NST current metering surprises

["Lau, Gary" <Gary.Lau-at-compaq-dot-com> (by way of Terry Fritz <twftesla-at-uswest-dot-net>)]

I finally got around to measuring my NST primary and secondary currents,
with the help of a hamfest 20A AC ammeter and a borrowed 250mA AC ammeter.

Briefly, my coil uses a 15KV/60mA NST.  Until recently, I had  been using a
.012uF cap, selected to be mains-resonant with the NST secondary, and I got
51" sparks.  My 1.6K/113W protection circuit resistors ran hotter than I
thought they should.  Now I run with a .022uF cap with even better 58"
sparks, and even hotter resistors, but now I'm tripping my 20 Amp circuit
breaker after about a minute of runtime.  This with 88uF of power factor
correction caps.

Today I metered the NST secondary current.  This is an unmodified NST whose
short-circuit current measures 72mA (-at-140VAC).  With the .022uF cap, it was
pushing a whopping 230mA!  Excluding power due to bypass cap discharging,
that accounts for over 84 Watts, finally enough to explain the hot
resistors.  When I went back to my .012uF cap, the current fell to 130mA.

Experimenting with the power factor correction caps, I couldn't tell if they
had any effect with the .022uF cap as the meter pegged both with and without
it.  Perhaps I should have measured how long it took my breaker to trip,
with and without the PFC.

With the .012 uF reso-cap, primary current was significantly less, dropping
from 14-20 to 11-14 Amps without and with the 88uF PFC.  Secondary current
was unchanged with or without the PFCs at 130mA.

I stated the primary currents as ranges because the reading is extremely
unstable, no doubt due to the chaotic firing of the static gap.  It's
curious that the secondary currents seem so much more stable.

One further interesting observation.  With the .012uF cap, spark output and
primary current increased monotonically with variac setting.  With the
.022uF cap, there was a distinct null in current and somewhat in
performance, at about 60% variac setting, and at this point, the primary
current became rock-steady.  The sound of the arc also became much more
stable.  I'll bet the charging was a non-chaotic 60 or 120Hz (didn't have
the scope out).  

With both the primary and secondary currents so much higher than simulations
predict, I now have to question the validity of that model.

Regards, Gary Lau
Waltham, MA USA