Re: X10 replacement?

[Joerg <notthisjoergsch@xxxxxxxxxxxxxxxxxxxxx>]

Hello Marc,

>
>>But even then it is a rather brazen assumption that
>>the loss along a power line would be less than 40dB or that the noise
>>won't swamp everything, unless there is a proper passband filter. One
>>that is only a few kHz wide and not tens of kHz.
>
> Who is making this assumption ?
>

The X10 designers, I would think. Between 50mV and 100mV most modules
typically quit receiving. Even if the transmitter is able to inject
120kHz at 5V against the low line impedance that's still less than 40dB
of headroom.


> At the transmitter end, and at the receiver end, there are low-current
> improvements that are practical in part because they involve physically small
> devices with concomitantly manageable costs. Here, the filters only need to
> pass the signal. not, generally, high current. And if they do, the current
> may be reduced (eg 5 amps as in a wall-wart filter) compared to the in-wall
> AC distribution lines (15, 20, 30 amps).
>
> As soon as one moves from the device level to the realm of permanently
> installed transmission medium which in US and Canada, is 110VAC is subject to
> electrical codes that (for permanently installed components of the
> transmission system) require a *minimum* 15 amp current-handling capability)
> the problem becomes quantitatively much different.
>
> How does one make a (c)UL-listed, low cost, 15+ amp, 120khz, multi-pole notch
> or bandpass filter ? If one can't, then the notion that robust distributed
> X-10 can be low cost is myth.
>

Just a few rough ideas:

There are several methods. A simple one would be to use a regular LC
circuit like now but couple it way more loosely so it runs at a much
higher Q. This also means the receiver input must be of high impedance.
The noise figure isn't important in this application. No problem to get
that bandwidth down to a few kHz. This will already be a whole lot
better than today's typical 120kHz "filter". Problem is you can't drive
the Q too high. While it isn't a really big deal to build and align a
120kHz circuit with a bandwidth of less than 1kHz there are,
unfortunately, legacy transmitting devices of rather sloppy build that
don't transmit on exactly 120kHz. The installed base just can't be
ignored, else consumers would bring it all back and say that it does not
work. The store clerks cannot be expected to understand the underlying
problems here, let alone explain them to the customer.

Another method is down-conversion to baseband. Either by mixing with
120kHz and full I/Q detection or by mixing with, say, 123kHz or so and
demodulating only the lower sideband. Now a simple zero-crossing gated
FFT would constantly run and the micro controller has to watch all it's
bins. Only if it finds a bin where the signal clearly "sticks out" would
it assume that a valid X10 transmission is occurring and latch onto that
frequency. Both methods can be done rather cheaply. The challenge is
that it must be realized with "jelly-bean" parts and a micro controller
that costs well under $1 in quantities. DSPs could do all this directly
at 120kHz but they are way out of price range here, at least for the
time being.

There are a few more esoteric methods such as a sweeping mixer. The uC
watches and the sweep function slams on the brakes the instant a signal
is detected and regulates into the maximum. One would have to take a
close look at the protocol to see if there is enough expendable prelude
for that. But whatever is done, it has to be cheap and a lot better than
what we have today.

The receive filter does not need to carry the full 15A or 20A here. Of
course, much of the above requires a thorough understanding and
application of "lean" filter algorithms. A cheap micro isn't a rocket.
It's more like a moped in terms of math horsepower so one would have to
employ techniques such as the wave digital filter. Seems like much of
that is a lost art at least with younger EE grads.

As to UL, that's a matter of proper sales forecasting. I do this kind of
mixed-signal processing all the time for a living and a full UL cert
typically runs into several $10k. More in my cases because it's medical
so it also needs to be tested to IEC601-1. All this needs to be
amortized over several years of sales, just like the costs for a redesign.

It doesn't have to retail at $12.99 a pop. $19.99 is usually the magic
threshold beyond which consumers become a bit hesitant. As long as those
extra $7 buy a substantial reliability improvement I believe this could fly.


> Lightolier introduced the concept and sold the Compose PLC (X-10) system
> based on large, expensive hardware to filter and segment the home AC
> distribution.
>
> http://www.lightolier.com/index.jsp?FLASH=1&FL=releases/PRComposePLC.htm&A=208&B=233
>
> It was (see this thread's subject) an evolutionary "X10 replacement".
>
> My limited perspective is that it seems to have had a short run in the
> marketplace and then seems to have mostly declined. I don't have any
> experience with it. Maybe someone in c.h.a else does and can pipe up.
>

That link didn't work but the Lightolier site is still up. For controls
it points to this site:
http://www.lolcontrols.com/

>
>>Many times X10 works ok, other times it doesn't. The worst case is when it
>>sometimes does and then quits. That drives ordinary customers nuts,
>>especially when they do  not understand why.
>
> It also drives  folks who _do_ understand to rip it up and install something
> "more reliable" (whatever that might mean ;-) Knowing that and *why*
> something doesn't work can also be a powerful driver to removing it. Depends
> in part on one's mindset.
>

Sure. But most of the folks who'd buy this are also married ;-)

If I came home with $1000 worth of stuff and tell her that it'll
(hopefully) work better than this X10 I'd be read the riot act...


> Thanks for the cogent discussion ... Marc


I'm enjoying it as well. Here's hoping that some day the manufacturers
of HA listen and be open for such discussions. In my world of medical
electronics at least the smaller companies do.

--
Regards, Joerg

http://www.analogconsultants.com