>1. Diode across relay coil (cheapest option) - Time
>honoured method. Must be suitably current rated
>( = coil current), voltage spikes don't exist here,

I think the discussion has determined that the voltage spikes do exist due
to diode turn on time - and most probably because of other factors such as
poor layout of the connections to the diode, which must be as directly
across the coil as possible.

>diode turn off time not an issue either,

agreed

>diode turn on time has been an interesting discussion
>but I am yet to be convinced that a lazy-as 1N4004 etc
>wouldn't be perfectly acceptable here.

Well I suspect that this may depend more on the Vce rating of the
transistor. And while a 1n400x diode may be described as lazy, the junction
capacitance may well have as much to do with limiting the voltage peak at
transistor turn off as the diode eventually conducting.

>3. RC on transistor base - Even just increacing base
>resistor value can slow down turn off time appreciably
>for no extra component cost, slower EMF spike => lower EMI.
>Works with BJT and MOSFET. Add C where necessary.

I suspect that when driving a BJT from a PIC port, the trick would be to
have a resistor BE and tristate the port. The resistor value would need to
be chosen to limit Vcb voltage shoot through from destroying the PIC though.
Same trick may work for a MOSFET.

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist