Weird safety gap behaviour

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

Original poster: "Gerry Reynolds" <gerryreynolds-at-earthlink-dot-net> 

This is a repost, the first one never came out of the pipe :-((

Hi All,

This is a follow on to the weird SRSG behavior and I thank everyone that 
responded to that post.

I'm getting a safety gap behavior that I don't understand.  I got the SRSG 
removed from the system so that added complexity is gone.  I have two 15/30 
NST's (magnetek) in parallel connected to the terry filter that in turn is 
connected to a 3 terminal safety gap (center terminal grounded).  One side 
of the safety gap is connected to Cp (2.5* Cres).   The other side of the 
safety gap is connected to the primary.  Cp is in series with the primary 
(standard TC topology for a two bushing power source).  The safety gap 
consist of brass heat sinks that are threaded for 3/8" carrage bolts.  The 
carrage bolts are adjusted to just not fire when the unloaded NST is driven 
with a variac at 140Vac.

I'm now measuring the total voltage across the hot terminals of the safety 
gap differentially (BTW getting the same answer as when measured single 
ended between a hot terminal and ground and then doubling the 
result).  Following are the measurements:

I slowly raise the variac voltage from 0 to 90V.  The peak differential 
voltage across the two hot safety gap terminals increases to about 
16KV.  With no further increase of the variac voltage, the 16KV starts to 
run away (exponentially it seems) and snaps to 30KV.  The safeties are now 
firing and healthy arcs are coming from the secondary top load.  The 
safetys dont fire until after the runaway so don't seem to cause the 
runaway.  It takes about one second to snap to 30KV.   In 30KV mode, the 
variac output voltage is still 90V so the variac doesn't seem to be part of 
the runaway.

The next interesting thing is that I start to lower the variac 
voltage.  The safety gap voltage stays locked on 30KV until the variac 
voltage is reduced to 70V.  At this point, the safety gaps stop firing and 
the voltage returns to normal.

The safety gap spacings measured 0.21 and 0.26 inches.

Next, I set both safeties to about 0.20 inches.  Results were the same.

Next, I set both safeties to about 0.17 inches.  The runaway again starts 
at 16KV but the peak voltage after runaway is now about 26KV

If the safeties fired first and didn't regulate that well, then I could 
understand that a transient response would be superimposed on top of the 
steady state response.  But the runaway happens first and then the safeties 
fire.

Any ideas what is causing the runaway??  (maybe an engineering explaination)

Also, could someone explain reverse voltage mode from an engineering point 
of view??

Many thanks for any responses,

Gerry R.