I realize that this is an old thread, but I thought I would mention something that we discovered many years ago when implementing this sort of setup.
When you use a commutating diode across the coil of an interposing DC relay, it does a good job of protecting the contacts (or transistor) in the output device driving that relay.
But because that commutating diode provides a low-loss current path for the coil being driven, the magnetic field in that coil (solenoid) collapses far more slowly than it normally would when your system tries to switch it off.
The result is that the contacts of the interposing relay open much more slowly than they might otherwise. And that leads to more arcing and damage to THAT relay’s contacts if IT is being used to drive an inductive load. As its contacts slowly open, and they attempt to interrupt the current through ITS inductive load, their slow separation encourages the development of an arc, establishes the conductive plasma, and burns those contacts far worse than if the contacts separated rapidly so that the arc gets quenched faster with the contacts opening far enough to overcome the voltage being generated by the load.
Of course, that load or the interposing relay’s contacts should have their own snubber or protection.
But the thing is: If you use an MOV instead of a diode across the interposing relay’s coil, that coil energy is dissipated far more rapidly, and the coil current interrupted faster, allowing the magnetic field to collapse far more quickly, and letting that relay’s contacts pop open faster so that IT doesn’t suffer as much contact damage if it is called upon to control an inductive load.
This can also be important when driving DC solenoid actuators. If you don’t want them to release in a sluggish way, use an MOV around their coil, and not a diode.
MOVs are also good for use around AC coils, motors, solenoids, etc. Sometimes this is better than using a snubber.