# Re: DC Supply

```> Original Poster: Jim Lux <jimlux-at-jpl.nasa.gov>
> But what if the primaries were fed in Delta?  120/208Y is quite common
> in light industrial/office areas, because you get the 120V to run
> everything. However, in heavier industrial, you see more 240 delta, with
> no neutral. Sometimes they ground the centertap of one of the feeder
> transformers, so that two of the legs are 120V relative to ground.
>
> Then, you could connect the secondaries of the NST's in delta, with no
> problem, but I am not sure about the centertap (having been too busy to
> calculate it or try it)

I'll discuss with the voltages available here but it scales
to US voltages as well. It's the configuration that matters,
not the exact voltages.

The problem is that if you connect the neon primary with
N and L you get 240Vac across the primary. Now, if you were
to connect L1 and L2 to the same neon primary, you would get
240Vac * sqrt(3) = 415Vac across the primary. Nice if you
have a primary that is designed for 415Vac but _NOT_ nice if
you still have a primary that is for 240Vac. So you cannot
just deside whether to use wye or delta - you are forced to
use what the equipment can handle. If you use three ordinary
neons designed for N and L to be connected to primary, you
have to connect N to all of them and then L1 to first, L2
to second and L3 to third neon. This is wye. You CAN'T
use delta (L1 and L2 to first primary, L2&L3 to second,
L3&L1 to third) with this.

On the other hand, there is no such limit on the secondaries
as long as they don't need to be referenced to any voltage.
This implies no center-taps to be grounded and enough
primary->secondary insulation. The neons do need to have
the secondary center-tap referenced to ground (limited by
insulation - all parts of secondary can't handle the full
secondary voltage). So you _have to_ connect the center-taps
to ground and thus you will have 6 phases of wye. No decisions
to make here either.

Now you need 12 diodes to rectify this like GL said before.
Each secondary HV lead needs diode to + and diode to - ; so
6 phases times two is 12.

This WILL cost more for the same spark output than just
running the system with 1 phase AC.

One alternative would be, like said here, to use several
spark-gaps in parallel - in practise a special RSG - where
the firing of gaps cannot occur simultaneously for different
caps (and phases). The RSG will fire cap1 to primary, then
cap2 to primary, cap3 to primary, cap1 to primary etc. ad
infinitum.

This will enable you to use AC input to all caps. And you could
use a lot more caps with each having a transformer connected to
any phase. It will require more caps but the REAL thing to look
at here is that each cap will see only total breakrate divided
with the amount of caps. This means that by adding more caps
it is possible to achieve VERY high breakrates and still be
within design-limits of caps (each cap has a design-limit for
maximum pulses per second). Of course, the RSG for that will
be interesting a design, but only a matter of MECHANICAL design.
If you can get caps cheaply (yeah, right) and have the means to
to build the RSG (lathe, mill etc), this is IMHO the way to go,
rather than to run a single gap to the same break-rate. To
get still better control, you'd need to use DC for each cap -
otherwise there is a lower limit for breakrate like with all AC
systems (each cap has to fire at least once per half cycle).

Just some thoughts,

Kristian Ukkonen.

```