerlinenote3.htm#expectations

  http://www.caslon.com.au/powerlinenote.htm
  http://www.caslon.com.au/powerlinenote1.htm
  http://www.caslon.com.au/powerlinenote2.htm
  http://www.caslon.com.au/powerlinenote3.htm
  http://www.caslon.com.au/powerlinenote4.htm
  http://www.caslon.com.au/powerlinenote5.htm


You wrote, in part:

>> (I guess you could QAM-modulate 50Hz mains frequency to distribute
>>  a data signal to a long range, but I doubt country folks would
>>  appreciate the performance, or what this would do to the
>>  appliances that rely on a clean sinusoidal feed). Paul.
>
> Thanks for your expertise, Paul. So, what about up to say 6 klm
> from the drop-off point, suggested above. Is this possible with
> BPL, do you know?  Even a few klm beyond ADSL range might
> significantly improve matters for many country folk. (haha, eg
> me/us)

Paul's sarcasm went straight over your head because you apparently
lack any insight into how communication technologies such as radio
links, DSL, BPL etc. work.

Paul no-doubt chose 50Hz because that is the one thing power lines
are good at - it is all they are designed for: transmitting 50Hz
power signals with high efficiency over long distances such as
hundreds of metres for 240 volts or kilometres at 22,000 volts.

(In 120 volt countries, it is only practical to send it 50 metres or
so.  Therefore each home or two or three homes has its own
transformer.  In Australia, one large transformer feeds homes up to
150 metres away.)

Firstly, it is impossible to modulate the 50Hz mains signal without
constantly changing the frequency, amplitude and phase of the mains
according to the data.  So your light globes would be constantly
fluctuating in brightness, your electronic equipment would be all
over the place, the induction motor in your fridge would be playing a
kind of stochastic bass drone electronic music etc.  There would need
to be massive 100km electronic drivers for each street - the size of
a small building and costing a million dollars or so.

The real sarcasm is in the data rate.  Let's say you used QAM64
modulation, which is pretty fancy.  You have 8 levels of sine and 8
levels of cosine.  Every symbol period you carry 3 bits on the sine
level and 3 bits on the cosine level.  The symbol rate must be lower
than 50Hz for this to work. Let's make it 25Hz.

So what is the total bit rate for this unidirectional system?

  25 * (3 + 3) = 150 bits per second.

This is 1/10,000 of "broadband" - and still, you need to
time-division multiplex it to split it up between upstream and
downstream capacity.  You also need to allow for protocol overhead
and forward error correction.  So you could probably get 50 to 100
bits per second total, upstream and downstream.


The highest performance modulation technique OFDM (Orthogonal
Frequency Division Multiplexing - typically hundreds of channels of
QAM4, QAM16, QAM64 each on a separate frequency up to several MHz for
BPL) has been well developed for 10 years now. It is used in DSL and
all the radio technologies which have the highest spectral
efficiencies.  This has been applied to BPL and still the system is
close to useless, and can only be made to work with repeaters every
50 or 100 metres.

The thing to remember is that although you imagine signals going one
way or the other, the way you want the information to flow, the wires
are incapable of supporting this.  Injected signal goes all over the
place.  Therefore, repeaters need to use different frequencies.

Every repeater needs to operate on two or four frequencies.

It must receive on A for incoming downstream and then transmit on B.
 Then it must receive on C for incoming upstream and transmit on D.
Alternatively, you can do it simplex, using the same frequencies for
upstream and downstream.  However that presents nasty timing problems
since the timing has to depend on traffic, and there are multiple and
constantly changing sources of traffic.  Also, this reduces the time
available for upstream or downstream, so it doesn't get you any extra
bandwidth.  It also increases latency and makes the entire system
much harder to coordinate.

Whatever you are doing with a frequency at one part of the cable, you
must expect enough of that to travel 1 km or so away - so you can't
re-use that frequency anywhere nearby, due to the transmitted signal
from 1km away acting as noise and clobbering the ability for the
system to work at all.

Yet these frequencies can't be close together, because the
necessarily analogue filtering to separate the high-powered transmit
frequency from the very sensitive receiver can't be made to have
extremely sharp cut-offs.

So if you have a simple BPL link with a repeater, then you have
already radically divided the available frequencies in half,

If you need multiple repeaters, say 6, then you have divided the
available frequencies for any one repeater, and therefore the entire
data carrying capacity of the system, by at least 6 in theory.  But
it becomes untenable in practice, since you can't make analogue
filters with the sharp-cut-offs you need to slice and dice as you
would like.  Nor can you make them stable and tunable with
sufficiently low noise and capability to reject the high transmit
signals of the local transmitter from the sensitive local receiver.
You can't use digital filters - the filtering needs to be analogue to
reduce the local transmit signal bleeding into the receiver's
analogue to digital converter, so you can get a good signal to noise
ratio there for the receive signals you want to detect.

I could go on, but it is tiresome.

I challenge you to find a single person who really understands this
technology who can explain why BPL could ever be practical as a
cost-effective *broadband* service, despite there having been a
decade to create equipment and fine-tune it with no substantial
results yet.

This person would then have to come up with some explanation about
why forces other than technical impossibility have lead to BPL not
being developed properly and adopted widely.

Everywhere you look in the BPL field there are technical barriers due
to the nature of power cables.  You can't change that nature - you
can't change the cables, go installing filters here and there.  They
are built to handle thousands of amps at hundreds or thousands of
volts.  They are gutsy things ready to be hit by lightning etc.  You
can't just decide to break the cable and put a 48dB/octave filter at
1.2MHz somewhere so you can isolate one section of the cable from
another for the purpose of reusing in one street the frequencies you
are pumping into the cable in another.

Hopefully anyone reading your optimistic writings about BPL on Link
will read this and the other messages indicating that in this
particular field, your optimism has no basis in reality.

  - Robin