I have, on occasion, used a TVS from the snubber common lead to
> ground (rather than snubber common to Vdd) - but that was to achieve
> a faster release time on the relays.
>

We used to use the diode across coil method, but I've switched all our
designs over to the zener across transistor method.
Most of our products drive larger motors (100-300W) and they are heavily
inductive, so relay arcing is significant if the relays don't release
quickly.  Even for DC resistive loads, I don't see any reason to torture
the contacts. :)



> FWIW - most of our stuff uses 5V Vdd with an unregulated supply
> driving the relays - the unregulated supply comes from either 12Vac
> or 24Vac (nominal +16 or +32 Vdc on the relay coils).


Pretty similar to what we do.


> The excess coil voltage is dropped in the (series) indicator LED and
> either a
> series dropping resistor *or* by applying PWM to the relay coils to
> reduce the average voltage to the proper value.  PWM is nominally 1ms
> OFF and 4.1ms ON (~80%) - works very well.
>

So are you using relays rated for significantly lower voltage than your
relay supply voltage?



I have seen some pretty interesting failures.  One was a mosfet that was
failing with only 60V (peak) applied to a 70V part.  It turned out that the
reason that I only saw 60 was that the part was driven into avalanche so
fast that a 300MHz scope might have been able to see it.  Testing with a
1000V part in that circuit showed the real flyback energy that was killing
the parts.

Current probes are wonderful tools, in that you can see what is really
going on, as opposed to what ought to be going on.
--=20
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.