> many great answers, especially from Vern Jones

I think that the litmus test would be to destructively test your CAN
station (or other station) in the way the UL would do it. F.ex. you have
an intelligent effector/sensor group connected to mains, a heater, and a
bunch of sensors in a NEMA box. And the bus. Now comes the UL man and puts
2500V transients into the mains while snipping off diverse parts from the
input transient filter and finally puts mains into the bus input. If at
the end of this there is a path between the mains (the real one) and the
bus terminals the bus will be toast, together with whatever is at its
other end. Agree ?

Aside: It would be interesting to set up a set of tests (and document them
on a webpage) that would allow a hobbyist to check out his designs before
leaving the newest and untestedest gizmo operating in the living room
while going off to work. Like surge injection, fireproffing, proper fusing
of ALL inputs and outputs (even if by flameproof resistors), checking on
heat generated over time in final enclosure, such things. What do you
think ?

The question is, what next imho. I.e. if the garage
antifreeze/ventilation/door/alarm/light controller is destroyed because a
mojo decided to sit on the wrong part of the pcb will it take out the
whole house system ? Having a single dc coupled bus winding its way
everywhere is dangerous imho, if it is dc coupled. Opto is a way but it is
expensive in parts and labor and the floating part of the bus will blow up
taking a lot of parts with it (assuming bus wiring, as opposed to star or
tree).

As Vern Jones has pointed out the power/decoupling systems are very
important, but also the fact that his wiring is in fact 24V on/off signals
mostly (excepting the sensors I presume). This is an industrial standard
(see PLCs, many work with this standard of signalling). 20mA current loop
and RS232 current loop are in the same family of ultra-robust industrial
well-proven systems.

So if someone would propose, say, using fiber to wire the house I would be
all for it (using MOX?? modular transceivers and standard 3.5mm plugs).
Hey, you can always use the fiber for spdif and video later. Ditto carrier
current using CAT5 cabling (there is no law against sending X10 and like
signals on a separate wire). I think you can send CAN over fiber if you
want to.  RS232 certainly. I did it (multidrop using spliced plastic
fiber). It is very nice.

I have experience with carrier current remote measuring and control using
CMOS40xx series chips for implementation, with signals on the mains or on
a separate pair (partly outdoor phone line, partly buried, etc). Needless
to say this was not chosen out of fancy. It is interesting to see the neon
bulbs mounted as transient suppressors at the coupling transformers
flicker when certain transients are on the mains and have your circuit
continue to work ... (this with the insulated 'phone' pair, which was
grounded through resistors at both ends !).

I have also seen a lot of transient and lightning damage (and ground
faults etc) and I know what I am looking for in a bus system I would
trust over the long term. Maybe CAN will cut it too, I am sure that the
people who devised it knew what they were doing.

I know that there are lots of chips out there. It's just that I have the
habit of walking into the supplier's store and chatting up the owner or
the salesperson about this and that part, and I am pretty good at seeing
when his eyes glaze over although I am not very good at poker ;-) That
gives me clues about parts which not to order ... or not just right now.

Peter

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics