Improving spark gaps

To: tesla-at-pupman-dot-com
Subject: Improving spark gaps
From: "Jim Lux" <jimlux-at-jpl.nasa.gov> (by way of Terry Fritz <twftesla-at-uswest-dot-net>)
Date: Fri, 14 May 1999 22:10:36 -0600
Approved: twftesla-at-uswest-dot-net
Delivered-To: fixup-tesla-at-pupman-dot-com-at-fixme

A useful article that I ran across recently:

MacGregor, S.J., et.al, "Methods of Improving the Pulse Repetition Frequency
of High Pressure Gas Switches" in an IEEE publication of some kind (sadly,
my computer died, taking my bookmark with it, but I'll try to find it
again). They are at the Univ. of Strathclyde, so you Glaswegan coilers out
there might want to track these folks down.

They're talking PRFs of 100's to 5 kHz, tens of kV, etc. which is right up
the TC alley. High pressure means 1 bar (1 atmosphere) The paper talks about
the variations in breakdown recovery for both pulsed and DC gaps as you
change pressures and gases. All of this is for static (sealed gaps),
although they talk about applicability to flowing gas gaps.

Interesting stuff:
1) They look at the recovery of a gap in Air and in SF6. There is sort of a
plateau effect. After the gap fires, the breakdown voltage starts to come
back up and is at about 50% after 1 mSec, then holds there until about
200-300 mSec, when it comes back to 100%. As you might expect SF6 is a bit
better than air, but not much (5-10%, reading from the graph in the paper).

2) Changing the pressure changes the height of the plateau, but not the
time. At 1.5 bar, the plateau is at 90% of eventual gap strength, but still
takes more than a mSec to get there.

3) They did some experiments with mixes of He and SF6, because they thought
that high thermal conductivity of the He would help, and apparently He/SF6
blends have been used in flowing gaps. It didn't. It's actually worse in the
static gap.

The most important thing they found is that putting  a field on the gap
after discharge helps to "sweep" the ions out (as would fast (i.e.
supersonic) gas flow, like in a Fruengel argon gap). Then, they go through
and talk about how you can stabilize a gap, increasing its recovery rate by
adding corona in a SF6 gap. The corona "inhibits the breakdown in the gap
until the voltage reaches the required value". They have a histogram of gap
breakdown times where the corona stabilized version is tightly packed around
1 mSec and the unstabilized one is all over from 0 to 2 mSec, with a big
hump at <100 uSec. They used a rod/plane gap, which has a lot of corona, as
opposed to a uniform field gap. Their gap was 20 mm and the rod was 2 mm in
diameter in SF6 at 0.5 bar. Finally, they show a graph of a relaxation
oscillation where the period is 200 uSec (5 kHz) and the voltage on the gap
is 36kV. I assume from the waveform shown that it is a resonant charging of
a capacitor (at least that's what the shape of the rise looks like).