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

Re: TESLA COIL REVISED



Original poster: Paul Nicholson <paul-at-abelian.demon.co.uk> 

Hi All,

Let's just nail this pointless issue of 'resonant rise':

Jaro wrote:
 > most people here claim that the output voltage is to be
 > determined by the L2/L1 ...

 > If it was due to resonant rise, then you could greatly increase
 > output voltage by using a THICK
 > So you guys better get this straight, it CAN'T BE BOTH.

It can.
In one case the voltage resonantly rises to a max of

  Vin * sqrt( Cs/Cp)

and in the other case it resonantly rises to a max of

  sqrt(Q*Pin/(pi*F*Cs))

In the first case, a fixed, finite amount of energy, 0.5*Cp*Vin^2
is available for transfer, in the second a fixed finite amount of
power, Pin, is available.  I've discussed this in an earlier post
but you don't seem to be taking it on board.

Both cases can be described as 'resonant rise' but the term
isn't very useful - it's too broad!

(Jaro is using the term resonant rise specifically for the 2nd case,
which is causing some confusion.)

Another distinction - the first involves two operating frequencies,
roughly equally sharing the stored energy, the second only a
single operating frequency.

In the first case, the secondary only needs to store energy for
a few tens of microseconds, in the second, storage time can be
tens of milliseconds.

Jaro, the reports by the experienced coilers on this list of poor
performance of HF coils are valid, and should not be discounted.
The reasons are, variously,
a) Poor match of output discharge to driver via the resonator
    characteristic impedance,
b) Lossy, unshielded coils, giving low efficiency.
c) A popular preference for long streamers as opposed to intense
    brush discharges.

Operating Q (ie loaded Q) is (ideally) determined by the load
applied, and the operating Q in turn determines the output voltage.

The unloaded Q affects the power transfer efficiency,

  Eff% = 100% times (Q_unloaded - Q_loaded)/Q_unloaded

and has only a modest impact on output voltage.  Lack of shielding
allows the HF coil to couple strongly to its environment, heating
everything up and diverting power from the desired load.

I've drawn a few connections with RF engineering in these posts,
but now it's time to stress some important differences between say
the resonator in a PA tank circuit and a HF plasma-producing CW TC.

Ballpark figures...

          Valve PA tank,   Transistor PA tank,   HF CW TC

  Zin        2k ohms             1 ohm              1 ohm (MOSFET)
  Zout       50 ohms            50 ohms            50 k ohms
  Z of
  resonator   300 ohms           8 ohms            250 ohms
  L/C        100,000               50               50,000
  Loaded Q     10                  10               200

I've used 50k as a nominal figure for discharge load resistance,
but obviously it will vary widely.   The loaded Q has to be
roughly sqrt( Zout/Zin), and the resonator Z must be around
sqrt( Zout * Zin)

We can see that in terms of the L/C ratio, the HF TC is not too
dissimilar to a valve PA output resonator.  The most important
difference is in the required loaded Q factor, a couple of hundred
for the HF TC, but only around 10 for the PA.  Now its easy to
make an efficient PA tank circuit - with a loaded Q of 10 we only
need say 200 for the unloaded Q to get (200-10)/200 = 95% efficiency,
and transmitter output stages can easily do better than this (except
for mine, it seems!).

But it's a different matter for the HF TC - the high impedance
transformation demands a high loaded Q, and to get any reasonable
efficiency the unloaded Q must somewhat higher than this.  If we
could achieve an unloaded Q of 500, then the efficiency would be
around (500-200)/500 = 60%, which would be a reasonably challenging
target.  With an unloaded Q of 1000 you would be doing very well
indeed.

To get unloaded Q factors like this, coils would need to be made of
silver plated tape, and enclosed in good quality screening
enclosures.   This adds a lot of extra shunt C to the resonator,
and so reduces the available L/C ratio, but note the rather modest
L/C required of the HF CW TC - a very much lower impedance than the
typical few tens of k ohms for the LF cap discharge TC.

To lessen the severe impedance transformation required of the HF TC,
and thus ease both the loaded and unloaded Q factor requirements,
most CW coils are not directly connected to the low Z output of
the driver, but are instead coupled through a matching transformer.
Unfortunately the additional losses introduced by the coupling
transformer go a long way to offsetting the benefit of the reduced
impedance transformation ratio of the resonator coil.  Not surprising
really, considering that the overall impedance conversion ratio from
driver to discharge load remains the same.

I hope these comments illustrate that although the principles are
the same, building a HF TC is in practice a lot more demanding than
making say an output stage for a valve (I guess I mean a tube!) PA
operating at the same frequency.  And this is before we mention
anything about how variable the load resistance and resonator
frequency is during operation, and before we think about what
happens to transients, and what happens to the driver when the
load is removed.

Perhaps for these reasons it's not a widely explored area in the
coiling community.  Definately non-trivial and hardly a beginners
project!  If you're adept at building RF PAs at the kW power level
then this could be your next project, but otherwise it could become
a nightmare of trial and mostly error.

So Jaro, forgive the skepticism of many of the list members when
they see a newbie launching straight into the cutting edge of TC
technology.  It does seem from your questions and comments that
you do have a fair bit of theory to learn and perhaps only then
will you really grasp the difficulties.

I would recommend studying the theory as much as you can, and build
some small resonators operating at a few watts power level so that
you can get a feel for the territory.  That's enough power for you
to measure the impedances, efficiencies, and Q factors.  You need
to generate measurements which show that the required resonator
properties can be achieved.  Only then will people begin to sit
up and take notice of your proposals.  You'll have noticed from
the pseudoscience groups that claims, ideas, proposals, etc, are all
ten-a-penny.  Here in the real world it's working results that count.

BTW, Jaro, for some reason YOUR CAPS KEY KEEPS STICKING :)  I only
mention because it begins to read like a piece of spam, and my
spam blocks risk chucking your posts away 'cos it detects oddly
capitalised words!!
--
Paul Nicholson
--