Dereck: X transients arriving via the service is a differential mode rather than common mode, meaning the voltage is L-L and L-N. Earth impedance is of no value or importance, it is merly a reference point.
So you can then look at it as the equipment sees it: a connection to 3 points that are labelled,"L", "N", & "G". We know that the other ends of "N" & "G" were at the same potential at a main panel and that the equipment ends may or may not be. But if we look into the equipment for any components between N & G, there aren't usually many susceptable ones there. You might find a few HF bypass or decoupling caps or maybe a spark gap or two. But in the over 25 years that I have repaired a broad range of equipment, I have rarely had to replace components L-G or N-G. My forgetter is getting as good as my rememberer so it's possible I'm forgetting a few HV discs I replaced in the 80's and just remember the half dozen or so input bridges and switching mosfets I've replaced in the past year.
If I'm given a hypothetical choice to use 2 components to protect my gizwalter, I'll pick a bi-directional transorb and a fast acting fuse. The closer the clamp voltage above the highest L-N expected, the better. So once in a blue moon, your mains drift up and you pop your fuse. So what! My fuse is not going to be much higher than my max I and definately picked to blow before the surge and continuous ratings of the transorb are exceeded. If I'm allowed a 3rd component, I would pick series inductance in the line after the transorb before I would clamp L-G or N-G. I think any good electron lawyer would claim that surges delayed are surges denied. Scott posted several drawings of surge arrestors that were just pi-type LC input filters with MOVs in parallel with the caps.
I'm curious if other folks out there who have to fix things after they blow have had similar experiences with the components that they have had to replace.
Joe
