[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

A triggered-s.g. 1-turn primary



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

I take note of jimmy hynes/Terry Fritz's posting of 4/28 but submit this 
posting anyway, which I had been preparing.

In another periodic retreat from solid-state exasperations, I have come up 
with a notional spark-gap design, 
<http://hot-streamer-dot-com/temp/tspk13s2.pdf>http://hot-streamer-dot-com/temp/tspk13s2.pdf. 
You will need Acrobat Reader for that.  (I note that gif format yields a 
lousy image, derived using Photoshop either from my CAD program's exported 
pcx or from a scanned print of the drawing.  The pdf image seems just 
lovely--similarly derived from a print.  Computer mysteries...)

Referring to the drawing, six "6C"s, six "1/6 L"s, six "1/6 R"s and six 
spark gaps are connected in a circle on the secondary's nominal diameter 
(in my case that would be 12").  Six 6Cs in series yield C and six 1/6 Ls 
in series yield L.  L and C resonate at the secondary's Fr when the 6 spark 
gaps fire.  Each 1/6 L is merely the sum of a 6C's internal inductance + 
the spark gap's intrinsic inductance + the interconnection inductance of 
that segment of the primary loop, as arrayed on the nominal 12" 
diameter.  Each 1/6 R is the sum of spark-gap, 1/6 C and interconnection 
resistances in its segment.  All of those elements constitute a 1-turn 
primary.  The secondary is to sit right on top of it--resting, perhaps, 
directly on the capacitors.

Each 6C has applied to it switched AC-mains input, coupled via windings of 
a T1 and a T2 operating as chokes.  The AC is a sine wave that is 
interrupted using two power MOSFETs, two IGBTs or a triac, not shown.  The 
switching circuit is turned on during each mains 1/2 cycle at the 
approximate voltage peak, staying on for 1/4 cycle and thus delivering a 
step-voltage via the T coils to each 6C.  The coils in the Ts resonate with 
each 6C so that, a few milliseconds after each step, the voltage on each 6C 
reaches about 275 V (as simulated with MOSFET drivers) with 160 V peak from 
the mains.

Phasing of the T1s and T2s is such that the resonant-charging currents' 
induced voltages in their series-connected second windings are of opposite 
polarity & thus the series connections exhibit 0 V during the charging 
time.  All 6 second-winding pairs are connected to a bus-pair that is 
driven via a break-over diode from one of the 6Cs.  Because of the phasing, 
the break-over diode sees only the voltage on the 6C.  The Ts are to be 
physically located close to the physical center, to minimize the Fr flux 
intercepted by them and their interconnections.

During charging, each gap sees twice the voltage on a 6C.  Just prior to 
the resonant-peak (absent breakdown), the diode breaks down and applies the 
6C's voltage to all six T-winding pairs in parallel.  This action induces 
voltages into the charging windings such as to momentarily increase the gap 
voltages, sufficient to cause the gaps to break down.  Again, during the 
spark event, the coil polarities are such that each pair of trigger 
windings exhibits no induced voltage.  Note that alternating trigger-coil 
pairs are connected to the driving bus oppositely.  Thus the trigger 
voltages on alternating primary-loop segments are of opposite polarity, 
doubling the trigger voltage seen by each gap.

The turns ratio in each T might need be no larger than 1:2 for reliable gap 
firing; 1:1 and perfect transformers would cause the gap voltages to 
increase by a factor of 1.5.  Also, all the Ts could well be consolidated 
onto just two cores with one trigger winding on each core.  The 20 mH (in 
my simulation) is not critical; a whole lot smaller and the peak mains 
current becomes excessive while a whole lot larger, the charging time 
becomes excessive.

As soon as the gaps break down, the damped sine wave loop-current at Fr 
commences flowing.  The 6C-charging T-coils then act to isolate the mains 
from the spark gaps during the firing event while a simple clipping network 
at the MOSFET/IGBT/Triac output clamps the pk-pk voltage there to less than 
~+/- 250 V--again, as simulated.  I've added a small R-C damping network as 
well, to soak up most of the Fr signal there.

The gaps keep conducting until the Fr current becomes too low, which will 
occur well before each following initiation of capacitor charging, ~8 ms 
later at 60 Hz mains frequency.  They are to have extremely close spacing, 
perhaps 0.02" or so and thus will dissipate relatively little power.

Since the capacitor charges for each half mains-cycle are of opposite 
polarity, any tendency for metal to transfer across the gaps should be 
minimized because the DC component of the gap current will alternate from 
one mains half-cycle to the next.

I show 115 VAC input but 240 V could be applied just as well.  I've 
simulated the circuit using 40 uF for 6C, 33 nH for 1/6 L, 5 m-ohms for 
each 1/6 R and 20 mH for each charging-coil of T.  I've not simulated the 
trigger scheme except for using a "switch" for each gap and turning those 
on for 500 us after the delay time.  The scheme works just fine in that 
simulation: Fr is about 125 KHz and the peak first-half-cycle loop current 
is about 9 KA with no simulated additional switch voltage-drop.  At 120 V 
in, the line current is about 15 A RMS, 30 A peak.  The Fr current 
diminishes to 1 KA at the 4th cycle (no secondary present).

A 30 m-ohm total loop resistance is perhaps too optimistic.  Higher 
resistance will, of course, diminish the peak current and also the number 
of Fr cycles that will occur prior to gap-extinguishing.  With 100 m-ohms 
total, I get only ~7 KA peak Fr current and that diminishes to 500 A in 
just 1 1/2 cycles; likely not enough excitation to produce much of a 
spark.  Thinking that perhaps two rather than 1 turn would be better (if Q 
= X/R, X would be 4x, R would be 2x, perhaps, so Q would be about 
doubled--right?), I temporarily added 600 nH and 60 m-ohms into the 
loop.  I got 4 KA peak diminishing to 500 A in 2 1/2 cycles & whether 
that's an improvement, I don't know.  I didn't bother to alter the 6C 
values to bring Fr to the same frequency.

It does seem to me that with too-high a dv/dt in the first half-cycle, 
there might be the risk of secondary turn:turn voltage breakdown.  Would 
the lower dv/dt and higher Q be better?  Perhaps someone else has already 
considered that kind of thing & I've not paid attention.

Clearly more primary segments could be added for more power--but cramming 
more capacitors& gaps into the nominal diameter might present a problem.

I could use some informed comment on this.  Perhaps it's all too fanciful...

I'd be happy to email my SIMetrix simulation-schematic file to anyone with 
SIMetrix and a real interest in the design.  Or, perhaps someone would care 
to simulate it otherwise, using the pdf drawing as the source.

Anyone interested enough to consider building it??

Ken Herrick