Ungrounded coil/toasted filter

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "Ed by way of Terry Fritz <twftesla-at-qwest-dot-net>" <edtrind-at-hotmail-dot-com>

Hi, 
I was experimenting with different RF ground routes with my 12/60 NST powered
coil, and inadvertently tested a setup that involved none.
I have an 8' cu ground rod down 7 feet before hitting bedrock, .. which seemed
to work well, and recently moved my coil and can only sink the new rod 2'.
(this is in agreement with the poured depth of the house foundation btw).
I noticed no noticeable difference in spark output (about 30 inches) so I
decided to try running a longer line to the other one just for the heck of it.
No change there either.
I'm not sure what this means... probably that I could get better output with a
shorter better RF ground path or counterpoise?
 
Anyway, ..  with wanton disregard for the pre-launch checklist that I carefully
wrote up, I fired up my coil without hooking it up to either RF ground, ....
and....  still no change in spark length for 9 seconds... . and then, 
a bright glow out of my controller cabinet followed by a percussive pop that
shook the cabinet pretty well and then a fountain of sparks shooting out from a
small hole right next to one of the terminals of the 30 amp line filter before
blowing the fuse I had in there (I had the spark fountain image burned into my
eyes for a while and could see the exact pattern when I shut them for a few
hours... (photographic memory without burdening the brain)).
 
I think I have a bit of a grasp of what happened, ...   but, rather than give
my cloudy image of the source/path/mechanics of the failure, I was wondering if
I could throw out the details and get a comment or two from those more
knowledgeable than I.
 
details:
The mains ground stops in the cabinet.
The filter that fried is in the cabinet, hooked up in reverse, and grounded to
the mains.
There is another identical line filter (that didn't fry) just before the NST at
the coil, hooked up in the 'normal' direction (after reading Gary Lau's
analysis of filter hookup... .. )  - this filter is grounded to my RF 'Buss'
(can't call it a ground for this case).
The other things also hooked up to the RF buss: NST case, safety gap, strike
rail, bottom of secondary (and, unfortunately, not the RF ground line).
Again, ... no decrease in spark output for 9 seconds until the event, .. and
the failure is in a mains grounded EMI filter hooked up in reverse at the cab.
and it manifested itself as a small hole on the coil side of the filter
(labeled 'line' side) that spewed forth sparks, apparently as that line arced
to the case and ate a hole in it.
(the damaged filter is on the mains side of the variac for the record and I
have a 130V metal oxide varistor that didn't kick in across the terminals on
the same side that failed.
 
I get the feeling I should consider my house wiring and all the stuff connected
to it lucky.
The filter is not shorted with no load now, . but does the same thing, sparks
and all, when a bit of current is drawn through it.
 
Any chance anyone can go for a quick 'circuit analysis' approach of what
happened for me?
Thanks in advance, sorry for the loquacious post. 
 
Ed. (waiting for more cut-off wheels for my die grinder to finish opening the
deceased filter).