A Marx-like design

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

Original poster: "K. C. Herrick by way of Terry Fritz <twftesla-at-qwest-dot-net>" <kchdlh-at-juno-dot-com>

For some time I've had an idea for a spark-gap-type t.c. with a Marx-like
primary, I've fully laid it out on paper (on my HD, that is), and I await
a time when my inclination to loaf may diminish.  My idea is to
resonantly charge a set of capacitor-modules, from the mains and in
parallel, then discharge them in series through the primary.  I've
designed it with 20 sections, each to be charged to ~600 V, then
discharged in series via spark gaps, yielding ~12 KV across a 4-turn
primary.  With Terry's indulgence I post a simplified version of the
schematic at 

http://hot-streamer-dot-com/temp/marx-pri.gif

It should be
(just) readable printed on 8 1/2" x 14" paper.

Referring to the schematic, I envision the thing to be built on two flat
disks; commercial serving trays would be dandy for that purpose. 
Capacitors, gaps and the primary coil are all to be mounted on the
disks--in such a way that the top disk may be readily lifted completely
off the bottom one, exposing all the gaps for maintenance, etc.  Gap
spacing, quite small since each gap, when un-fired, sees only twice the
voltage to which the Cs are charged, is readily maintained by controlling
the spacing between the disks.  I incorporate a dummy gap (not shown) for
each active gap, closely-located and spaced with a shim-disk common to
all of them.  The dummy gaps are thermally connected to the active gaps
so that their respective thermal expansions match.  I would surround each
active gap closely with a resilient Si-rubber closed-cell foam ring to
keep out oxygen.

In the posted drawing, 8 capacitor groups and 8 spark gaps are arrayed in
a circle so as to contribute their magnetic flux as 1 of the 4 primary
turns; the other 3 turns are made from copper tubing.  Via small
inductors L and a simple connector (J & P) between the two disks, all
capacitors are connected in parallel.  That parallel set resonates, at a
bit below 60 Hz in the U.S., with L1 in series with L2.

At first turn-on, relay K is held closed momentarily & then released,
dc-charging C1.  Resonant amplitude builds up and at the half-cycle
following the time when K opens, the Q "Sidac"s (bilateral breakover
diodes) fire (because C1 is oppositely charged).  C1's charge is dumped
through T's primary.

A time comes when all the loop-Cs have become resonantly charged in
parallel, to about an instantaneous 600 V each, via the small Ls. 
Following around the loop, you can see that, if the gaps can be made to
fire at this time, all the C-groups will become connected in series
(disregarding the Ls) with the 3 turns of heavy conductor and will then
resonantly discharge.

The gaps are triggered by the output of T.  Visualize the set of Cs as
having an upper part and a lower part (in the drawing).  Both of those
parts are connected to near-ground potential (as respects high voltage)
at the left, via L1, L2 and the mains.  From the mains location, you can
trace around, clockwise & then counter-clockwise, exactly two primary
turns each way (including an upper or a lower portion of the Cs) until
you reach the spot where T's output is connected to the coil.  Those two
turns one way & two turns the other way constitute a (nearly) balanced
bridge, causing the potential of the T-connecting point to remain near
the mains potential at all times except when T delivers its output pulse.

But notice the impedance that T's H. V. pulse sees: all the gaps are off
so what it sees are only the Ls in series-parallel.  T's pulse can
develop across the Ls and become large enough in amplitude to start
firing gaps.  As gaps fire progressively, un-fired gap voltages increase,
due to the voltages in the Cs and the already-fired gaps, until, very
rapidly, all the gaps fire.  At that instant, the Cs and the primary
inductance resonate & off we go.  The impedances of the Ls are high with
respect to the oscillatory impedance so they will have no effect on the
Fr.

Properly choosing component values, the next firing event can be made to
occur at the 3rd 60 Hz half-cycle following the previous one.  That means
that the charges on all the Cs will have become reversed.  That means
that the dc component of gap current will be in the opposite direction. 
And that means that any metal-transfer across the gaps will be in the
other direction from previously.  That should minimize gap deterioration.

I contemplate a novel method of tuning:  Each C-group is divided into a
larger-capacitance and a smaller-capacitance part.  The
smaller-capacitance part is switched in and out using a common toggle
switch for each one.  But when the switch is open, that part is still
connected via a low-value resistance.  During charging, that resistance
is immaterial; during discharging, the resistance is large compared to
the primary's oscillatory impedance.  Thus, the effects of the
smaller-capacitance parts on the high-frequency Fr will depend on how
many of the switches are open & closed.  So voila!...tuning!  Manual- and
not self-tuning but perhaps I will work on that.

Because all the C-voltages exponentially decline at each spark event and
because of the presence of L1 & L2, the gaps should self-quench; they
surely will as the mains potential approaches each zero-crossing.

C1 and the Qs can have another function: to ensure that the average
resonant frequency of all the Cs with L1 and L2 remains below 60 Hz. 
When the Qs fire, C1's capacitance becomes added to the total of the Cs',
momentarily lowering the near-60 Hz resonant frequency.

The bypass gap serves the usual purpose, conducting inadvertent primary
strikes to earth ground.

It's great fun to dream this kind of thing up.  Is it viable?  Will I
build it?  Will anyone build it?  Stay tuned...

Ken Herrick