# Re: Mica Cap Question

```Original poster: "Dr. Duncan Cadd by way of Terry Fritz <twftesla-at-uswest-dot-net>" <dunckx-at-freeuk-dot-com>

Hi John!

>Hello Duncan,
>I am a learner so please humour me.

I am a learner too.  Join the club!

>Can you please go over part of your explanation above, as though
> you were addressing a dummy.
>From this part onward I get lost:
>"hence the break rate will be near 12500bps ( = 1000 / 0,08 )
>I am keen to learn and get a grip, but I do not understand about
>the spark gap breaks per second.
>Are you talking about a rotary gap or static gap?
>How does Voltage/joules give breaks per second?

OK, this part is not cast in stone, but it's quite simple.  If you
have one kilowatt, that means 1000 joules in one second.  The energy
stored in a 0,01uF cap charged to 4kV is 0,5CV^2 = 80mJ = 0,08J.  Now
divide 1000 joules by 0,08 joules and you get 12500.  In other words,
you can charge that cap to 4kV 12500 times per second if you have 1000
joules per second to play with.  This is going to be an approximation,
because the power delivered by the transformer is not going to be 1kW
precisely, also for a variety of reasons, the gap firing may not be at
sufficiently reproducible a voltage per bang for the energy on the cap
to be exactly 80mJ per bang, though it should tend to average out.
This is for a static gap set to fire at 4kV (or a toothless rotary as
per clifdengap.jpg at http://hot-streamer-dot-com/temp/ or in a
subdirectory.)

Now (slight digression to pick up on another post) as Malcolm remarks,
getting an effective quench at so high a break rate is likely to be
problematic.  My solution to this (tested at 6000bps) is to use the
phenomenon of resonance to assist.  If you guesstimate the break rate
at 12500 per second and you know the capacitance is 0,01uF, then you
can use a series charging choke of the appropriate inductance to
resonate with the cap at 12500c/s.  This will help ensure that the
voltage across the gap cycles at 12500c/s and results in much cleaner
quenching.  This is what I did when putting my tank caps in series for
the "equidrive" configuration, and in that case I split the resonant
choke in two, putting a half-size choke in each HT leg to keep the
whole circuit symmetrical.  The choke inductance can be calculated
thus:

L = 1 / ( 4.pi^2.f^2.C )

where L is the inductance; pi = 3,142; f is the break rate 12500; C is
the capacitance 0,01uF.

We thus have L = 16,2mH.  Either a single choke of 16,2mH or two
chokes of 8,1mH one in each HT rail, would do the job.  My experiments
suggest this choke value is not terribly critical.  Ballpark +/- 20%
is plenty good enough.  I used iron fencepost wire, 1mm diameter,
annealed in a fire, for the choke cores (under a thick insulating tube
which extended an inch clear of the iron at either end of the choke)
both to reduce the number of turns needed and also to provide
additional losses at rf.  I did notice that with the lopsided
arrangement and a NST, the safety gap fired quite often, whereas in
the symmetrical split-choke / equidrive configuration, the safety
hardly fired at all.  Maybe someone adept at Microsim can say why this
should be.

The resonant chokes have a very high voltage across them and need to
be mounted well clear of each other and surrounding objects.  Also, it
helps if they are wound in several separate "pies" like standard rf
chokes as opposed to one continuous winding, as this separates the
start and end of the choke physically, thus increasing the breakdown
voltage.  Mine were impregnated with beeswax - however you do it, they
will need first rate insulation to avoid breakdown.  The transformer
protection filter needs to be a good 'un and the safety gap correctly
set.

With a close coupled transformer (k very close to 1) like an MOT there
is another way to do this using a choke in the primary, but this tends
not to work quite as well as the choke in the secondary, maybe leakage
inductance is one reason and frequency response of the transformer is
another (which if I remember my Langford Smith, basically boils down
to leakage inductance and lamination characteristics) so I won't weary
you with the details, but you will see it in books on spark
transmitters and possibly in old ones on Tesla coils.  Certainly,
major leakage inductance is the reason it doesn't work with NSTs.

>> Alternatively, using the late Philip Kemp's way of reckoning for a
>> spark transmitter circuit:
>> If the break rate is 12500bps, the circuit resistance is 4 ohms
(Terry
>> has guessed this is close) and we assume the voltage is 4kV then
>> I = E SQRT (n C / 2R) = 4000 SQRT (12500.1.10^-8 / 2.4)
>> I = 4000 SQRT (1,56.10^-5) = 15,8A which is within ratings.
>>
>> [ in the above, I is current, E applied voltage, n the breaks per
>> second, C capacitance, R resistance ]

This all comes from a book on ac - "Alternating Current Electrical
Engineering" by Philip Kemp, publ. Macmillan, 6th edn reprinted 1944,
p593 (from chapter 37 entitled "Oscillatory Circuit" p 586 et seq.)  I
did post Philip Kemp's derivation of the equation used some time ago
and it's doubtless lurking in the archives somewhere (title: Re: Fw:
Primary RMS Current calculation, date 1st March 2001).  It simply
allows you to determine the rms current flowing in a spark transmitter
primary circuit (or Tesla primary) given the capacitance, voltage,
break rate and resistance.

>I am sorry if I am asking something that is common
>knowledge among coilers but I am a bit lost.

Hope this helps.  If not, shout again!

>If I do not ask, I will never know.

Truer words seldom emailed ;-)  The more I find out, the more I
realise how little I know.  Opportune moment to say "thanks" to other
list denizens (I won't name names as it is guaranteed I'll leave
someone out, also the list will be rather long) for stuff on Microsim,
triggered gaps, ballasting and other things I don't know much about.

Dunckx
Geek#1113 (G-1)

```