Re: Preventing Random X-10 Code interference...

["Robert Green" <ROBERT_GREEN1963@xxxxxxxxx>]

Dave Houston" <nobody@xxxxxxxxxxxx> wrote in message

<stuff snipped>

>> As soon as I figure out how to use the noise analysis mode
>> of the meter, I can evaluate both filters for noise
>> rejection from a device like a compact fluorescent.

> The filters are designed for 120kHz so they should be most efficient at
that
> frequency. All should be better at controlling signal suckers since, in
this
> case, they need to block 120kHz from reaching the signal sucker. They may
be
> less efficient at blocking noise since the noise may not be 120kHz but may
> still cause trouble for X-10 receivers.

The Monterey has a noise analysis mode.  It has a two line display of noise
at the .5ms point and then an average (I assume) from the range of .2ms to
.9ms.  It reads in both millivolts and volts, since a press on a nearby
Maxicontroller button will show up on both lines as "4.+" since, as you
know, it has a top-end limit of 4 volts.  I've moved it around the house and
see general "background" levels of 5-10 mV in both ranges.  I have not yet
tested the evil CF - I wanted to get a feel for what might be generating
noise.   While it varies around the house, I suspect without filters I
recently installed I would see a lot of noise.  Something got the Elk up to
one segment before I began applying filters.  I'm so far behind on my
honeydew list I'm due retribution so I have to "time slice" my hobbywork.
:-(  So that and a few other experiments have to wait their turn.

> In this particular case, the monitors may be putting noise on the line
> directly (from their power supplies) or indirectly via RFI.

I thought about asking the OP if he had any X-10 gear near any of the new
equipment.

> Most, if not all, X-10 _brand_ PLC receivers (i.e. switches, lamp and
> appliance modules) count the number of cycles in the "acceptance window"
> between ZC+250µS and ZC+900µS.

Ah ha!  THAT'S what the .2mS to .9mS is looking at.  I assume that if there
a voltage above a certain range in that window, it means the X-10 signal is
being "stepped on."  Is that correct?  What are they looking at the .5mV
range?  It that where the 120kHz X-10 burst is optimally supposed to be
found?

> Noise that is greater than ~80kHz can cause a receiver to think it's
> seeing X-10 carrier while noise lower than ~80kHz
> might mask the carrier. So far, nobody has determined the
>  upper/lower frequency/amplitude limits although
> Dan Lanciani has documented a case where a weak +200kHz
> was jamming X-10 signals.

The documentation says they are looking for noise greater than 110kHz.  Is
it reasonable to conclude that since their lower range is 10kHz lower than
the target value that their upper range might be 10kHz over, namely 130kHz?
Or would it be just as likely that they are looking well above 120kHz?

> The problem gets more complicated now that other manufacturers are making
> devices that use the X-10 PLC protocol but may or may not use cycle
counting
> to determine carrier presence/absence. There are schematics available for
> many X-10 made devices but not for those made by SmartHome, Leviton, ACT
and
> others.

Those obscurationist bastards!  :-)

> Some X-10 made devices are vulnerable to spikes and brownouts which can
> result in unwanted on/off events but will not cause valid X-10 PLC signals
> that will be sensed/recorded by PLC interfaces.

Interestingly enough, and I am not sure why this happens, but when I plug
and unplug certain equipment, particularly things identified as signal
suckers, the Monterey often indicates BSC.  There's clearly something the
Monterey "sees" as these items are plugged and unplugged.  I have not tried
this same experiment with the Elk, but as soon as I remember where I left
it, I'm going to try that.

> BTW, most X-10 made PLC interfaces like the TW523, CM11A, CM15A, etc. do
not
> use cycle counting but sample the demodulated data envelope. They might
not
> react exactly the same as those that count cycles.
>
> When there are valid X-10 PLC signals on the powerline, the source must be
> something that's capable of sending X-10 PLC signals - they are not going
to
> be created out of whole cloth by a noise source. A likely source is the
> SmartHome two-way switches and modules but your guess is as good as mine
as
> to the exact mechanism whereby "noise" triggers them to transmit.

Every human endeavor has its "UFO"s - but I think a lot of phantom X-10
activity is related to short powerline blips.  It's a guess as to which
stock X-10 modules I have that are going to snap on after a few flicks from
a thunderstorm, but it's a guarantee that some will.  I'd guess it's
probably "current sensing" at work.

> Receiver sensitivity is another factor. Older X-10 devices needed ~100mVpp
> but newer devices and those made by others may respond to much lower
signal
> (or noise) levels.

I'll finally be able to measure that threshold when I find branches that
read about .05v of X-10 signal.  I've got some AGC equipment that's
supposedly more sensitive than stock X-10 and now I can see for myself!

I've seen some other interesting readings, BTW.  Sometimes the bathroom
light (on a Hawkeye) command reads in lower case, which the documentation
says "one good start code and one good message block (of the two identical
ones) has been received ."  Does that mean the code was mangled but not
badly enough to prohibit firing?  Would this explain why people report a
slight delay in when the bathroom light fires?

Thanks again for the detailed explanation!

--
Bobby G.