> ... but MOV's are normally associated with rather high voltages.  I
> don't think you get 5V MOV's.  You could try using a zener diode
> instead.

  The two COMMONEST varieties of MOV floating around (I use a curious
criterion: those most commonly appearing "disposals"; as well as those
most frequently listed in catalogues) are 275V and 7 (seven) Volt at the
"knee" point (a little arbitary).  Note that possibly even MORE common,
but rarely advertised, are those used in telephones, presumably rated
about 50 to 100V, to protect your ears!  Remember that the telephone is
the single most mass-produced electronic item in your country - nearly
everyone has at least one, and most are the same model.

  The 275V model is used to suppress surges on 220(/240/250)V mains, (I
daresay another size used in the «States, but I have yet to see them)
while the 7V model is indeed used on 5V logic supplies.  MOVs are
supposed to conduct faster than Zeners and are far more rugged.

> Optoisolators can also work quite well as far as isolation is
> concerned, but they will affect you voltage measurement.

  The ordinary ones are anything but linear.  You therefore need to use
a linear one, (a speciality of Analog Devices) or various devious tricks
based on feedback through a second coupler in the same pack (which is, as
it happens, the real reason they market twin-packs and quads).  I trust
this is self-explanatory.

  If your analogs have a common ground it is much simpler to use MOVs to
protect the inputs and opto-couple the digital data stream.  If separate,
probably remains simpler to use separate ADCs (in PICs, presumably) and
opto-isolate them.

  My suggestion is to chain the output data through successive PICs
rather than use a "combiner", using only one opto-coupler per PIC and
limiting the number of serial channels each has to process. Think in
terms of token-passing from one to the next.

  Just musing,

  Paul B.