[Prev][Next][Index][Thread]
Modeling a magnifier
----------
From: Antonio C. M. de Queiroz [SMTP:acmq-at-compuland-dot-com.br]
Sent: Tuesday, March 10, 1998 12:29 PM
To: Tesla List
Subject: Re: Modeling a magnifier
Jim McVey wrote:
> I think we all agree that there will be a beat frequency with the
> L2C2 and L3C3 connections you proposed. It still seems to me that the
> C2 will be of an impractical size to tune L2c2 near (L2+L3)*c3. I suspect
> an actual system would have large radiation loses in the L2C2 and
> virtually no storage capability in C3. The sparks would be very thin.
> (Ok, I'm guessing here....)
You rise an important point. Really, if L2<<L3 the necessary C2 for
optimal tuning is impracticable as a distributed capacitance. In my test
setup I used L3=32 mH and L2=188 uH. As L3 resonates at 385 kHz, my C2
goes to 914 pF. But the system would also work with L2 much greater,
and C2 much smaller. Actually, in this mode of operation, there is
no advantage in using L2<<L3. Systems using L2<<L3 probably work in
some other, maybe as good as this, way.
An interesting thing that I observed is that it is perfectly possible
to tune L2C2 to the higher resonance modes of L3 (L3 works as a transmission
line). This requires smaller C2, but imposes greater voltage gradients over
L3, increases the losses due to operation at higher frequency, and traps
a lot of energy in coil instead of at the top load. This may be the reason
for resonators being covered by corona, as I sometimes see described here.
Radiation losses are small due to the size of the system compared to
the wavelength corresponding to the frequency of operation. The optimal
tuning causes the maximum possible concentration of energy in C3, and
the sparks would be as strong as possible, I believe.
> The experiment is flawed if you are using a variable capacitor instead of
> an isotropic capacity as in the real thing. There is no actual ground
> connection as your model posseses.
A distributed capacitor (a ball/toroid/etc. terminal on L2) works essentially
as a capacitance to ground. The self-capacitance of L2 would also be connected
as shown.
> That is what I was trying to say
> in my last post referring to the " out of tune" impedance reflected into
> the primary. With your latest post , i see that you did not use a primary,
> therefore you were unable to see this effect. Check this out on your
> grounded C2 experiment. Replace it with an isotropic capacity of the same
> value, and I'm sure there will be a marked difference in dynamics.
I will have to use another L2, as 914 pf is rather big for a top load.
I was considering that the primary would be active for a very short time,
used only to transfer energy to the system. But I see now that there is
a problem. This would work well with L2<<L3, because the energy transfer
from the L2C2 tank to the (L2+L3)C3 tank would take many cycles. But with
L2 and L3 in the same magnitude order, the energy transfer is faster,
and it is not anymore possible to ignore the effect of the primary.
Would the fast quenching rotary gaps used in magnifiers be an attempt to
make the primary-secondary system act as a high-voltage CW source? It is
necessary only to syncronize the rotary to a precise fraction of the
resonance frequency of the system to get appreciable energy accumulation
in the third coil. With high-Q coils, low operating frequency, and high
break rates? I know that this is not the right idea in a conventional coil,
but in a magnifier?
> In my opinion, a model closer to the actual system dynamics
> (simulation-wise) might look more like this:
>
> L2------+----------L3------C3----+
> | C2 Rr
> | Rr |
> | | |
> V V V
>
> The "V" represents ground, Rr is radiation resistance among other
> things
> such as ion channel absorbtion etc. There's been books written on the
> nature
> of Rr...buts thats another topic. At low voltage- very much below
> breakout,
> it's pretty much a radiator. This model will behave as Richard hull
> desribed his results, where C2 is more of a tuning issue for the L1C1
> L2C2 coupled system than it is for the secondary. The Q of the L2C2
> section is impaired by Rr. Depending on the degree that C2 is
> compensated for in the primary, there will be leway for some tuning
> combinations and beats from these interactions.
The main resistive losses in these systems are due to skin effect and
proximity effect in the wires of the coils, with some extra losses due
to dielectric losses in the coil forms. Due to the dimensions of the
system, radiation losses are small. My test system behaves precisely
as if there were a 1 kOhm resistance in series with L3 (or C3) (the
other coil is much smaller and uses thicker wire). The DC resistance of
my L3 is only 150 Ohms. The presence of (low) resistive losses adds
damping to the system (good to allow the observation of the transients
in the oscilloscope), but does not change appreciably the resonance
frequencies.
> For C2<<C3 as in Richard's system, It can be thought of like a
> current
> divider, where the most current will want to follow the (L2+L3)*C3 path.
> Again, the Thumbnail sketch to get in the ballpark is:
>
> L1C1=(L2+L3)*(C2+C3)
Close to this, certainly. I didn't look yet at the dynamics of three
coupled coils.
Comments are welcome,
Antonio Carlos M. de Queiroz
http://www.coe.ufrj.br/~acmq