Not the old-fashioned fluorescent fixtures, but some of the new ones with electronic ballasts. This is due to some leakage current slowly charging up the ballast until it reaches the point where it can flash on for a fraction of a second. This fools the load sensing circuit into thinking that the user has closed the switch.
The fixes involve opening the modules, then cutting specific leads or desoldering. It worked for me in one case. The Smarthome Appliancelinc allows one to programmatically turn off the load sensing, and has a quieter relay. I have one that works reliably (more reliably than X10 pro modules) with a fluorescent load having an electronic ballast. I couldn't say that it is immune to that problem though.
Yes, a simple latching relay. Noisy but effective. To save you the trouble of taking one apart to check it out:
With relatively small fixtures (15/20 watts) with a conventional starter and balast, I have had situations where I had a hard time turning them off and KEEPING them off.
That sounds about right except that it happens with CONVENTION flash tube "starters" also.
One would like to think that "they" would market a "dumb" module that would completely ignore the load and only respond to X10 commands. Or, perhaps, have an extra switch that does that. But NOOOOOO.
Perhaps YOU should check it out. One of the pictures jogged my memory. It's a STEP relay: each time the coil is pulsed it opens if it was closed and closes if it was open. The electronics "senses" the voltage drop across the contacts to get feedback so that the electronics "knows" whether to pulse the coil if it decodes an "on" of "off" command. If the module "knows" that the contact is closed and receives an "on" command, it does NOTHING.
Trouble is, of course, is that certain "light" loads may make the closure detection circuitry not work reliably: the electronics may "think" that an "open" contact is actually closed and give it an extra pulse.
A "dumb" module would actually require an additional contact on the relay to determine it's position regardless of load. Since most folks don't really care about such stuff, a "dumb" module might well cost 4 to 5 times the cost of the standard appliance module.
(A latching relay has TWO coils: one operates the relay and the other "unlatches" the relay. These are MUCH more complicated that a simple sequencing relay. These "sequencing relays" used to be commonly used in toy electric trains and let the operator reverse the direction of AC powered toy trains. DC operated trains reversed by just changing the polarity.)
No kidding ? I don't have many of those around and never tried automating one so I missed the experience. My image was the magnetic rapid start or instant start ballast.
I remember those two coil relays but haven't seen one in years. For amusement I checked the IEEE dictionary to see what they call the appliance module relays. The closest that I find is a "latch-in relay", defined as "a relay that maintains its contacts in the last position assumed without need of maintaining coil energization".
Your point about the module needing to sense the contact state was helpful. I never had thought about their need to do that and the extra ways it can misfire as a result. One has to give the original X10 people credit for figuring out how to make these things so inexpensive. But it would be nice if someone like PCS made a higher-end appliance module that we could buy when something more reliable is needed.
=>Does the appliance module use a mechanical relay to close the circuit?
Guys, thanks much for all the good info! That Geocities site is full of great stuff. Will have to lookover carefully.
So far, all modules are working fine, even after a lightning hit to the pole outside. The minicontroller showed a fixed colon and two zeroes afterwards. Disconnected the keep-alive battery, reconnected. Voila! Strangely believe it!
The appliance module was to control a small water-fountain pump at the end of the pool.